Manufacturing Processes: A Practical Reference
Manufacturing is a chain of processes: producing raw stock, shaping parts, joining assemblies, modifying properties and surfaces, and verifying the result. This guide catalogs 212 processes across seven layers, from continuous casting to statistical process control.
Continuous Stock Production
make continuous sheet, bar, film, and fiber stock14 processesMolten metal is continuously solidified into billets, blooms, slabs, rods, or strip and cut to length. It is the dominant upstream route for producing feedstock for later rolling, extrusion, forging, or machining.
Strengths & weaknessesThroughput is very high, consistency is good, and cost per ton is low. The downside is that a caster takes an enormous capital investment, needs high utilization to pay for itself, requires continuous process control, and is generally limited to constant cross-sections.
When to useThis is a feedstock decision, not a part-making one. If you are buying commodity-scale stock that will later be rolled, extruded, forged, or machined, specify continuously cast billet, bloom, slab, or rod. Operating a caster yourself only makes sense at integrated-mill scale, meaning hundreds of thousands of tons per year and near-continuous utilization. If you need discrete shaped parts, or short runs of special alloys, look at shape casting or conventional ingot casting instead.
Key numbersSlab sections roughly 200–300 mm thick and 0.8–2.5 m wide, billets 100–200 mm square · casting speed 0.8–2.5 m/min on slabs and 4–6 m/min on small billets · line output typically 1–4 million tons per year · yield from liquid metal about 95–98%, against 85–90% on the ingot route · caster capex in the hundreds of millions of dollars.
ExamplesSteel slabs, aluminum billets, copper rod, steel blooms.
Economic profileCapex is extremely high and throughput is commodity-scale. The economics are dominated by utilization, energy, raw-material spread, and yield.
VideosIntroduction to Continuous Casting (Continuous Casting Consortium, Univ. of Illinois) · Continuous Casting Course (steeluniversity, World Steel Association)
Metal passes between rotating rolls to reduce thickness or create a continuous cross-section. Major forms include hot rolling, cold rolling, plate rolling, foil rolling, and shape rolling.
Strengths & weaknessesRolling is extremely productive and cheap for sheet, plate, rail, and structural sections. It is largely restricted to continuous or gradually changing geometry, and it needs large, specialized equipment.
When to useChoose rolling for flat products and long structural sections that get consumed by the ton (sheet, plate, strip, foil, rail, and beams). Nothing else comes close on cost per kilogram. Rolling in-house is a mill-scale decision, so unless you have commodity volume and dedicated high-utilization lines, buy rolled stock instead. If you need a complex constant cross-section in aluminum, extrusion is usually better. If you need discrete shaped parts, look at forging or casting.
Key numbersHot rolling of steel at 1,100–1,250 °C · reduction 20–50% per pass hot and 20–40% cold · hot strip mills exit at up to about 20 m/s · cold-rolled thickness tolerance around ±1%, with foil down to roughly 6 µm · cold-rolled surface Ra 0.2–1.5 µm · mill output measured in millions of tons per year.
ExamplesSteel sheet, aluminum foil, rails, I-beams, plate, automotive body stock.
Economic profileCapex is very high and utilization needs to stay high to justify it. Output is typically measured in tons rather than parts.
VideosHot Rolling Course (steeluniversity, World Steel Association)
A billet is forced through a shaped die to create a continuous product with a constant cross-section. Extrusion can be hot, warm, or cold.
Strengths & weaknessesExtrusion produces complex constant cross-sections efficiently and with relatively little waste. The profile has to stay constant along its length, and difficult alloys may require high forces, slow speeds, or specialized dies.
When to usePick extrusion for constant-cross-section lengths with complex profiles, especially in aluminum: a die costs only a few thousand dollars, and runs from a few hundred kilograms upward are economical. It is the default for heat sinks, frames, and structural profiles that would waste a lot of material if machined from solid. If the section has to vary along the length, machine or fabricate the part instead. For steels and other high-strength alloys, hot-rolled standard sections or roll forming are usually cheaper.
Key numbersAluminum billet at 400–500 °C · extrusion ratio 10:1–100:1 · press force 500–10,000 tons · standard profile tolerance about ±0.25 mm on dimensions under 25 mm, minimum wall around 1 mm · dies $1,000–10,000 at 2–4 week lead time · economical from a few hundred kilograms per run.
ExamplesAluminum window frames, heat sinks, tubing, rails, channels, structural profiles.
Economic profileTooling and press costs are moderate to high, and the process is attractive at medium to very high length volumes.
VideosBasics of the Aluminum Extrusion Process (Aluminum Extruders Council) · Extrusion Design Tips (Aluminum Extruders Council)
Metal rod or wire is pulled through progressively smaller dies to reduce diameter and improve finish and strength.
Strengths & weaknessesDrawing gives precise dimensions, excellent surface finish, and very high throughput. It is limited to continuous products, and the wire may need intermediate annealing to keep it from cracking.
When to useSpecify drawn wire when you need round stock below hot-rolled rod sizes, or with tight diameter tolerance, elevated tensile strength, or a bright finish (springs, cable, conductors, welding wire). This is a continuous commodity operation, so buy from a wire drawer unless you consume tons per month. If you need non-round profiles or discrete parts, or if work hardening is unacceptable, use shaped-profile rolling or machining from bar instead.
Key numbersWire rod feedstock 5–12 mm, drawn down to about 0.01 mm on the finest sizes · area reduction 15–30% per pass over 5–15 passes · line speed 1–5 m/s on heavy wire and up to 30 m/s on fine wire · diameter tolerance roughly ±0.5–1% · drawn surface Ra 0.1–0.8 µm · high-carbon steel wire reaching 2,000–4,000 MPa tensile after full drawing.
ExamplesElectrical wire, steel cable, welding wire, springs, piano wire.
Economic profileThe process is capital-intensive and continuous, so lines run at high utilization. Cost per unit length is low.
VideosWire Drawing Process: History and Fundamentals (Expometals) · Wire Technology: Process Engineering and Metallurgy, R. N. Wright (Elsevier)
Tubing is pulled through a die, often over a mandrel, to reduce diameter, wall thickness, or both.
Strengths & weaknessesDrawing makes precise, thin-wall tubing with good mechanical properties. Geometry is limited to tube-like products, and large reductions often take several drawing and annealing steps.
When to useUse tube drawing when off-the-shelf welded or extruded tube can't meet your wall tolerance, concentricity, surface finish, or fine diameter (hypodermic sizes, heat-exchanger tube, precision instrumentation lines). For large reductions, plan on multiple draw-and-anneal passes, and remember that you are restricted to tubular geometry. If standard-tolerance tube is good enough, buy welded or seamless extruded stock and skip the cost of drawing.
Key numbersOD from about 0.2 mm on hypodermic sizes up to roughly 150 mm · wall down to about 0.05 mm · area reduction 20–40% per pass, with an anneal every few passes · OD tolerance around ±0.05 mm and wall tolerance ±5% on precision tube, against ±10% on standard stock · drawn surface Ra 0.2–0.8 µm · dies, plugs, and mandrels for a size family typically under $50,000.
ExamplesHypodermic tubing, hydraulic lines, heat-exchanger tubes, precision mechanical tubing.
VideosTube drawing principles (The Fabricator) · Aspects of wire drawing and tube drawing (NPTEL)
Molten polymer is continuously forced through a die to produce a profile, pipe, sheet, film, coating, or compounded pellet. Major variants include profile extrusion, pipe extrusion, cast film, blown film, sheet extrusion, wire coating, and twin-screw compounding.
Strengths & weaknessesExtrusion is one of the highest-throughput and lowest-cost polymer processes. Geometry has to stay substantially constant along the production direction, and quality depends on melt rheology, cooling, line stability, die swell, and feedstock consistency.
When to useMake extrusion your default for any thermoplastic product that is long and constant in section (pipe, profile, sheet, tubing, wire insulation) at medium volumes and above. Nothing else matches its cost per meter. Startup scrap and die tuning make short runs expensive, so if you have a small lot or a discrete three-dimensional part, use injection molding instead. If the cross-section has to change along the length, extrusion is out entirely.
Key numbersMelt temperature 180–250 °C for commodity thermoplastics · single-screw output 50–1,000 kg/h at 24:1–32:1 L/D · line speed 1–20 m/min on pipe and profile, far higher on wire coating · wall tolerance roughly ±5% and sheet gauge ±2–5% · profile dies $2,000–20,000 at 2–6 week lead time · tens to hundreds of kilograms of startup scrap per changeover.
ExamplesPipe, medical tubing, window profiles, wire insulation, plastic sheet, packaging film, weather seals, resin pellets.
Economic profileCapex is moderate to high and the economics are excellent at scale. Startup scrap and line utilization matter a lot.
VideosPlastic Extrusion Process Guide (Xometry) · Extrusion Know-How Hub (Plastics Technology)
Molten polymer exits an annular die as a tube, is inflated into a bubble, cooled, flattened, and wound into film.
Strengths & weaknessesFilm blowing is efficient for thin, flexible films, and it can produce multilayer structures. The hard parts are keeping the bubble stable and controlling thickness uniformity, cooling, and orientation.
When to useChoose blown film for thin flexible packaging that needs balanced machine- and transverse-direction strength, tube or bag formats, and easy width changes (grocery bags, liners, agricultural film, and most polyethylene film generally). If optical clarity, tight gauge uniformity, or maximum line speed drives your spec, use cast film instead.
Key numbersFilm gauge roughly 8–250 µm · blow-up ratio 1.5:1–4:1 · gauge uniformity typically ±5–10% · layflat widths of several meters at 100–1,500 kg/h · 3 to 9 coextruded layers on multilayer lines · die and air-ring tooling $50,000–250,000 for a multilayer head.
ExamplesGrocery bags, agricultural film, food packaging, liners, shrink film.
VideosMolten polymer is extruded through a flat die onto chilled rolls and wound as a continuous film.
Strengths & weaknessesCast film has high clarity, good thickness control, and high line speed. It usually has less biaxial strength than blown film, and it needs expensive flat dies and chill-roll systems.
When to usePick cast film when clarity, gloss, and tight thickness control matter (stretch wrap, food packaging, medical packaging) and your volumes justify the flat-die and chill-roll investment. These lines run faster than any other film process. If you need balanced biaxial strength or tear resistance, or you have short runs and frequent width changes, blown film handles those better.
Key numbersFilm gauge roughly 10–250 µm · gauge uniformity about ±1–3%, tighter than blown film · line speed often 300–600 m/min, the fastest of the film processes · die widths 1–5 m · chill rolls held at 10–30 °C · flat die and chill-roll tooling $250,000 to $1M.
ExamplesFood packaging, stretch wrap, hygiene films, medical packaging.
VideosBase resin is mixed with fillers, fibers, pigments, flame retardants, plasticizers, stabilizers, or other additives, typically using twin-screw extrusion.
Strengths & weaknessesCompounding lets you tune material properties for a specific application. The things that determine whether it works are dispersion, heat history, moisture, contamination, and batch consistency.
When to useCommission custom compounding when no stock resin hits your property targets (filled, reinforced, flame-retardant, conductive, or precisely matched color grades) and your annual usage justifies a dedicated formulation, which typically means tons rather than kilograms. If you only need color, masterbatch let-down at the molding machine is cheaper. At modest volumes, an existing commercial compound usually beats developing your own.
Key numbersTwin-screw extruders at 32:1–52:1 L/D and 200–1,200 rpm · throughput 20–5,000 kg/h by screw diameter · melt residence time 10–60 s · filler and fiber loadings commonly 10–50% by weight · minimum economic run 500 kg to several tons · toll compounding adds roughly $0.20–1.00 per kg over the base resin.
ExamplesGlass-filled nylon pellets, flame-retardant plastics, conductive polymers, colored resin.
VideosTwin Screw Extruders as Continuous Mixers for Thermal Processing (AAPS PharmSciTech, open access)
Molten glass floats on molten tin, forming a flat sheet with highly uniform thickness and smooth surfaces.
Strengths & weaknessesThis is the most efficient way to make large, optically smooth flat glass. It cannot directly produce shaped three-dimensional parts.
When to useSpecify float glass for any flat glazing need (windows, mirrors, automotive and architectural stock), because nothing else produces large optically smooth sheet at commodity cost. This is strictly a feedstock decision. Float lines cost hundreds of millions and run continuously for years, so everyone except the largest producers buys rather than makes. If you need shaped or hollow glassware, use pressing, blowing, or slumping instead.
Key numbersRibbon thickness 0.4–25 mm, with architectural stock at 2–19 mm · thickness tolerance roughly ±0.1–0.3 mm · jumbo sheet about 3.2 m by 6 m · line output 500–1,000 tons per day · furnace campaign roughly 12–15 years between rebuilds · fire-polished surface below 1 nm Ra · line capex $150–400M.
ExamplesWindows, mirrors, architectural glazing, automotive glass feedstock.
Economic profileCapex for a continuous line is enormous, and the economics depend heavily on keeping furnace utilization high.
VideosPolymer is converted into continuous filaments by extrusion through spinnerets, with solidification by cooling, coagulation, solvent evaporation, or electrostatic drawing depending on the variant.
Strengths & weaknessesMelt spinning is fast, solvent-free, and cheap. Solution-based spinning extends the process to materials that cannot be melted, and electrospinning reaches micro- and nanoscale fibers. The variants that use solvent are constrained by solvent recovery, environmental controls, and low throughput.
VariantsMolten polymer is extruded through spinnerets into filaments that cool and solidify. It is fast, solvent-free, and cheap, but it only works for polymers that can be melted without degrading. Examples: polyester, nylon, polypropylene fibers.
Polymer solution is extruded into a coagulation bath, where the polymer precipitates into fibers. It works for materials that cannot be melt-spun, but solvent recovery, washing, environmental controls, and slower line speeds all add cost. Examples: rayon, acrylic, specialty cellulose fibers.
Polymer solution is extruded into heated gas so solvent evaporates and leaves a filament. It avoids a liquid coagulation bath, but it needs careful solvent recovery and explosion control. Examples: spandex, acetate fibers.
An electric field draws very fine fibers from a polymer solution or melt. The fibers are micro- and nanoscale with very high surface area, but throughput and scale-up are still difficult. Examples: filtration media, tissue scaffolds, battery separators, wound dressings.
Default to melt spinning for any fiber whose polymer melts cleanly (polyester, nylon, polypropylene), because it is the cheapest option at commodity scale. If the polymer degrades before it melts, as cellulosics, acrylics, and spandex do, use wet or dry spinning and plan for the solvent-recovery costs that come with them. Use electrospinning only when micro- or nanoscale diameters and very high surface area are worth the very low throughput, as in filtration media, scaffolds, and separators. It is not a way to make bulk textile fiber.
Key numbersMelt spinning at 500–6,000 m/min, against 50–150 m/min for wet spinning · filament diameter 10–50 µm, or 0.5–20 denier · spinneret packs with tens to several thousand holes · draw ratio 3:1–6:1 after spinning · electrospun fibers 50–1,000 nm at under about 1 g/h per needle · spinneret packs from thousands to tens of thousands of dollars.
ExamplesPolyester, nylon, rayon, spandex, filtration media, battery separators.
VideosHow Manufactured Fibers Are Made (FiberSource, American Fiber Manufacturers Association)
Cellulose fibers are suspended in water, deposited as a continuous web, pressed, dried, and often coated or calendered.
Strengths & weaknessesThroughput is high and the feedstocks are renewable. Water, energy, drying, fiber quality, and machine utilization dominate the economics.
When to useThis applies when your product is a web of renewable cellulose fiber (paper, board, tissue, packaging stock) at machine-scale volumes where cost per ton decides the business. For essentially everyone downstream it is a buy-not-make process, since mills cost hundreds of millions and live or die on utilization, energy, and fiber cost. If you need barrier or structural performance beyond what coated paper can deliver, look at polymer film or laminate constructions instead.
Key numbersMachine speed 300–2,000 m/min at 2–11 m width · basis weight 15–500 g/m² · output 200–2,000 tons per day · furnish enters the headbox at 0.2–1% solids and leaves the dryers at 92–95% · water use roughly 10–50 m³ per ton · machine capex in the hundreds of millions of dollars.
ExamplesPaper, paperboard, tissue, packaging stock.
VideosPaper Primer: Pulp and Paper Manufacturing Overview (NC State Paper Science and Engineering)
Flexible material moves continuously between rolls while being printed, coated, patterned, laminated, dried, cured, or cut.
Strengths & weaknessesThroughput is very high and unit cost is low for flexible products. The difficult parts are tension control, registration, defect propagation, coating uniformity, and line yield.
When to useChoose roll-to-roll when your product is thin, flexible, and needed in very large areas (coated films, battery electrodes, tapes, flexible electronics). Once a line is dialed in, nothing matches its area throughput and unit cost. It does require volume and a mature process, because a defect propagates down the whole web. If you have low volumes, a rigid substrate, or a process still in development, use sheet-fed or batch coating instead to keep scrap and capital contained.
Key numbersWeb width typically 0.3–3 m · line speed 10–300 m/min for coating, higher for printing · coated layer 0.1–500 µm at ±1–3% uniformity · registration accuracy roughly ±10–100 µm on patterned webs · tension control within a few percent · line capex about $1M for a narrow coater, $20M and up for a wide multi-station line.
ExamplesBattery electrodes, flexible electronics, packaging films, solar films, adhesive tape.
VideosWeb Handling Basics: Processes and Applications (Maxcess) · Economic Analysis of Technology Infrastructure Needs for Advanced Manufacturing: Roll-to-Roll Manufacturing (NIST)
High-purity semiconductor material is grown into a crystal, sliced into wafers, ground, lapped, polished, cleaned, and prepared for device fabrication.
Strengths & weaknessesIt produces the ultra-flat, low-defect substrates that semiconductor manufacturing depends on. The downside is that crystal growth, contamination control, polishing, and defect control all require very high capital spending and tight process discipline.
When to useTreat this as a sourcing decision. In almost all cases you should buy polished prime wafers to the diameter, orientation, doping, and flatness spec your device requires, because merchant supply is far cheaper than growing your own. Growing and finishing crystal in-house only makes sense for substrates you can't source at all (novel compound semiconductors, unusual orientations, captive SiC capacity), and it takes cleanroom-grade capital plus years of process development.
Key numbersPrime wafer diameters 100–300 mm at 525–775 µm thick · total thickness variation under 1 µm and polished roughness below 0.1 nm Ra · polysilicon purity 9N to 11N · Czochralski pull rate roughly 0.5–2 mm/min, giving 300 mm ingots of 250–450 kg · wire-saw kerf and polishing lose 30–50% of the ingot · 300 mm prime silicon wafers roughly $60–150 each, with silicon carbide an order of magnitude more.
ExamplesSilicon wafers, silicon-carbide wafers, gallium-arsenide substrates.
VideosCzochralski and Float-Zone Silicon Crystal Growth (MicroChemicals) · Single Crystal Silicon Wafer Manufacturing (MKS Instruments)
Metal Casting
shape parts by solidifying molten metal in molds11 processesMolten metal is poured into a disposable sand mold, which is broken apart after solidification.
Strengths & weaknessesHandles large parts, complex shapes, and many alloys with relatively inexpensive tooling. Surface finish, dimensional accuracy, minimum wall thickness, and defect control are weaker than in precision casting processes.
When to useDefault to sand casting for metal parts from one-offs to a few thousand per year, especially large or heavy parts (kilograms to many tons), cored internal passages, and iron or steel that die casting can't handle. Expect machining allowances on functional surfaces, walls no thinner than roughly 4 mm, and general tolerances of a millimeter or worse. Move to investment casting when fine detail and finish matter, or to permanent-mold and die casting when aluminum volumes climb into the tens of thousands.
Key numbersPart weight from under 1 kg to several hundred tons · minimum wall about 4 mm · linear tolerance roughly ±1 mm on small features and ±1–3% on large castings · as-cast surface Ra 6.3–25 µm · patterns and coreboxes $500–20,000 at 2–6 week lead time · automatic green-sand lines run 100–400 molds per hour · economical from one-offs to a few thousand parts a year.
ExamplesEngine blocks, pump housings, machine bases, valves, manifolds.
Economic profileTooling costs are low to moderate, which makes it a good fit from prototypes through medium volume.
VideosSand Casting: How It Works, Advantages, and Disadvantages (Xometry) · What Is Sand Casting and How Does It Work? (MetalTek)
A wax pattern is coated with ceramic, the wax is removed, and molten metal is poured into the resulting shell.
Strengths & weaknessesProduces intricate shapes, thin walls, fine details, and high-performance alloy parts. Slower and more expensive than sand casting and has practical size limits.
When to useChoose investment casting for complex, thin-walled (down to about 1.5 mm), fine-finish parts in hard-to-machine alloys (superalloys, titanium, stainless) at tens to tens of thousands of pieces, typically under about 50 kg. You get a near-net shape with detail that would otherwise take a lot of machining. Skip it for simple geometry or large parts, where sand casting is much cheaper, and for high-volume aluminum, where die casting has much shorter cycle times.
Key numbersWall thickness down to about 1.5 mm · linear tolerance roughly ±0.13 mm on the first 25 mm plus ±0.05 mm per 25 mm after that · as-cast surface Ra 1.6–3.2 µm · part weight from a few grams to about 50 kg · wax dies $5,000–50,000 at 8–16 week lead time · shell built from 5–10 dip-and-dry coats over 1–2 days · economical from tens to tens of thousands of pieces.
ExamplesTurbine blades, aerospace brackets, surgical implants, impellers, jewelry.
VideosInvestment Casting: How It Works and Advantages (Xometry) · Introduction to Investment Casting (Investment Casting Institute)
A foam pattern is embedded in sand and vaporized by molten metal as the cavity fills.
Strengths & weaknessesIt produces complex near-net-shape castings with fewer cores and parting lines. The hard parts are foam quality, gas evolution as the foam vaporizes, pattern handling, and getting repeatable results.
When to useConsider lost foam at automotive-scale volumes for complex castings that would otherwise need many sand cores (blocks, heads, manifolds), because one foam pattern consolidates the internal passages and removes parting lines and draft. It only pays off if your volumes cover the foam-pattern tooling and you've thoroughly debugged the process. Avoid it for short runs, or if your shop can't hold tight pattern-quality control. The usual fallback is conventional cored sand casting.
Key numbersLinear tolerance roughly ±0.25 mm per 25 mm, better than green sand · as-cast surface Ra 6–12 µm · minimum wall about 3 mm · EPS pattern density 20–25 g/L · pattern tooling $50,000–250,000 at 8–16 week lead time · part weight typically 0.5–100 kg · economical from roughly 10,000 parts a year upward.
ExamplesEngine blocks, cylinder heads, manifolds, pump housings.
VideosLost Foam Casting: Process, Benefits, and Comparisons (The Federal Group) · Lost Foam Casting Step-by-Step Guide (Supreme Petrochem)
Molten metal flows under gravity into a reusable metal mold.
Strengths & weaknessesProvides better consistency, finish, and mechanical properties than sand casting. Tooling is more expensive, geometry is less flexible, and the process is mainly used for lower-melting alloys.
When to useUse permanent-mold casting for aluminum or magnesium parts in roughly the 1,000 to 100,000 per year band. In that range the finish, tolerances, and mechanical properties are better than sand casting, the tooling costs a fraction of a die-cast cell, and the parts stay fully heat-treatable. Below that volume you won't recover the mold cost, so stay with sand casting. Above it, or if your walls are under about 3 mm, use high-pressure die casting instead. Ferrous alloys are generally not an option here.
Key numbersMinimum wall about 3 mm · linear tolerance roughly ±0.25–0.5 mm · as-cast surface Ra 3.2–6.3 µm · cycle time 2–10 minutes per mold · mold life 20,000–100,000 shots · tooling $40,000–200,000 at 8–16 week lead time · economical from about 1,000 to 100,000 parts a year.
ExamplesAluminum wheels, pistons, cookware, brackets, housings.
VideosPermanent Mold Casting: How It Works and Advantages (Xometry) · A Primer on Permanent Mold Casting (Modern Casting)
Molten metal is injected into a hardened steel mold at high speed and pressure.
Strengths & weaknessesProduces thin walls, fine detail, excellent repeatability, and short cycles. Tooling is expensive, production volumes must be high, and trapped gas or porosity can impair welding, heat treatment, fatigue strength, and pressure tightness.
When to useChoose HPDC for aluminum, magnesium, or zinc parts at tens of thousands per year and up, with thin walls (1–3 mm), fine detail, and near-net finish. Cycles run in seconds, so unit cost is usually lower than any other casting route at those volumes. Don't specify it if you need welding, solution heat treatment, or pressure tightness, unless you pay for a vacuum-assisted variant, because some entrapped gas comes with the process. At lower volumes, or for structural parts that need higher integrity, fall back to permanent-mold or low-pressure die casting.
Key numbersWall thickness 1–3 mm · linear tolerance about ±0.1–0.25 mm on the first 25 mm · as-cast surface Ra 0.8–3.2 µm · injection pressure 30–150 MPa at gate velocities of 30–60 m/s · cycle time 15–90 s · die life 100,000–200,000 shots in aluminum and up to a million in zinc · dies from tens of thousands of dollars on small zinc parts to several million on large structural castings, at 12–20 week lead time.
ExamplesTransmission housings, electronics enclosures, appliance components, structural aluminum castings.
VideosDie Casting: Types, Materials, and Design (Xometry) · Die Casting Frequently Asked Questions (NADCA)
Controlled gas pressure pushes molten metal into a reusable die.
Strengths & weaknessesCreates calmer filling and generally lower porosity than high-pressure die casting. Slower and less suited to extremely thin walls.
When to usePick LPDC for structural aluminum parts that need better metallurgical integrity than HPDC and full heat-treatability (wheels, suspension components, cylinder heads), at volumes in the tens of thousands where a cycle time measured in minutes is acceptable. Avoid it for very thin walls, or when cycle time is the main cost driver, since HPDC is better on both. For short runs the die cost is hard to justify, so go back to permanent-mold gravity or sand casting.
Key numbersFill pressure 0.2–1 bar · cycle time 3–10 minutes per mold · minimum wall about 3 mm · linear tolerance roughly ±0.4 mm with as-cast Ra 3.2–6.3 µm · metal yield above 90%, against 50–60% for gravity pouring · die life 50,000–100,000 shots · tooling $150,000–800,000.
ExamplesWheels, suspension components, cylinder heads, pressure housings.
VideosLow-Pressure vs. High-Pressure Die Casting (Kurtz Ersa) · Die Casting of Aluminum Alloys: Part Two (Total Materia)
Molten or partially solid metal solidifies under substantial mechanical pressure inside a die.
Strengths & weaknessesProduces dense, strong parts with low porosity. Equipment, dies, and process development are expensive and the operating window can be narrow.
When to useReserve squeeze casting for safety- and fatigue-critical aluminum parts (suspension arms, knuckles, pistons) that need forging-like density and full heat-treat response in a geometry that's too complex or too costly to forge. Your volumes need to be high enough to cover the dies plus the process development that the narrow operating window requires. If ordinary casting integrity is good enough, use LPDC. If the shape can be forged, closed-die forging often gets you the same properties for less money.
Key numbersApplied pressure 50–150 MPa held through solidification · porosity typically under 1%, low enough for full T6 heat treatment · elongation 5–15% in aluminum, against 1–3% from conventional high-pressure die casting · cycle time 1–3 minutes · part weight usually 0.5–20 kg · tooling $200,000–800,000.
ExamplesPistons, wheels, suspension components, aerospace aluminum parts.
VideosSqueeze Casting: How It Works and Advantages (Xometry) · Squeeze Casting Process: Part One (Total Materia)
Metal is formed while containing both solid and liquid phases. Major variants include thixomolding, thixocasting, and rheocasting.
Strengths & weaknessesReduces turbulence and porosity and can produce strong, thin-walled components. Feedstock preparation, alloy compatibility, thermal control, and specialized machinery increase complexity.
When to useConsider semi-solid processing for thin-walled, high-integrity aluminum or magnesium parts at high volume (electronics housings, structural brackets), when conventional die-cast porosity would rule out heat treatment, welding, or pressure tightness. For magnesium, thixomolding is the practical entry point. The specialized feedstock and machines only pay off at scale, and only if you have real process expertise on staff. If you don't, vacuum-assisted HPDC or squeeze casting get you similar integrity on more conventional equipment.
Key numbersSolid fraction 30–60% at forming · wall thickness down to 0.5–1 mm in thixomolded magnesium · porosity typically under 1%, low enough to weld and heat treat · cycle time 30–90 s · part weight usually under about 5 kg · tooling $200,000 to $1M, close to high-pressure die casting.
ExamplesMagnesium electronics housings, automotive structural components, precision aluminum enclosures.
VideosSemi-Solid Processing of Alloys: Thixocasting and Rheocasting Routes (Total Materia) · Thixocasting (Total Materia)
A rotating mold forces molten material against the mold wall.
Strengths & weaknessesCreates dense cylindrical parts without a central core. Mainly useful for axisymmetric geometry and may require machining of the inner surface.
When to useChoose centrifugal casting for axisymmetric hollow parts (pipe, cylinder liners, bushings, large rings), where spinning the mold gives you dense walls with no core, at anywhere from one-off to steady production. Budget for machining the as-cast bore, since the lighter impurities collect there. It doesn't work for non-round geometry, and for small bushings at high volume, powder metallurgy or drawn tube plus machining is usually cheaper.
Key numbersMold rotation 300–3,000 rpm, giving 60–150 g at the mold wall · outside diameter from about 25 mm to 3 m, lengths to roughly 6 m · wall thickness 5–125 mm · bore machining allowance 3–6 mm, where the lighter inclusions collect · OD tolerance roughly ±0.5–2 mm · molds $5,000–50,000 per size and good for thousands of pours.
ExamplesPipes, cylinder liners, bushings, bearing rings.
VideosCentrifugal Casting: How It Works and Advantages (Xometry) · What Is Centrifugal Casting and How Does It Work? (MetalTek)
Molten metal is poured into a plaster or ceramic mold capable of reproducing finer features than ordinary sand.
Strengths & weaknessesOffers improved surface quality and detail. Mold production is slower, molds are fragile, and temperature compatibility is limited.
When to useUse plaster or ceramic mold casting for prototype and short-run aluminum, zinc, or magnesium parts that need die-cast-like finish and detail before production tooling exists. Typical runs are ones to hundreds of pieces, and it's the usual bridge to die casting. The molds are single-use and slow to make, so the process doesn't scale to higher volumes. Plaster molds can't take the temperature of ferrous alloys, so iron and steel go to ceramic molds (the Shaw process handles tool steel routinely), investment casting, or sand casting.
Key numbersLinear tolerance roughly ±0.13 mm on the first 25 mm · surface Ra 1.6–3.2 µm, close to a die-cast finish · minimum wall about 1.5 mm · plaster molds need 12–24 hours of drying and are used once · a silicone master pattern costs $2,000–15,000 and yields dozens to a hundred molds · lead time 2–4 weeks · typical runs of one to a few hundred parts.
ExamplesPrototype aluminum parts, aerospace castings, tooling components.
VideosPlaster Casting: How It Works, Uses, and Advantages (Xometry) · Ceramic Mold Casting: How It Works and Applications (Xometry)
Material solidifies against the mold wall before the remaining liquid is drained, producing a hollow shell.
Strengths & weaknessesSimple and useful for decorative hollow objects. Wall thickness and structural quality are difficult to control.
When to useChoose slush casting only for decorative hollow parts in low-melting alloys (zinc, pewter, spelter figurines and lamp bases) at craft-to-modest volumes, where simple, cheap molds are the main thing you care about. Don't use it for structural or precision parts, because wall thickness is essentially uncontrolled. If you need functional hollow metal parts, use centrifugal or die casting instead. The plastic equivalent is rotational molding.
Key numbersPouring range roughly 180–420 °C for the pewter, tin, and zinc alloys it suits · wall thickness 1.5–5 mm, controlled to no better than about ±30% · dwell of 5–30 s before the mold is inverted · cycle time under a minute · part weight from grams to a few kilograms · bronze or steel molds good for thousands of pours.
ExamplesFigurines, lamp bases, hollow decorative shells.
VideosPolymer & Elastomer Molding
melt and mold plastics into net-shape parts12 processesMolten thermoplastic is injected under pressure into a metal mold, cooled, and ejected.
Strengths & weaknessesProduces complex parts rapidly with excellent repeatability and very low unit cost at scale. Molds are expensive, design changes are costly, and parts must satisfy rules involving draft, shrinkage, gating, wall thickness, cooling, and ejection.
When to useInjection molding is the default for thermoplastic parts once lifetime volume clears roughly 5,000–10,000 units and the design follows the molding rules (uniform walls, draft, manageable undercuts). At that point you get ±0.05–0.1 mm repeatability at cents per part. Below that volume the mold cost dominates, so use machining, 3D printing, or vacuum casting instead. For very large parts, look at RIM, thermoforming, or rotomolding. Freeze the design before you cut steel, because mold changes are expensive.
Key numbersWall thickness typically 1–3 mm · dimensional repeatability ±0.05–0.1 mm · cycle time 10–60 s · injection pressure 50–150 MPa at clamp forces of 20–4,000 tons · mold shrinkage 0.3–2.5% by resin · steel molds $20,000–200,000 at 6–12 week lead time and 500,000 to over a million shots of life · breakeven against machining or printing usually around 5,000–10,000 parts.
ExamplesElectronics housings, bottle caps, medical disposables, toys, clips, appliance parts.
Economic profileTooling costs are high, so the process usually makes sense for thousands to millions of parts.
VideosInjection Molding Design Guidelines (Protolabs) · Injection Molding: The Manufacturing and Design Guide (Hubs)
Reactive thermosetting material is injected into a mold and chemically cured.
Strengths & weaknessesProvides heat resistance, dimensional stability, and electrical insulation. Cycles can be longer, scrap cannot be remelted, and cure chemistry must be controlled.
When to useUse thermoset injection molding for parts that have to hold their shape and keep insulating at temperatures and arc conditions thermoplastics can't sustain (connectors, breaker bodies, appliance handles), at medium-to-high volumes on BMC, phenolic, or epoxy compounds. If a glass-filled high-temperature thermoplastic meets the spec, use standard injection molding instead, since cycles are faster and the scrap can be remelted. Runners and rejects here can't be recycled, so design your gating to keep them small.
Key numbersBarrel held at 60–100 °C with the mold at 150–200 °C · cure time 20–120 s, so cycles run longer than thermoplastic molding · mold shrinkage 0.1–0.8%, tighter than thermoplastics · continuous service temperature 150–250 °C · runner and reject scrap 5–20%, none of it recyclable · tooling $25,000–150,000, with abrasive-filled compounds cutting mold life to a few hundred thousand shots.
ExamplesElectrical connectors, circuit breakers, appliance handles, high-temperature housings.
VideosThermoset Processing Guide: Injection, Compression, and Transfer Molding (Plenco)
Low-viscosity reactive liquids are mixed and injected into a mold, where they polymerize.
Strengths & weaknessesAccommodates large parts, thick sections, and lower-cost tooling because mold pressure is relatively low. Cycle times, surface quality, and material selection may be inferior to thermoplastic injection molding.
When to useChoose RIM for large, lightweight polyurethane parts (machine covers, fascias, panels at sizes where thermoplastic molds get too expensive) at hundreds to roughly 10,000 units per year. Cavity pressure is low, so the tooling can be inexpensive aluminum or even composite. Avoid RIM for small parts, tight tolerances, or six-figure volumes; thermoplastic injection molding is faster and cheaper per part there. Plan on painting the parts, because the as-molded surface isn't very good.
Key numbersCavity pressure 0.3–1 MPa, against 50–150 MPa for thermoplastic injection molding · cycle time 2–10 minutes · wall thickness 3–12 mm, and it can vary widely across one part · part size up to 3–4 m and 0.5–50 kg · molded density 200–1,100 kg/m³ by foam level · aluminum or composite tooling $5,000–50,000 at 4–8 week lead time · economical from a few hundred to about 10,000 parts a year.
ExamplesAutomotive fascias, machine covers, medical-device housings, polyurethane panels.
VideosWhat Is Reaction Injection Molding? (Romeo RIM) · RIM Design Guide: Part, Tool, and Material Design (Romeo RIM)
A measured charge is placed in a heated mold and compressed until it conforms and cures.
Strengths & weaknessesWorks well for thermosets, rubber, and fiber-reinforced compounds. Slower and less geometrically flexible than injection molding, though tooling may be simpler.
When to useChoose compression molding for rubber, thermoset, and fiber-reinforced parts (SMC panels, seals, electrical housings) at medium-to-high volumes, where simple tooling and low material shear matter. Low shear preserves fiber length, which is why compression molding is the default for large composite panels. Avoid it for intricate geometry, tight flash control, or fragile inserts; transfer molding or injection molding handle those better. For plain thermoplastics, injection molding is faster.
Key numbersMolding pressure 2–20 MPa on presses of 100–3,000 tons · mold temperature 140–180 °C with a 1–5 minute cure · wall thickness 1.5–25 mm · fiber length preserved at 12–50 mm in SMC, against under 1 mm after injection molding · dimensional tolerance roughly ±0.1–0.5 mm · panels up to about 3 m across · tooling $10,000–100,000, well under an equivalent injection mold.
ExamplesElectrical housings, rubber seals, composite panels, melamine products.
VideosCompression Molding: Process, Materials, and Applications (Xometry) · SMC and BMC Thermoset Composites Technology (IDI Composites)
Thermosetting material is forced from a chamber into a closed mold cavity and cured.
Strengths & weaknessesHandles inserts and more complex geometry than compression molding. Creates runner waste and requires more complex tooling.
When to useUse transfer molding when a thermoset or rubber part needs delicate inserts held in a closed cavity, or finer geometry than compression molding can fill. Semiconductor encapsulation and insert-heavy connectors are the usual cases. In exchange you accept cured runner scrap and more expensive tooling. If the shape is simple and has no inserts, stay with compression molding. At very high volumes with complex geometry, thermoset injection molding is usually the better fit.
Key numbersTransfer pressure 20–70 MPa, between compression and injection molding · mold temperature 150–190 °C, cure in 1–5 minutes or 60–120 s for semiconductor encapsulation · dimensional tolerance roughly ±0.1–0.25 mm · cull and runner scrap 10–30%, none of it reusable · epoxy molding compounds usually get a 4–6 hour post-mold cure at about 175 °C · tooling $15,000–150,000.
ExamplesEncapsulated electronics, semiconductor packages, connectors, rubber parts.
VideosTransfer Molding: Process, Uses, and Comparison to Other Molding (Xometry) · Thermoset Processing Guide: Transfer Molding Sections (Plenco)
A heated plastic tube or preform is inflated inside a mold. Variants include extrusion blow molding, injection blow molding, and stretch blow molding.
Strengths & weaknessesHighly efficient for hollow containers. Wall thickness can vary, fine structural features are difficult, and geometry must support inflation and release.
When to useBlow molding is the right choice for hollow thermoplastic containers at six-figure volumes and up: stretch blow for PET beverage bottles, extrusion blow for handled and irregular containers like detergent bottles and fuel tanks, injection blow for small precise-neck vials. Below tens of thousands of units, or for very large tanks at modest volume, rotational molding is usually cheaper because its tooling costs so much less. If the part needs tight wall control or structural detail, use injection molding instead.
Key numbersDimensional tolerance roughly ±0.5 mm on small blown features and looser as parts scale up · wall thickness variation ±10–25% · extrusion blow molds roughly $5,000–50,000, 4–10 weeks lead time · cycle about 2 s per cavity on PET stretch blow, 30–120 s on large extrusion-blown tanks · part volumes from a few milliliters to about 1,000 L · economic lot size 100,000 units and up.
ExamplesBeverage bottles, fuel tanks, detergent containers, drums, ducts.
VideosBlow Molding vs. Injection Molding: Differences and Comparison (Xometry)
Polymer powder coats the inside of a heated rotating mold and forms a seamless hollow part.
Strengths & weaknessesExcellent for large hollow parts and low-to-medium volumes because tooling is relatively inexpensive. Cycles are slow, tolerances are loose, and material choices are limited.
When to usePick rotomolding for large seamless hollow parts (tanks, kayaks, bins from tens of liters to several cubic meters) at dozens to a few thousand units per year, where its cheap cast-aluminum or fabricated tooling costs far less than the alternatives. In exchange you get cycles of 30–60 minutes, loose tolerances, and a short list of materials led by polyethylene. At high volumes, or for small containers, use blow molding instead. If you need precision or structural detail, use injection molding.
Key numbersDimensional tolerance roughly ±1% of the dimension, wall thickness ±10–20% · walls typically 3–12 mm · cycle 30–60 minutes per part · cast-aluminum or fabricated-steel molds $3,000–25,000, 3–8 weeks lead time · part volumes from tens of liters to several cubic meters, almost all polyethylene · economic lot size dozens to a few thousand a year.
ExamplesWater tanks, kayaks, playground equipment, coolers, barriers.
VideosRoto 101: Introduction to Rotomolding (Association of Rotational Molders) · ARM Design Guide: Designing Products for Rotational Molding (Association of Rotational Molders)
Plastic sheet is heated, stretched over or into a mold, and trimmed. Vacuum forming and pressure forming are common variants.
Strengths & weaknessesTooling is relatively inexpensive and large parts can be formed quickly. Geometry is mostly one-sided, wall thinning occurs, and trimming generates scrap.
When to useThermoforming suits shell-like one-sided parts made from sheet (trays, liners, panels, enclosures), from prototypes on cheap wood or epoxy molds up to very high-volume thin-gauge packaging on dedicated lines. Tooling costs a fraction of an injection mold, so thermoforming also covers the large-part and mid-volume range that injection molding can't serve economically. Avoid it when you need two-sided features, uniform wall thickness, or tight tolerances, and budget for trim scrap. Injection molding covers those needs.
Key numbersSheet thickness 0.2–1.5 mm thin gauge, 1.5–12 mm heavy gauge · dimensional tolerance about ±0.5 mm thin gauge and ±1–1.5 mm heavy gauge · corners thin to 30–50% of starting gauge · tooling from a few hundred dollars in wood or epoxy to $20,000 in cast aluminum, 2–6 weeks lead time · cycle a few seconds per part on roll-fed thin-gauge lines, 1–6 minutes on heavy gauge · economic lot size from single prototypes to millions.
ExamplesPackaging trays, refrigerator liners, aircraft interior panels, signs, shower enclosures.
VideosThermoforming Overview and Design Guide (Ray Products) · Heavy Gauge Thermoforming Process and Design Guide (Productive Plastics)
Liquid polyurethane or similar resin is poured under vacuum into a flexible silicone mold, usually copied from a master pattern.
Strengths & weaknessesUseful for prototypes and short production runs without hard tooling. Silicone molds wear quickly and material properties only approximate production polymers.
When to useUse vacuum casting for 10–100 production-like polyurethane parts (functional prototypes, pilot builds, marketing samples) delivered in days without hard tooling. Each silicone mold is good for roughly 15–25 casts. Treat the material properties as approximations of the production polymer rather than equivalents. This is a bridge process. Past a few hundred parts, cut a mold and switch to injection molding; for a handful of pieces, 3D printing is cheaper still.
Key numbersDimensional tolerance roughly ±0.3% of the dimension, about ±0.25 mm minimum · silicone molds $500–2,000 and 3–7 days to make, good for 15–25 casts · cycle 1–4 hours per casting including cure · parts up to roughly 500 mm in typical chambers · polyurethane resins from Shore A 30 to Shore D 80 · economic lot size 10–100 pieces.
ExamplesPrototype housings, automotive mockups, medical-device enclosures.
VideosPolymer expands inside a mold through gas, steam, or chemical blowing.
Strengths & weaknessesProduces lightweight, insulating, impact-absorbing structures. Surface finish, structural consistency, and recyclability may be limited.
When to useChoose foam molding when you need lightness, thermal insulation, or controlled energy absorption (helmet liners, protective packaging, automotive absorbers) at medium volumes and up. EPS or EPP steam-chest molding is the standard route. Structural-foam variants stiffen large parts while cutting weight and clamp tonnage. Avoid foam molding when you need fine cosmetic surfaces, tight tolerances, or easy recyclability; use solid injection molding instead.
Key numbersMolded density 15–100 kg/m³ for EPS and 20–120 kg/m³ for EPP · dimensional tolerance roughly ±1% · steam-chest cycle 1–3 minutes plus drying · aluminum tooling $5,000–40,000, 4–10 weeks lead time · structural foam variants cut part weight 10–30% and run at well under half the clamp tonnage of solid injection · economic lot size tens of thousands and up.
ExamplesHelmets, protective packaging, insulation, automotive energy absorbers.
VideosFoam Injection Molding: Process, Benefits, and Materials (Xometry)
A prefabricated metal or other insert is placed in the mold and encapsulated by plastic.
Strengths & weaknessesConsolidates assembly and creates integrated threaded, electrical, or structural interfaces. Insert handling, thermal expansion, and molding forces add complexity.
When to useSpecify insert molding when a threaded, conductive, or load-bearing metal feature has to be captured in plastic at medium-to-high volume (connectors, sensor bodies, fastener bosses). It removes assembly steps and performs better than a glued or pressed joint. You need enough volume to justify insert-loading automation or the added cycle labor. At low volumes, mold plain bosses and install heat-set or ultrasonic inserts afterward, which keeps tooling simple and the design flexible.
Key numbersDimensional tolerance ±0.05–0.15 mm on molded features, insert position within about ±0.1 mm · cycle 20–60 s, roughly 10–30 s longer than the same part molded without inserts · tooling $10,000–80,000, plus insert-loading automation · metal-to-polymer thermal expansion mismatch of 3–6x at the interface · economic lot size tens of thousands and up.
ExamplesThreaded housings, electrical connectors, sensor bodies, surgical handles.
VideosOvermolding and Insert Molding Design Guidelines (Protolabs) · Insert Molding vs. Overmolding: What's the Difference? (Xometry)
A second material is molded over an existing substrate or first molded component.
Strengths & weaknessesCreates grips, seals, soft-touch areas, insulation, and multi-material components. Adhesion, shrinkage, thermal compatibility, and extra tooling must be controlled.
When to useChoose overmolding for soft-touch grips, integrated seals, and two-material parts (toothbrushes, tool handles, gasketed housings) at volumes that justify either two-shot tooling (high volume, best per-part economics) or a transfer process between two single molds (moderate volume). Check the adhesion of the material pair early, because an incompatible pair needs mechanical interlocks. Below a few tens of thousands of units, molding the parts separately and joining them by assembly or adhesive is usually cheaper.
Key numbersOvermold layer 1–3 mm thick, 0.5 mm minimum · TPE grades from Shore A 20 to Shore A 90 · two-shot tooling $30,000–150,000 against $15,000–60,000 for a pair of single-shot molds, both 8–14 weeks · cycle 30–60 s on two-shot, longer through a transfer process · dimensional tolerance roughly ±0.15–0.3 mm on the soft layer · economic lot size tens of thousands through a transfer process, 100,000 and up for two-shot.
ExamplesTool grips, toothbrushes, sealed electronics, cable connectors.
VideosOvermolding and Insert Molding Design Guidelines (Protolabs) · TPE Overmolding, Bonding and Substrate Considerations (RTP Company)
Glass & Ceramic Forming
shape molten glass and ceramic powders into parts6 processesHeated glass is inflated and shaped manually or mechanically, often inside a mold.
Strengths & weaknessesReadily creates seamless hollow objects. Dimensional precision, thick structural sections, and sharp internal features are difficult.
When to useMachine blowing is the route for narrow-necked hollow glassware at volume (bottles, jars, bulbs, globes). Hand blowing covers low-volume laboratory, lighting, and artistic ware, where tooling would never pay for itself. Expect loose dimensional control, and avoid thick structural sections or sharp internal features. If the piece is open, thick, or precise, press it instead. Flat stock comes from float lines.
Key numbersGob delivered at 1,050–1,150 °C, annealing lehr around 500–560 °C · IS machines run roughly 200–700 containers per minute · container walls 1.5–4 mm, weights around 100–800 g · dimensional tolerance ±0.5–1 mm on the body and about ±0.25 mm at the neck ring · cast-iron mold sets a few thousand dollars per cavity · hand blowing takes minutes to hours per piece.
ExamplesBottles, laboratory glassware, lighting globes, vessels.
VideosA measured gob of molten glass is pressed into shape in a mold.
Strengths & weaknessesRepeatable and efficient for relatively thick glass components. Undercuts, thin walls, and large thickness variation are difficult.
When to useChoose pressing for open, relatively thick glass shapes at medium-to-high volume (lenses, insulators, dishes, headlamp optics), wherever a plunger can reach every surface and repeatability matters more than delicacy. Pressing can't make narrow-necked hollowware, thin uniform walls, or undercuts; those need blowing. For precision optics, plan on grinding and polishing downstream unless you invest in precision molding.
Key numbersGob pressing at 1,000–1,150 °C; precision glass molding at 500–700 °C with low-Tg glasses · sections 2–10 mm, thicker than blowing can hold · cycle 5–15 s per station on automated lines · dimensional tolerance ±0.2–0.5 mm on ordinary pressed ware, near ±0.02 mm center thickness on precision-molded optics · mold sets $5,000–30,000, more for tungsten-carbide precision molds · economic lot size tens of thousands and up.
ExamplesLenses, dishes, insulators, headlamp optics, tiles.
VideosFlat glass is heated until it softens and conforms to a mold or bends under gravity.
Strengths & weaknessesProduces curved glass without cutting and reassembly. Cycle times are long, optical distortion can occur, and shape complexity is limited.
When to useUse slumping and bending to curve flat float glass (architectural panels, curved glazing, lighting shades) at one-off to medium volumes. Molds are cheap and the feedstock is commodity sheet. Expect long kiln cycles, gentle curvature only, and some optical distortion on demanding contours. If you need a deep or complex three-dimensional shape, use pressing or blowing instead.
Key numbersSlumping at roughly 620–700 °C for soda-lime float, below the 780–820 °C full-fuse range · kiln cycle 6–24 hours including the anneal · glass thickness 3–19 mm · minimum bend radius usually a few hundred millimeters, growing with thickness · steel, ceramic-fiber, or refractory-board molds $500–10,000 · economic lot size one-off to a few thousand panels.
ExamplesArchitectural panels, automotive glazing, lighting components.
VideosCeramic slurry is poured into a porous mold, which removes liquid and leaves a consolidated layer.
Strengths & weaknessesProduces complex hollow ceramics with inexpensive tooling. Slow and subject to drying and firing shrinkage.
When to useSlip casting is the usual pick for hollow or complex ceramic shapes (sanitary ware, tubes, artware), from single pieces to thousands, because plaster molds cost almost nothing. Budget generous tolerances for drying and firing shrinkage, and expect cast cycles that run for hours. If the shape is simple and solid and the volume is high, pressing is far faster. For small intricate technical ceramics at scale, use ceramic injection molding.
Key numbersCast wall builds a few millimeters per hour, thickness rising with the square root of dwell time · total drying and firing shrinkage 10–15% linear · dimensional tolerance about ±1–2% of the dimension · plaster molds $100–1,000, good for 50–150 casts · firing 1,100–1,400 °C depending on the body · economic lot size one piece to a few thousand.
ExamplesSanitary ware, pottery, ceramic tubes, shell forms.
VideosSlip Casting: Technical Principles and Control (Digitalfire) · Ceramic Casting Methods (Fraunhofer IKTS)
Ceramic powder is compacted in a rigid die or through isostatic pressure and then sintered.
Strengths & weaknessesProduces consistent, high-volume parts. Undercuts, major thickness variation, and density gradients can cause defects.
When to useChoose die pressing for simple, prismatic ceramic parts at high volume (tiles, substrates, cutting inserts, seal faces), where uniaxial compaction and sintering give the lowest unit cost. Switch to isostatic pressing when you need uniform density, or for elongated shapes such as tubes and rods. Avoid undercuts, large thickness variation, and delicate features, which press poorly and crack in sintering. If the shape is complex and hollow, use slip casting instead; if the parts are fine and intricate, use ceramic injection molding.
Key numbersUniaxial compaction at 50–200 MPa, cold isostatic pressing at 100–400 MPa · length-to-diameter under about 2–3 for uniaxial pressing before density gradients show up · sintering shrinkage 12–20% linear · as-fired tolerance roughly ±1–2%, down to ±0.01 mm if you grind after firing · carbide die sets $5,000–30,000 · press rate 10–100 small parts per minute.
ExamplesTiles, cutting inserts, electronic substrates, insulators, wear parts.
VideosUniaxial and Isostatic Pressing (Fraunhofer IKTS) · Technical Ceramics Handbook (CeramTec)
Ceramic powder mixed with binder is injection-molded, debound, and sintered.
Strengths & weaknessesProduces intricate small ceramic components at scale. Tooling, debinding, sintering shrinkage, and process development are demanding.
When to useUse CIM for small, intricate technical ceramics (dental brackets, watch cases, nozzles), typically under about 100 g with sections below roughly 10 mm so that debinding stays clean. Volumes of tens of thousands and up are what amortize the hard tooling and the process development. Sintering shrinkage runs near 20%, and you have to dial it in for each geometry. If your volumes are lower or your parts are larger, press-and-sinter or green machining of pressed blanks is the practical route.
Key numbersParts usually under 100 g with sections below about 10 mm · binder 40–50% of the feedstock by volume · sintering shrinkage near 20% linear · as-sintered tolerance roughly ±0.3–0.5% of the dimension · molding cycle 15–60 s, with debinding and sintering adding 1–3 days · tooling $20,000–100,000, 8–16 weeks lead time · economic lot size tens of thousands and up.
ExamplesDental brackets, watch parts, medical instruments, electronic insulators.
VideosBulk Metal Deformation
forge and squeeze solid metal into shape8 processesMetal is compressed between simple dies that do not fully enclose the workpiece.
Strengths & weaknessesProduces very large, strong parts with favorable grain flow and relatively modest tooling. Accuracy is low and substantial machining is usually required.
When to useOpen-die forging is for very large or very low-volume parts that have to be metallurgically sound (turbine shafts, vessel rings, rotors from hundreds of kilograms to hundreds of tons), where no closed die could exist. Treat the forging as a rough envelope and plan substantial machining allowances. If you need thousands of identical net-shape parts, move to closed-die forging. If your strength requirements allow it, a large casting is cheaper still.
Key numbersPart mass from a few hundred kilograms to several hundred tons · forged at 1,100–1,250 °C in steel · dimensional tolerance roughly ±1–2% of the dimension, so 6–25 mm of machining stock per surface · presses of 1,000–15,000 tons cover most industrial work · flat, V, and swage dies costing a few thousand dollars, usually already on the floor · cycle minutes to hours per piece · economic lot size 1 to a few hundred.
ExamplesTurbine shafts, pressure-vessel rings, generator rotors, large gears.
VideosOpen Die Forging Process and Advantages (Scot Forge) · Product Design Guide for Forging (Forging Industry Association)
Metal is compressed inside shaped dies that substantially enclose the workpiece.
Strengths & weaknessesProduces strong, fatigue-resistant parts with directional grain flow. Dies are expensive, draft and flash are common, and internal cavities are limited.
When to useSpecify closed-die forging for fatigue- and impact-critical parts (connecting rods, crankshafts, landing gear, hand tools) once volumes reach roughly 10,000 pieces, which is about where die cost is amortized. The directional grain flow gives properties castings cannot match. Design around draft, flash, and the absence of internal cavities, and expect to finish machine the critical surfaces. Below tooling-payback volume, machine from bar or use open-die forging. If the loads permit it, casting is cheaper.
Key numbersPart weight typically 0.1–50 kg · dimensional tolerance ±0.25–1.5 mm depending on size, with 3–7° draft on hot-forged surfaces · die sets $10,000–100,000, 6–16 weeks lead time, roughly 10,000–50,000 parts before resink · cycle 5–30 s per piece · flash and scale take 10–30% of the input material · economic lot size around 10,000 pieces and up.
ExamplesConnecting rods, crankshafts, landing-gear components, hand tools.
VideosOpen Die vs. Impression Die Forging (Trenton Forging) · Product Design Guide for Forging (Forging Industry Association)
Wire, rod, or billet is compressed axially so material flows outward and forms a head or enlarged section.
Strengths & weaknessesExtremely fast, chipless, and produces strong parts with good grain flow. Geometry and deformation are constrained by material ductility and die access.
When to useCold heading is the automatic choice for fasteners and similar headed axisymmetric parts in ductile metals, running at hundreds of pieces per minute. The economics need runs in the hundreds of thousands to millions. The process is chipless, so it strengthens the material and wastes almost nothing, which is why it beats machining at volume. Below tens of thousands of pieces, or for low-ductility alloys and features the dies cannot reach, screw-machine turning is the fallback.
Key numbersRate 100–400 pieces per minute · wire and rod stock 2–25 mm diameter · unsupported upset length limited to about 3 stock diameters per blow · diameter tolerance ±0.05–0.15 mm · material utilization above 95%, against 40–60% for screw machining · tool sets $2,000–20,000, good for 100,000 to 1,000,000 hits · economic lot size in the hundreds of thousands.
ExamplesBolts, screws, rivets, valve heads, pins.
VideosWhat Is Cold Heading? Key Processes Explained (TFG USA) · Cold Heading and Cold Forming (IQS Directory)
A billet passes through grooved rolls that progressively redistribute material.
Strengths & weaknessesEfficient for elongated parts and forging preforms. Geometry must be compatible with progressive rolling and dedicated tooling is required.
When to useUse roll forging for elongated parts whose cross-section varies along their length (axle and leaf-spring blanks, tapered levers). Most often it is used as a preform step feeding closed-die forging, where a good preform cuts flash losses and die wear. It needs medium-to-high volumes to pay for the dedicated grooved rolls. For one-offs or short runs, open-die drawing-out or machining from bar does the same job without tooling.
Key numbersCross-sectional reduction 20–50% per pass, usually 2–4 passes · cycle 5–20 s per piece · billet stock roughly 20–150 mm diameter · grooved roll sets $10,000–40,000, 6–12 weeks lead time · preforming cuts the flash loss on the finish forging, which otherwise runs 10–30% of input mass · economic lot size tens of thousands and up.
ExamplesAxles, spring blanks, connecting-rod preforms, tool blanks.
VideosA pierced ring is rolled to increase diameter and reduce wall thickness.
Strengths & weaknessesCreates strong seamless rings with excellent grain orientation and material efficiency. Limited to annular geometry.
When to useRing rolling is the sensible route to seamless rings from roughly 200 mm to several meters in diameter (bearing races, jet-engine and flange rings) at one-off to medium volumes. It gives circumferential grain flow and far better material yield than machining from plate or a forged block. It handles annular geometry only. If the rings are small and the volume is high, cold forming or machining from tube is usually better, and non-round frames have to be forged or fabricated.
Key numbersRing diameters from about 200 mm to several meters, to 10 m on the largest mills · wall thickness 20–500 mm · ring mass from tens of kilograms to over 50 tons · rolled at 1,100–1,250 °C in steel · cycle 1–5 minutes per ring · material yield 60–90%, against roughly 30–50% cutting the ring from plate · machining allowance 5–15 mm per surface.
ExamplesBearing races, jet-engine rings, wind-turbine rings, flanges.
VideosSeamless Rolled Ring Forging Process (Scot Forge) · Product Design Guide for Forging (Forging Industry Association)
Repeated radial hammering or die action reduces or shapes rods, tubes, or wire.
Strengths & weaknessesCreates tapered or stepped axisymmetric parts with good finish and strengthened material. Complex non-axisymmetric geometry is impractical.
When to useChoose rotary swaging for tapering, pointing, or stepping round bars and tubes (cable fittings, tapered shafts, tube-end reductions) at medium-to-high volumes. The dies are cheap, the process is chipless so it saves material, and the surface comes out work-hardened and burnished, which is what makes it beat turning here. It also attaches end fittings by swaging directly over them. Anything non-axisymmetric is out of scope, and for one-offs or complex internal profiles CNC turning is the simpler answer.
Key numbersBar and tube diameters from about 1 mm to 50 mm · area reduction 20–50% per pass · diameter tolerance ±0.02–0.1 mm · burnished finish around Ra 0.2–0.8 µm · cycle a few seconds per piece, with the dies striking 1,500–4,000 times a minute · die sets $1,000–5,000 · economic lot size tens of thousands and up.
ExamplesCable fittings, tapered shafts, medical tubing, tool handles.
VideosRotary Swaging Technology Overview (Felss) · Rotary Swaging Applications from Aerospace to Automotive (FENN)
A metal slug is struck by a punch and flows around the punch or through a die.
Strengths & weaknessesProduces thin-walled seamless parts rapidly with little waste. Requires ductile materials and relatively simple geometry.
When to useImpact extrusion makes seamless deep thin-walled cans and tubes in ductile metals (aluminum, zinc, copper) at high volumes: aerosol cans, battery cases, collapsible tubes. A single blow replaces multiple draw stages. It requires soft alloys and simple axisymmetric geometry. For steels, complex shapes, or shallower parts, deep drawing is the standard alternative. At low volumes, machine or spin the part instead.
Key numbersWall thickness 0.2–2 mm at length-to-diameter ratios up to about 10:1 · 50–95% area reduction in a single blow · rate 30–150 pieces per minute · diameter tolerance ±0.05–0.15 mm · tool sets $5,000–30,000 · limited to soft metals such as aluminum, zinc, copper, and tin · economic lot size 100,000 pieces and up.
ExamplesAerosol cans, battery cases, collapsible tubes, aluminum housings.
VideosWedge-shaped dies redistribute material along a rotating cylindrical billet.
Strengths & weaknessesProduces stepped shafts quickly with low waste and favorable grain flow. Tooling is specialized and internal cracking can occur.
When to useCross-wedge rolling suits stepped shafts and forging preforms at automotive volumes, typically 100,000-plus per year. The cycle takes seconds, material loss is minimal, and the grain flow is better than what you get turning from bar. Tooling is specialized, and internal Mannesmann-effect cracking has to be engineered out during development. Below tooling-payback volume, CNC turning or roll forging covers the same parts. For the largest shafts, use open- or closed-die forging.
Key numbersCycle 3–15 s per shaft · billet 20–100 mm diameter, lengths to roughly 800 mm · area reduction 30–75% at the rolled steps · material utilization 90–95%, against 50–70% turning the same shaft from bar · diameter tolerance ±0.5–1 mm, so 1–3 mm of machining stock · wedge die sets $30,000–100,000, 10–20 weeks lead time · economic lot size 100,000 pieces a year and up.
ExamplesAxles, transmission shafts, stepped pins, connecting-rod preforms.
VideosAnalysis of a Cross Wedge Rolling Process for Producing Drive Shafts (Int J Adv Manuf Technol, open access) · Material Fracture Susceptibility in Cross-Wedge Rolling with Concave Tools (Materials, open access)
Sheet-Metal Forming
bend and stretch flat sheet into 3D shapes11 processesSheet metal is cut or formed in a press using dies. It is an umbrella family covering blanking, piercing, bending, drawing, embossing, and related operations.
Strengths & weaknessesProvides extremely high production rates and low unit cost at volume. Presses and dies are expensive, design changes are costly, and tearing, springback, wrinkling, and tool wear constrain designs.
VariantsA punch cuts a desired shape from sheet; the removed piece becomes the product. Fast and repeatable but produces burrs and surrounding scrap.
A punch removes holes or slots from sheet. Highly productive but creates burrs and requires compatible tool access.
Sheet is sheared under strong compressive constraint using specialized tooling, producing smooth, nearly full-thickness cut edges. It works best within a moderate thickness band, so don't think of it as a "thick-sheet" process.
Sheet is plastically bent along a line using a punch and die. Springback, bend radius, grain direction, and tool interference govern feasibility.
A flat blank is drawn into a die to create a cup, shell, or box. Draw depth, ductility, wrinkling, tearing, and corner radius are the main constraints.
Material is compressed into fine die details or surface patterns. Creates texture, stiffness, or precise local features but imposes high die loads.
Stamping is the default for sheet-metal parts at high volume, meaning tens of thousands per year and up. Progressive dies cost tens to hundreds of thousands of dollars, and they pay back through cents-per-part pricing at dozens of strokes per minute. Lock the design early around springback, tearing, and die constraints, because changes after die tryout are expensive. Until annual volume reaches tens of thousands, laser cutting plus press-brake forming is almost always cheaper.
Key numbersSheet thickness usually 0.2–6 mm · dimensional tolerance ±0.05–0.25 mm, down to ±0.025 mm with fine blanking · punch-to-die clearance 5–10% of stock thickness per side · progressive dies $20,000–300,000, 8–20 weeks lead time, 1–10 million hits of die life · press rate 20–100 strokes per minute, into the hundreds on small high-speed work · economic lot size tens of thousands a year and up, at piece prices in cents.
ExamplesAutomotive panels, brackets, appliances, electrical contacts, washers.
VideosMetal Stamping Design Standards and Considerations (Xometry) · AHSS Application Guidelines: Forming (WorldAutoSteel)
Sheet or plate is bent using a programmable punch-and-die press.
Strengths & weaknessesIt's flexible and cheap for low-to-medium volume fabrication. It's also slower than dedicated stamping, and it's subject to springback, operator variation, and tool-access constraints.
When to useUse a press brake for bent sheet-metal parts (enclosures, brackets, chassis) from one-off prototypes up to a few tens of thousands per year. Standard tooling means near-zero tooling cost and same-day turnaround from any fab shop. Design for tool access and bend order, and expect roughly ±0.25 mm and a degree on bends. If annual volume climbs into the high tens of thousands and the design is stable, dedicated stamping dies will beat a press brake on unit cost.
Key numbersSheet thickness 0.5–12 mm, bend lengths to 3–4 m on a standard machine · bend angle within about ±1° and flange dimensions within ±0.25 mm · inside bend radius roughly equal to material thickness in mild steel · minimum flange length about 4 times thickness · cycle 5–20 s per bend · standard tooling, so near-zero tooling cost and same-day turnaround · economic lot size one piece to a few tens of thousands a year.
ExamplesEnclosures, brackets, cabinets, channels, structural panels.
VideosSheet Metal Fabrication Design Guide (Hubs) · The Basics of Bend Radii in Sheet Metal (Protolabs)
Continuous sheet passes through successive rolls that gradually bend it into a constant profile.
Strengths & weaknessesIt's very efficient for long, high-volume components. Tooling is expensive, and variable cross-sections are difficult.
When to usePick roll forming for constant-cross-section profiles produced from coil at high annual volumes (tens of thousands of linear meters per year or more). It's especially good for lengths beyond press-brake capacity and for parts that need in-line punching and cutoff. It also handles high-strength and pre-painted stock gently, because the bending is distributed over many stands. Avoid it below the volume needed to amortize a roll set (often tens of thousands of dollars per profile), for cross-sections that vary along the length, or for short discrete parts. In those cases, press-brake forming or stamping is the default.
Key numbersLine speed 10–100 m/min through 6–20 forming stands · strip thickness 0.3–6 mm, profile widths to about 600 mm · cross-section tolerance ±0.2–0.5 mm, cut length within about ±1 mm · roll sets $15,000–100,000 per profile, 8–16 weeks lead time · economic lot size tens of thousands of linear meters a year and up.
ExamplesRoofing, door tracks, automotive rails, shelving uprights, gutters.
VideosSheet is held under tension and stretched around a die.
Strengths & weaknessesIt produces large smooth curves with little wrinkling. It's inefficient for small intricate parts, and it causes edge waste and thinning.
When to usePick stretch forming for large, gently contoured panels (meter-scale skins with shallow compound curvature) at tens to a few thousand pieces. At those volumes a single form die plus a wrinkle-free class-A surface costs less than matched stamping dies. Budget for the gripped edge trim, often 50–150 mm per side, and check that thinning stays within skin gauge allowances. Avoid it for sharp features, deep draws, or small parts. Stamping handles those at volume, and incremental sheet forming covers die-less one-offs.
Key numbersSheet stretched 2–5% past yield, thinning it 3–10% · contour tolerance ±0.5–1.5 mm · gripped edge trim 50–150 mm per side · form dies in steel, kirksite, or epoxy at $5,000–50,000 · presses pulling 100–2,000 tons on tables to about 15 m · cycle 1–5 minutes per part · economic lot size tens to a few thousand pieces.
ExamplesAircraft skins, railcar panels, architectural surfaces.
VideosStretch Forming and Thermal Forming of Aerostructures (Ducommun)
Pressurized fluid forces sheet or tube into a die.
Strengths & weaknessesIt produces complex seamless structures and can cut part count. The equipment and tooling are expensive, pressure control is demanding, and the material has to be formable enough to survive the process.
When to usePick tube hydroforming when one seamless hollow member can replace a welded assembly of stampings and stiffness per kilogram justifies the press investment. The usual case is automotive frames and exhausts at tens of thousands of units per year, in ductile aluminum or in mild and stainless steel. Sheet hydroforming makes sense at lower volumes, because using a single rigid die half cuts tooling cost roughly in half versus matched dies. Avoid it for prototypes and short runs (fabricate and weld instead), for simple open sections that stamping or roll forming handles better, and for low-elongation alloys that split under internal pressure.
Key numbersForming pressure 30–100 MPa for sheet and 100–400 MPa for tube · press capacity 1,000–10,000 tons · tube circumference expansion up to 20–50% in steel, less in aluminum · cycle 20–60 s per part · dimensional tolerance ±0.5–1 mm · tooling $50,000–500,000, 12–24 weeks lead time · economic lot size tens of thousands a year.
ExamplesAutomotive subframes, exhaust components, bicycle frames, aerospace ducts.
VideosHydroforming 101: Process, Calculations, and Applications (Macrodyne) · What Is Hydroforming? (Jones Metal Products)
A rotating sheet blank is progressively formed over a mandrel using a roller.
Strengths & weaknessesTooling is inexpensive, which suits prototypes and low-volume axisymmetric parts. Production is slower than stamping, and the geometry has to be rotationally symmetric.
When to usePick spinning for round hollow shapes (cones, domes, hemispheres, flanged shells) from one-offs up to a few thousand pieces per year. A mandrel costs hundreds to a few thousand dollars and replaces a deep-draw die costing fifty times that. It scales from cookware diameters to multi-meter tank heads, and it handles thicknesses that conventional presses struggle with. Avoid it for anything non-axisymmetric. If you need sustained volumes above roughly 5,000–10,000 pieces per year, deep drawing or stamping is better on cycle time and consistency.
Key numbersBlank diameters from about 50 mm to 5 m · thickness 0.5–6 mm on manual work, to 25 mm with power spinning · dimensional tolerance ±0.25–1 mm · surface finish Ra 0.4–3.2 µm · mandrels $500–5,000 in wood, aluminum, or steel · cycle 20–60 s on a CNC spinner, minutes per part by hand · economic lot size one-off to roughly 5,000–10,000 pieces a year.
ExamplesNose cones, cookware, reflectors, pressure-vessel ends.
VideosMetal Spinning Design Guidelines (Church Metal) · Metal Spinning Technology (Leifeld)
Rollers compress a rotating cylindrical preform over a mandrel, reducing wall thickness and extending its length.
Strengths & weaknessesIt produces strong, precise, thin-walled axisymmetric parts. The equipment is specialized, and the geometry you can make is restricted.
When to usePick flow forming for seamless thin-walled cylinders and cones at hundreds to low tens of thousands of pieces, when you need wall accuracy of a few hundredths of a millimeter and cold-worked strength. Typical parts are rocket and motor cases, high-pressure cylinders, and premium wheels. The 20–50% strength gain from cold work often lets you thin the walls and drop weight compared with machined or drawn equivalents. Avoid it for non-rotational geometry, for very short simple tubes that are cheaper to draw, and for one-offs that don't justify the mandrel and machine setup. Conventional tube drawing or spinning covers those cases.
Key numbersWall reduction 50–90%, stretching the preform 3–8 times its starting length · wall thickness tolerance ±0.02–0.05 mm · surface finish Ra 0.4–1.6 µm · yield strength up 20–50% from the cold work · diameters roughly 25–1,000 mm, lengths to about 4 m · cycle 2–15 minutes per part · economic lot size hundreds to low tens of thousands.
ExamplesRocket motor cases, pressure cylinders, wheels, drive shafts.
VideosA CNC-controlled tool progressively deforms sheet without a full matched die.
Strengths & weaknessesIt avoids expensive tooling, which makes it useful for prototypes and custom parts. It's slow, and dimensional accuracy is usually modest.
When to usePick incremental forming for one-offs and runs under roughly 100 pieces (prototype panels, patient-specific implants, replacement parts for out-of-production vehicles), where die cost would dominate and a CNC toolpath is the only tooling you need. Expect cycle times of minutes to hours per part and accuracy around ±0.5–1 mm, unless you add a partial die and compensation passes. Avoid it for production volumes or tight-tolerance skins. Stamping takes over once quantities justify dies, and stretch forming handles large smooth aerospace contours.
Key numbersSheet thickness 0.5–3 mm · maximum wall angle 60–70° from the sheet plane, which leaves a third to half the starting gauge · accuracy roughly ±0.5–1 mm · hemispherical tools 6–30 mm across, stepping down 0.2–1 mm per pass · cycle minutes to hours per part · tooling is a forming tool and a backing plate, a few hundred dollars · economic lot size 1 to about 100 pieces.
ExamplesCustom medical implants, prototype body panels, low-volume enclosures.
VideosSingle Point Incremental Forming: State-of-the-Art and Prospects (Duflou et al., KU Leuven)
An explosive-generated pressure wave drives metal into a die, often through water.
Strengths & weaknessesIt can form very large parts with relatively simple tooling. Safety, regulation, facility requirements, repeatability, and cycle time all limit where it gets used.
When to usePick explosive forming only for very large parts in single digits to low tens of pieces: multi-meter domes, vessel heads, and panels that exceed any available press. One female die and a water tank replace forming capacity you couldn't buy at a reasonable price. It also suits hard-to-form alloys, because the high strain rate can improve formability. Avoid it for anything a press or spinning machine can reach. Explosive licensing, remote facilities, and one shot per cycle make it uncompetitive at any real volume, and stretch forming or superplastic forming handles most large aerospace contours.
Key numbersStrain rates of 100 to 10,000 per second, against roughly 1 per second in a press · charges from tens of grams to a few kilograms, fired at a standoff in a water tank · peak pressures of hundreds of MPa at the sheet · parts from about 1 m to several meters across · one female die in concrete, ductile iron, or steel at $10,000–100,000, instead of a matched set · one shot per cycle, so hours between parts · economic lot size single digits to low tens.
ExamplesLarge aerospace panels, vessel heads, ship structures.
VideosExplosive Forming Techniques for Saturn V Components (NASA NTRS) · Explosive Forming for Fastening and Joining Components (NASA NTRS)
A pulsed magnetic field induces forces in conductive metal and rapidly deforms it.
Strengths & weaknessesIt's clean, fast, and contactless. It mainly works on conductive materials, and it requires high-voltage equipment.
When to usePick electromagnetic forming for high-conductivity alloys (aluminum and copper above roughly 1–2 mm skin depth equivalents) in fast, repeatable crimping, flanging, and joining operations at medium to high volume. It's especially useful for dissimilar-material joints such as aluminum to steel or metal to composite, where you can't weld. Cycle times run a second or less, and because no tool touches the part it works well on cosmetic surfaces. Avoid it for steels and other poor conductors unless you add a driver sheet, for large-area panel forming, and in shops that aren't willing to maintain pulsed high-voltage capacitor banks. The fallback is mechanical crimping or conventional press forming.
Key numbersCapacitor bank 5–100 kJ at 3–25 kV · discharge in 10–100 µs · workpiece velocity up to about 300 m/s · sheet and tube wall typically 0.5–3 mm in aluminum or copper · cycle time under 1 second.
ExamplesTube crimping, aluminum forming, connector joining.
VideosElectromagnetic Forming — A Review, Psyk et al. (Univ. of New Hampshire Scholars Repository)
Specialized alloys are heated into a superplastic regime and slowly formed into a die.
Strengths & weaknessesIt can produce complex one-piece lightweight structures. Cycles are slow, the list of usable alloys is short, and temperature control is demanding.
When to usePick superplastic forming for deep, complex thin-wall shells in fine-grain titanium or aluminum at aerospace volumes (tens to low thousands per year). It's especially useful combined with diffusion bonding, which turns a multi-part stiffened structure into one piece. Expect cycle times of 20 minutes to several hours, and plan for thinning gradients in deep sections. Avoid it for cost-driven or high-volume work and for alloys that don't come in superplastic grades. Conventional stamping or hydroforming covers those, and hot stamping handles high-strength steels.
Key numbersGrain size below about 10 µm · forming at roughly 900 °C for Ti-6Al-4V and 450–520 °C for aluminum · strain rates of 10⁻⁴ to 10⁻² per second, giving elongations of 200–1,000% · argon forming pressure 0.3–3 MPa · cycle time 20 minutes to several hours · wall thinning commonly 30–50% in the deepest sections.
ExamplesTitanium aerospace panels, ducts, lightweight shells.
VideosSuperplastic Forming 101 (Macrodyne) · A Review on Superplastic Forming of Ti-6Al-4V Alloy (arXiv)
Conventional Machining
cut parts to size with rotating or linear tools11 processesA rotating workpiece is cut by a stationary or traversing tool.
Strengths & weaknessesIt's accurate and efficient for rotational geometry. It wastes material, and it's inefficient for features that aren't axisymmetric.
When to usePick turning as the default for anything rotationally symmetric (shafts, bushings, fittings) from one-off prototypes through mid volumes. Tolerances to ±0.01 mm are routine, and live tooling covers incidental flats and cross-holes. Swiss-type and multi-spindle lathes extend it economically to high-volume small parts. Avoid it when the part is fundamentally prismatic, and mill it instead. If volumes get high enough that a net-shape process like cold heading or powder metallurgy is cheaper per part, turning stops being competitive because of the chips and the cycle time.
Key numbersTolerances ±0.01 mm routine and ±0.005 mm with a good machine and a finishing pass · surface finish 0.4–3.2 µm Ra · no dedicated tooling, so lot size 1 is economic · setup 15 minutes to a few hours · cycle time seconds to minutes per part · US shop rates roughly $60–120 per hour for CNC turning.
ExamplesShafts, bushings, fittings, pistons, rollers.
VideosGeneral Turning Knowledge (Sandvik Coromant) · CNC Turning Design Guidelines (Protolabs)
A rotating multi-edge cutter removes material while the workpiece and tool move along controlled axes.
Strengths & weaknessesIt's very flexible and can produce complex three-dimensional geometry. It wastes material, it needs tool access to every feature, and it gets expensive for high-volume parts or deep internal features.
When to usePick milling for prismatic and freeform parts from prototypes through a few thousand units, whenever the geometry is still changing or volumes can't justify tooling. It holds ±0.01–0.05 mm with no fixed tooling beyond workholding. It's also the standard finishing step after castings, forgings, and metal additive parts. Avoid it as the primary process at high volumes, where casting, molding, or stamping plus finish machining is cheaper. For deep narrow cavities or sharp internal corners in hardened material, use EDM instead.
Key numbersTolerances ±0.01–0.05 mm · surface finish 0.8–3.2 µm Ra, down to 0.4 µm with a finishing pass · no dedicated tooling beyond workholding, so lot size 1 is economic · material removal in aluminum from about 50 cm³/min on a general machining center to several hundred on a high-speed spindle · setup 30 minutes to several hours · US shop rates roughly $60–120 per hour for 3-axis and $100–200 for 5-axis.
ExamplesBrackets, molds, aerospace structures, housings, engine parts.
VideosMilling Knowledge (Sandvik Coromant) · How to Design Parts for CNC Machining (Hubs)
A rotating drill creates a round hole axially.
Strengths & weaknessesIt's simple and available everywhere. Deep, small, high-aspect-ratio, or very precise holes need specialized tooling or a different process.
When to useDrilling is the default for round holes in any machinable material at any volume, and it's worth questioning only in unusual cases. A good rule of thumb is to treat a drilled hole as roughly ±0.1 mm on position and IT11-12 on diameter. If size, finish, or location matter more than that, plan a reaming or boring pass after it. Beyond about 10:1 depth-to-diameter, move to gun drilling or fast-hole EDM. For very small, hard, or oddly angled holes, use EDM or laser drilling.
Key numbersPosition roughly ±0.1 mm and diameter IT11-12, about 0.09–0.15 mm of spread on a 10 mm hole · hole finish 1.6–6.3 µm Ra · standard twist drills from 0.5 mm to 50 mm · depth-to-diameter up to about 10:1 conventionally and past 100:1 with gun drilling · a few seconds per hole · drills cost a few dollars to a few tens of dollars.
ExamplesFastener holes, fluid ports, oil passages, mounting holes.
VideosAn existing hole is enlarged and trued using a single-point tool.
Strengths & weaknessesImproves diameter, alignment, straightness, and concentricity. Cannot initiate a hole and can be vibration-prone in deep bores.
When to usePick boring when an existing hole needs its location corrected or its diameter brought to IT6-7 accuracy: bearing seats, cylinder bores, and any diameter too large or too non-standard for a reamer. A single-point boring bar cuts a hole concentric to the spindle axis, so unlike reaming it can fix a drilled hole that wandered. Avoid it for long slender bores where bar overhang exceeds roughly 4–6 diameters, unless you have damped bars. If a drilled-and-reamed hole already meets spec, that is usually the cheaper route. If you need a finer surface than boring leaves, honing is the next step.
Key numbersDiameter accuracy IT6-7, roughly 0.015–0.025 mm on a 50 mm bore · roundness and concentricity within a few micrometers · surface finish 0.8–3.2 µm Ra · finish passes remove 0.1–0.5 mm on diameter · bar overhang up to 4–6 bore diameters, more with damped bars · bore diameters from about 6 mm to over 1 m on a horizontal boring mill.
ExamplesEngine cylinders, bearing seats, valve bodies, machine housings.
VideosA multi-edge tool removes a small amount from an existing hole to improve size and finish.
Strengths & weaknessesProduces accurate smooth holes efficiently. Cannot correct major location or straightness errors.
When to usePick reaming to bring drilled holes to H7-class fits with a fine finish in seconds: dowel holes, bushing bores, and any standard diameter you make repeatedly. A reamer costs little and runs at production rates. Leave the right stock (typically 0.1–0.4 mm on diameter) and it will hold size all day. Avoid it when position or straightness is wrong, because a reamer follows the existing hole. If the location needs correcting, bore first. For non-standard or very large diameters, single-point boring is the better tool.
Key numbersHolds H7, about 0.015 mm of tolerance on a 10 mm hole · stock allowance 0.1–0.4 mm on diameter · surface finish 0.4–1.6 µm Ra · common diameters 2–50 mm · a few seconds per hole · reamers cost tens to a few hundred dollars.
ExamplesDowel holes, precision bushings, valve guides, bearing bores.
VideosA toothed tool with progressively larger cutting edges passes through or across the workpiece.
Strengths & weaknessesProduces repeated profiles extremely quickly and accurately. Broaches are expensive and geometry is constrained by straight-line tool travel.
When to usePick broaching for internal keyways, splines, and polygonal holes at volumes of thousands per year and up. A single stroke of a few seconds finishes the full profile, so the multi-thousand-dollar broach amortizes quickly. It is the standard answer for production splines and firtree slots. Avoid it for blind holes and for profiles the tool cannot pass straight through. If you only need a few parts, wire EDM, gear shaping, or slotting will do the job without dedicated tooling.
Key numbersTolerances ±0.01–0.025 mm · surface finish 0.8–1.6 µm Ra · rise per tooth 0.02–0.1 mm · a few seconds per stroke, finishing the full profile in one pass · broaches cost roughly $2,000–20,000 with 6–16 week lead times · economic from a few thousand parts a year upward.
ExamplesKeyways, splines, turbine-disc slots, polygonal holes.
VideosBroaching Fundamentals (Gear Technology) · Broaching Fundamentals Course (General Broach)
A toothed blade cuts stock into pieces or rough shapes.
Strengths & weaknessesVersatile and inexpensive for material preparation. Accuracy and finish are generally insufficient for final precision surfaces.
When to usePick sawing for cutoff and stock preparation (billets, bar, tube, structural sections, lumber) at any volume. Cost per cut is lower than any other separation method and kerf loss is small. Treat sawn length as roughly ±0.5–1 mm, and leave facing stock when the end matters. Don't rely on it for finished surfaces or profiles. If you need accurate sheet and plate contours, use laser, waterjet, or plasma. If you need precise faces, follow the saw with machining.
Key numbersLength tolerance roughly ±0.5–1 mm · kerf 1–3 mm depending on blade type · cut rate roughly 10–100 cm² per minute in steel · sawn face 3.2–12.5 µm Ra · band saw blades cost $30–300 · capacity from small bar up to sections around 1 m on large machines.
ExamplesBillets, tube, structural steel, lumber, plastic stock.
VideosGuide to Band Sawing (LENOX) · Bandsaw Blade Guide: TPI, Bi-Metal vs Carbide and Selection (AIMS Industrial)
A single-point tool moves linearly relative to the workpiece.
Strengths & weaknessesCan produce large flat surfaces with simple tooling. Slow, and has largely been displaced by milling and grinding.
When to useReach for planing or shaping only in niche situations: machine beds and guideways too large for the milling envelopes you have, internal keyways or corners a rotating cutter cannot reach, or a legacy shop where a cheap single-point tool beats buying inserts. The single-point cut can also leave a good finish on long ways. For essentially all new work, milling is faster and more flexible, and grinding holds tighter flatness. Use those by default and treat planing and shaping as the exception.
Key numbersFlatness roughly 0.02–0.05 mm per meter · surface finish 1.6–6.3 µm Ra · cutting speeds 5–30 m/min, with the return stroke cutting nothing · shaper strokes up to about 900 mm, planer tables from 1 m to over 10 m · tooling is a single HSS or carbide bit costing tens of dollars · one part per setup, so it fits prototype and low volume only.
ExamplesMachine beds, guideways, large plates, keyways.
VideosCutting Tool Applications, Chapter 7: Shaping & Planing (American Machinist) · Construction, working principle and applications of shaping, planing and slotting machines (NPTEL, IIT Kharagpur)
Gear teeth are generated by hobbing, shaping, or skiving, depending on geometry and production rate.
Strengths & weaknessesMature, accurate, and productive for external and internal gears. Each variant trades flexibility against speed and machine rigidity requirements.
VariantsA rotating hob progressively generates external gear teeth. It is highly productive, but it works poorly on internal gears or teeth near shoulders.
A reciprocating gear-shaped cutter produces internal or external teeth. It is more flexible than hobbing, but usually slower.
A continuously rotating cutter and workpiece interact at crossed axes to generate internal or external gears rapidly. Productivity is high, but it requires rigid, accurate machines and careful process control.
Pick hobbing as the default for external spur and helical gears at production volumes. Use shaping where a hob cannot run (internal gears, teeth close to a shoulder), and skiving when you need internal gears at scale and have rigid modern machines. Cut gears typically reach AGMA 8-10 as-machined, so if the drawing calls for better, plan on gear grinding or honing after hardening. Avoid dedicated gear cutting for low-load or low-precision gears at very high volume, where powder metallurgy, cold forming, or plastic molding produces the tooth form without cutting. For one-off prototypes, five-axis milling of the flanks is now practical.
Key numbersAs-cut quality AGMA 8-10, so anything better needs grinding or honing · module typically 0.5–25 mm · tooth flank finish 0.8–3.2 µm Ra as cut · hobbing cycle roughly 1–10 minutes per gear · hobs cost $500–5,000 · economic from a few hundred gears a year upward.
ExamplesTransmission gears, ring gears, splines, sprockets.
VideosThreads are created using taps, dies, single-point tools, thread mills, or grinding wheels.
Strengths & weaknessesFlexible and suitable for custom or low-volume threads. Removes material and generally produces weaker fatigue behavior than rolled threads.
When to usePick thread machining for internal threads, custom or large-diameter forms, hard or pre-hardened materials, threads next to shoulders, and any low-volume or prototype work. A tap, single-point tool, or thread mill needs no dedicated tooling and handles odd pitches freely. Thread milling is the safe choice in expensive parts, since a broken tap in a finished component often means scrap. Avoid it for mass-produced external fasteners and fatigue-critical studs. There, thread rolling is faster, wastes no material, and leaves compressive residual stress that machined threads lack.
Key numbersCut threads typically hold class 2A/2B or metric 6H/6g, with 3A/3B possible at extra cost · thread grinding reaches pitch-diameter tolerances near 0.005 mm · flank finish usually 1.6–3.2 µm Ra · taps cost $10–100 and thread mills $50–300 · a tapped hole takes seconds and a milled thread 10–60 seconds · diameters from M1 to over 1 m with single-point tooling.
ExamplesLead screws, pipe fittings, custom fasteners, threaded shafts.
VideosThreading Knowledge (Sandvik Coromant) · Threading Considerations for CNC Machining (Protolabs)
Hardened dies plastically deform a blank to form threads without cutting.
Strengths & weaknessesFast, chipless, and creates strong fatigue-resistant threads. Requires ductile material, accurate blanks, and dedicated dies.
When to usePick thread rolling for external threads on ductile metals (typically below about 40 HRC) at high volumes. It runs at parts per minute, makes no chips, and gives fatigue life several times that of cut threads. On critical bolts, roll after heat treatment so the compressive layer survives. Blank diameter must be held to roughly ±0.02–0.05 mm, because the material is displaced rather than removed. Avoid it for internal threads, brittle or fully hardened materials, thin-walled parts that collapse under die force, and short runs that cannot pay for dies. In those cases, use tapping or single-point threading.
Key numbersWorks on ductile material below roughly 40 HRC · blank diameter held to ±0.02–0.05 mm · production rates from tens to a few hundred parts per minute · fatigue life typically 2–5× that of cut threads · die sets cost $500–3,000 and run 100,000 parts and up between regrinds · no chips, so material use drops roughly 15–20% against cutting.
ExamplesBolts, studs, screws, threaded rods.
VideosFundamentals of Thread Rolling (CJWinter) · Precision Thread Rolling: How It Works (Horst Engineering)
Abrasive & Precision Machining
grind and polish to fine finish and tight tolerance9 processesAbrasive grains remove small chips to produce accurate dimensions and fine surfaces. Variants include surface, cylindrical, internal, centerless, creep-feed, jig, gear, and profile grinding.
Strengths & weaknessesWorks on hardened materials and achieves excellent tolerances and finishes. Heat, wheel wear, burn, residual stress, and low removal rates must be controlled.
When to usePick grinding whenever the part is hardened past practical cutting-tool range (roughly 45–50 HRC and up), or the drawing demands tolerances tighter than about ±0.005 mm or finishes below 0.4 micrometers Ra. Typical parts are bearing surfaces, dies, gauges, and cutting tools. Centerless grinding turns it into a genuine high-volume process for small cylindrical parts. Don't use it for bulk stock removal. The usual sequence is to machine soft, heat treat, then grind the last tenths of a millimeter (creep-feed grinding of superalloy slots is the notable exception). If the geometry is simple, hard turning with CBN now covers some of the same work.
Key numbersTolerances ±0.002–0.005 mm · surface finish 0.1–0.4 µm Ra, down to 0.05 µm with fine wheels · works above 45–50 HRC, where cutting tools stop · leave 0.05–0.5 mm of stock for the grind · wheel speeds 30–60 m/s, and 100 m/s and up with CBN · centerless machines run hundreds of small parts per hour.
ExamplesBearing races, shafts, cutting tools, dies, engine components.
VideosAbrasive Processes: Grinding Courseware Module (NPTEL / IIT Kharagpur)
Abrasive stones move in a controlled rotating and reciprocating pattern against a bore.
Strengths & weaknessesCorrects bore geometry and creates a functional crosshatch surface. Removal rates are low and a near-size hole is required.
When to usePick honing as the final operation on bores that seal or slide and must be round, straight, and sized to a few micrometers: engine cylinders, hydraulic cylinders, valve and bearing bores. It is especially useful where the crosshatch pattern has to retain oil. It expects a bored or reamed hole within roughly 0.02–0.1 mm of final size, and it fixes geometry that reaming cannot. Skip it when a reamed or bored finish already meets spec. For external surfaces, grind or superfinish instead. Honing follows the existing bore position rather than moving it, so it won't correct location.
Key numbersBore roundness and straightness within 1–5 µm · surface finish 0.1–0.8 µm Ra · crosshatch angle usually 30–60 degrees included · removes 0.02–0.1 mm on diameter · cycle time 20 seconds to a few minutes per bore · bore diameters from a few millimeters to over 1 m.
ExamplesEngine cylinders, hydraulic cylinders, valve bores, barrels.
VideosLoose abrasives between a lap and workpiece remove microscopic material.
Strengths & weaknessesAchieves exceptional flatness, contact, and surface finish. Slow and unsuitable for substantial stock removal.
When to usePick lapping when flatness or fit has to reach the light-band level: mechanical seal faces, gauge blocks, valve seats, optics, and wafers. It reaches finishes down to 0.01–0.05 micrometers Ra, which grinding cannot deliver. It removes only micrometers to a few hundredths of a millimeter, so parts have to arrive already ground to size. It is slow and messy, so skip it wherever ground tolerances suffice. For bores use honing, and for contoured surfaces use polishing.
Key numbersFlatness to one helium light band, about 0.3 µm, and below 0.1 µm on optical work · surface finish 0.01–0.05 µm Ra · abrasive grit 3–30 µm · removal rate roughly 1–10 µm per minute · total stock removed from a few micrometers to a few hundredths of a millimeter · cycle time minutes to about an hour per load.
ExamplesMechanical seals, gauge blocks, optical surfaces, wafers.
VideosFine abrasives smooth a surface using belts, wheels, pads, compounds, or robots.
Strengths & weaknessesImproves appearance, friction, cleanliness, and coating performance. Can be labor-intensive and may round edges or change dimensions.
When to usePick polishing and buffing when you need cosmetic or functional surface quality rather than dimensional control: mirror trim, cookware, mold cavities that must release cleanly, and surfaces being prepared for plating. Budget real labor hours unless volume justifies robotic cells, or the parts are small enough for mass finishing. Avoid it on tight-tolerance features and crisp edges, because it rounds them. If you need to finish internal passages or large batches of stainless parts, electropolishing does the same job chemically. If you just need bulk deburring, vibratory finishing is much cheaper.
Key numbersMirror finishes of SPI A-1 to A-3, roughly 0.012–0.1 µm Ra · abrasive progression from about 120 grit to 3,000 grit, then compounds · removes only a few micrometers of material · hand work runs minutes to tens of minutes per part · hand polishing stays cheaper below roughly a few thousand parts a year, robotic cells above that.
ExamplesMedical tools, cookware, molds, automotive trim.
VideosThe workpiece acts as the anode in an electrolytic bath, preferentially dissolving microscopic peaks.
Strengths & weaknessesImproves cleanliness, corrosion resistance, fatigue behavior, and finish, including on difficult internal surfaces. Requires conductive materials, chemical handling, and careful edge control.
When to usePick electropolishing for stainless and nickel-alloy parts in hygienic or high-purity service (pharmaceutical vessels, medical implants, semiconductor gas lines), and wherever internal passages, batches of small parts, or delicate features rule out mechanical polishing. The same bath also deburrs and improves corrosion resistance. Design for the 5–50 micrometers it removes, and expect it to dissolve sharp edges faster than flat areas. Avoid it on parts with tight dimensional stack-ups or mixed alloys in one assembly. It smooths microscopic peaks rather than deep scratches, so a gouged surface just comes out shinier. If you need a specific directional finish or heavy stock removal, use mechanical polishing.
Key numbersRemoves 5–50 µm of material · typically halves the starting Ra, so a 0.8 µm surface comes out near 0.4 µm · current density 5–100 A/dm² · phosphoric and sulfuric acid bath at roughly 50–75 °C · cycle time 2–20 minutes.
ExamplesSemiconductor gas lines, stainless medical devices, pharmaceutical equipment, additive parts.
VideosWhat Is Electropolishing and How Does It Work? (Able Electropolishing) · Electropolishing User's Guide (Delstar Metal Finishing)
Fine abrasive stones oscillate against a rotating surface to remove surface peaks.
Strengths & weaknessesReduces friction, noise, and wear and improves fatigue life. Cannot correct major geometric errors.
When to useUse superfinishing as the last operation on ground rotating contact surfaces (bearing races, cams, crank and cam journals, gear flanks) when friction, noise, or contact fatigue drives the spec and the finish has to drop below roughly 0.1 micrometers Ra. It removes only the peaks left by grinding, a few micrometers at most, and it does that in seconds per part at production rates. Don't use it as a corrective step: size, roundness, and position have to already be right from grinding. If the surface is decorative, or a standard ground finish is good enough, skip it.
Key numbersSurface finish 0.01–0.1 µm Ra · removes 2–10 µm, just the peaks left by grinding · stones at 400 grit and finer · oscillation 200–3,000 strokes per minute at 1–5 mm amplitude · cycle time 10–60 seconds per part · economic from thousands of parts a year upward.
ExamplesCrankshafts, camshafts, gears, bearing races.
VideosViscous abrasive media is forced through or across complex geometry.
Strengths & weaknessesReaches internal passages inaccessible to conventional tools. Material removal can be difficult to localize.
When to useUse abrasive-flow machining to deburr, radius, and polish internal geometry that no tool can reach: cross-drilled intersections, turbine cooling passages, extrusion dies, fuel injector interiors, and the rough as-built channels of metal additive parts. It's a batch process, so it works well on families of similar parts once the fixtures and media are dialed in. Avoid it if removal has to be confined to one spot or a tight tolerance has to be held, because the media polishes everything it flows past. If the surface is external and accessible, ordinary polishing or vibratory finishing costs far less.
Key numbersMedia pressure roughly 7–20 MPa · abrasive from about 8 to 500 grit · removes 0.01–0.1 mm per surface · cuts Ra by 50–90%, for example 3 µm down to about 0.4 µm · cycle time 1–10 minutes per load · fixtures cost roughly $1,000–10,000 per part family.
ExamplesTurbine passages, extrusion dies, fuel components, manifolds.
VideosAbrasive Flow Machining Overview (Extrude Hone) · Abrasive Flow Machining Systems (Winbro Flow Grinding)
Parts are processed in rotating or vibrating media to deburr, clean, radius, or polish them.
Strengths & weaknessesEconomical for batches of small parts. Offers limited control over specific surfaces and can damage delicate features.
When to useUse vibratory finishing or tumbling for bulk deburring, edge radiusing, cleaning, and pre-plate smoothing of batches of small robust parts (stampings, machined fittings, castings, fasteners). Cost per piece drops to cents because no operator touches individual parts, and the cycle time, measured in hours, matters less than that near-zero labor. Avoid it for delicate thin walls, for threads and sharp edges that have to survive, and for parts that nest or dent each other. If one specific surface needs selective treatment, use hand deburring, brush deburring, or abrasive-flow machining instead.
Key numbersCycle times from about 15 minutes for deburring to 24 hours for a polish · edge radii of 0.05–0.5 mm · removes 0.01–0.05 mm from surfaces · bowls and tubs of 30–1,000 L, holding hundreds to thousands of parts per load · media vibrates at 900–3,000 cycles per minute with 2–6 mm amplitude · cost per part in cents once the batch is loaded.
ExamplesFasteners, jewelry, cast fittings, stamped brackets.
VideosParticles are propelled against a surface to clean, texture, or prepare it.
Strengths & weaknessesRapidly removes rust, scale, coatings, and contamination. May alter dimensions, embed media, or damage delicate surfaces.
When to useUse abrasive blasting to strip scale, rust, and old coatings, and to build the anchor profile that paint, powder coat, and thermal spray need. It's the standard prep on castings, weldments, and structural steel, and you pick the media to suit the job (steel grit for profile, glass bead for cosmetic matte, plastic or soda for delicate substrates). It's also the cheap way to get a uniform non-directional matte finish. Avoid it on precision surfaces and on thin panels, which peen and warp, and on parts where embedded media would contaminate later processing. Mask anything that has to stay sharp or stay on size. If you want controlled compressive strengthening, use shot peening instead, which is the same physics run to a specification.
Key numbersBlast pressure 0.3–0.7 MPa (about 40–100 psi) · anchor profile 25–100 µm, set to match the coating spec · cleanliness graded against SSPC SP 6 commercial or SP 10 near-white · hand blasting covers roughly 10–50 m² per hour · media from 60 mesh glass bead to 16 mesh steel grit.
ExamplesCastings, structural steel, automotive bodies, refurbished machinery.
VideosNontraditional Material Removal
cut with jets, sparks, beams, and chemistry11 processesHigh-pressure water, often carrying abrasive particles, cuts material without meaningful heat input.
Strengths & weaknessesCuts nearly any material and avoids heat-affected zones. Slower than laser on thin sheet, produces wet abrasive waste, and can create taper.
When to useUse waterjet when the part can't take heat or the material can't be cut thermally: titanium and hardened plate up to 150–200 mm, composites and laminates that delaminate under a laser, glass, stone, and pre-hardened parts that you need to cut without tempering the edge. Expect roughly ±0.1–0.25 mm, a 1 mm kerf, and some edge taper unless the machine compensates for it. For thin mild steel and stainless at volume, laser cutting is several times faster and cheaper. For fine features and tight tolerances in conductive metals, use wire EDM instead.
Key numbersPump pressure 400–600 MPa (60,000–90,000 psi) · orifice 0.25–0.4 mm, giving a kerf near 1 mm · tolerance ±0.1–0.25 mm with some edge taper · cuts up to 150–200 mm thick in most materials · garnet consumption 0.2–0.7 kg per minute · cut edge roughly 3.2–6.3 µm Ra.
ExamplesTitanium plate, composites, stone, armor, glass, gaskets.
VideosLaser and Waterjet Cutting Design Guide (Xometry) · Waterjet Cutting Tolerances Explained (Techni Waterjet)
Fine abrasive particles carried by high-speed gas erode material.
Strengths & weaknessesCreates delicate features without substantial heat or force. Removal rates are low and nozzle wear is significant.
When to useUse abrasive-jet machining for small, delicate work on brittle materials: etching and frosting glass, trimming fired ceramics, cutting fine slots and micro-holes, and deburring where cutting forces would crack the part. It suits low and medium volumes of small features rather than bulk shaping. Avoid it for substantial material removal or large cuts (waterjet handles those) and for ductile metals, which resist erosion. If you need deep accurate cavities in brittle stock, ultrasonic machining is the better choice. Plan on containing the dust and replacing nozzles regularly.
Key numbersNozzle bores 0.2–1 mm, cutting features down to about 0.1 mm · abrasive grit 10–50 µm aluminum oxide or silicon carbide · gas pressure 0.2–1 MPa, giving jet velocities of 150–300 m/s · removal rate around 15 mg per minute · tolerance roughly ±0.05–0.1 mm · nozzle life from tens of hours in carbide to a few hundred in sapphire.
ExamplesGlass etching, ceramic trimming, micro-holes, deburring.
VideosA shaped tool vibrates ultrasonically while abrasive slurry erodes a brittle workpiece.
Strengths & weaknessesMachines hard, brittle, nonconductive materials with low thermal stress. Slow, and causes tool wear.
When to useUse ultrasonic machining for shaped cavities, holes, and profiles in hard brittle nonconductive materials (glass, sapphire, quartz, fired technical ceramics) where diamond grinding can't reach the geometry and the material isn't conductive enough for EDM. It puts almost no heat or lateral force into the part, so fragile thin sections survive. For conductive hard metals and carbides, EDM is faster. Avoid it on soft ductile materials, which just absorb the abrasive, and on deep cavities, where tool wear degrades accuracy. If volumes justify it, rotary ultrasonic (diamond-tool) variants give higher removal rates.
Key numbersTool vibrates at 19–25 kHz with 10–50 µm amplitude · boron carbide or silicon carbide slurry at 15–150 µm grit · removal rate roughly 1–10 mm³ per minute in glass · tolerance ±0.025–0.05 mm · surface finish 0.2–0.8 µm Ra · practical cavity depth around 3 tool diameters before accuracy drops off.
ExamplesGlass, sapphire, ceramics, carbide dies.
VideosA focused laser melts, burns, or vaporizes material along a programmed path.
Strengths & weaknessesFast, precise, and highly automated for sheet material. Reflective metals, thick sections, capital cost, gas consumption, and heat-affected zones can be limiting.
When to useLaser cutting is the default for 2D profiles in sheet metal from roughly 0.5 to 20 mm. It needs no tooling, holds ±0.1 mm, and stays economical from a single prototype through high-volume nested production, which is why job shops quote it first. Fiber lasers handle reflective aluminum, brass, and copper reasonably well now. Above about 25 mm of steel, laser speed and edge quality drop off, so use plasma or oxy-fuel. For composites, heat-sensitive alloys, and glass, use waterjet. If the part is stable and runs more than tens of thousands of pieces per year, switch to stamping, because die amortization beats per-meter cutting cost at that point.
Key numbersFiber lasers of 1–20 kW, with 3–6 kW covering most sheet work · kerf 0.1–0.5 mm · tolerance about ±0.1 mm · thickness 0.5–20 mm in steel, falling off past 25 mm · above 20 m/min in 1 mm mild steel and 1–2 m/min at 10 mm · heat-affected zone 0.05–0.5 mm · machines cost roughly $200,000 to over $1 million.
ExamplesEnclosures, brackets, signage, gaskets, sheet-metal components.
VideosGuide to Sheet Metal Laser Cutting (Xometry) · Laser Cutting: Processes and Parameters (TRUMPF)
A high-temperature plasma jet melts electrically conductive metal.
Strengths & weaknessesCuts thick metal rapidly and economically. Edge quality and precision are generally inferior to laser or waterjet.
When to useUse plasma for conductive plate in the roughly 6–50 mm range (structural steel, heavy-equipment blanks, ship plate), where it's the cheapest fast option and a ±0.5–1 mm edge is fine because the edge gets welded or machined anyway. High-definition plasma tightens that tolerance usefully. Capital cost is a fraction of a laser with the same thickness capacity, and handheld units cover field work. Avoid it for thin sheet and precision profiles, where laser is faster and cleaner, and for nonconductive materials. If the edge has to go into service as-cut with no heat-affected zone, use waterjet instead.
Key numbersTorch current 30–800 A, with 100–400 A covering most plate work · kerf 1.5–4 mm with a 1–5 degree bevel · tolerance ±0.5–1 mm, tightening to ±0.2–0.4 mm on high-definition systems · thickness 6–50 mm at production speeds · cut speed 1–6 m/min in 10 mm steel · electrodes and nozzles last a few hours of arc-on time.
ExamplesHeavy-equipment parts, structural plate, ship components.
VideosWhat Is a Plasma Cutter? (Hypertherm) · Plasma Arc Cutting: Process and Equipment (TWI)
Fuel gas preheats carbon steel and oxygen rapidly oxidizes and removes the metal.
Strengths & weaknessesPortable, inexpensive, and can cut extremely thick steel. Slow, thermally aggressive, and unsuitable for many nonferrous alloys.
When to useUse oxy-fuel for carbon and low-alloy steel from about 25 mm up to 300 mm and beyond (thicknesses plasma can't reach), and for field work, demolition, and shops where a torch set costing a few hundred dollars is the entire capital outlay. Multiple-torch machines still bevel and strip heavy plate economically. It only works where the oxidation chemistry works, so stainless, aluminum, and copper alloys are out. Below roughly 25 mm, plasma cuts faster with less distortion, and on thin sheet laser is better still.
Key numbersCarbon and low-alloy steel only, and above roughly 0.3% carbon the plate needs preheat · cuts from a few millimeters to 300 mm, and past 900 mm with special torches · economical against plasma above about 25 mm · cut speed 300–700 mm/min in 25 mm plate · kerf 1.5–5 mm with a 1–5 mm heat-affected zone · tolerance roughly ±1–3 mm · a complete torch set costs a few hundred dollars.
ExamplesStructural fabrication, demolition, shipbuilding, heavy plate.
VideosControlled sparks erode conductive material through a dielectric gap.
Strengths & weaknessesMachines hard materials, delicate features, and sharp internal geometry without cutting forces. Only works on conductive materials and is relatively slow.
VariantsUses a continuously fed wire for through-cut profiles.
Uses a shaped electrode to create blind cavities.
Produces small, deep cooling or fluid holes.
Creates precise starter and high-aspect-ratio holes.
Use EDM when hardness, geometry, or fragility rules out cutting tools: hardened tool steel and carbide after heat treatment, near-sharp internal corners, thin fragile ribs, and deep narrow cavities. Wire EDM holds ±0.005 mm or better on through profiles and is the standard for punches, dies, and splines in hardened stock. Sinker EDM is the usual choice for blind mold cavities, and fast-hole EDM produces turbine cooling holes at high aspect ratio. Avoid it whenever the material is nonconductive or still soft enough to mill. Removal rates are low, so a good rule of thumb is to machine the part soft and use EDM only for the features that heat treatment or geometry force you into. For high-volume shaped cavities in hard alloys, consider electrochemical machining instead.
Key numbersWire EDM holds ±0.005 mm or better · wire 0.1–0.33 mm, so the kerf runs 0.3–0.4 mm and internal corners bottom out near 0.15 mm radius · surface finish 0.1–3.2 µm Ra depending on skim passes · recast layer 2–25 µm, thinner on finishing settings · sinker removal 5–400 mm³/min, well below milling · fast-hole EDM drills 0.3–3 mm holes past 100:1 aspect ratio.
ExamplesMolds, dies, turbine cooling holes, punches, medical tools.
VideosMaterial is dissolved anodically using electrolyte while a shaped electrode approaches without touching.
Strengths & weaknessesRemoves hard conductive materials rapidly without burrs, tool wear, or thermal damage. Electrolyte handling, localization, capital cost, and environmental controls are difficult.
When to useUse ECM for repeating shaped features in hard conductive alloys at real volume (blisks, turbine airfoils, fuel-system components). Removal rate doesn't depend on hardness, the electrode never wears, and every part comes off burr-free with no recast layer or residual stress, which matters for fatigue-critical hardware. The catch is up-front cost: electrode development, electrolyte management, and machine capital only pay back across thousands of identical parts. For prototypes, one-offs, or geometry that changes often, EDM or milling gets you there without the development cycle. It doesn't work at all on nonconductive materials.
Key numbersCurrent density 20–300 A/cm² at 5–25 V · working gap 0.05–0.5 mm · feed rate 0.5–5 mm/min, roughly 1.5–2 cm³ per minute per 1,000 A in steel · surface finish 0.1–0.8 µm Ra with no recast layer · tolerance ±0.02–0.13 mm, tighter with pulsed ECM · electrolyte 10–20% sodium chloride or nitrate pumped through the gap at 10–60 m/s.
ExamplesTurbine blades, blisks, fuel-system parts, aerospace passages.
VideosMasked areas are protected while exposed material is chemically dissolved.
Strengths & weaknessesRemoves material uniformly from broad, thin surfaces without mechanical stress. Undercutting, chemical handling, environmental burden, and modest precision limit use.
When to useUse chemical milling to thin large, already-formed skins and shells, such as pocketing aircraft panels for weight after stretch forming. That's the case where no cutter could fixture the floppy contoured sheet without distorting it. Depth control is modest (roughly ±0.05–0.1 mm), and the etchant undercuts the mask about one-to-one with depth, so keep pockets shallow and generous. If the part is rigid enough to hold in a machine, use high-speed CNC pocketing, which has displaced most of this work and gives better tolerances with no chemical waste stream. For fine flat detail in thin sheet, photochemical machining is the right variant.
Key numbersEtch depth control roughly ±0.05–0.1 mm · undercut roughly 1:1 with etch depth · etch rates around 0.02–0.03 mm per minute on aluminum · pocket depths usually under 12 mm · surface finish Ra roughly 1–5 µm · maskant and scribe tooling in the low thousands of dollars, ready in days.
ExamplesAircraft skins, etched panels, lightweight aerospace structures.
VideosLaser Scribing and Chemical Milling of Aerospace Parts: Open Review (Micromachines via PMC)
Photoresist patterns are applied to thin metal sheet, and exposed regions are chemically etched away.
Strengths & weaknessesCreates intricate burr-free flat parts without hard tooling. Limited to thin sheet and subject to lateral undercutting.
When to usePick photochemical machining for intricate flat parts in metal sheet from roughly 0.01 to 1.5 mm: shims, screens, encoder discs, flexures, EMI shields. Phototooling costs hundreds of dollars and arrives in days, and parts come out burr-free and stress-free with no edge deformation, which suits springy or magnetic alloys that stamping would work-harden. It stays economical from prototype quantities into the hundreds of thousands. Avoid it above about 2 mm thickness, or where edge walls must be square (etching leaves a characteristic undercut profile). Very high volumes of simple shapes eventually justify a stamping die, and if you need thick precision flats, look at fine-blanking instead.
Key numbersSheet thickness 0.01–1.5 mm · tolerance typically ±10% of thickness, around ±0.02 mm at the thin end · minimum feature width about 1.0–1.2× the thickness · phototooling a few hundred dollars, ready in days · surface finish Ra roughly 0.5–2 µm and burr-free · economical from single prototypes to hundreds of thousands of parts.
ExamplesShims, filters, encoder discs, lead frames, flexures.
VideosChemical Etching: Process and Capabilities (Precision Micro) · Photo Chemical Machining Institute (PCMI)
Accelerated ions sputter atoms from a surface. Focused ion beam systems provide highly localized nanoscale removal.
Strengths & weaknessesProvides exceptional precision across many materials. Extremely slow, expensive, and limited to small areas.
When to usePick FIB when the feature is measured in nanometers to a few micrometers and nothing else can address it: semiconductor circuit edit and failure analysis, TEM lamella preparation, nanoscale prototyping, and trimming thin-film devices. Broad ion-beam milling suits gentle large-area material removal such as cross-section polishing. Removal rates are cubic micrometers per second, so treat this as laboratory and one-off work and never plan it as a production removal step. If the feature is visible to the naked eye, use laser micromachining, lithography-plus-etch, or conventional precision machining instead.
Key numbersFeature sizes from about 10 nm to a few micrometers · gallium beams at 5–30 keV and 1 pA to 65 nA · removal rates on the order of cubic micrometers per second · milled areas typically under 100 µm across · dual-beam FIB-SEM systems roughly $500k to $2m.
ExamplesSemiconductor repair, microscopy sample preparation, thin-film devices, nanoscale prototypes.
VideosIntroduction to the Focused Ion Beam System, Nan Yao (Cambridge University Press excerpt) · How FIB Instruments Transformed Sample Preparation (AZoM)
Additive Manufacturing
build parts layer by layer from digital models11 processesMelted thermoplastic or paste is deposited layer by layer through a nozzle. Fused-filament fabrication is the most common form.
Strengths & weaknessesEquipment and feedstock are inexpensive and ideal for prototypes, fixtures, and custom parts. Surface finish, anisotropy, layer adhesion, accuracy, and production speed are weaker than molding.
When to usePick FFF for form-and-fit prototypes, jigs, fixtures, and one-off functional parts. A few hundred dollars of machine and commodity filament turn CAD into a part overnight, which makes it the cheapest route to a physical object below roughly a few hundred units. Design around the weaknesses: ±0.2 mm typical accuracy, visible layers, and Z-direction strength a fraction of in-plane. If you need fine detail and smooth cosmetic surfaces, use vat photopolymerization; for isotropic production-grade nylon parts, use powder-bed fusion. Past roughly 500–1,000 units, injection molding's per-part economics take over.
Key numbersTolerance roughly ±0.2 mm or ±0.5% · layer heights 0.1–0.3 mm through a 0.4 mm nozzle · surface finish Ra roughly 10–25 µm from visible layer lines · Z-direction strength 30–70% of in-plane · commodity filament $20–50/kg · desktop machines from a few hundred dollars, with injection molding cheaper past roughly 500–1,000 units.
ExamplesJigs, fixtures, prototypes, housings, tooling aids.
VideosHow to Design Parts for FDM 3D Printing (Hubs) · Material Extrusion (Loughborough AMRG)
Light selectively cures liquid photopolymer resin. Major variants include stereolithography and digital light processing.
Strengths & weaknessesCreates excellent detail and smooth surfaces. Resins may be brittle, UV-sensitive, temperature-limited, or poor representations of production polymers.
When to usePick vat photopolymerization when detail and surface finish drive the job. It resolves features down to 25–100 micrometers with near-injection-molded smoothness, which suits dental models and aligner molds, jewelry and investment-casting patterns, hearing-aid shells, microfluidics, and master patterns for silicone molding. Desktop machines make it the best-looking prototype per dollar. Avoid it for load-bearing parts in service, because most resins are brittle, creep, and degrade under UV and heat. If you need functional end-use parts, use powder-bed fusion or machining; beyond low volumes, use injection molding.
Key numbersLayer thickness 25–100 µm · tolerance roughly ±0.1 mm or ±0.2% · surface finish Ra typically 0.5–2 µm · resin roughly $50–200 per liter for common grades · desktop machines a few hundred to a few thousand dollars, industrial SLA $50k and up · build envelopes about 150 mm on desktop machines and 500 mm or more on industrial ones.
ExamplesDental models, jewelry patterns, hearing-aid shells, microfluidic parts.
VideosIntroduction to SLA 3D Printing (Hubs) · Ultimate Guide to Stereolithography (Formlabs)
Print heads deposit droplets of resin or wax that are cured or solidified.
Strengths & weaknessesOffers high resolution, smooth surfaces, multiple materials, and full-color capability. Equipment and materials are expensive and long-term part durability may be limited.
When to usePick material jetting when a single build must combine materials, durometers, or full color: anatomical models with soft and rigid tissue, photorealistic marketing prototypes, overmold simulations. It also fits wax patterns for investment casting that need jetted precision. Resolution and finish rival vat photopolymerization, with easier soluble supports. Avoid it for functional parts, because jetted photopolymers age and creep worse than most resins. Machines and per-kilogram material prices are among the highest in additive, so it is a poor fit for cost-sensitive work. For single-material mechanical prototypes, vat photopolymerization or powder-bed fusion delivers more part per dollar.
Key numbersLayer thickness 14–32 µm · tolerance roughly ±0.1 mm or ±0.1% · surface finish Ra under 1 µm on glossy faces · resins roughly $300–600 per kilogram, among the highest in additive · machines from about $20k to over $300k · build envelopes up to roughly 500 × 400 × 200 mm.
ExamplesAnatomical models, realistic prototypes, casting patterns.
VideosIntroduction to Material Jetting 3D Printing (Hubs) · Material Jetting (Loughborough AMRG)
Liquid binder selectively joins powder particles layer by layer.
Strengths & weaknessesFast, requires no separate support structures in the powder bed, and can process metal, ceramic, or sand. Green parts are fragile and sintering causes shrinkage and distortion.
When to usePick binder jetting for printed sand molds and cores (its most mature use, delivering casting tooling in days with no pattern). It also fits batches of small complex metal parts in the hundreds to tens of thousands, where full-bed nesting without supports makes it the cheapest metal additive route per part. Design for sintering: roughly 15–20% linear shrinkage, distortion on unsupported spans, and final density a shade below wrought. Avoid it for fatigue-critical or highly loaded metal parts, where laser powder-bed fusion or MIM gives better properties, and for one-off metal prototypes where the sintering furnace cycle erases the speed advantage.
Key numbersLayer thickness 30–100 µm for metal and 200–400 µm for sand · sintering shrinkage 15–20% linear, held to about ±1–2% on final dimensions · tolerance roughly ±0.2 mm on sintered metal parts · sintered density 95–99% of wrought · sand build boxes up to about 4 × 2 × 1 m · economical from hundreds to tens of thousands of parts.
ExamplesSand molds and cores, metal components, ceramic parts.
VideosWhat Is Binder Jetting? (TWI) · Binder Jetting Process, Materials and Machines (Loughborough AMRG)
A laser fuses polymer powder in a bed, with unfused powder supporting the part.
Strengths & weaknessesProduces complex functional polymer parts without dedicated supports. Surface finish is grainy, machines are expensive, and powder management is substantial.
When to usePick polymer powder-bed fusion for functional end-use nylon parts (snap fits, ducts, brackets, orthotics) from prototypes into bridge production of hundreds to a few thousand units. Parts nest densely in the bed with no supports, and properties are far more isotropic than FFF. It is the standard way to ship real plastic parts before or instead of cutting a mold. If you need smooth cosmetic surfaces and fine crisp detail, vat photopolymerization does better; dyeing and vapor smoothing mitigate the grainy finish but don't erase it. Beyond roughly 5,000–10,000 units of a stable design, injection molding's unit cost is very hard to beat.
Key numbersLayer thickness 100–120 µm · tolerance roughly ±0.3 mm or ±0.3% · surface finish Ra 10–15 µm, matte and grainy · minimum wall about 0.7–1 mm · PA12 powder $50–100/kg, with 30–50% virgin refresh per build · machines $20k to $250k with build envelopes of 300–550 mm.
ExamplesDucts, orthotics, clips, housings, aerospace polymer parts.
VideosWhat Is SLS 3D Printing? A Design Guide (Protolabs Network) · Selective Laser Sintering: Complete Guide (Formlabs)
A laser or electron beam selectively melts metal powder layer by layer.
Strengths & weaknessesCreates internal channels, lattices, consolidated assemblies, and complex metal geometry impossible through ordinary machining. Slow, expensive, size-limited, sensitive to defects and residual stress, and usually requires heat treatment, support removal, inspection, and machining.
When to usePick metal powder-bed fusion when the geometry justifies it: conformal cooling channels, internal passages, lattices, and part consolidation in high-value low-volume hardware (rocket injectors, implants, heat exchangers, aerospace brackets). That typically means ones to hundreds of units where the design could not be machined or cast at all, or where lead time on castings and forgings is the bottleneck. Budget the full chain: stress relief, support removal, HIP or heat treatment, finish machining of interfaces, and inspection often cost more than the print. Avoid it for geometry a lathe or mill handles directly, for parts larger than roughly 400–800 mm build envelopes, and for volumes in the thousands, where casting plus machining or MIM is far cheaper.
Key numbersLayer thickness 20–60 µm · tolerance roughly ±0.1 mm or ±0.2% before finish machining · as-built surface finish Ra 8–20 µm · density above 99.5% of theoretical · build rates roughly 5–30 cm³ per hour per laser · Ti-6Al-4V powder $200–400/kg · machines $500k to $2m, with build envelopes commonly 250–400 mm and up to about 800 mm.
ExamplesRocket injectors, implants, heat exchangers, turbine components, aerospace brackets.
VideosWhat Is Metal 3D Printing and How Does It Work? (Protolabs Network) · Powder Bed Fusion: DMLS, EBM, SLM and SLS Compared (Loughborough AMRG)
Powder or wire is fed into a laser-, electron-beam-, plasma-, or arc-generated melt pool and deposited onto a substrate. Variants: laser powder DED, laser wire DED, electron-beam DED, plasma DED, and wire-arc additive manufacturing.
Strengths & weaknessesRepairs high-value parts, adds features, and builds large structures. Accuracy and finish are relatively poor, distortion can be substantial, and machining is usually required.
When to usePick DED to repair and remanufacture high-value metal parts (turbine blade tips, worn shafts, damaged molds) and to add bosses, flanges, or cladding to existing forgings and castings, since depositing onto an existing substrate is exactly what the process does. Wire-arc variants build meter-scale near-net structures at kilograms per hour, which is attractive when a forging has a year of lead time. Treat every build as near-net: plan finish machining of all functional surfaces and manage distortion. Avoid it for small intricate new parts with internal features, where powder-bed fusion holds far finer detail, and for anything a standard forging or casting can supply on acceptable lead time.
Key numbersDeposition rates 0.5–2 kg/hr for laser powder DED and 1–10 kg/hr for wire-arc · near-net tolerance roughly ±0.5–1 mm, with 1–3 mm of machining stock per surface · as-deposited Ra 20–50 µm · powder capture 40–90% for blown powder, near 100% for wire · working envelopes to several meters on gantry and robot cells.
ExamplesTurbine repair, large aerospace structures, mold repair, added bosses and flanges.
VideosWhat Is Directed Energy Deposition? (TWI) · Directed Energy Deposition Process and Applications (Loughborough AMRG)
Sheets are cut and bonded layer by layer.
Strengths & weaknessesCan combine dissimilar materials and embed sensors. Interlayer bond strength and removal of internal waste are limiting.
When to usePick sheet lamination for the jobs its layer-stacking nature uniquely enables: embedding sensors, wiring, or dissimilar-material layers mid-build, and cheap large-format visual models from paper or polymer sheet. Ultrasonic metal variants join aluminum and copper foils at low temperature around embedded electronics. Design so loads run in-plane, because the interlayer bond is the weak axis, and avoid enclosed cavities, which trap waste sheet you cannot dig out. For general functional prototyping or production metal parts, FFF, powder-bed fusion, or machining are all stronger defaults. Treat sheet lamination as a niche process you pick for embedding and laminating rather than as a general-purpose printer.
Key numbersLayer thickness set by the sheet: 0.1–0.2 mm for paper and about 150 µm for metal foil · ultrasonic bonding stays under roughly 150 °C, low enough to embed electronics and fiber · XY tolerance roughly ±0.1–0.3 mm · build envelopes up to about 1.8 × 1.8 × 0.9 m on large ultrasonic machines · ultrasonic metal machines several hundred thousand dollars.
ExamplesLaminated models, embedded-sensor structures, tooling.
VideosSheet Lamination: LOM and Ultrasonic AM (Loughborough AMRG) · What Is Ultrasonic Additive Manufacturing? (Fabrisonic)
Metal powder particles strike a surface at supersonic speed and bond through severe plastic deformation without bulk melting.
Strengths & weaknessesProvides low thermal damage and low oxidation. Geometry is line-of-sight constrained and final machining is often required.
When to usePick cold spray when material must go onto a part that cannot tolerate heat: dimensional restoration of worn or corroded aerospace and defense components, repairs on magnesium and aluminum housings where welding would distort or crack, and thick copper or oxidation-sensitive deposits laid down without melting. Deposition rates of kilograms per hour make large-area buildup practical, and machining to final dimension is assumed. Avoid it for geometry off the line of sight of the nozzle, for fine internal features, and for freestanding complex parts, where powder-bed fusion or DED is the better additive route. If you need simple hardfacing and heat is acceptable, conventional thermal spray or weld overlay is cheaper.
Key numbersParticle velocities 500–1,200 m/s · gas preheat 300–1,100 °C, below the powder's melting point · deposition rates roughly 1–10 kg/hr on high-pressure systems · deposit porosity typically under 1–2% · deposits from 0.1 mm to tens of millimeters thick · as-sprayed Ra roughly 10–30 µm.
ExamplesCorrosion repair, copper deposition, dimensional restoration, aerospace repair.
VideosCold Spraying: Process and Coating Deposition (TWI) · Cold Spray Additive Manufacturing Systems and Applications (Impact Innovations)
Cementitious material is robotically deposited layer by layer.
Strengths & weaknessesReduces formwork and enables unusual architecture. Reinforcement, building codes, interfaces, surface quality, and curing remain difficult.
When to usePick concrete printing where formwork dominates cost or geometry: curved and non-repeating walls that would need one-off molds, landscaping and marine structures, and demonstration housing in regions where skilled formwork labor is scarce. A wall system can go up in days with a small crew. Structural spans still need conventional reinforcement strategies, and printed surfaces show layer lines unless finished. Avoid it for standard rectilinear buildings, where crews with plywood forms or precast panels remain cheaper and code-approved, and for load paths a local jurisdiction will not certify. Today it usually complements conventional concrete construction rather than replacing it.
Key numbersBead widths 25–60 mm laid in 10–30 mm layers · nozzle travel roughly 50–250 mm/s · printable mortar compressive strength typically 30–60 MPa · a single-story wall shell in 24–48 hours of printing · gantry and boom printers from roughly $200k to over $1m.
ExamplesWalls, small buildings, landscaping structures, bridge elements.
Videos3D Construction Printing Technology and Printer Systems (COBOD) · Review of Material Processing Technology for 3D Concrete Printing (Materials, MDPI)
Cells, hydrogels, and biomaterials are deposited in controlled spatial arrangements.
Strengths & weaknessesEnables complex tissue-like structures. Vascularization, cell survival, maturation, scale, mechanical strength, and regulation remain major constraints.
When to usePick bioprinting for research applications where spatial arrangement of cells is the experiment: tissue models for drug screening and toxicity testing, patterned co-cultures, and engineered skin or cartilage constructs. In those cases a printed model can reduce animal testing or improve on flat cell culture. It is a laboratory tool rather than a manufacturing process. Avoid planning it for implantable or production tissue, because vascularization, maturation, and regulatory approval remain unsolved at scale. If you need actual implants or surgical models today, machined or printed biocompatible metals and polymers are the working answer.
Key numbersExtrusion nozzles 100–800 µm, giving features around 100–500 µm · inkjet and laser-assisted methods reach 10–50 µm · post-print cell viability typically 80–95% · constructs stay within roughly 200 µm of a nutrient source without vasculature · research bioprinters $10k to $200k.
ExamplesTissue models, drug-screening platforms, experimental skin and cartilage.
VideosBioprinting Methods: Extrusion, Light-Based and Biodispensing (CELLINK) · Advances in 3D Bioprinting of Vascularized Constructs (Biology Direct)
Powder Metallurgy
press and sinter powders into dense parts5 processesMetal powder is compacted in a die and heated below its melting point so particles bond.
Strengths & weaknessesThe process gives you high material utilization, controlled porosity, and low unit cost at volume. Density and properties may be lower than wrought material, and shapes must permit ejection.
When to usePick press-and-sinter for small-to-medium parts with essentially 2.5D geometry (gears, sprockets, bushings, structural brackets) at tens of thousands of pieces per year and up. At those volumes near-net shaping with over 95% material utilization undercuts machining from bar. It also helps when controlled porosity is a feature you want, as in self-lubricating bearings and filters. The shape has to press and eject along one axis, which means uniform sections and no undercuts or cross-holes without secondary ops. If your volumes are too low to amortize the die, or the part is fatigue- or impact-critical and needs full wrought properties, use forging plus machining instead. If you need intricate small 3D geometry at volume, step up to metal injection molding.
Key numbersCompaction pressures 400–800 MPa · ferrous sintered density 6.6–7.2 g/cm³, roughly 85–93% of wrought · tolerance roughly ±0.05–0.13 mm radially and looser along the press axis · parts usually under 2 kg, pressed at 5–30 per minute · die sets $10k–$100k with 8–16 week lead times · economical from about 10,000 parts a year.
ExamplesGears, bushings, filters, magnets, automotive components.
VideosPowder Metallurgy Processes (MPIF) · Design Considerations with Powder Metallurgy (PickPM, MPIF)
Fine metal powder mixed with binder is injection-molded, debound, and sintered.
Strengths & weaknessesMIM produces intricate small metal parts at scale. Tooling and development are expensive, sintering causes large shrinkage, and part size is limited.
When to usePick MIM for small, genuinely complex metal parts (ideally under about 100 grams and 6 mm wall thickness) at tens of thousands of pieces per year and up. You get injection-molding geometric freedom, meaning undercuts, thin walls, and fine features, along with near-wrought density in stainless, low-alloy steel, or titanium. Surgical instruments, firearm components, and watch parts are the classic wins. Plan for tooling and development in the tens of thousands of dollars, months of qualification, and roughly 15–20% sintering shrinkage held to about ±0.3–0.5%. For simple shapes, press-and-sinter or screw machining is cheaper. For large or thick parts, use casting. If volumes are low or the design is still changing, CNC machining or metal additive gets you parts without the tooling commitment.
Key numbersParts typically under 100 g with walls of 0.3–6 mm · sintering shrinkage 15–20% linear, held to about ±0.3–0.5% · sintered density 96–99% of wrought · surface finish Ra roughly 0.8–1.6 µm · tooling in the tens of thousands of dollars with months of qualification · economical from tens of thousands of parts a year.
ExamplesSurgical tools, watch parts, small gears, electronics hardware.
VideosMetal Injection Molding Guide (Xometry) · Designing with Metal Injection Molding (MIMA)
Powder or a preformed component is exposed to high temperature and uniform gas pressure.
Strengths & weaknessesHIP creates dense material and closes internal porosity in cast or printed parts. Equipment and cycles are expensive, and external geometry generally requires a shaped container or later machining.
When to useSpecify HIP when internal porosity limits fatigue life, which covers cast or powder-bed-fusion titanium and nickel parts for rotating aerospace hardware, medical implants, and similar critical service. It also makes sense for consolidating superalloy powder into segregation-free billets. Treat it as a densification step you buy from a toll processor. It only fixes the inside of the part, so you still need to budget for canning or finish machining, and for batch cycles measured in hours. If the part is statically loaded or cost-driven, skip HIP; vacuum impregnation or simply accepting the porosity is the norm there.
Key numbersPressures 100–200 MPa in argon · temperatures 900–1,250 °C for steel, nickel, and titanium work · hold times 2–4 hours inside 8–12 hour furnace cycles · residual porosity closed to above 99.9% density · vessel working zones from about 200 mm to 2 m across.
ExamplesTurbine disks, aerospace castings, powder-superalloy billets, additive parts.
VideosHot Isostatic Pressing: Process and Applications (Quintus Technologies) · Hot Isostatic Pressing for Fatigue-Critical Additively Manufactured Ti-6Al-4V (Materials, MDPI)
Powder in a flexible mold is compressed uniformly by high-pressure liquid and then sintered.
Strengths & weaknessesCIP creates relatively uniform density in large preforms. Precision is modest, and the flexible tooling deforms.
When to useChoose CIP for large, long, or thick-section powder preforms such as ceramic tubes, refractory billets, and sputtering targets, where uniaxial die pressing would leave density gradients. Cheap elastomer tooling also suits low-to-medium volumes and frequent shape changes. Plan on green machining or generous sintering allowances, since as-pressed tolerances are loose. For small parts at high volume, automated uniaxial die compaction is the default. If you need full density in one step, go to HIP instead.
Key numbersPressures 100–400 MPa · green density 55–70% of theoretical and uniform through thick sections · as-pressed tolerance roughly ±1–2% · elastomer tooling in the hundreds to low thousands of dollars, ready in days · vessels up to about 1.2 m across and 3 m deep · wet-bag cycles 5–30 minutes, dry-bag production much faster.
ExamplesCeramic tubes, refractory parts, powder billets, grinding wheels.
VideosCold Isostatic Pressing: Systems and Applications (Quintus Technologies) · Powder Metallurgy Consolidation Processes (PickPM, MPIF)
Powder is consolidated using pressure and rapid electrical heating.
Strengths & weaknessesSPS densifies powder quickly and keeps microstructures fine. Size, tooling geometry, equipment cost, and scale-up are all still limiting.
When to useReserve SPS for research and pilot quantities of materials that conventional sintering degrades or cannot densify, such as nanostructured and transparent ceramics, thermoelectrics, and ultra-hard composites. It works because minutes at temperature preserve fine grains that hours in a furnace would coarsen. Expect part geometry limited to graphite-die pucks, disks, and simple cylinders up to a few hundred millimeters. If you need production volumes or complex shapes, switch to conventional press-and-sinter, hot pressing, or HIP.
Key numbersHeating rates 100–1,000 °C per minute · sintering to about 2,000 °C at 20–100 MPa through the graphite die · hold times 5–20 minutes inside cycles under an hour · relative density above 99% with grain size near the starting powder · disc diameters typically 20–100 mm, up to roughly 300 mm on the largest presses · machines $200k to over $1m.
ExamplesAdvanced ceramics, thermoelectrics, hard metals, research alloys.
VideosFAST/SPS Technology Explained (FCT Systeme) · Spark Plasma Sintering Capabilities and Fundamentals (Fraunhofer IFAM)
Composite Manufacturing
lay up and cure fiber-reinforced parts11 processesFiber reinforcement is manually placed in a mold and impregnated with resin.
Strengths & weaknessesTooling is inexpensive and very large parts are possible. Labor, voids, repeatability, resin content, and worker exposure are disadvantages.
When to useChoose hand layup for one-offs through a few hundred parts a year, especially large ones like hulls, tanks, and architectural shapes. It makes sense when a mold costing a few thousand dollars matters more to you than laminate quality, and when 30–40% fiber content is acceptable. If void content, repeatability, or styrene exposure are controlled requirements, use vacuum infusion instead; it gives a better laminate from the same single-sided tool. Once volumes reach the thousands, RTM or SMC take over.
Key numbersFiber volume fraction 30–40% · void content typically 2–5% · thickness tolerance around ±10–20%, with one finished side · single-sided tooling from a few hundred to a few thousand dollars · room-temperature cure in about 24 hours · one-offs to a few hundred parts a year, at sizes up to tens of meters.
ExamplesBoat hulls, covers, tanks, prototype panels.
VideosHow to Hand Laminate and Vacuum Bag a Part (Easy Composites) · Open Molding Processes Primer (ACMA Discover Composites)
Chopped fiber and resin are sprayed into an open mold and compacted.
Strengths & weaknessesSpray-up is faster than hand layup for large simple parts. Fiber control and mechanical properties are worse than continuous-fiber methods.
When to useUse spray-up for large, simple, gelcoat-finished glass parts at hundreds to a few thousand a year, such as tubs, shower stalls, and hull liners. On the same cheap open molds it cuts labor well below hand layup. Treat the result as non-structural. Chopped fiber at low volume fraction rules it out for load-bearing parts, carbon fiber, or anything with property specs, and open-mold styrene emissions increasingly rule it out under air-quality regulation. In those cases, step up to infusion or closed molding.
Key numbersFiber volume fraction 15–25% from 25–50 mm chopped strands · deposition rates 10–30 kg/hr, several times hand layup · laminate thickness typically 2–6 mm · tensile strength roughly 60–100 MPa, well below woven laminates · single-sided tooling a few hundred to a few thousand dollars · hundreds to a few thousand parts a year.
ExamplesBathtubs, boat hulls, shower stalls, RV panels.
VideosOpen Molding Processes Primer (ACMA Discover Composites) · Composite Fabrication Methods Overview (CompositesWorld)
Pre-impregnated fiber sheets are placed in a mold, vacuum-bagged, and cured under heat and pressure.
Strengths & weaknessesThe process produces very high-quality composites with low void content. Materials, labor, cold storage, tooling, autoclaves, and cycle times are all expensive.
When to useThis is the default for certified primary aerostructure, spacecraft, and top-tier motorsport. Pick it when sub-1% void content, maximum fiber volume, and an established certification basis justify autoclave time, frozen-material logistics, and high-end tooling, at rates from one-offs to a few thousand parts a year. If your part does not need aerospace laminate quality, infusion or RTM will do the job at a fraction of the price. If the part outgrows the autoclave chamber, OOA prepreg removes the autoclave entirely.
Key numbersFiber volume fraction 55–65% with void content under 1% · autoclave cure at 120–180 °C and 3–7 bar for 2–8 hours · prepreg $40–150/kg, stored at -18 °C with 10–30 days of out-life · invar or steel tooling $50k to several hundred thousand dollars · autoclaves $1m and up · one-offs to a few thousand parts a year.
ExamplesAircraft wings, fuselage sections, satellites, racing structures.
VideosWhat Are Prepregs? (Fibre Glast) · Prepreg Technology Handbook, PDF (Hexcel)
Prepreg is vacuum-bagged and cured in an oven rather than a pressurized autoclave.
Strengths & weaknessesSkipping the autoclave lowers capital cost and loosens part-size constraints. Void control and laminate quality get harder.
When to usePick OOA prepreg when you want near-autoclave laminate quality without owning an autoclave, or when the part is simply too large to fit in one. Typical work includes drone and motorsport structures, aerospace secondary structure, and large one-offs cured in an oven under vacuum bag alone. Keeping voids down takes disciplined debulking and bagging, so if you need certified primary structure with guaranteed sub-1% void content, use an autoclave instead. If the part has no prepreg-level property requirements, vacuum infusion is cheaper still.
Key numbersCure at 80–130 °C under vacuum alone, roughly 1 bar of consolidation · void content typically 1–2% against under 1% from an autoclave · fiber volume fraction 50–60% · ovens $10k–$100k against $1m and up for an autoclave · intermediate debulks every 3–5 plies · part size limited only by the oven, so parts of several meters are routine.
ExamplesDrone structures, automotive panels, aerospace secondary structures.
VideosIntroduction to Out-of-Autoclave Prepreg (Easy Composites) · Out-of-Autoclave Prepregs: Hype or Revolution? (CompositesWorld)
Dry reinforcement is placed in a closed mold and resin is injected before curing.
Strengths & weaknessesRTM gives you finished surfaces on both sides and supports moderate production volumes. Mold sealing, wet-out, void prevention, and flow design are all hard to get right.
When to useChoose RTM at roughly one to thirty thousand parts a year when you need both surfaces finished, net molded edges, and repeatable dimensions. Automotive structural parts, bicycle frames, and aerospace components qualified to it are the usual cases. Below that volume the matched-mold investment and flow-development effort go unamortized, so use infusion or hand layup instead. For very large parts like hulls and blades, infusion is the practical route regardless of volume. At automotive takt times, HP-RTM with presses and fast-cure resins extends the process to minutes-scale cycles.
Key numbersFiber volume fraction 50–55% with void content 1–2% · injection at 1–10 bar for standard RTM and 30–120 bar for HP-RTM · cycle times 20–60 minutes, down to 2–5 minutes with HP-RTM · matched tooling $50k to $500k · tolerance roughly ±0.2–0.5 mm on both molded faces · economical from about 1,000 to 30,000 parts a year.
ExamplesAutomotive structures, aircraft parts, bicycle frames, housings.
VideosClosed Molding Processes Primer (ACMA Discover Composites) · The Rise of HP-RTM (CompositesWorld)
Vacuum draws resin through dry reinforcement under a flexible bag.
Strengths & weaknessesInfusion produces very large composite parts at lower tooling cost than closed molding. Leaks, dry spots, infusion timing, and resin-flow design are the main risks.
When to useDefault to vacuum infusion for very large composite parts at low-to-mid volume, such as hulls, wind blades, and one-piece decks. A single-sided tool and room-temperature cure keep tooling cost an order of magnitude below closed molds, and you still beat hand layup on fiber fraction and emissions. If both surfaces must be finished, or if cycle time matters, use RTM instead. Treat every new part as a flow-design exercise, because a leak or dry spot on a hull-sized infusion scraps an expensive stack of material in one shot.
Key numbersFiber volume fraction 50–60% with void content 1–2% · consolidation capped at one atmosphere, roughly 0.9–1 bar · single-sided tooling about an order of magnitude cheaper than matched molds · infusion in minutes to a couple of hours, then a room-temperature cure overnight · parts up to 100 m long on wind blades · one-offs to a few thousand a year.
ExamplesWind-turbine blades, boat hulls, large aerospace panels.
VideosBeginner's Guide to Resin Infusion (Easy Composites) · Composite Fabrication Methods Overview (CompositesWorld)
Resin-impregnated fibers are wound under tension around a mandrel.
Strengths & weaknessesWinding produces strong, lightweight pressure-containing structures and places fiber efficiently. Geometry is largely limited to convex rotational forms.
When to useChoose filament winding for pressure-containing or torsion-loaded bodies of revolution: CNG and hydrogen tanks, rocket motor cases, pipes, and drive shafts. Controlling the winding angle puts fiber exactly along the hoop and axial load paths, and buying tow keeps material cost low. The process needs an extractable, collapsible, or sacrificial mandrel and mostly convex geometry. If your part is flat, concave, or highly contoured, use AFP; for general shapes, use layup or RTM. If you need a torsion-dominated tube with good damage tolerance, consider braiding instead.
Key numbersFiber volume fraction 60–70% · winding angles from about 10° to 90° off the axis · deposition rates 1–10 kg/hr at tow speeds of 1–3 m/s · large-tow carbon $15–30/kg, the cheapest form of the fiber · hydrogen tanks wound for 700 bar service with a 2.25× burst margin · convex bodies of revolution from millimeters to several meters.
ExamplesPressure vessels, rocket cases, hydrogen tanks, pipes, shafts.
VideosWhat Is the Filament Winding Process? (Addcomposites) · Filament Winding, Reinvented (CompositesWorld)
Continuous fibers are pulled through resin and a heated die to form a continuous composite profile.
Strengths & weaknessesPultrusion gives you high throughput and very good longitudinal properties. The cross-section cannot vary, and transverse properties are weaker.
When to useUse pultrusion when the part is a constant cross-section produced by the kilometer, such as structural profiles, ladder rails, composite rebar, and cable tray. At that scale the die amortizes and the line runs continuously, which makes it the cheapest continuous-fiber process per kilogram. Any varying cross-section rules it out, and transverse strength stays weak unless you add mats or fabrics to the roving stack. If you need hoop-loaded tubes, use filament winding. If you need discrete parts, use molding.
Key numbersLine speeds 0.3–3 m/min · fiber volume fraction 60–70% · longitudinal tensile strength typically 400–1,000 MPa against 30–70 MPa transverse · heated dies run 120–200 °C · dies $10k–$50k with 4–12 week lead times · profiles up to about 1 m wide, produced by the kilometer.
ExamplesBeams, ladder rails, rods, cable trays, window frames.
VideosThe Pultrusion Process (Strongwell) · Technical Principles of Pultrusion (European Pultrusion Technology Association)
Robots place narrow prepreg tows along programmed paths.
Strengths & weaknessesAutomates large high-performance composite structures and enables optimized fiber direction. Machines are expensive and gaps, overlaps, and tow defects require control.
When to useAFP makes sense on large, high-value, contoured structures built at aerospace rates (fuselage sections, wing skins, rocket tanks). Steered tows, low scrap, and repeatable placement beat manual prepreg layup once parts reach meters in scale and programs run for years, which is what justifies the multimillion-dollar machines. If your parts are small, the program is short, or the budget is modest, use hand prepreg layup instead. If the surface is flat or gently curved, ATL lays material down several times faster.
Key numbersTows 3.2–12.7 mm wide, 8–32 per head · lay-down rates 10–50 kg/hr · gap and overlap control around ±0.5 mm · minimum steering radius roughly 400–1,000 mm · scrap under 5%, against 30–50% for hand prepreg layup · machines $2m to over $10m.
ExamplesFuselages, wings, rocket tanks, large aerospace structures.
VideosOverview of the Automated Fiber Placement Process (Addcomposites) · Automated Composites Manufacturing Knowledge Center (CompositesWorld)
Robots lay wide composite tape over relatively flat or gently curved molds.
Strengths & weaknessesFaster than fiber placement for broad simple surfaces. Less capable on highly curved or intricate geometry.
When to useChoose ATL over AFP when the surface is flat or gently curved (wing skins, tail planes, spar caps, large panels). Tape 150–300 mm wide lays down material several times faster than narrow tows, so throughput on big laminates justifies the aerospace-scale machine cost. The wide tape cannot steer or conform to tight double curvature. If the contour is complex, use AFP instead; if the part is small or a one-off, use manual prepreg layup.
Key numbersTape 150–300 mm wide against 3–13 mm tows on AFP · plies about 0.13–0.25 mm thick · lay-down rates several times AFP's, commonly above 30 kg/hr · gap and overlap control around ±0.5 mm · machines in the millions of dollars · parts measured in square meters over multi-year programs.
ExamplesWing skins, tail structures, large panels.
VideosAutomated Composites Manufacturing Knowledge Center: ATL and AFP (CompositesWorld)
Premeasured chopped-fiber resin sheets are compressed and cured in a heated matched die.
Strengths & weaknessesSupports medium-to-high-volume structural composite parts. Fiber orientation is less favorable than continuous-fiber laminates and tooling is expensive.
When to useSMC is the usual composite choice at automotive volumes. Pick it for roughly 10,000–100,000+ parts a year when 1–3 minute cures in matched steel dies, molded-in ribs and bosses, and paintable class-A-capable surfaces matter more than peak properties (body panels, battery enclosures, electrical boxes). Below a few thousand parts a year the die cost is hard to justify, so use RTM or infusion instead. If you genuinely need continuous-fiber stiffness or strength, use prepreg or RTM laminates.
Key numbersGlass content 25–50% by weight · molded at 5–15 MPa and 140–160 °C, curing in 1–3 minutes · presses of 500–3,000 tons · matched steel dies $100k to $500k with 12–20 week lead times · tolerance roughly ±0.1–0.5 mm on molded features · economical from about 10,000 to over 100,000 parts a year.
ExamplesAutomotive panels, battery enclosures, electrical housings.
VideosCompression Molding of SMC (Molded Fiber Glass Companies) · Closed Molding Processes Primer (ACMA Discover Composites)
Wood, Paper & Textile
convert fiber and wood into everyday goods8 processesSaws, routers, planers, drills, lathes, and shapers remove material from lumber or engineered wood.
Strengths & weaknessesWood is easy to machine and aesthetically attractive. It is anisotropic, moisture-sensitive, variable, combustible, and dimensionally unstable.
When to useWood machining is the default for furniture, cabinetry, millwork, and instrument making from prototypes through mid volumes. Tooling is nearly free, CNC routers make complex 2.5D work repeatable, and the material cuts fast and finishes easily. Design around the material rather than the process. Avoid wood where tolerances must hold across humidity swings, where fire or rot resistance is required without treatment, or at volumes where molded plastics or stamped metal become cheaper per part.
Key numbersTolerance about ±0.25 mm on CNC-routed parts, with seasonal moisture movement often larger than that · kiln-dried moisture content 6–8% for interior work and 12–15% outdoors · movement across the grain roughly 0.2% per 1% change in moisture content · router spindles 12,000–24,000 rpm at 5–15 m/min feed · tooling under $5k, mostly bits and fixtures · economical from lot size 1 up to a few thousand parts.
ExamplesFurniture, cabinetry, framing, musical instruments, flooring.
VideosWood Handbook: Wood as an Engineering Material (USDA Forest Products Laboratory)
Thin wood or sheet layers are bonded together for stability, shape, or appearance.
Strengths & weaknessesConserves valuable material, enables curved structures, and improves dimensional stability. Adhesive durability and delamination are concerns.
When to useUse veneering to put premium species on stable, economical substrates. Use lamination to build curved parts that solid wood cannot hold, or engineered structural members (LVL, glulam) with better consistency and span than sawn lumber. Bent lamination over a simple form is the standard route to curves at moderate volume. Avoid deep compound curves, which need molded plywood pressing or a different material entirely. Match the adhesive class to the moisture and heat the part will see, because delamination is the characteristic failure mode.
Key numbersFace veneer 0.5–0.9 mm thick, rotary-peeled plywood plies 1.5–3 mm · hot-press platens 110–140 °C at 0.7–1.4 MPa · press time 3–10 minutes hot and 1–4 hours cold · glulam laminations 20–45 mm thick in beams spanning 30 m and more · forms and cauls $5k–50k · economical from a few hundred panels a year upward.
ExamplesPlywood, furniture surfaces, laminated beams, curved chairs.
VideosPlywood Basics and Applications (APA – The Engineered Wood Association) · Wood-Based Composite Materials, Wood Handbook Chapter 11 (USDA Forest Products Laboratory)
Fiber slurry is deposited onto a porous mold, dewatered, dried, and sometimes hot-pressed.
Strengths & weaknessesUses renewable or recycled fibers and works well for protective packaging. Precision, moisture resistance, and structural performance are limited.
When to usePick pulp molding for high-volume protective and food packaging where sustainability or recycled content is a requirement and tolerances of a millimeter or more are acceptable: egg cartons, cushioning inserts, and, in the hot-pressed thermoformed-fiber grades, smooth premium electronics trays. Tooling is cheap relative to injection molds, so it also suits packaging that changes shape often. Avoid it for parts needing tight dimensions, wet strength, or structural stiffness. Where those dominate and regulation still permits it, thermoformed plastic is the alternative.
Key numbersWall thickness 1–5 mm, down to about 1 mm on hot-pressed thermoformed grades · tolerance roughly ±1 mm, looser on unpressed grades · slurry consistency 0.5–1.5% fiber · cycle time 20–60 seconds per mold set · tooling $5k–50k per tool with lead times of a few weeks · economical above roughly 100,000 units.
ExamplesEgg cartons, food trays, electronics packaging.
VideosPaperboard is cut, creased, folded, glued, laminated, printed, or coated into finished products.
Strengths & weaknessesInexpensive, fast, and highly scalable. Moisture, strength, and durability are limited.
When to usePaper converting is the default for high-volume, low-cost packaging (cartons, corrugated shippers, labels, cups). Steel-rule cutting dies cost little, changeover is fast, and unit costs fall to fractions of a cent at line speeds of hundreds of meters a minute. Avoid it wherever the product must bear sustained load, stay wet, or survive repeated use. Coatings and laminations extend moisture life only modestly before molded pulp or plastics take over.
Key numbersLine speeds 200–400 m/min on corrugators and faster still on folder-gluers · paperboard 200–600 g/m², corrugated flutes 1.6 mm (E) to 4.8 mm (A) · cut and crease tolerance around ±0.5 mm · steel-rule dies $500–5k, rotary dies higher · unit cost from fractions of a cent to a few cents · economical from tens of thousands of units upward.
ExamplesCartons, corrugated boxes, labels, cups.
VideosWhat Is Corrugated: Board, Flutes, and Box Styles (Fibre Box Association)
Two perpendicular yarn systems are interlaced into fabric.
Strengths & weaknessesCreates stable, strong textiles. Usually provides less stretch and conformability than knitting.
When to useWeave when the fabric must hold its shape and carry load (upholstery, belting, airbags, tarpaulins, 0/90 composite reinforcement) and when you want high loom throughput at commodity cost. If you need stretch, drape, or seamless three-dimensional shapes, use knitting instead. If the product is disposable and cost per square meter is the only metric that matters, use nonwovens.
Key numbersAir-jet looms run 1,000–1,500 picks per minute and rapier looms 500–700 · loom width usually 1.5–3.6 m, wider for technical fabric · fabric weight 100–1,000 g/m² · output roughly 30–60 m per loom-hour at 10–20 picks per cm · looms $50k–150k each · warp changeover takes hours, so economical runs start in the hundreds of meters.
ExamplesClothing, upholstery, industrial belts, composite reinforcement.
VideosWeaving Basics (CottonWorks, Cotton Incorporated) · Weaving Technologies and Structures lecture (Australian Wool Education Trust)
Yarn is formed into interlocking loops.
Strengths & weaknessesCreates stretchable, conformable textiles and seamless structures. Dimensional stability may be lower than woven fabric.
When to useKnit when the product must stretch, conform, or come off the machine as a net-shape three-dimensional piece: apparel, shoe uppers, compression and medical textiles, covers for complex forms. Circular and flatbed machines cover high volumes economically, and 3D knitting eliminates cut-and-sew waste. Knits stretch, distort, and can run under tension, so if dimensional stability and strength under sustained load matter more than drape, use weaving instead.
Key numbersGauge 3–18 needles per inch on flat-bed machines and 18–40 on circular · circular machines 25–40 rpm with 90–120 feeders, roughly 20–30 kg of fabric per hour · whole-garment flat-bed knitting 20–60 minutes per piece · fabric weight 100–400 g/m² · machines $30k–80k circular, $100k and up for whole-garment flat-bed · knitting to shape avoids the 15–20% fabric waste of cut-and-sew.
ExamplesApparel, shoe uppers, medical textiles, filters.
VideosKnit Basics (CottonWorks, Cotton Incorporated) · Formation and Properties of Knitted Structures lecture (Australian Wool Education Trust)
Fibers are bonded mechanically, thermally, or chemically without weaving or knitting.
Strengths & weaknessesProvides very high throughput and tunable filtration, absorption, and barrier behavior. Strength and durability may be lower than woven textiles.
When to useChoose nonwovens when the product is roll goods sold by the square meter and the performance requirement is filtration, absorption, or barrier rather than strength (hygiene products, wipes, masks, insulation, geotextiles). Lines are capital-intensive and pay off only at continuous commodity volumes. Avoid the process for durable or load-bearing textiles and for anything without large steady demand; woven or knit fabric covers those cases.
Key numbersBasis weight 10–100 g/m² for hygiene webs, up to 800 g/m² for geotextiles · fiber diameter 15–25 µm spunbond and 1–5 µm meltblown · line width 3.2–5 m at 200–800 m/min · spinneret and bonding-roll tooling in the hundreds of thousands of dollars · line capex $10m–40m for 5,000–20,000 tonnes a year · spunbond polypropylene roughly $2–3/kg.
ExamplesMasks, diapers, wipes, insulation, filters, geotextiles.
VideosHow Nonwovens Are Made: Web Formation and Bonding (EDANA) · About Nonwovens (INDA)
Multiple fibers or yarns are intertwined diagonally around a core or path.
Strengths & weaknessesProduces flexible, damage-tolerant tubular or structural reinforcement. Variable geometry and fiber-path control can be difficult.
When to useBraid for tubular, torsion-tolerant, damage-tolerant structures produced by the continuous meter: ropes, hose and cable reinforcement, catheter shafts and stents, and ±45-dominated composite tubes braided over cores or mandrels. For flat broad goods, weave instead. For hoop-pressure-dominated vessels, filament winding places fiber more efficiently. Braiders control fiber angle only within limits, so avoid braiding for parts that need precise fiber angles varying sharply along the length.
Key numbersCarrier count 16–144 on most machines and several hundred on large ones · braid angle ±20° to ±80°, with most parts near ±45° · line speed 0.5–5 m/min · fiber volume fraction roughly 45–55% in braided composite preforms · mandrels and fixtures $5k–50k · machines from about $50k for rope work to several hundred thousand for radial composite braiders.
ExamplesRopes, hose reinforcement, vascular stents, composite tubes.
VideosBraid 101: Biaxial and Triaxial Braid Fundamentals (A&P Technology) · Effect of Braiding Architectures on 3D Braided Composites (Polymers, open access)
Rubber Manufacturing
compound, shape, and cure elastomer products5 processesElastomer is mixed with fillers, curing agents, plasticizers, stabilizers, pigments, and other additives.
Strengths & weaknessesAllows extensive tuning of hardness, grip, abrasion, weather resistance, conductivity, and cost. Mixing history, contamination, dispersion, and heat buildup strongly affect quality.
When to useCompounding is where a rubber part's properties are actually set. Specify a custom compound whenever hardness, abrasion, weather, chemical, or electrical requirements are real, and buy it from a custom mixer unless annual consumption reaches the hundreds of tonnes that justify in-house Banbury capacity. For undemanding parts, standard off-the-shelf compounds are cheaper and better characterized. Once a recipe and mixing procedure are qualified, lock them; casual substitutions in fillers or cure package are a classic source of field failures.
Key numbersCarbon black 30–80 phr, sulfur 0.5–3 phr, accelerators 0.5–2 phr · hardness range 20–90 Shore A · internal mixer batches 50–400 kg on 2–5 minute cycles, dumping at 100–160 °C · custom mixed compound roughly $2–6/kg · in-house mixing pays off above a few hundred tonnes a year.
ExamplesTire compounds, seals, hoses, belts, vibration isolators.
VideosFormulation 101: Basics of Rubber Compounding (ChemCeed) · Compounding Technical Papers Library (Akrochem)
Uncured rubber is forced through a die into a continuous profile before vulcanization.
Strengths & weaknessesEfficiently creates long constant-cross-section products. Die swell, temperature, dimensional control, and later curing must be managed.
When to useExtrude any elastomer product that is a constant cross-section sold by the meter (seals, tubing, hose bodies, tread stock). Dies cost hundreds rather than tens of thousands of dollars, and continuous cure lines (salt bath, microwave, hot air) sustain high throughput. If the part is a discrete three-dimensional shape, or its tolerances are tighter than die swell allows, use compression, transfer, or injection molding instead.
Key numbersDies cost a few hundred to a couple of thousand dollars · tolerance about ±0.25 mm on small sections at RMA class E1 and ±0.5 mm or looser at commercial class · die swell 10–40% · line speeds 5–60 m/min at 100–1,000 kg/hr · continuous cure lines at 200–250 °C · economical from a few hundred meters per run.
ExamplesWeather seals, hose, tubing, tire tread, gaskets.
VideosRubber Extrusion Process and Materials Overview (IQS Directory) · RMA Extrusion Tolerance Tables (ELBEX)
Rubber compound passes through heated rolls to create sheet or coat reinforcement.
Strengths & weaknessesProduces controlled thickness and reinforced laminates. Equipment is large and temperature and thickness control are demanding.
When to useCalender when you need wide, thickness-controlled rubber sheet or rubber-coated fabric in continuous quantity (tire plies, conveyor belting, roofing membrane), holding gauge tolerances of a few percent across meters of width. The multi-roll line only pays for itself at plant-scale volumes. If you need narrow profiles, use extrusion instead. If you need moderate quantities of sheet, extrusion through a sheet die or press-cured sheet is usually the cheaper route.
Key numbersSheet 0.2–5 mm thick held to roughly ±2% of gauge · web width 1.5–2.5 m · roll temperatures 60–120 °C · line speed 10–60 m/min · a four-roll calender line costs several million dollars · economical only at thousands of tonnes a year.
ExamplesTire plies, belts, roofing membranes, flooring.
VideosAP-42 Section 4.12: Manufacture of Rubber Products (US EPA) · Using Rubber Extrusion to Drive Quality in EPDM Manufacture (IIBEC)
Rubber is heated with sulfur or another crosslinking system so polymer chains form an elastic network.
Strengths & weaknessesCreates resilience, durability, and dimensional stability. Vulcanized rubber cannot simply be remelted, and cure conditions strongly affect performance.
When to useChoose vulcanized thermoset rubber wherever compression set, heat resistance, and dynamic fatigue life matter: seals, tires, engine mounts, and anything loaded continuously or running hot. Sulfur cure is the general-purpose system for dynamic service, while peroxide and specialty cures give higher temperature capability and lower compression set. If parts are lightly loaded and you care more about recyclability or fast injection-molding cycles, thermoplastic elastomers skip the cure step and the scrap that comes with it.
Key numbersSulfur cures run 140–180 °C, with cure time roughly halving per 10 °C rise · sulfur 0.5–3 phr in conventional systems and under 0.5 phr in efficient ones · press cure 2–20 minutes for typical parts, 10–20 minutes for a passenger tire and 40 minutes or more for truck tires · compression set 10–30% after 22 hours at 100 °C, lower with peroxide cure · service temperature 80–100 °C for natural rubber and 150–200 °C for peroxide-cured EPDM and silicone.
ExamplesTires, seals, hoses, bushings, belts.
VideosCuring with Sulfur and Sulfur Donor Systems (Akrochem) · Peroxide Curing of Rubber (Akrochem)
Liners, plies, beads, belts, sidewalls, and tread are assembled into a green tire and then molded and vulcanized.
Strengths & weaknessesIntegrates many materials into a highly engineered composite. Placement, contamination, cure, and inspection are demanding, and small defects can have severe consequences.
When to useTire building is product-specific: every pneumatic tire, from passenger to aircraft, goes through ply assembly, shaping, and mold vulcanization. The plant plus a segmented mold for each size only pay off at hundreds of thousands of units per SKU per year. Below that, don't try to run it yourself. Niche and small-wheel needs are usually better served by molded solid or polyurethane tires, or by contracting capacity at an existing tire plant.
Key numbers20–30 separate components and 10–20 compounds in a passenger tire · green-tire build 30–60 seconds per tire on a modern machine · cure 10–20 minutes at 150–180 °C for passenger sizes, longer for truck · segmented molds roughly $30k–100k each, one per size and tread pattern, with lead times of a few months · greenfield plant capex in the hundreds of millions for a few million tires a year · economical at hundreds of thousands of units per SKU per year.
ExamplesPassenger, truck, aircraft, and industrial tires.
VideosWelding & Joining
join parts by fusion, adhesion, or fasteners14 processesAn electric arc melts base and filler metal to fuse a joint. Major variants differ in electrode, shielding, and deposition rate.
Strengths & weaknessesBroadly applicable, from portable field repair to automated production. Speed, quality, fume generation, and operator skill vary strongly by variant.
VariantsA flux-coated consumable electrode creates the arc and deposits weld metal. Inexpensive and portable but slow and operator-dependent.
A continuously fed wire melts under shielding gas. Fast and automatable but sensitive to wind, fit-up, and cleanliness.
A nonconsumable tungsten electrode creates a clean, precise arc. High-quality welds but relatively slow and skill-intensive.
A tubular flux-filled wire provides high deposition rates. Works well on heavy steel but creates more fumes and slag.
A wire welds beneath granular flux. Very high deposition rates but mainly suits long flat or horizontal seams.
Arc welding is the default for structural steel and general fabrication, from one-off field repair through mid-volume production. Pick stick for portability and outdoor work, MIG for shop throughput and robotic automation, TIG for thin sections and critical or exotic-alloy joints, and flux-cored or submerged arc for heavy plate deposition. If you are joining high-volume thin sheet, resistance or laser welding is faster and distorts less. If distortion matters on aluminum seams, friction-stir welding is often the better answer.
Key numbersDeposition rate roughly 1–3 kg/hr for stick, 3–8 kg/hr for MIG and 10–45 kg/hr for submerged arc · welding current 50–400 A on most shop work and over 1,000 A submerged arc · heat input 0.5–3 kJ/mm with a 1–5 mm heat-affected zone · travel speed 100–500 mm/min · weldment tolerance ±1–3 mm before machining · power sources $500–5k, a robotic cell $100k and up.
ExamplesStructures, vehicles, pipelines, pressure vessels, heavy equipment.
VideosElectric current and clamping force locally fuse metal at the contact interface, without filler.
Strengths & weaknessesRequires no filler and is highly automatable. Mainly suits sheet, often needs two-sided access, and electrodes wear.
VariantsOverlapping sheets are clamped between electrodes and locally fused. Extremely fast and common in automotive assembly.
Rotating wheel electrodes create overlapping welds along a seam. Useful for leak-resistant sheet-metal containers.
Resistance spot welding is the automatic choice for lap-joined steel sheet at automotive volumes. Cycles run under a second, no filler or shielding gas is needed, robots handle it easily, and a single car body takes thousands of welds. Seam welding extends the same idea to liquid-tight containers. The process needs two-sided access and overlapping flanges, so if you have single-sided joints, thick sections, or butt joints, use arc or laser welding instead. Laser is also the answer when flange width has to shrink or the joint has to be invisible.
Key numbersWeld time 0.2–1 second per spot · current 5–15 kA on steel and 20–40 kA on aluminum, at 2–5 kN electrode force · nugget diameter 4–7 mm, roughly 4 to 5 times the square root of sheet thickness in mm · sheet 0.5–3 mm per layer · electrode tips last 2,000–10,000 welds between dressings · 3,000–5,000 spots in a typical car body.
ExamplesAutomotive body assembly, sheet-metal containers, appliances.
VideosResistance Welding Fundamentals Learning Center (AMADA WELD TECH) · What Is Spot Welding? (TWI)
A focused laser melts a narrow joint region, often at high speed.
Strengths & weaknessesProvides low heat input, narrow welds, and excellent automation. Joint fit-up must be precise and equipment is expensive.
When to useChoose laser welding for high-rate precision joining where heat input must stay low (battery tabs and cans, medical devices, powertrain components, hairpin stators) and where fixturing can hold fit-up gaps around 0.1 mm. The equipment cost only pays off at mid-to-high volumes or on genuinely precision work. If you have poor fit-up, thick multi-pass sections, or field conditions, arc welding is still the right tool. If you need very deep welds in reactive metals, electron-beam welding is the specialist alternative.
Key numbersFit-up gap around 0.1 mm, or under 10% of sheet thickness · laser power 1–10 kW for most production welding · penetration roughly 1 mm per kW in keyhole mode on steel · travel speed 1–10 m/min, up to 20 m/min on thin sheet · weld width 0.2–2 mm with a heat-affected zone under 1 mm · cell capex $150k–1m.
ExamplesBatteries, automotive bodies, medical devices, aerospace assemblies.
VideosLaser Welding Applications and Methods (TRUMPF) · Laser Welding Fundamentals Learning Center (AMADA WELD TECH)
A focused electron beam melts the joint, generally in vacuum.
Strengths & weaknessesCreates deep, narrow, clean welds with low distortion. Vacuum chambers, radiation controls, and part-size limits increase cost.
When to useSpecify EB welding for deep, narrow, single-pass welds (tens of millimeters in one shot) in critical rotating hardware and in reactive metals like titanium and zirconium, where the vacuum protects weld purity and distortion has to stay minimal. Batch pumping and chamber size limit both throughput and part dimensions. Unless the depth, purity, or material genuinely requires it, laser welding gives you similar precision faster and cheaper in open air.
Key numbersAccelerating voltage 60–150 kV at 3–100 kW · single-pass penetration 25–50 mm in steel, more on the largest machines · depth-to-width ratio 20:1 to 50:1 · chamber vacuum of 0.0001 mbar or better, with 1–15 minutes of pump-down per cycle · travel speed 0.5–5 m/min · machines $500k to several million.
ExamplesTurbine parts, aerospace gears, nuclear components.
VideosRelative motion generates interface heat, after which pressure forges the parts together without bulk melting.
Strengths & weaknessesCreates strong repeatable joints and can join some dissimilar metals. Parts and fixtures must support the required motion and axial force.
When to useUse rotary friction welding for axisymmetric joints at medium-to-high volume: axles, drill pipe, engine valves, and bimetallic transitions like aluminum-to-steel or copper-to-aluminum that fusion welding cannot make. It runs like a machine-tool process and gives forged-quality joints with no filler, gas, or spatter. Parts have to tolerate high axial force and one side has to spin, which rules out thin sheet and non-round geometry. If the joint is not round, linear friction welding handles it at aerospace-level cost. If you need long seams, use friction-stir welding.
Key numbersCycle time 1–30 seconds, with most joints under 15 · 1,000–3,000 rpm on direct-drive machines and higher on inertia welders · forge pressure 30–200 MPa at machine thrusts of 50–500 kN · upset consumes 3–10 mm of part length · bar diameters 5–150 mm on common machines · machines $150k–1.5m.
ExamplesDrill pipe, axles, cutting tools, valves.
VideosTypes of Friction Welding (MTI) · What Is Friction Welding? (TWI)
A rotating tool plastically stirs material along a seam below the melting point.
Strengths & weaknessesProduces strong low-distortion welds, especially in aluminum. Requires rigid fixturing, substantial reaction force, and tool access.
When to useFSW is the process of choice for long, straight or gently curved seams in aluminum plate and extrusions (battery trays, rail and ship panels, rocket tank barrels), including the 2xxx and 7xxx alloys fusion welding cannot handle. Distortion stays low enough to hold machining-grade flatness afterward. It needs rigid backing, clamping that can react large forces, and tool access along the whole seam. If the joints are complex and three-dimensional, or you are running thin sheet at high rate, use laser or resistance welding instead. Steel is still marginal for FSW because of tool wear.
Key numbersSingle-pass thickness 1–25 mm in aluminum, more when welded from both sides · tool rotation 200–2,000 rpm at 0.5–3 m/min travel · axial force 5–50 kN, reacted through the fixture and backing bar · joint efficiency 70–90% of parent metal in 2xxx and 7xxx alloys · tool life hundreds to thousands of meters in aluminum and far less in steel · machines $250k–2m, or roughly $100k for a head on an existing machining center.
ExamplesAircraft panels, rocket tanks, railcars, ship panels, battery trays.
VideosFriction Stir Welding Technology Overview (MTI) · Friction Stir Welding Process and Properties (TWI)
High-frequency mechanical vibration and clamping pressure create a joint. In thermoplastics, viscoelastic and interfacial heating melts the mating surfaces; in metals, plastic deformation and oxide disruption create a solid-state metallurgical bond without bulk melting.
Strengths & weaknessesExtremely fast, clean, and requires no filler. Thermoplastic welding requires compatible materials and suitable joint geometry; metal ultrasonic welding is primarily limited to thin sheets, foils, wires, and tabs.
When to useUltrasonic welding is the default for mass-produced thermoplastic assemblies. Cycles run under a second, there are no consumables, and the process is clean enough for medical disposables, provided the joint is designed with an energy director, sits near-field under the horn, and joins compatible materials. In metals it is the standard for battery tab and foil-stack joints and wire splices, where laser and resistance welding struggle with thin dissimilar stacks. If the plastic parts are large, use vibration or hot-plate welding instead, and if the metal is thicker than foils and wires, look elsewhere.
Key numbersFrequency 20–40 kHz at 20–60 µm horn amplitude · weld time 0.1–1 second, full cycle under 2 seconds · stack power 1–5 kW · energy-director ribs 0.3–0.6 mm on thermoplastic joints · metal joints limited to foils, tabs, and wires under roughly 0.5 mm · horns and fixtures $2k–10k per part.
ExamplesPlastic medical disposables, electronics housings, wire harnesses, battery tabs, foil stacks, semiconductor wire bonds.
VideosUltrasonic Plastic Welding Technology (Branson by Emerson) · What Is Ultrasonic Welding? (TWI)
Clean surfaces are held under heat and pressure so atoms diffuse across the interface.
Strengths & weaknessesCreates high-integrity low-distortion joints and complex laminated structures. Slow, expensive, and highly sensitive to surface preparation.
When to useReserve diffusion bonding for joints that have to be metallurgically invisible, or for parts that contain thousands of joints buried inside them: stacked-plate microchannel heat exchangers and printed-circuit heat exchangers, titanium SPF/DB aerostructures, and nuclear hardware. It fits low volumes where you can afford hours at temperature under press or HIP load. Success depends mostly on how well the surfaces are prepared, so budget time for that. For ordinary dissimilar or multi-joint assemblies, brazing gets you most of the benefit at a fraction of the cost and time.
Key numbersBonding temperature 50–80% of the material's absolute melting point, around 900 °C for titanium · pressure 1–10 MPa in a press and 100–200 MPa under hot isostatic pressing · hold time 1–4 hours at temperature · mating surfaces prepared to about 0.4 µm Ra and kept clean · joint strength within roughly 10% of parent metal · printed-circuit heat exchanger channels 1–2 mm wide, stacked hundreds of plates deep.
ExamplesAerospace heat exchangers, titanium structures, nuclear components.
VideosDiffusion Bonding for Microchannel Heat Exchangers (Vacuum Process Engineering) · Diffusion Bonding Process and Parameters (TWI)
Filler metal melts above roughly 450 °C and flows into a close-fitting joint while the base materials remain solid.
Strengths & weaknessesJoins dissimilar materials, thin parts, and complex assemblies with limited distortion. Joint strength and high-temperature performance may be lower than welding, and gap control is critical.
When to useBraze when an assembly has many joints to make in one heat, thin sections, or dissimilar pairs that fusion welding would damage: plate-fin and tube heat exchangers, carbide inserts on steel bodies, honeycomb panels. Use furnace brazing at volume, and torch or induction brazing for job work. Design the joints for capillary gaps of a few hundredths to a tenth of a millimeter, since gap control is what determines whether the joint holds. If service temperature approaches the filler's melting range, or you need full parent-metal strength, weld instead. If temperatures and loads are lower still, solder.
Key numbersFiller melts above 450 °C: 590–620 °C for aluminum brazing, 620–800 °C for silver fillers, 1,000–1,200 °C for nickel fillers · capillary joint gaps 0.02–0.10 mm · lap length roughly 3 times the thinner member's thickness · furnace cycles 1–4 hours per load, with hundreds of joints made in one heat · joint shear strength 100–300 MPa on silver-brazed steel · fixtures and furnace tooling $5k–50k.
ExamplesHeat exchangers, carbide tools, HVAC tubing, honeycomb structures.
VideosBrazing Fundamentals: The Six Steps (Lucas-Milhaupt) · What Is Brazing? (TWI)
Low-melting filler metal bonds components without melting the base materials.
Strengths & weaknessesInexpensive, electrically conductive, and suitable for delicate components. Structural strength and temperature resistance are limited.
When to useSolder wherever you need an electrical connection made at low process temperature (it is the interconnect default for all electronics), and for leak-tight copper plumbing. Keep it away from structural loads and from sustained service much above roughly 150 °C, where creep and softening set in. Brazing is the same capillary process with far higher strength and temperature capability, so use it instead when the parts can take the heat.
Key numbersEutectic tin-lead melts at 183 °C and SAC305 lead-free at 217–220 °C · iron tips run 320–370 °C for lead-free work · joint shear strength roughly 20–50 MPa · service temperature limited to about 150 °C before creep matters · 2–5 seconds per hand-soldered joint · equipment from a $20 iron to a $500 temperature-controlled station.
ExamplesPrinted circuit boards, plumbing fittings, electrical terminals.
VideosPolymeric adhesive joins surfaces through chemical adhesion and mechanical interlocking.
Strengths & weaknessesJoins dissimilar materials, spreads load, seals joints, and avoids thermal distortion. Surface preparation, cure time, environmental degradation, inspection, and peel loads are concerns.
When to useBond when you are joining dissimilar or heat-intolerant materials (composites to metals, thin skins, glass), or when you want the joint to seal as well as spread load over an area instead of concentrating it at fasteners or welds. Budget for surface preparation, fixturing, and cure time, and design the joint so the adhesive is loaded in shear. Avoid peel-dominated loads, service beyond the adhesive's temperature rating, and safety-critical joints you cannot inspect or process-control. For those, combine adhesive with mechanical fastening or use fastening alone.
Key numbersLap shear strength 20–40 MPa for structural epoxies and 5–15 MPa for polyurethanes · bond line 0.1–0.5 mm · room-temperature cure about 24 hours, aerospace film adhesive roughly 1 hour at 120–180 °C · service temperature 80–120 °C for common epoxies and up to about 200 °C for high-temperature systems · adhesive $20–200/kg · fixtures and cure tooling usually under $5k.
ExamplesAircraft panels, automotive structures, composites, electronics.
VideosControlled explosive energy accelerates one metal plate into another, creating a solid-state metallurgical bond.
Strengths & weaknessesJoins otherwise incompatible metals across large areas without bulk melting. Requires specialized remote facilities, explosive controls, and post-process flattening.
When to useUse explosion welding for large-area bonds between metallurgically incompatible metals (titanium- or nickel-clad steel tubesheets and pressure-vessel plate, aluminum-steel structural transition joints), where fusion processes would form brittle intermetallics. It is bought as a toll service from a handful of specialist sites, and it is economical from single plates up to modest lot sizes. If you need continuous bimetal strip at volume, roll bonding is far cheaper. For small parts, look at brazing or diffusion bonding instead.
Key numbersPlates up to roughly 30 m² per shot · cladding layer 0.5–25 mm on backer plate up to 300 mm thick · detonation velocity 1,500–3,000 m/s at 5–20 mm standoff · interface waves 0.05–1 mm in amplitude · bond shear strength above 140 MPa, the ASME minimum for clad plate · economical from a single plate up to lots in the low hundreds.
ExamplesTitanium-clad steel, aluminum-steel transitions, corrosion-resistant plate.
VideosMetal layers are compressed together through rolling, often with heat, to form a bonded laminate.
Strengths & weaknessesEfficiently combines corrosion resistance, conductivity, strength, and cost properties. Surface cleanliness, deformation compatibility, intermetallic growth, and downstream forming behavior must be controlled.
When to useRoll bond when you need bimetal strip or sheet by the coil (stainless- or nickel-clad steel, copper-aluminum busbar and battery strip, cookware blanks). A continuous mill gives the lowest cost per square meter of any cladding route. The mill investment requires high steady volume and strip-like geometry, so for one-off thick plates or refractory combinations use explosion welding instead. If a thin surface layer will do rather than a solid clad layer, weld overlay or thermal spray is cheaper.
Key numbersSingle-pass reduction of 50–70% needed to break the surface oxide and bond · clad layer usually 5–20% of total thickness · strip 0.1–50 mm thick and 0.5–2 m wide · hot roll bonding of stainless-clad steel at 1,100–1,250 °C, with cold-bonded bimetal strip annealed at 200–400 °C afterward · mill capex in the tens of millions · economical at thousands of tonnes a year.
ExamplesStainless-clad steel, copper-aluminum strip, cookware, battery materials.
VideosHow Is Clad Metal Made? (Materion) · What Is Roll Bonding? (TWI)
Parts are joined with discrete hardware or interlocking geometry rather than fusion or adhesive.
Strengths & weaknessesBroadly applicable, often reversible and inspectable. Adds weight or hardware, requires holes or precise tolerances, and joints can loosen, creep, or fatigue.
VariantsReversible and inspectable, but they add weight, require holes, and can loosen.
Reliable for thin sheet and difficult-to-weld materials, but requires hole preparation and often two-sided access.
Locally deforms sheet layers into an interlock without separate fasteners. Fast but lower-strength and generally requires two-sided tools.
Flexible features engage mating geometry without separate hardware. Reduce assembly cost but are sensitive to creep, fatigue, and tolerance.
Dimensionally overlapping components are forced together. Compact but require tight tolerances and may be difficult to disassemble.
Heating or cooling temporarily changes dimensions before assembly. Strong concentric joints but added thermal process complexity.
Fasten mechanically when joints must come apart for service, when you are joining dissimilar or heat-intolerant materials, or when assembly happens in the field without process control. It is also the default whenever inspectability and repairability matter more than weight. Within the family, choose rivets for thin sheet and airframes, snap fits for high-volume plastics, and press or shrink fits for concentric hubs. Avoid mechanical fastening where sealing, weight, or fatigue at hole stress concentrations dominates, since welding or adhesive bonding eliminates both the holes and the hardware.
Key numbersBolt preload usually 65–75% of proof load, scattered ±25% or worse by plain torque control · roughly 50% of applied torque goes to head friction and 40% to thread friction, leaving 10% as preload · property class 8.8 bolts at 800 MPa tensile and 12.9 at 1,200 MPa · rivet pitch 4–6 diameters at about 2 diameters edge distance · stress concentration around 3 at an open hole · fasteners $0.05–5 commercial and $1–20 each in aerospace grades.
ExamplesAircraft structures, machinery, consumer products, electronics.
VideosSnap-Fit Joint Design (Xometry) · Fastener Design Manual, NASA RP-1228 (NASA NTRS)
Electronics Packaging & Assembly
mount, connect, and protect chips and boards6 processesAutomated equipment positions electronic components onto solder-paste-coated circuit boards.
Strengths & weaknessesPlaces thousands of small components per hour with high accuracy. Equipment, programming, feeder systems, and board design are capital- and process-intensive.
When to useIf you are populating surface-mount boards beyond a handful of prototypes, pick-and-place is the only realistic option. It scales from small-batch runs on desktop machines to production lines placing tens of thousands of components per hour, and programming and feeder setup are the fixed cost that batches amortize. Hand placement still makes sense for one-off prototypes and rework. For extreme-density bare-die interconnect, use dedicated flip-chip and advanced-packaging equipment instead.
Key numbersPlacement rate 3,000–5,000 components per hour on desktop machines and 20,000–100,000 on production chip shooters · placement accuracy ±25–50 µm, or ±10–20 µm on fine-pitch machines · components from 0201 imperial (0.6 by 0.3 mm) up to 50 mm connectors · 100–200 feeder slots per machine on 8 mm tape pitch · machines $30k for a desktop unit to $1m for a high-speed line · changeover and programming 30 minutes to a few hours per board.
ExamplesComputers, phones, industrial controllers, vehicle electronics.
VideosSolder paste and components pass through a controlled thermal profile until joints form.
Strengths & weaknessesJoins many surface-mount components simultaneously. Warpage, oxidation, voids, thermal damage, and profile control are key risks.
When to useReflow is the default soldering step for any surface-mount board, because every joint forms in one oven pass. Thermal profiling is the main engineering effort, and lead-free peaks around 245–250 °C set what the components have to survive. Use wave or selective soldering when through-hole connectors dominate the board, hand or hot-air work for prototypes and rework, and vapor phase or vacuum reflow when void-sensitive power or RF parts require it.
Key numbersPeak temperature 245–250 °C for lead-free alloys, which melt around 217 °C · 60–90 seconds above melting and 4–6 minutes total through the oven · ramp rates 1–3 °C/s with a 150–200 °C soak for 60–120 seconds · 7–12 heating zones at conveyor speeds 0.5–1.5 m/min · BGA voiding accepted to about 25% of ball area under IPC-A-610 · ovens $30k–150k.
ExamplesPrinted circuit boards, LED assemblies, electronic modules.
VideosFine gold, copper, or aluminum wire is bonded between a die and package using ultrasonic energy, heat, pressure, or combinations of them.
Strengths & weaknessesMature, flexible, inexpensive, and well understood. Adds electrical inductance and package height and is limited at extreme interconnect density.
When to useWire bonding is the default die-to-package interconnect for cost-sensitive and moderate-pin-count devices (analog, sensors, LEDs, MEMS), and heavy aluminum wire is still standard in power modules. It needs no wafer bumping, and a routing change takes a program edit rather than a new mask. Move to flip-chip when pin counts run into the high hundreds, when interconnect inductance limits high-frequency or high-current performance, or when package height is critical.
Key numbersWire 15–50 µm gold or copper for ball bonding, 100–500 µm aluminum in power modules · 10–20 wires per second on production bonders · pad pitch down to about 35 µm at 100–300 µm loop height · thermosonic bonding at 125–175 °C with ultrasonic drive near 60 kHz · loop inductance roughly 1 nH per mm of wire · bonders $100k–300k.
ExamplesAnalog ICs, sensors, LEDs, MEMS, power devices.
VideosWire Bonding: Ball and Wedge Processes (Semiconductor Digest)
Solder bumps, copper pillars, or similar interconnects on a die are bonded directly to a substrate.
Strengths & weaknessesEnables high density, short electrical paths, and strong thermal performance. Alignment, warpage, underfill, bump reliability, and inspection are difficult.
When to useChoose flip-chip when wire bonds cannot deliver what the die needs: interconnect counts from the high hundreds into the thousands, minimal inductance for high-frequency or high-current paths, or heat extraction through the exposed die back. Typical parts are CPUs, GPUs, RF front ends, and chiplet assemblies. Flip-chip requires wafer bumping, fine-alignment assembly, and underfill development, so for low-pin-count cost-driven devices wire bonding is still the economical default.
Key numbersBump pitch typically 130–200 µm with solder bumps and down to about 40 µm with copper pillars · interconnect counts from several hundred to over 10,000 on large processors · placement accuracy ±5–10 µm on mass-reflow lines and 1–3 µm on thermocompression bonders · throughput roughly 2,000–6,000 dies per hour on mass reflow, well under 1,000 on thermocompression · underfill cure 30–60 minutes at about 150 °C · die sizes from a few mm on a side up to roughly 30 mm.
ExamplesCPUs, GPUs, RF devices, image sensors, chiplets.
VideosSemiconductor dies are attached to packages or heat spreaders using solder, silver sinter, eutectic bonds, or adhesive.
Strengths & weaknessesProvides structural support, thermal conduction, and sometimes electrical connection. Voids, intermetallic growth, contamination, and thermal mismatch can cause failure.
When to useDie attach is a mandatory packaging step, so the real decision is which material to use. Use solder or silver-sintered attach when the die dissipates a lot of power or needs a low-resistance electrical path (power modules, high-brightness LEDs), with sintering preferred for high-temperature and automotive reliability. Conductive or plain epoxy is the low-cost default for logic, sensors, and anything thermally undemanding, and film adhesives suit thin-die stacking in memory packages.
Key numbersBondline thickness roughly 20–75 µm · thermal conductivity about 1–3 W/mK for conductive epoxy, 30–60 W/mK for solder, and 150–250 W/mK for sintered silver · placement accuracy ±10–25 µm on production die bonders and a few um on precision machines · throughput 5,000–20,000 dies per hour · voiding usually specified under 5% of the die area · silver sintering at about 250 °C and 10–30 MPa, epoxy cure 30–60 minutes at 150–175 °C.
ExamplesPower modules, LEDs, processors, sensors.
VideosEutectic Die Bonding (Palomar Technologies) · Eight Steps of Assembling Conventional Packages (SK hynix)
Epoxy molding compound, underfill, glob top, or hermetic sealing protects the die and interconnects.
Strengths & weaknessesImproves mechanical and environmental protection. Cure stress, moisture absorption, delamination, and trapped voids are risks.
When to useEvery packaged die needs protection, so choose the method by environment and volume. Transfer-molded epoxy compound is the high-volume default for ICs, glob top is the cheap answer for chip-on-board, and underfill is effectively mandatory under flip-chip dies. Reserve hermetic metal or ceramic packages for aerospace, implantable, and other applications where molded plastic's moisture uptake and delamination risk are unacceptable; they cost roughly an order of magnitude more.
Key numbersTransfer molding at 170–180 °C with 60–120 seconds in the mold, then 4–6 hours of post-mold cure near 175 °C · molding pressure 5–10 MPa · epoxy molding compound 70–90% silica filler by weight, giving a CTE of roughly 8–16 ppm/K · moisture sensitivity level 1 to 3 under JEDEC J-STD-020 · underfill cure 30–60 minutes at about 150 °C · hermetic ceramic or metal packages at roughly 10x the cost of molded plastic.
ExamplesIC packages, controllers, MEMS, power electronics.
VideosBattery Manufacturing
turn electrode materials into finished cells9 processesActive material, conductive additive, binder, and solvent are mixed into a controlled slurry.
Strengths & weaknessesDetermines dispersion, rheology, coating quality, and electrochemical performance. Sequencing, shear, contamination, temperature, and scale-up are critical.
When to useSlurry mixing is the mandatory first step of any slurry-cast electrode line, so the decisions are equipment and solvent. Batch planetary mixers suit pilot lines and multi-product plants, while continuous twin-screw mixing is better at GWh scale on footprint, consistency, and cost. Water-based processing is standard for graphite anodes, but most cathodes still require NMP and its recovery loop. If solvent handling and drying dominate your economics, dry-electrode processing is the emerging alternative that skips this step entirely.
Key numbersFormulation roughly 90–97% active material, 1–5% binder, 1–3% conductive carbon · solids loading about 65–75% by weight for NMP cathode slurries and 45–55% for water-based anode slurries · slurry viscosity roughly 2,000–10,000 cP at coating shear rates · batch planetary cycles of 1–4 hours in 500–2,000 L vessels · continuous twin-screw residence time of seconds to a few minutes.
ExamplesAnode and cathode slurries for lithium-ion cells.
VideosImpact of Formulation and Slurry Properties on Electrode Manufacturing (Battery Design) · Production Process of a Lithium-Ion Battery Cell (PEM RWTH Aachen and VDMA)
Slurry is continuously applied to metal foil, often by slot die, and dried to remove solvent.
Strengths & weaknessesProvides high throughput and precise areal loading. Thickness variation, cracking, edge behavior, dryer length, solvent recovery, and line speed are major constraints.
When to useSlot-die coating on continuous roll-to-roll lines is the standard for any serious cell production. Choose it whenever areal loading has to hold to a percent or two at tens of meters per minute. Line speed is bought with dryer length, so ovens dominate plant footprint and energy cost, and NMP cathode lines add solvent-recovery capital on top. Below pilot scale, doctor-blade and lab coaters are enough. At giga-scale, dry-electrode coating is the developing alternative that removes the ovens.
Key numbersLine speeds of roughly 30–80 m/min on production coaters · web widths 300–1,500 mm · dry coating 50–150 µm per side, giving areal loadings around 2–4 mAh/cm² · areal loading held to about ±1–2% across the web · dryers 30–100 m long · NMP recovery systems reclaiming typically over 90% of the solvent.
ExamplesLithium-ion anode and cathode webs.
VideosElectrode Coating Techniques: From Lab to Gigafactory (Battery Design) · Slot-Die Coating: Theory, Design, and Applications (Ossila)
Dried electrodes pass through precision rolls to reach target thickness, density, and porosity.
Strengths & weaknessesImproves particle contact, adhesion, and energy density. Excessive compression reduces transport and can damage particles or foil.
When to useCalendering is mandatory to hit target density, so the decision is the setpoint rather than whether to run the step. Compress energy cells toward roughly 25–30% porosity for maximum volumetric capacity, and leave power and fast-charge designs more open, since over-densification restricts electrolyte transport and cracks particles. High-nickel cathodes and silicon-bearing anodes tolerate the least compression. As webs get wider, watch roll deflection and foil elongation, because uneven density across the web shows up later as lithium plating.
Key numbersTarget porosity roughly 25–30% for energy cells and 35% or more for power and fast-charge designs · electrode density about 1.5–1.7 g/cm³ for graphite anodes and 3.2–3.6 g/cm³ for NMC cathodes · thickness held to roughly ±1–2 µm across the web · coating compressed 20–40% from the as-dried state · line speeds up to roughly 80–100 m/min · hot-calendering rolls at 60–120 °C.
ExamplesDensified anode and cathode webs.
VideosHot versus Cold Roll Calendering (Battery Design) · Production Process of a Lithium-Ion Battery Cell (PEM RWTH Aachen and VDMA)
Coated rolls are cut into narrow strips and shaped to create tabs or cell-specific profiles.
Strengths & weaknessesThroughput is high, but the process is sensitive to burrs, dust, edge cracks, and registration. Small defects can cause internal shorts.
When to useSlitting and notching are mandatory between coating and cell assembly, so the real decision is which cutting technology to use. Mechanical slitting is cheapest and is the standard choice for plain strip at fixed widths, but blades wear and burrs grow as they wear. If tab geometries change often or the formats are complex, use laser notching instead: you get tool-free flexibility, and in exchange you accept heat-affected edges and the need for fume extraction. Either way, burr height and particle cleanliness are the specs that matter, because edge defects become internal shorts.
Key numbersSlitting line speeds roughly 50–150 m/min · laser notching at roughly 60–120 m/min with no tooling change between formats · strip width tolerance about ±0.1–0.3 mm · burr height usually specified under 10 µm, against a separator only 12–25 µm thick · heat-affected zone of tens of um on laser-cut edges.
ExamplesCell-width electrode strips, tabbed electrodes.
VideosProduction Process of a Lithium-Ion Battery Cell (VDMA / RWTH Aachen PEM) · Advanced Li-Ion Battery Manufacturing Equipment for Gigafactories (iScience)
Anode, separator, and cathode webs are wound into a compact jelly roll.
Strengths & weaknessesWinding is fast and material-efficient. Tension, alignment, wrinkling, telescoping, and contamination must be controlled.
When to useWind whenever the format allows it. It is the fastest, cheapest, and most mature assembly route, it is the only practical one for cylindrical cells, and winders comfortably outpace stacking machines per unit of capital. You do have to live with its limits: winding puts curvature stress on the electrodes, it puts corner strain into flattened prismatic rolls, and it fits thick or fragile electrodes poorly. If you are building large pouch or prismatic formats where packaging efficiency and uniform stack pressure drive performance, or solid-state and lithium-metal designs, use stacking instead.
Key numbersWinding roughly 1–3 seconds per jelly roll on cylindrical machines · electrode alignment held to about ±0.2–0.5 mm against a designed anode overhang of 1–2 mm · separator 12–25 µm thick · standard cylindrical formats of 18650 (18 x 65 mm), 21700, and 4680 · cell capacities from roughly 2.5–3.5 Ah in 18650 to 25–30 Ah in 4680.
ExamplesCylindrical and wound prismatic cells.
VideosProduction Process of a Lithium-Ion Battery Cell (VDMA / RWTH Aachen PEM) · Current and Future Lithium-Ion Battery Manufacturing (iScience)
Alternating electrode and separator sheets are aligned and stacked.
Strengths & weaknessesStacking supports rectangular formats and uniform compression. The machinery is mechanically complex and sensitive to alignment.
When to useStack when the format is rectangular and the performance gain justifies slower, costlier machinery. Pouch and large prismatic EV cells get packaging efficiency, uniform pressure, and better swelling tolerance from flat stacks, and solid-state and lithium-metal designs with fragile layers essentially require them. Z-folding the separator is the common compromise between speed and alignment. If you are making cylindrical cells, or assembly cost per amp-hour dominates, winding is the faster default.
Key numbersStacking rates of roughly 0.1–0.3 seconds per layer on current machines · 20–60 electrode layers in a large pouch or prismatic cell, so 5–15 seconds per cell · layer alignment held to about ±0.2–0.3 mm · anode overhang of 1–2 mm per edge to cover alignment error · cell capacities of 50–120 Ah for EV pouch cells and 100–300 Ah for large prismatic.
ExamplesPouch, prismatic, solid-state, and lithium-metal cells.
VideosAdvanced Li-Ion Battery Manufacturing Equipment for Gigafactories (iScience) · Current and Future Lithium-Ion Battery Manufacturing (iScience)
Electrolyte is introduced, often under vacuum, and allowed to penetrate porous electrodes and separator.
Strengths & weaknessesProper wetting is essential for impedance, capacity, and safety. Moisture sensitivity, soak time, incomplete wetting, and trapped gas are major challenges.
When to useEvery liquid-electrolyte cell line needs this step, so the real choices are how you dose and how you wet. Use vacuum filling with multi-step dose-and-soak cycles for large-format prismatic and pouch cells, for high-loading or highly calendered electrodes, and for any design where trapped gas or dry spots would show up as impedance outliers. Budget hours of soak time, and hold the fill room below roughly -40 °C dew point. Single-shot filling with short soaks is acceptable only for small cylindrical cells with thin, porous electrodes. If fill and wetting time is dominating line takt, fix it through electrode porosity, electrolyte formulation, or a warm soak, not by shortening the soak and shipping poorly wetted cells.
Key numbersElectrolyte charge of roughly 1.5–3 g per Ah of cell capacity · dosing accuracy around ±1% of fill mass · filling in 2–5 dose-and-soak cycles under vacuum · wetting soak from several hours to more than a day on large-format cells · dry-room dew point held below about -40 °C.
ExamplesLithium-ion cells of all formats.
VideosProduction Process of a Lithium-Ion Battery Cell (VDMA / RWTH Aachen PEM) · Advanced Li-Ion Battery Manufacturing Equipment for Gigafactories (iScience)
Newly assembled cells undergo controlled initial charging and discharging to form stable interfacial layers.
Strengths & weaknessesFormation strongly affects lifetime, impedance, self-discharge, and safety. It also requires extensive power electronics, long residence time, heat management, floor space, and working capital.
When to useFormation is unavoidable for lithium-ion, so the decision is how much capex and residence time to spend on it. Plan for it early in factory design, since formation and aging typically occupy a quarter or more of plant footprint and capex, and size charger channels with regenerative discharge to recover energy at scale. Use slow, multi-step protocols with elevated-temperature holds when you are qualifying new chemistries or when warranty life dominates. Move to compressed fast-formation recipes only after cycle-life and self-discharge data prove them out on the exact cell design. Do not shorten formation to relieve a capacity bottleneck without that validation, because latent SEI defects show up as field failures rather than line rejects.
Key numbersFirst charge at roughly C/20 to C/10, with total formation time from about 12 hours to several days per cell · first-cycle irreversible capacity loss of roughly 5–10% to SEI formation · elevated-temperature holds usually 40–60 °C · formation channel current and voltage accuracy of roughly 0.05% of full scale · formation and aging together taking a quarter or more of a cell plant's floor space and capex.
ExamplesAll lithium-ion cell production.
Economic profileOften one of the largest capex and cycle-time bottlenecks in a battery factory.
VideosPredicting the Impact of Formation Protocols on Battery Lifetime (arXiv) · Current and Future Lithium-Ion Battery Manufacturing (iScience)
Cells rest and are measured for capacity, voltage retention, impedance, leakage, and self-discharge before sorting.
Strengths & weaknessesAging and grading identify latent defects and let cells be matched into packs. They also consume time, floor space, equipment, and inventory capital.
When to useHold cells through a high-temperature or room-temperature aging period, typically one to three weeks, whenever self-discharge screening matters. That covers any application where an internal micro-short becomes a field safety event, and any pack with many cells in series where capacity and impedance mismatch drives balancing loss. Grade tightly for EV and grid packs that string dozens to thousands of cells. Loose binning is tolerable for single-cell consumer devices. Shorten aging only if you have statistical evidence from delta-OCV data on a mature, stable process, and treat rising bin spread as an upstream process-control alarm rather than a sorting problem.
Key numbersAging hold of roughly one to three weeks, often a 1–3 day stage at 45–60 °C followed by room-temperature rest · self-discharge screened from OCV drops of a few mV, resolved to about 0.1 mV · capacity bins typically ±1–2% wide and impedance bins around ±5% · direct leakage-current measurement resolving down to roughly 1 uA in hours rather than weeks · weeks of finished-cell inventory held as working capital.
ExamplesCell binning and pack matching.
VideosProduction Process of a Lithium-Ion Battery Cell (VDMA / RWTH Aachen PEM) · High-Precision Detection of Cell-to-Cell Variation in Li-Ion Batteries (Scientific Reports)
Heat Treatment
tune metal hardness and toughness with heat8 processesMaterial is heated and cooled under controlled conditions to soften it, relieve stress, or modify microstructure.
Strengths & weaknessesAnnealing improves ductility, machinability, and dimensional stability. It may reduce strength and can cause oxidation or distortion.
When to useAnneal when accumulated cold work is about to cause cracking: between wire-drawing or deep-drawing passes, before severe forming of sheet, and after welding or heavy machining when residual stress would distort a later precision operation. Use full annealing to maximize softness and machinability of high-carbon and alloy steels. If you only need stress removal without a strength penalty, the cheaper stress-relief or process anneals below the transformation temperature will do. Avoid annealing on parts whose final strength comes from work hardening or prior heat treatment, since it erases both. If the steel just needs a uniform, moderately tough structure at lower furnace time, normalize instead. Use a protective atmosphere or vacuum when scale and decarburization on finished surfaces are unacceptable.
Key numbersFull annealing of carbon steel at roughly 760–900 °C, subcritical and process anneals at 600–700 °C, stress relief at 550–650 °C · soak time about 1 hour per 25 mm of section · furnace cooling at roughly 10–30 °C per hour, so total cycles of 8–24 hours · aluminum annealed at 340–415 °C, soda-lime glass near its 550 °C annealing point · annealed carbon steel typically 120–200 HB.
ExamplesSheet before forming, wire between draws, castings, glass.
VideosAnnealing Process Guide (Xometry) · Overview of Steel Heat Treatment Processes (tec-science)
Steel is heated above a transformation temperature and air-cooled to refine grain structure.
Strengths & weaknessesNormalizing improves uniformity and toughness. Property control is less precise than with quenching and tempering.
When to useNormalize carbon and low-alloy steel forgings, castings, and hot-worked stock when you want a uniform, refined grain structure and a predictable response to later machining or hardening. It is faster and cheaper than full annealing because parts air-cool outside the furnace. It is also the right call for structural components where moderate strength and good toughness are enough and a quench would risk distortion or cracking in heavy or complex sections. If the application needs a specific hardness or strength class, or fatigue performance beyond what a pearlitic structure delivers, use quenching and tempering. If you want maximum softness for machining, use full annealing.
Key numbersAustenitizing at roughly 830–950 °C, about 30–60 °C above the Ac3 temperature · soak about 1 hour per 25 mm of section · still-air cooling, so cycles of a few hours against the 8–24 hours a full anneal takes · normalized carbon steel typically 150–250 HB · tensile strength around 550–700 MPa on normalized medium-carbon grades.
ExamplesForgings, castings, structural components.
VideosNormalizing Heat Treatment Guide (Xometry) · Normalizing of Steel (tec-science)
Steel is transformed, rapidly cooled to create hard martensite, and reheated to restore toughness.
Strengths & weaknessesQuenching and tempering produce high strength and hardness with adjustable toughness. Quenching can cause distortion, cracking, and residual stress.
When to useChoose quench and temper when a medium-carbon or alloy steel part needs through-thickness strength and hardness (roughly 30 to 60 HRC, dialed in by tempering temperature), as in shafts, fasteners above class 8.8, tools, and gears. Match hardenability to section size: plain carbon steels only harden thin sections in aggressive quenches, while alloy grades harden deep sections in oil with less distortion risk. Leave grinding stock and plan for straightening on slender or asymmetric parts. If distortion budgets are tight, switch to a milder quench, austempering, or a pre-hardened steel. If only the surface needs hardness over a tough core, use induction hardening or carburizing instead of hardening the whole part.
Key numbersAustenitizing at roughly 800–900 °C, then quenching in water, oil, polymer, or gas · as-quenched martensite of about 55–65 HRC depending on carbon content · tempering at 150–650 °C, giving 30–60 HRC · plain carbon grades through-hardening only to roughly 15–25 mm of section, alloy grades to 50–100 mm in oil · 0.2–0.5 mm of grinding stock left for quench distortion.
ExamplesGears, tools, shafts, armor, fasteners.
VideosQuenching Process Guide (Xometry) · Quenching and Tempering of Steel (tec-science)
An alloy is heated to dissolve strengthening phases, quenched to retain a supersaturated solid solution, and aged so fine precipitates form.
Strengths & weaknessesThe treatment can greatly increase yield strength, creep resistance, or elevated-temperature capability while preserving the alloy's underlying density and corrosion characteristics. It only works for suitable alloy systems, and poor temperature, quench, or aging control can cause distortion, residual stress, underaging, or overaging.
When to useThis is the default strengthening route whenever the alloy supports it: 2xxx, 6xxx, and 7xxx aluminum, PH stainless grades, beryllium copper, nickel superalloys, and alpha-beta titanium. Specify it when you need the highest strength-to-weight the alloy family offers (a T6 temper takes 6061 from about 55 MPa yield annealed to about 275 MPa, roughly 5x). Sequence forming and welding before aging, since both are far easier in the solutionized condition and welding locally destroys the temper. Machine distortion-critical parts after solution quench but before or between aging steps, or use low-distortion tempers. For non-heat-treatable alloys such as 5xxx aluminum or austenitic stainless, this route does nothing, so rely on work hardening. If a steel needs bulk hardness, use quench and temper instead.
Key numbersAluminum solution treated at roughly 465–540 °C, quenched within seconds, then aged 8–24 hours at 120–190 °C · 17-4 PH stainless solution treated near 1,040 °C and aged 1–4 hours at 480–620 °C for 30–44 HRC · nickel superalloys solution treated at 1,050–1,200 °C and aged 8–24 hours at 700–850 °C · 6061-T6 yielding about 275 MPa and 7075-T6 about 500 MPa · natural aging to T4 taking 4–5 days at room temperature.
ExamplesAluminum aircraft structures, nickel-superalloy turbine components, titanium aerospace parts, copper-alloy springs.
VideosHeat Treatment of Aluminum: Introduction (Thermal Processing) · Heat Treatment of Aluminum: Artificial Aging (Thermal Processing)
Carbon, or carbon plus nitrogen, diffuses into steel before hardening.
Strengths & weaknessesCarburizing creates a hard wear-resistant case over a tough core. Treatment is slow and may cause distortion.
When to useCarburize low-carbon and low-alloy steels when high contact stress demands a deep hard case (roughly 0.5 to 2 mm at 58–62 HRC) over a ductile core. That is the classic recipe for gear teeth, cam lobes, and pins that see both rolling contact and shock. Carbonitriding runs cooler and suits thin cases on cheap plain-carbon parts like fasteners and stampings at high volume. Because parts are quenched from temperature, expect distortion and plan a post-hardening grind on precision features. If the part has to go in finished-machined, or the alloy already contains nitride formers, use nitriding instead for its lower temperature and minimal distortion. If you need selective hardening of medium-carbon parts without furnace batch times, induction hardening is the usual alternative.
Key numbersGas carburizing at 900–950 °C, low-pressure vacuum carburizing up to about 1,050 °C, carbonitriding cooler at 780–870 °C · case depth 0.5–2 mm, roughly 4 hours per 0.5 mm at 925 °C and growing with the square root of time · surface carbon of about 0.8–1.0%, giving 58–62 HRC after quench · core hardness typically 30–45 HRC · carbonitrided cases much shallower at roughly 0.075–0.75 mm · furnace cycles of 4–30 hours.
ExamplesGears, pins, camshafts, transmission components.
VideosCarburizing Process Guide (Xometry) · Case Hardening of Steel (tec-science)
Nitrogen diffuses into the surface and forms hard nitrides.
Strengths & weaknessesNitriding provides wear and fatigue resistance with relatively low distortion and often no quench. Case depth is limited, and alloy compatibility matters.
When to useNitride when the part has to be hardened after finish machining. The process runs around 500–550 °C with no quench, so precision gears, crankshafts, and mold cavities come out essentially at size. It requires steels with nitride-forming alloying (Cr-Mo grades, Nitralloy, tool steels, some stainless), and it delivers very high surface hardness with excellent fatigue and scuffing resistance. The diffusion case is shallow, though, typically under 0.5 mm even after long cycles. If heavy Hertzian contact or impact needs a deep supporting case, carburize instead. Plain low-carbon steels respond poorly, and long furnace times make nitriding expensive for deep cases.
Key numbersProcess temperature 500–550 °C with no quench · case depth typically 0.1–0.5 mm · surface hardness roughly 700–1,200 HV depending on the alloy's nitride formers · compound (white) layer 5–20 µm thick, often lapped or ground off · furnace cycles of 10–100 hours, with case growth following the square root of time · dimensional growth of only a few um.
ExamplesGears, crankshafts, molds, dies.
VideosNitriding Process Guide (Xometry) · Nitriding of Steel (tec-science)
Electromagnetic induction rapidly heats selected surface regions before quenching.
Strengths & weaknessesInduction hardening is fast, local, and controllable. Coil design is geometry-specific, and the material has to be conductive and hardenable.
When to useUse induction hardening for medium- and high-volume production of medium-carbon steel parts (roughly 0.4–0.5% C or hardenable alloy grades) where a specific surface, such as a journal, tooth flank, or raceway, needs 55–62 HRC while the rest of the part stays tough and soft for machining. It is the usual choice for shafts and axles at volume, because cycle times run in seconds, the equipment integrates in-line, and frequency selection gives precise control over case depth. The coil and process development cost only pays off on repeating geometry, so use flame hardening for one-offs, very large parts, or field work. For complex 3D surfaces that a coil cannot couple to uniformly, or for low-carbon steels, use carburizing.
Key numbersFrequencies from about 1 kHz for 3–6 mm cases up to 200–400 kHz for cases under 1 mm · case depth typically 0.5–6 mm · surface hardness 55–62 HRC on 0.4–0.5% carbon steels · surface power density roughly 1–5 kW/cm² · heat-and-quench cycles of 1–10 seconds per feature · machine ratings from about 10 kW to several hundred kW.
ExamplesGear teeth, shafts, rails, bearing surfaces.
VideosInduction hardening applications guide (ENRX) · Induction Hardening: Understanding the Basics (Heat Treat Today)
A flame rapidly heats selected steel surfaces before quenching.
Strengths & weaknessesEquipment is simple and useful for large parts. Uniformity and control are weaker than with induction hardening.
When to usePick flame hardening for large or one-off medium-carbon steel parts where an induction coil is impractical or unjustifiable (machine-tool ways, big ring gears, rolls, sprockets repaired in a maintenance shop). Tooling is little more than a torch, so it works well at low volumes and on parts too big for a furnace or coil, and case depths of 1–6 mm are achievable. Accept looser control of case depth and hardness uniformity, and keep it away from thin sections that overheat easily. If volumes climb or the case-depth tolerance tightens, use induction hardening instead. If you need to harden after finish machining with minimal distortion, nitriding is the better route.
Key numbersCase depth typically 1–6 mm · surface hardness roughly 50–60 HRC on 0.4–0.6% carbon steels · oxy-acetylene flame at about 3,100 °C · progressive traverse speeds of roughly 50–300 mm/min · torch equipment at a few thousand dollars against six figures for an induction cell · no practical upper limit on part size.
ExamplesLarge gears, rollers, machine ways.
VideosFlame Hardening of Steels, ASM Handbook Vol. 4A (ASM International)
Surface Stress Treatment
add compressive stress to resist fatigue2 processesSmall spherical media bombard a surface and create compressive residual stress.
Strengths & weaknessesImproves fatigue and stress-corrosion resistance. Poor control can damage surfaces or alter dimensions.
When to useSpecify shot peening on any steel or aluminum part whose life is fatigue-limited at the surface (springs, gear roots, welded joints, machined fillets). A compressive layer of roughly 0.1–0.3 mm typically raises fatigue strength by tens of percent, which at a fixed stress level buys 2–10x the fatigue life, for pennies per part. It is cheap enough to apply broadly at medium and high volume, but ask for Almen-strip intensity and coverage control on anything critical, since uncontrolled peening can do more harm than good. Avoid it on surfaces with tight finish or dimensional requirements unless a post-peen lapping step is planned, and on thin sections that will distort. If you need a deeper compressive layer of 1 mm or more, as on turbine airfoils or aircraft primary structure, use laser peening instead.
Key numbersCompressive layer roughly 0.1–0.3 mm deep · peak compressive stress around 50–60% of the material's tensile strength · Almen A-strip intensity typically 0.006–0.024 in of arc height, at 100–200% coverage · shot diameter 0.1–1.5 mm · surface roughness rising to roughly Ra 2–6 µm · cost of pennies to a few dollars per part at volume.
ExamplesSprings, gears, turbine blades, landing gear.
VideosShot Peening Overview (Electronics Inc. / shotpeener.com) · Shot peening technical library (The Shot Peener)
High-energy laser pulses generate shock waves that create deep compressive stress.
Strengths & weaknessesProvides deeper fatigue enhancement than shot peening. Equipment and processing costs are high.
When to useReserve laser peening for high-value, fatigue- or FOD-critical components where the 1–2+ mm deep compressive layer (several times deeper than shot peening) measurably extends life or crack tolerance: turbine blade leading edges, aircraft structural details, nuclear and stress-corrosion-critical welds. It also leaves a smoother surface and holds its residual stresses better at elevated temperature, which matters for engine hardware. The cost per treated area is orders of magnitude above shot peening, so it only makes economic sense where the cost of failure is extreme or where shot peening has already proven insufficient. For routine springs, gears, and general fatigue enhancement, use controlled shot peening instead.
Key numbersPulse energy of roughly 3–10 J in 8–30 ns, giving power densities of 5–10 GW/cm² · shock pressures of about 1–10 GPa at the surface · compressive layer 1–2 mm deep, several times what shot peening reaches · spot sizes of 3–7 mm applied at a few to 10 pulses per second · cost per treated area one to two orders of magnitude above shot peening.
ExamplesTurbine blades, aircraft structures, nuclear components.
VideosCoating & Surface Conversion
deposit or grow protective and decorative layers9 processesAn electrochemical process grows a controlled oxide layer on aluminum, titanium, or similar metals.
Strengths & weaknessesImproves corrosion resistance, wear, paint adhesion, insulation, and appearance. Changes dimensions slightly and may reduce fatigue performance.
When to useAnodizing is the default finish for machined and extruded aluminum exposed to handling or weather: Type II sulfuric for corrosion protection and dyed cosmetics, Type III hardcoat at 25–75 µm for wear surfaces, and chromic or thin-film types for fatigue-sensitive aerospace parts. Account for growth of roughly half the coating thickness on toleranced features, and mask threads and bearing bores. Avoid anodizing where electrical grounding or EMI bonding is needed, since the oxide is an insulator; use a chromate conversion coating there instead. Derate fatigue-critical parts or specify a thin coating, because hardcoat in particular reduces fatigue strength. Highly loaded 2xxx alloys and castings with high silicon anodize poorly and look blotchy when dyed.
Key numbersType II coatings 5–25 µm thick, Type III hardcoat 25–75 µm · dimensional growth of about half the coating thickness per surface · sulfuric baths at 15–20% acid, near 20 °C for Type II and 0–5 °C for hardcoat, at 1–4 A/dm² for 20–60 minutes · hardcoat surface hardness of roughly 400–600 HV · dielectric strength around 500–2,000 V · fatigue strength penalty of roughly 10–60% on thick hardcoat.
ExamplesAerospace parts, enclosures, bicycles, architectural aluminum.
VideosWhat is anodizing and how does it work? (Hubs / Protolabs Network) · Anodizing 101 design guide (Aluminum Anodizers Council)
A chemical reaction converts the surface into a protective compound layer.
Strengths & weaknessesImproves corrosion resistance and paint adhesion at low cost. Protection may be limited without a topcoat, and some chemistries create environmental burdens.
When to useUse conversion coating as the standard pretreatment under paint, powder coat, or e-coat (zinc phosphate or zirconium on steel bodies and panels, chromate or trivalent chem film on aluminum). As a standalone finish, use it only for mild indoor exposure or where electrical conductivity must be preserved, as with chem-filmed aluminum electronics chassis. It is cheap, adds no meaningful thickness, and runs at high volume in dip or spray lines. Do not rely on it alone in outdoor or marine service; bare conversion coatings fail salt-spray tests that anodize or galvanizing pass easily, so specify a topcoat or switch processes. Check REACH and customer restrictions before specifying hexavalent chromate.
Key numbersCoating weights of roughly 1.5–4 g/m² for zinc phosphate, 0.3–0.8 g/m² for iron phosphate, and 20–100 mg/m² for zirconium pretreatments · film thickness typically under 1 µm, with no meaningful dimensional change · zinc phosphate baths at 45–60 °C with 1–3 minute dwell, zirconium systems at ambient · MIL-DTL-5541 chem film on aluminum passing 168 hours of neutral salt spray bare · Class 3 chem film holding contact resistance under 5,000 microhms per square inch.
ExamplesAutomotive bodies, fasteners, appliance panels.
VideosWhat is chem film / chromate conversion coating? (Best Technology) · Phosphate Conversion Coatings (Products Finishing)
Electric current deposits metal ions onto a conductive surface.
Strengths & weaknessesProvides decorative, corrosion-resistant, conductive, or wear-resistant surfaces. Bath control, waste treatment, thickness uniformity, and hydrogen embrittlement are concerns.
When to useElectroplate when a conductive part needs a thin functional metal layer at low cost per piece: zinc or zinc-nickel at 5–25 µm for fastener corrosion protection, nickel-chrome for decorative hardware, hard chrome for wear, and gold, silver, or tin for contacts and solderability. Barrel plating makes it extremely cheap for small parts at high volume, and rack plating handles larger geometry. Current-density hot spots mean edges and corners plate thick while recesses starve, so if you need uniform coverage on complex shapes or internal passages, use electroless plating instead. On high-strength steels above about 1000 MPa or 32 HRC, require a post-plate hydrogen embrittlement bake within hours of plating, or pick a mechanical or vapor-deposited coating.
Key numbersTypical thicknesses of 5–25 µm for zinc, 0.05–2.5 µm for gold contacts, and 25–500 µm for hard chrome · current density 1–10 A/dm², giving deposition rates around 10–50 µm per hour · cathode efficiency from about 95% on acid zinc down to 10–25% on hard chrome · thickness on high-current edges commonly 2–5 times that in recesses · post-plate embrittlement bake at 190–220 °C for 4–24 hours, started within 4 hours of plating · zinc-nickel reaching 720 hours or more of neutral salt spray before red rust, against roughly 96–200 hours for plain zinc.
ExamplesZinc fasteners, copper interconnects, nickel coatings, chrome hardware.
VideosZinc Electroplating (Products Finishing) · Nickel Electroplating (Products Finishing)
Chemical reduction deposits metal without externally applied current.
Strengths & weaknessesCoats complex geometry more uniformly than electroplating and can coat activated nonconductors. Baths are expensive and chemistry-sensitive.
When to useChoose electroless nickel when thickness uniformity on complex geometry is the requirement. Internal passages, threads, deep recesses, and valve bodies plate to within a few microns everywhere, which no electroplating bath can match. It works on nonconductors too, like ABS or a printed circuit board that must be metallized. High-phosphorus EN also brings excellent corrosion resistance and, after a 400 °C bake, hardness approaching hard chrome. The bath costs several times more per micron than electrolytic nickel and plates slowly, so for simple external geometry at volume, plain electroplating is cheaper. For thick wear coatings on accessible surfaces, hard chrome or thermal spray is usually the better choice.
Key numbersDeposition rate of roughly 10–25 µm per hour in baths held at 85–95 °C · thickness typically 5–50 µm, uniform to within a few um over threads, bores, and recesses · phosphorus content of 1–4% for hardness and 10–13% for corrosion resistance · as-deposited hardness 500–600 HV, rising to 900–1,050 HV after an hour at 400 °C · bath life of roughly 4–8 metal turnovers · cost per micron several times electrolytic nickel.
ExamplesNickel-coated plastic, circuit boards, internal passages.
VideosAn Overview of Electroless Nickel Plating (Products Finishing) · Process Specification for Electroless Nickel Plating, PRC-5007 (NASA)
Charged polymer powder is sprayed onto a part and baked into a continuous film.
Strengths & weaknessesProduces durable, thick, low-solvent finishes with good material utilization. Cure temperature and coating thickness in recesses can be limiting.
When to usePowder coating is the default finish for fabricated steel and aluminum that can survive a 150–200 °C oven: enclosures, frames, wheels, and outdoor furniture get a tough 60–120 µm film with near-zero VOCs and 95+ percent material utilization on a reclaim line. It does better than liquid paint on chip resistance, edge coverage, and cost per part once a line is running. Rule it out for heat-sensitive substrates like most plastics, wood, and assemblies with seals or electronics, for parts needing films under about 25 µm or Class A automotive color-matched finishes, and for deep recesses where Faraday-cage effects starve coverage. Liquid spray painting handles all of those. Very large one-off structures also favor liquid paint applied in place.
Key numbersFilm thickness typically 60–120 µm, against a practical floor near 25 µm · cure at 150–200 °C metal temperature for 10–20 minutes · material utilization above 95% with powder reclaim, and effectively zero VOC · powder at roughly $4–10 per kg covering 8–12 m² per kg at 60 µm · salt-spray life of roughly 500–1,000 hours over a proper pretreatment.
ExamplesAppliances, outdoor furniture, wheels, enclosures.
VideosPowder Coated Tough magazine (Powder Coating Institute) · User's Guide to Powder Coating (SME, via P2 InfoHouse)
Liquid coating is atomized or otherwise applied and then dried or cured.
Strengths & weaknessesOffers broad material, color, and substrate compatibility. Solvents, overspray, curing, surface preparation, and durability add complexity.
When to useReach for liquid paint when the substrate cannot take a powder-cure oven (plastics, composites, finished assemblies), when the job needs thin films, precise color matching, metallics, or Class A gloss, or when coating must happen in the field or on structures too large for any booth. Two-component urethanes and epoxies give good chemical and UV durability when specified properly, and low-temperature or air-dry systems cover almost any material. Accept the costs: surface prep drives most of the outcome, overspray wastes 30–50% of material without electrostatic assist, and VOC permitting constrains shop setup. For simple durable coverage of bare metal parts at volume, powder coating is cheaper and tougher, so use it whenever the part tolerates the oven and the finish requirements allow.
Key numbersFilm builds of roughly 25–100 µm per coat, with multi-coat systems totaling 100–300 µm · transfer efficiency about 50–70% with conventional spray and 70–95% with electrostatic assist · cure from ambient air dry to a 30–60 minute force cure at 60–80 °C · VOC limits on industrial coatings commonly 250–420 g/L · abrasive blast to a 25–75 µm anchor profile before high-performance coatings · automotive body systems of four to five layers totaling roughly 100 µm.
ExamplesVehicles, aircraft, furniture, machinery.
VideosThe Basics of Airless Spraying (Graco) · Spray Application: Techniques to Optimize Film Build (KTA-Tator)
Molten or softened particles are propelled onto a surface. Variants include plasma spray, flame spray, arc spray, and high-velocity oxygen-fuel spraying.
Strengths & weaknessesApplies thick wear-, heat-, or corrosion-resistant coatings without heating the full part. Bond strength, porosity, line-of-sight access, and surface preparation are critical.
When to useUse thermal spray when a part needs a thick engineered surface (0.1 to several millimeters of carbide, ceramic, or metal) that plating or PVD cannot build, or when restoring worn dimensions on expensive shafts and rolls is preferable to replacing them. HVOF tungsten carbide is the standard hard-chrome replacement on hydraulic rods and landing gear, plasma spray is the usual choice for ceramic thermal-barrier and dielectric coatings, and arc spray covers cheap large-area zinc and aluminum corrosion protection. The substrate stays below about 150 °C, so heat-sensitive parts are safe. Avoid it for internal bores and shadowed geometry the torch cannot see, for thin precise films under 50 µm where PVD or plating is the right tool, and for coatings that must survive point impact or bending, since spray coatings are mechanically bonded and can spall.
Key numbersCoating thickness from 0.1 mm to several mm · particle velocities of roughly 50–100 m/s for flame spray, 200–400 m/s for plasma, and 600–1,000 m/s for HVOF · porosity around 0.5–2% for HVOF, 1–5% for plasma, and 5–15% for arc and flame spray · bond strength of roughly 20–40 MPa for flame spray and 70 MPa or more for HVOF · deposition rates of 2–6 kg/h for HVOF and 10–50 kg/h for arc spray · substrate held below about 150 °C.
ExamplesTurbine coatings, hydraulic rods, engine cylinders, repaired shafts.
VideosWhat Is Thermal Spray? (Oerlikon Metco) · ASM Handbook Vol. 5A: Thermal Spray Technology (ASM International)
Steel is immersed in molten zinc.
Strengths & weaknessesProvides robust long-term corrosion protection. Coatings are thick, appearance may be uneven, and bath size limits component size.
When to useGalvanize structural steel destined for decades of outdoor exposure with no maintenance (guardrail, transmission towers, fence posts, embedded anchors). The 50–100+ um zinc layer and its sacrificial protection outlast any paint system at similar cost per square meter. Design for the process: drainage and vent holes, parts that fit the kettle, and clearance on threads, which typically need retapping or oversize taps. Avoid it on precision or tight-tolerance parts, thin sheet that warps at the 450 °C bath temperature, and cosmetic surfaces where the spangled, uneven finish is unacceptable. Small fasteners at moderate exposure use zinc electroplating, sheet products use continuous mill galvanizing, and appearance-critical work gets duplex paint-over-galvanizing or powder coat instead.
Key numbersCoating thickness of 50–100 µm and above, or 610 g/m² for ASTM A123 grade 100 · bath at about 445–455 °C with 3–10 minutes of immersion · zinc corroding at roughly 1–2 µm per year in rural air and 4–8 µm per year in industrial or marine air, so 30 to 70+ years to first maintenance · vent and drain holes sized at 25–30% of tube diameter · kettles up to roughly 15–20 m long setting the maximum single-dip part size.
ExamplesGuardrails, structural steel, utility hardware.
VideosDesign of Products to be Hot-Dip Galvanized After Fabrication (American Galvanizers Association) · Hot-Dip Galvanizing for Corrosion Protection: A Specifier's Guide (American Galvanizers Association)
Chemical treatment removes free iron and promotes a protective oxide layer, especially on stainless steel.
Strengths & weaknessesImproves corrosion resistance without materially changing dimensions. Cannot compensate for poor alloy choice, severe contamination, or bad surface finish.
When to usePassivate every machined, ground, or welded stainless part before service, because free iron picked up from tooling otherwise seeds rust spots. In medical, food, pharma, and aerospace work it is mandatory, spec-driven practice (ASTM A967/AMS 2700). Citric acid chemistry now covers most needs with fewer safety and disposal issues than nitric. Passivation is a cleaning and oxide-optimization step rather than a coating: dimensions are untouched, so it is safe on finished precision parts. Do not expect it to compensate for 400-series or free-machining grades in chloride service, or to fix heat tint and weld scale, which need pickling or electropolishing first. If the alloy is marginal for the environment, change the alloy instead of relying on passivation.
Key numbersCitric acid 4–10% by weight or nitric acid 20–50% by volume · bath 20–70 °C · immersion 4–30 minutes depending on the ASTM A967 method · passive oxide film 1–3 nm thick · dimensional change well under 1 µm · batch tanks hold dozens to hundreds of parts at once.
ExamplesMedical tools, food equipment, pharmaceutical vessels.
VideosWhat Is the Passivation Process? (Astro Pak) · ASTM A967 Passivation Standard Explained (Able Electropolishing)
Glass Strengthening
toughen glass against impact and breakage2 processesGlass is heated near its softening point and rapidly cooled, placing its surfaces in compression.
Strengths & weaknessesSignificantly increases strength and causes safer fragmentation. Tempered glass cannot be cut afterward, and optical distortion may increase.
When to useTemper flat soda-lime glass 3 mm and thicker whenever codes demand safety glazing or the design needs roughly four times annealed strength (doors, shower enclosures, vehicle side windows, facades). It is by far the cheapest strengthening route at volume. All cutting, drilling, and edging must happen before tempering, so finalize geometry first. It cannot handle thin glass, which lacks the thermal mass to build the stress profile, and the quench pattern and roller wave rule it out for display covers and precision optics; use chemical strengthening for those. If post-breakage retention matters, as in overhead glazing, use laminated or heat-strengthened constructions rather than fully tempered lites, and specify heat soaking when spontaneous nickel-sulfide breakage is a liability concern.
Key numbersThickness 3–19 mm, with about 3 mm the practical minimum · surface compression at least 69 MPa fully tempered and 24–52 MPa heat-strengthened · roughly 4× the bending strength of annealed glass · furnace around 620–650 °C followed by an air quench · heating time roughly 40 seconds per mm of thickness · roller wave distortion typically 0.05–0.15 mm peak to valley.
ExamplesArchitectural glazing, side windows, shower doors, oven doors.
VideosHeat-Strengthened vs. Tempered Glass (Vitro Glass Education Center) · TD-138: Heat-Treated Glass for Architectural Glazing (Vitro Architectural Glass)
Larger ions replace smaller ions near the glass surface through ion exchange, generating compression.
Strengths & weaknessesProvides high strength in thin or complex glass with relatively little optical distortion. Slower, more expensive, and creates a shallow strengthened layer vulnerable to deep scratches.
When to useChemically strengthen when the glass is too thin to temper (anything under about 3 mm, including the 0.4–1 mm covers on phones and wearables), or when optical quality and complex 2.5D or curved shapes rule out a tempering quench. The process needs ion-exchangeable aluminosilicate or similar compositions plus hours in a hot salt bath, so cost per part is much higher than tempering. The compression layer is only tens of microns deep, so design the part to avoid scratches and expect deep gouges to penetrate past it. If you have flat soda-lime glazing 3 mm and up where cost dominates and code-compliant fragmentation is required, thermal tempering is still the default; chemical strengthening does not produce safety-glass dicing.
Key numbersGlass thickness 0.3–3 mm, below the 3 mm floor for tempering · molten potassium nitrate bath at 380–450 °C for 4–16 hours · surface compression 500–1,000 MPa, several times what thermal tempering reaches · compression depth 20–50 µm in one step, up to about 100 µm with two-step exchange · cost per part several times thermal tempering at the same size.
ExamplesSmartphone glass, watches, avionics displays, medical devices.
VideosIon-Exchange Strengthening Process Overview (Saxon Glass Technologies)
Semiconductor & Microscale
pattern and build features at chip scale13 processesLight transfers a geometric pattern through a mask onto photosensitive resist.
Strengths & weaknessesEnables extremely small, repeatable features and parallel fabrication. Equipment, masks, cleanrooms, process control, and defect prevention are all very expensive.
When to usePhotolithography is the only economical way to pattern micron-and-below features across a whole wafer at once, so use it whenever device counts reach the thousands and geometry fits in stacked planar layers (ICs, MEMS, photonics, microfluidic molds). Contact and proximity printing with cheap chrome or even film masks serves university-scale MEMS down to a few microns; projection steppers and scanners take over below that, with mask sets running from thousands of dollars to millions for advanced nodes. If you are building one-offs or the design is still changing, the mask cost is hard to justify, and maskless direct-write laser or e-beam lithography gives up throughput in exchange for zero tooling. For microfluidic prototypes, write one master and replicate it in PDMS. Nonplanar substrates and features above ~100 µm are usually better served by machining or printing processes.
Key numbersFeature sizes from a few microns with contact printing down to about 13 nm half-pitch with EUV · exposure wavelengths 365 nm i-line, 193 nm ArF immersion, 13.5 nm EUV · overlay 1–3 nm on leading-edge scanners · throughput 150–300 wafers per hour · tool cost from a few hundred thousand dollars for a contact aligner to $150–350M for an EUV scanner · mask sets from a few thousand dollars for MEMS work to well over $10M at advanced nodes.
ExamplesIntegrated circuits, MEMS, sensors, microfluidics.
VideosSemiconductor Lithography Basics (Chris Mack, Lithoguru) · Lithography Principles (ASML)
Liquid chemicals selectively dissolve exposed material. Wet etching may be isotropic or crystallographically anisotropic.
Strengths & weaknessesInexpensive, high-throughput, and chemically selective. Isotropic etchants cause lateral undercutting; anisotropic KOH or TMAH silicon etching is constrained by crystal planes and produces characteristic angled sidewalls.
When to useUse wet etching when features are larger than a few microns and undercut of roughly the etch depth is tolerable (blanket film stripping, PCB and metal patterning, oxide removal). It is cheap because batch immersion processes dozens of wafers or panels at once for the cost of chemicals, and selectivity between materials can be effectively infinite. KOH or TMAH bulk micromachining is the cheap route to deep cavities, membranes, and V-grooves when the 54.7-degree sidewalls of (100) silicon are acceptable or even wanted. If the design needs vertical sidewalls, sub-3 µm features, tight CD control, or geometry independent of crystal orientation, move to dry plasma etching, and to DRIE for deep high-aspect-ratio silicon.
Key numbersPractical feature floor around 3 µm · isotropic undercut roughly equal to the etch depth · KOH etches (100) silicon at 1–1.4 µm/min at 80 °C, TMAH at about half that · KOH (100):(111) selectivity 100–400:1, which is what produces the 54.7-degree sidewall · buffered oxide etch removes thermal SiO2 at roughly 100 nm/min · batch immersion runs 25–50 wafers at once, with chemicals the main cost.
ExamplesPrinted circuit boards, silicon bulk micromachining, glass channels, metal patterning.
VideosWet-Chemical Etching of Silicon and SiO2 (MicroChemicals) · KOH Anisotropic Etch Rate Data (BYU Cleanroom)
Reactive plasma chemically and physically removes exposed material.
Strengths & weaknessesCreates anisotropic, high-resolution features with near-vertical walls. Equipment is expensive and plasma damage, selectivity, and contamination can be problematic.
When to useChoose plasma etching whenever pattern fidelity matters: submicron features, vertical sidewalls, and tight critical-dimension control that wet chemistry's undercut makes impossible. It is required for transistor-scale patterning and is the default for MEMS and photonic features below a few microns, with RIE and ICP chemistries tunable per material stack. In exchange you get single-wafer processing on tools costing hundreds of thousands to millions, plus development time to balance selectivity against the mask and underlying layers and to manage plasma-induced damage on sensitive devices. If features are tens of microns, sidewall profile does not matter, or you are stripping blanket films, wet etching does the same job at a fraction of the cost. For silicon structures deeper than ~20 µm at high aspect ratio, use DRIE.
Key numbersFeatures from sub-10 nm at leading-edge nodes up to tens of microns · etch rate 0.1–1 µm/min, higher with ICP sources · sidewall angles 88–90 degrees · selectivity to photoresist usually 3:1 to 10:1 · chamber pressure 1–100 mTorr · single-wafer tools $300k to several million dollars.
ExamplesTransistor gates, MEMS structures, microfluidics, optical gratings.
VideosReactive Ion Etching Technology Overview (Oxford Instruments)
Alternating etch and passivation steps create deep high-aspect-ratio structures.
Strengths & weaknessesEnables deep, nearly vertical silicon features. Sidewall scalloping, cost, process complexity, and material limitations remain concerns.
When to useUse DRIE when silicon has to be etched tens to hundreds of microns deep with near-vertical walls and aspect ratios of 10:1 to 30:1 or beyond (through-silicon vias, MEMS proof masses, comb drives, microphone diaphragms, and through-wafer release etches). No other process combines that depth with lithographic lateral precision and orientation independence. Budget for slow, single-wafer processing on expensive ICP tools. If surfaces matter optically or fluidically, you also have to manage the Bosch process's periodic sidewall scallops (typically 100–500 nm), and across mixed feature sizes you have to watch aspect-ratio-dependent etch lag. For shallow features under ~20 µm use standard RIE. If angled sidewalls and crystal-plane geometry are acceptable and cost dominates, KOH or TMAH wet etching is far cheaper. DRIE is essentially a silicon process; glass and compound semiconductors need other chemistries.
Key numbersEtch depth from tens of microns to through-wafer at 500–775 µm · aspect ratios 10:1 to 30:1 routinely, past 50:1 with tuned recipes · etch rate 2–10 µm/min, dropping in narrow trenches · Bosch sidewall scallops 100–500 nm · selectivity 50–100:1 to photoresist and 100–200:1 to oxide masks · single-wafer ICP tools roughly $500k to $2M.
ExamplesMEMS sensors, microphones, through-silicon vias.
VideosDRIE and the Bosch Process (Oxford Instruments) · Bosch and Cryo ICP-RIE for Silicon (Caltech KNI Lab)
Silicon is exposed to oxygen or steam at high temperature to grow silicon dioxide.
Strengths & weaknessesProduces high-quality dielectric interfaces and uniform films. Consumes silicon, requires high temperature, and is substrate-specific.
When to useGrow thermal oxide whenever the electrical quality of the Si-SiO2 interface matters (gate and tunnel oxides, high-quality passivation, and any dielectric where defect density must be minimal), and for cheap, uniform hard masks on bare silicon early in a flow. Dry oxidation gives dense, slow-growing films for thin critical oxides; wet oxidation grows thick field and masking oxides up to a couple of microns in reasonable furnace time. It runs at 900–1200 °C and consumes the substrate, so it is only usable on exposed silicon before metallization or any dopant profile that cannot tolerate the thermal budget. If either constraint applies, deposit oxide by CVD instead, or use ALD for thin conformal dielectrics on arbitrary materials.
Key numbersFurnace temperature 900–1,200 °C · dry oxidation grows 1–100 nm films at tens of nm per hour · wet oxidation reaches 0.3–2 µm at a few hundred nm per hour · silicon consumed equals about 0.44 of the oxide thickness grown · interface trap density around 10^10 cm^-2 eV^-1, the lowest of any oxide route · batch furnaces hold 50–200 wafers.
ExamplesGate oxides, isolation layers, masks, passivation.
VideosDeal-Grove Oxide Growth Calculator (BYU Cleanroom) · Micro/Nano Processing Technology (MIT OpenCourseWare)
Accelerated dopant ions are driven into a semiconductor substrate and later activated by annealing.
Strengths & weaknessesProvides precise control of dose and depth. Equipment is expensive and implantation damages the crystal lattice.
When to useImplantation is the standard doping method for any modern device flow. Choose it when dose must be controlled to a few percent and depth set independently by beam energy, when doping through a masking layer at room temperature, or when the profile you need (a retrograde well, a shallow source/drain extension, a buried layer) cannot be obtained by diffusion from the surface. Every implant needs a subsequent activation anneal, so check that the thermal budget fits the rest of the stack, and use rapid thermal annealing when shallow junctions must not diffuse. For deep, cheap, uniform doping where profile shape is uncritical (power-device drift regions, solar cell emitters at commodity cost), furnace diffusion still has the better throughput and the lower capital cost. Heavy-dose amorphization damage in compound semiconductors may also push you to doped epitaxial growth instead.
Key numbersBeam energy 0.2 keV for shallow extensions up to about 3 MeV for deep wells · dose 10^11 to 10^16 ions per cm2 · dose uniformity typically within 1% across a wafer · projected range 10 nm to a few um, set by energy alone · activation anneal 900–1,100 °C, seconds or less on an RTA · implanters roughly $3–8M each.
ExamplesSource/drain regions, wells, threshold adjustment, power devices.
VideosIon Implantation in Semiconductor Processing (MKS Instruments)
A crystalline layer is grown on a crystalline substrate in a defined orientation.
Strengths & weaknessesEnables high-purity, compositionally engineered semiconductor layers. Growth is slow, expensive, and sensitive to defects and lattice mismatch.
When to useUse epitaxy when the device physics lives in the grown layer: heterostructures and quantum wells for LEDs and lasers, GaN-on-Si or SiC drift layers for power devices, SiGe and III-V layers for RF, and lightly doped device layers over heavily doped substrates that no implant or diffusion can replicate. It is the only way to get single-crystal material with independently chosen doping and composition. MOCVD is the standard production tool for compounds, and MBE is mostly reserved for research-grade abruptness. Watch the lattice and thermal mismatch budgets: if you exceed the critical thickness, dislocations form and yield drops. If the layer only needs to be a dielectric, polycrystalline, or amorphous film, ordinary CVD or PVD is faster and far cheaper, so use epitaxy only where crystal quality matters to the device.
Key numbersLayer thickness 0.1–100 µm · silicon CVD epitaxy grows 0.1–5 µm/min at 900–1,150 °C, MOCVD 1–5 µm/hr, MBE around 1 µm/hr · doping controllable from 10^14 to 10^20 cm^-3 · thickness and composition uniformity typically within 1–2% across a wafer · threading dislocations below 1 cm^-2 for homoepitaxial silicon and 10^8–10^9 cm^-2 for GaN on silicon · reactors roughly $2–5M.
ExamplesLEDs, lasers, RF devices, power semiconductors, advanced logic.
VideosMaterial is physically vaporized and condensed as a thin film.
Strengths & weaknessesProduces high-purity metal, ceramic, and optical films. Vacuum-intensive and often line-of-sight.
When to useReach for PVD when you need a pure, dense, well-adhered thin film (nanometers to a few microns) of a metal, nitride, or optical material at low substrate temperature: sputtered metallization and barrier stacks, TiN and DLC tool coatings, evaporated optical and decorative layers. Sputtering handles alloys and compounds with good composition control and works on plastics and other temperature-limited substrates where CVD cannot go. Deposition is line-of-sight, so step coverage is poor; avoid PVD for deep trenches, internal surfaces, and high-aspect-ratio features and use CVD or ALD instead. Deposition rates are also low, so films beyond ~10 µm are uneconomical and electroplating or thermal spray is the better choice.
Key numbersFilm thickness 10 nm to about 5 µm, with anything past 10 µm uneconomical · deposition 0.1–1 µm/min in production sputter tools, an order of magnitude slower in small research systems · chamber pressure 10^-3 to 10^-2 mbar for sputtering and 10^-6 mbar for evaporation · substrate temperature from near room temperature to about 500 °C · tool coatings 1–5 µm thick at 2,300–3,300 HV for TiN and TiAlN · step coverage under 20% in high-aspect-ratio features.
ExamplesSemiconductor metallization, tool coatings, mirrors.
VideosWhat Is Sputtering? (Semicore Equipment) · Magnetron Sputtering Explained (Angstrom Engineering)
Reactive gases form a solid film on a heated substrate.
Strengths & weaknessesProduces high-quality conformal coatings. Processes may require high temperatures, hazardous gases, and complex equipment.
When to useChoose CVD when the film must be conformal over topography or must be a material best formed from gas-phase chemistry: polysilicon, silicon nitride, oxide interlayers, tungsten plugs, SiC, and diamond coatings. Batch furnace LPCVD coats hundreds of wafers uniformly and cheaply; PECVD trades some film quality for deposition at 200–400 °C when metallized or temperature-limited substrates rule out the 600–900 °C thermal processes. Handling toxic and pyrophoric precursors is a facility-level commitment, so plan for it. If the substrate cannot tolerate even plasma-assisted temperatures, or the film is a simple metal, sputtered PVD is easier. If conformality requirements reach extreme aspect ratios or thickness control has to be sub-nanometer, use ALD instead and accept its slow deposition rate.
Key numbersFilm thickness 10 nm to several microns · LPCVD deposits 5–20 nm/min at 600–900 °C and 0.1–2 Torr, PECVD 50–300 nm/min at 200–400 °C · step coverage 80–100% for LPCVD versus 40–80% for PECVD · batch LPCVD furnaces hold 100–200 wafers · CVD diamond grows at 0.1–10 µm/hr · tools roughly $500k to $3M.
ExamplesSemiconductor films, diamond coatings, silicon carbide.
VideosChemical Vapour Deposition Overview (Oxford Instruments) · PECVD Technology Explained (Oxford Instruments)
Sequential self-limiting reactions deposit one atomic-scale layer at a time.
Strengths & weaknessesProvides very tight thickness control and excellent conformality. Deposition is slow and precursor chemistry is expensive.
When to useUse ALD when the film is thin (typically 1 to 50 nm) and has to be perfectly conformal or thickness-controlled at the single-nanometer level: high-k gate dielectrics, liners and barriers in high-aspect-ratio vias and 3D NAND structures, pinhole-free moisture barriers on OLEDs, and protective coatings inside porous electrodes. Its self-limiting chemistry coats aspect ratios beyond 100:1 that no CVD process reaches, and low process temperatures suit polymers and sensitive devices. Growth runs at roughly 0.1 nm per cycle, so anything beyond ~100 nm is very slow and expensive in precursor. Use CVD for thick conformal films and PVD for simple planar metals, and consider spatial ALD only when a specific thin film has to cover large areas quickly.
Key numbersFilm thickness typically 1–50 nm · growth about 0.1 nm per cycle with cycles of 0.5–10 seconds, so roughly 30–300 nm per hour · conformality above 95% at aspect ratios past 100:1 · thickness uniformity within about 1% across a wafer · deposition temperature 50–350 °C, low enough for polymers · tools roughly $300k to $2M, with hafnium and zirconium precursors at thousands of dollars per kilogram.
ExamplesGate dielectrics, battery coatings, nanoscale barriers.
VideosHow ALD Works (Beneq) · New Development of Atomic Layer Deposition: Processes, Methods and Applications (Science and Technology of Advanced Materials)
A wafer is polished using a reactive slurry and mechanical pad to create a highly flat surface.
Strengths & weaknessesEnables multilayer semiconductor structures. Scratching, dishing, erosion, contamination, consumable cost, and uniformity are major challenges.
When to useCMP is required wherever a process flow stacks patterned layers that must each start flat. Copper damascene interconnects, shallow-trench isolation, tungsten plugs, and the bonding-grade surfaces needed for hybrid wafer bonding all depend on it, and any design with more than a couple of metal levels cannot avoid it. Use it, too, whenever a later lithography step's depth of focus cannot absorb accumulated topography. It is expensive in consumables and sensitive to layout, so enforce layout density rules and dummy fill to control dishing and erosion, and budget for slurry, pads, and metrology. If the flow is simple (few layers, relaxed lithography, MEMS that tolerate topography), skip CMP and use resist etch-back or spin-on planarization for mild smoothing instead.
Key numbersRemoval rate 100–500 nm/min on oxide and 300–1,000 nm/min on copper · post-polish roughness under 0.5 nm RMS, below 0.2 nm for bonding-grade surfaces · within-wafer thickness non-uniformity typically 2–5% · dishing and erosion 20–100 nm on wide features · pads last 300–1,000 wafers and consumables run several dollars per wafer pass · tools $2–5M.
ExamplesCopper interconnects, dielectric planarization, shallow-trench isolation.
VideosChemical-Mechanical Polishing Process Overview (Fraunhofer ISIT)
Two wafers are joined through fusion, anodic, adhesive, or metallic bonding.
Strengths & weaknessesEnables sealed cavities, multilayer structures, and complex devices. Surfaces must be extremely flat and clean.
When to useBond wafers when the device fundamentally requires joined substrates: hermetic vacuum cavities for gyros and resonators, SOI material, capping of MEMS before dicing, backside-illuminated image sensor stacks, and 3D die stacking via hybrid bonding. Pick the variant by constraint. Fusion bonding gives the highest strength when both wafers tolerate high anneal temperatures and sub-nanometer roughness is achievable. Anodic bonding suits glass-to-silicon MEMS sealing at moderate temperature. Adhesive or eutectic bonding works when surfaces are rough, patterned, or temperature-limited. The flatness and particle requirements are severe: a single micron-scale particle voids square millimeters, so budget for CMP-grade surfaces and meticulous cleaning. If you only need package-level sealing or attachment of a finished die, conventional packaging and die-attach processes are far cheaper than wafer-level bonding.
Key numbersSurface roughness under 0.5 nm RMS and bow within a few tens of microns for fusion bonding · anneal temperature 1,000–1,100 °C for fusion, 300–450 °C at 200–1,000 V for anodic, 363 °C for Au-Si eutectic, 200–250 °C for adhesive · bond surface energy 1–2.5 J/m2 after a full fusion anneal · alignment 0.5–2 µm on MEMS aligners and under 200 nm for hybrid bonding · a single micron-scale particle voids square millimeters of bond area · bonders $1–3M.
ExamplesMEMS sensors, microfluidics, image sensors, 3D integration.
VideosWafer Bonding Technology Portfolio (EV Group) · Hybrid and Fusion Bonding Explained (EV Group)
Finished wafers are separated into individual dies by saw, laser, plasma, or stealth dicing.
Strengths & weaknessesMechanical sawing is mature and inexpensive; laser and plasma methods support thinner wafers and narrower streets. Chipping, kerf loss, debris, and thermal damage can reduce yield.
When to useDefault to blade sawing. For standard-thickness silicon with streets of 50 µm or more it is the cheapest, best-understood option. Switch to stealth (laser-internal) dicing for thin wafers below roughly 100 µm, for memory and stacked-die products where edge chipping costs too much yield, and for dry processes where cooling water would damage MEMS. Use ablation lasers for hard or brittle materials like SiC and glass, where blades wear fast. Plasma dicing is worth its cost when die are tiny and numerous, because narrow streets and simultaneous singulation of the whole wafer reclaim several percent of area and improve die strength. Avoid blade sawing on cantilevered or released MEMS structures and on wafers where microcracks would propagate in service. Choose the singulation method before street width is frozen in layout.
Key numbersBlade kerf 20–50 µm, needing streets of at least 50 µm, while stealth and plasma dicing work with 5–20 µm streets · blade feed 20–100 mm/s, while plasma singulates a whole wafer in 10–30 minutes regardless of die count · wafer thickness above 100 µm for blades and 20–100 µm for stealth dicing · front-side chipping under 20 µm on a well-set blade, near zero with stealth or plasma · die break strength roughly 300–500 MPa after sawing versus 700–1,000 MPa after plasma dicing.
ExamplesLogic chips, sensors, MEMS, power devices.
VideosPrecision Optics
figure and finish surfaces to nanometer accuracy4 processesAbrasives progressively generate and refine optical surfaces.
Strengths & weaknessesSupports many glass and crystal materials and can reach excellent figure and finish. Slow, and depends heavily on metrology and process stability.
When to useThis is the default route for spherical lenses, flats, prisms, and windows in any glass or crystal from prototype through medium volume: tooling is modest, nearly every optical material is workable, and lambda/10 figure with angstrom-level roughness is routine in a competent shop. Choose it whenever material choice matters more than geometry, since CNC generation plus pitch or pad polishing handles the vast majority of catalog optics. It gets expensive for aspheres and freeforms, where sub-aperture CNC polishing helps but deterministic MRF finishing is usually the final step. It also cannot compete at consumer volumes, where precision glass molding or polymer injection molding make optics for cents. For infrared crystals and metal mirrors, single-point diamond turning is often faster to a finished asphere.
Key numbersFigure accuracy lambda/4 commercial, lambda/10 precision, lambda/20 or better on high-end work at 633 nm · surface roughness a few nm RMS commercial, under 1 nm on precision work and 0.3 nm superpolished · scratch-dig 80-50 commercial down to 10-5 on laser optics · radius tolerance ±0.5% commercial and ±0.1% precision · diameters from a few millimeters to several meters · hours to days per surface, so lot sizes of one to a few thousand.
ExamplesLenses, telescope mirrors, prisms, windows.
VideosUnderstanding Optical Specifications (Edmund Optics) · Optical Manufacturing Tolerance Chart (Optimax)
A monocrystalline diamond tool machines optical-quality surfaces on a precision lathe.
Strengths & weaknessesProduces nanometer-scale finishes and complex aspheric or freeform geometry. Ferrous materials degrade diamond tools, and vibration and temperature control are demanding.
When to useDiamond turn when the optic is metal, crystal, or polymer and the geometry is aspheric, freeform, or off-axis. Aluminum and copper mirrors, germanium and zinc selenide infrared lenses, and the mold inserts behind molded polymer optics all come off the machine with single-digit-nanometer roughness and submicron figure, with no polishing lab required. At prototype and low-to-medium volume it is usually the fastest path from CAD to a finished asphere. Chemical tool wear rules out steel and most ferrous alloys; if you need a ferrous part, plate it with electroless nickel and turn the plating instead. Glass and other brittle materials generally still need grinding and polishing. If the system works at visible wavelengths, residual turning marks will scatter light, so plan a post-polish or use MRF. For high-volume polymer optics, use SPDT for the mold and let injection molding make the parts.
Key numbersSurface roughness 2–5 nm Ra, good enough for infrared use without polishing · form accuracy 0.1–0.5 µm peak to valley across the aperture · part diameters up to about 300–700 mm on production machines · air-bearing spindles at 1,000–10,000 rpm, feeds of 1–10 µm per revolution, depths of cut 1–20 µm · machines $400k–$1.5M and diamond tools $500–5,000 each · hours per surface, so prototype through low-thousands volume.
ExamplesInfrared optics, mirror substrates, laser components, optical molds.
VideosOverview of Single-Point Diamond Turned Optics (Apollo Optical Systems) · All About Aspheric Lenses (Edmund Optics)
Magnetically stiffened abrasive fluid removes microscopic material from selected optical regions.
Strengths & weaknessesProvides deterministic, localized correction with low subsurface damage. Removal rates are low and the method is mainly economical for high-value optics.
When to useUse MRF as the finishing step on optics whose figure spec exceeds what conventional polishing reliably delivers: aspheres and freeforms needing lambda/20 or better, laser optics where subsurface damage limits damage threshold, and imaging systems with tight mid-spatial-frequency requirements. Fed by interferometric maps, it converges in one or two deterministic runs, while pitch polishing can iterate for days. That is why it pays on high-value glass. It only removes microns, so the incoming surface must already be within a few microns of final figure from grinding and pre-polishing; it is not a bulk process. Skip it for commodity spherical lenses, where conventional polishing meets spec at far lower cost. For the very last nanometers on meter-class or EUV optics, use ion-beam figuring instead.
Key numbersFigure convergence to lambda/20 or better, often under 10 nm peak to valley, in one or two runs · total material removal 0.5–5 µm · surface roughness held under 1 nm RMS, typically 0.3–0.8 nm · part sizes from a few millimeters to about 1 m on the largest machines · run time 30 minutes to several hours per surface · machines roughly $1–2M, plus magnetorheological fluid as a consumable.
ExamplesAspheric lenses, lithography optics, telescope mirrors.
VideosAn ion beam sputters tiny quantities of material from selected areas to correct optical figure.
Strengths & weaknessesNoncontact and capable of nanometer-level correction. Slow, expensive, and vacuum-intensive.
When to useIBF is the final-correction process for the most demanding optics. Use it when the figure error budget is in single-digit nanometers RMS or below (EUV lithography mirrors, space telescope segments, synchrotron and high-energy laser optics), and when a contact process would print through on lightweighted or thin substrates. The ion beam applies no force and produces no edge roll, which is why it works on those substrates. It removes only tens to hundreds of nanometers per run inside a vacuum chamber, so the optic must arrive already polished to excellent finish and near-final figure by conventional means or MRF. If you don't need that last bit of figure correction, MRF is faster and cheaper. IBF also does not improve roughness, so pair it with smoothing steps when microroughness matters too.
Key numbersResidual figure error 0.5–2 nm RMS after correction · removal of tens to hundreds of nanometers per run · ion beam diameter 1–30 mm, which sets the smallest correctable spatial period · 80–90% of the remaining figure error removed per iteration · run time hours to tens of hours in a chamber at about 10^-4 mbar · part sizes to roughly 1.5 m, machines several million dollars.
ExamplesSpace telescope mirrors, EUV optics, high-energy laser optics.
VideosInspection & Process Control
find flaws and keep processes on spec12 processesA tactile or optical probe measures three-dimensional coordinates relative to a calibrated frame.
Strengths & weaknessesProvide accurate, traceable dimensional inspection of complex parts. Can be slow and require controlled environments, fixtures, and skilled programming.
When to useUse a CMM when tolerances are in the single-digit-micron range, when GD&T callouts (true position, profile, datum-referenced features) must be verified with metrological traceability, and for first-article and PPAP inspection where the report is the deliverable. Tactile probing is still the reference method for bores, planes, and anything where a scanner's optical noise floor is marginal. It is the wrong tool for 100% inspection at production takt unless you dedicate a shop-floor CMM to the line, for soft or flexible parts that deflect under probe force, and for dense freeform surfaces where you need millions of points. For those cases use structured-light or laser scanning, and use CT for internal features a stylus cannot reach.
Key numbersLength measurement error typically 1.5–3.5 µm + L/300 on shop-floor machines and under 1 µm + L/1,000 on lab-grade ones, with L in mm · probing error 0.3–2.5 µm depending on class · measuring volumes from 500 mm cubes to 5 m gantries · probe force 0.05–0.3 N, enough to deflect thin or soft parts · minutes to an hour per part for a full GD&T report · machines $50k to well over $500k.
ExamplesAerospace parts, molds, precision machined components, medical implants.
VideosCoordinate Measuring Machines Textbook (Mitutoyo) · CMM Measurement Strategies Good Practice Guide (NPL)
Cameras, structured light, interferometers, confocal sensors, or laser scanners measure dimensions and surfaces without contact.
Strengths & weaknessesFast and suitable for soft, delicate, or densely featured parts. Reflectivity, transparency, line of sight, and calibration can limit performance.
When to useChoose optical measurement when speed and coverage matter more than ultimate point accuracy. Structured-light and laser scanners capture millions of points in seconds for full-surface comparison against CAD, which suits stamped, molded, and additive parts, soft or flexible components a touch probe would deflect, and inline 100% inspection at production rate. Vision systems are the usual choice for small 2D features, and interferometers for surface topography. Expect tens-of-microns practical accuracy from scanners rather than the microns of a CMM, and plan on matte spray or exposure tuning for shiny, transparent, or dark surfaces. If a datum-referenced GD&T callout at single-digit-micron tolerance must be certified, or deep bores and undercuts block line of sight, fall back to a tactile CMM or industrial CT.
Key numbersStructured-light and laser scanners hold 10–50 µm accuracy and capture 12–16 million points in a few seconds per scan · 2D vision systems reach a few microns on small features with telecentric optics · interferometers and confocal sensors resolve sub-nanometer height at 0.1–1 µm lateral resolution · measurement volumes from a few millimeters to several meters · seconds per part, fast enough for 100% inline inspection · $10k for a handheld scanner to $300k for a metrology-grade interferometer.
ExamplesElectronics, molded plastic, additive parts, stamped components, optics.
VideosMeasurement System Selection Library (Keyence) · Optical Testing Course Notes, OPTI 513 (University of Arizona)
X-ray projections are reconstructed into a three-dimensional model of internal and external geometry.
Strengths & weaknessesReveals internal passages, porosity, inclusions, wall thickness, and assembled interfaces without destroying the part. Expensive, scans can be slow, and dense or large parts are difficult.
When to useUse CT when the feature of interest is internal and the part must survive: porosity mapping in castings and additively built parts, internal channels no probe can reach, wall thickness of hollow components, and assembled products (connectors, batteries, medical devices) inspected without teardown. It is also usually the fastest way to root-cause a field failure before committing to destructive sectioning. Resolution scales inversely with part size, and X-ray penetration limits how dense a metal you can scan; expect trouble beyond roughly 50 mm of steel-equivalent path. For large or dense parts, fall back to ultrasonic testing or 2D radiography. Scan cost and time make 100% CT inspection viable only for high-value or safety-critical production. If you only need external geometry, optical scanning or a CMM is far cheaper.
Key numbersVoxel size from about 1 µm on small samples to 200 µm on large ones, roughly the part's largest dimension divided by 1,000–2,000 · X-ray tubes 90–450 kV, with practical penetration around 50 mm of steel-equivalent path · scan time 5 minutes to several hours · part envelopes from 100 mm on desktop systems to 1–2 m in walk-in bays · dimensional accuracy typically 5–30 µm when calibrated against a CMM · systems roughly $200k to well over $1M, and 2–50 GB of data per scan.
ExamplesAdditive parts, castings, batteries, turbine blades, composites.
VideosWhat Is Industrial CT? (Lumafield) · CT 101 Introductory Course (Lumafield)
High-frequency sound waves detect internal flaws, interfaces, or thickness changes.
Strengths & weaknessesDetects cracks, lack of fusion, delamination, and wall loss without radiation. Geometry, grain structure, coupling, and interpretation can complicate results.
When to useMake UT the first choice for volumetric inspection of thick sections (forgings, plate, pressure-vessel welds, composite laminates), especially where single-side access rules out radiography and where the concern is planar defects like cracks and lack of fusion, which X-rays often miss. Phased-array systems produce recordable, code-accepted weld inspections without radiation permits, and thickness gauging of corroded pipe and tanks is a one-sided, in-service job that nothing else does as well. Avoid it on coarse-grained austenitic and cast structures that scatter sound, on very thin sections below a couple of millimeters, and on complex geometry where beam paths are hard to interpret; radiography handles those better. Results depend heavily on operator skill and couplant, so certified technicians and reference blocks are part of the real cost.
Key numbersFrequency 0.5–25 MHz, with 2–10 MHz standard for weld inspection · measurable wall thickness from about 1 mm with a delay-line probe up to several meters of fine-grain steel forging at low frequency · smallest reliably detected flaw about half a wavelength, so 0.5–1 mm at 5 MHz in steel · thickness gauging to ±0.1 mm, and ±0.01 mm with precision gauges · phased-array probes carry 16–128 elements · instruments $5k for a thickness gauge to $80k for a phased-array system.
ExamplesWelds, forgings, composites, pressure vessels, rails.
VideosUltrasonic Flaw Detection Tutorial (Evident) · What Is Ultrasonic Testing? (TWI)
X-rays or gamma rays create images based on material thickness and density.
Strengths & weaknessesReveals internal voids, inclusions, assembly errors, and some cracks. Radiation control, access, interpretation, and poor sensitivity to some planar defects are drawbacks.
When to useUse radiography for volumetric defects that displace material (gas porosity, shrinkage, and inclusions in castings, missing or misplaced components in assemblies and electronics) and for code-mandated weld inspection where a permanent film or digital image is required. Gamma sources cover field work on pipelines and structures without power, and digital detectors enable high-throughput inline inspection of castings and PCBs. The main blind spot is planar defects: cracks and lack of fusion oriented off the beam axis simply do not show, so use ultrasonic or surface methods for fatigue-critical crack detection. Radiography needs two-sided access, exclusion zones, and licensed operators, and thick dense sections demand high-energy sources. If 2D projections cannot resolve overlapping features, step up to CT.
Key numbersX-ray tubes 50–450 kV and linacs 1–15 MeV, with Ir-192 suiting roughly 20–65 mm of steel and Co-60 50–200 mm · flaw sensitivity 1–2% of section thickness · exposure from a few seconds on a digital detector to tens of minutes on film · digital detector pixel pitch 50–200 µm against 20–100 line pairs per mm on film · radiation exclusion boundaries of tens of meters for field gamma work · $15–60k for a portable X-ray set, $100–500k for a digital inline system.
ExamplesCastings, welds, batteries, electronics, aerospace assemblies.
VideosRadiography Fundamentals, Job Knowledge 124 (TWI) · Radiography Part 2: Techniques and Image Quality (TWI)
Visible or fluorescent liquid penetrates surface-breaking flaws and is drawn out against a developer.
Strengths & weaknessesInexpensive and sensitive on many nonporous materials. Only detects flaws open to the surface.
When to usePenetrant testing is the cheap default for surface-crack detection on nonmagnetic materials (aluminum, titanium, austenitic stainless, nickel alloys, even ceramics) where magnetic-particle inspection cannot work. Use fluorescent systems for critical aerospace hardware and simple visible-dye kits for weld shops and field checks. It inspects whole complex surfaces at once with minimal equipment and modest training. It finds nothing below the surface, and it finds nothing in a crack whose opening has been closed off. Smeared metal from machining, peening, or blasting seals crack openings, so etch first or inspect before those operations. Porous surfaces, such as castings with connected porosity and as-printed additive parts, give overwhelming false indications. On ferromagnetic steel, magnetic-particle inspection is faster and also catches slightly subsurface flaws. For anything internal, use ultrasonic or radiographic methods.
Key numbersDetects surface cracks around 1 µm wide and 10 µm deep with high-sensitivity fluorescent systems, coarser with visible dye · penetrant dwell 10–30 minutes and developer dwell at least 10 minutes, so 30–60 minutes total per batch · five sensitivity levels under AMS 2644, from 1/2 to 4 · working temperature about 4–52 °C · aerosol kits $50–150, production fluorescent lines $50k–300k · consumable cost well under a dollar per part.
ExamplesTurbine blades, castings, forgings, welds.
VideosLiquid Penetrant Inspection FAQs (Magnaflux) · Basic Knowledge of Dye Penetrant Testing (Karl Deutsch)
A ferromagnetic part is magnetized and particles accumulate around flux leakage caused by flaws.
Strengths & weaknessesFast and sensitive to fine surface and near-surface cracks. Only works on ferromagnetic materials.
When to useMake MPI the default surface-crack check on carbon and alloy steel. It is faster than penetrant, tolerates thin coatings, and unlike penetrant it also reveals flaws just below the surface, which is why it is the usual method for forgings, weld toes, gear teeth, and in-service items like crane hooks and landing gear. Wet fluorescent benches suit production, and portable yokes handle field welds. Magnetize in two perpendicular directions, since indications only form for flaws roughly perpendicular to the flux, and demagnetize parts headed for machining or plating. It does not work at all on aluminum, austenitic stainless, titanium, and other nonmagnetic materials, so use dye penetrant there. For genuinely internal flaws, move to ultrasonic testing.
Key numbersDetects surface cracks and subsurface flaws to roughly 1–3 mm depth · tolerates nonmagnetic coatings up to about 50 µm thick · tangential field 2.4–4.8 kA/m, from head-shot currents of 500–6,000 A or a portable yoke lifting 4.5 kg on AC and 18 kg on DC · fluorescent inspection needs UV-A of at least 1,000 uW/cm2 at the part · seconds to a minute per part, faster than penetrant · portable yokes $500–2,000, wet fluorescent benches $15k–100k.
ExamplesGears, steel forgings, rails, landing gear.
VideosMagnetic Particle Inspection FAQs (Magnaflux) · Basic Knowledge of Magnetic Particle Crack Detection (Karl Deutsch)
Electromagnetic fields induce currents in conductive material, and changes reveal defects or material variation.
Strengths & weaknessesFast, contactless, and sensitive to cracks, conductivity, coating thickness, and heat-treatment condition. Penetration is limited and calibration is geometry-specific.
When to useUse eddy current where speed matters and surface preparation is not an option: in-service aircraft crack checks around fasteners and through paint, 100% inline inspection of tube, bar, and wire at meters per second, heat-exchanger tubing surveys with internal bobbin probes, and quick sorting by alloy, conductivity, or heat-treat condition, which no other NDT method does as easily. It needs no couplant or chemicals and automates cleanly. Skin effect confines sensitivity to the surface and a few millimeters below at best, so use ultrasonic testing for deep volumetric flaws. Ferromagnetic steels require magnetically biased or specialized probes, and plain MPI is often simpler there. Every geometry change demands recalibration against reference standards with artificial defects, so eddy current suits repetitive inspection of consistent parts and works poorly for one-off jobs.
Key numbersFrequency 100 Hz to 10 MHz, with 100 kHz to 1 MHz usual for surface cracks · standard depth of penetration about 0.26 mm in aluminum at 100 kHz, so usable depth runs from under 0.1 mm to a few mm · detects surface cracks from about 0.5 mm long and 0.1 mm deep · works through 0.5–2 mm of paint or lift-off · inline tube and bar inspection at 1–3 m/s · handheld instruments $5–25k, array systems $30–100k.
ExamplesAircraft skins, tubing, wheels, fastener holes.
VideosPressure decay, tracer gas, bubble detection, mass spectrometry, or flow measurement identifies leakage.
Strengths & weaknessesDirectly tests functional sealing performance. Fixture leakage, temperature stabilization, test time, and defect localization are difficult.
When to useLeak test any product whose function depends on containment, and pick the method by the acceptable leak rate. Pressure decay is the cheap production default down to roughly 10^-3 mbar·l/s, tracer-gas sniffing reaches 10^-5 to 10^-6, and helium mass-spectrometer vacuum testing covers the 10^-7 and tighter rates demanded of refrigeration circuits, battery cells, implants, and vacuum hardware. Specify the numeric leak-rate limit from physics (allowable pressure or fluid loss over service life) rather than defaulting to "bubble tight". Pressure decay gets unreliable on large-volume or flexible parts, because temperature drift and creep produce signals bigger than the leak, so move those to tracer gas. These are pass/fail global tests. If a failed part has to be repaired, add a localization step like sniffing or bubble immersion. Leak testing also says nothing about structural integrity, so pair it with proof or burst testing where rupture matters.
Key numbersPressure decay resolves to about 10^-3 mbar·l/s, tracer-gas sniffing 10^-5 to 10^-6, helium mass spectrometry to 10^-12 under vacuum · cycle time 10–120 seconds per part including fill and stabilization · test pressures typically 0.5–10 bar · tracer mixes from 5% helium in nitrogen up to pure helium · $5–20k for a pressure-decay instrument, $25–80k for a helium leak detector.
ExamplesVacuum systems, batteries, fuel systems, refrigeration, medical devices.
VideosLeak Detection Knowledge Hub (Leybold) · Fundamentals of Leak Detection e-book (Leybold)
Sensors measure force, vibration, acoustic emission, temperature, dimensional drift, melt-pool behavior, pressure, or other signals during production.
Strengths & weaknessesCan detect defects early, reduce scrap, support traceability, and enable closed-loop control. Signals must be validated against actual part quality, often process by process.
When to useInvest in in-process monitoring where post-process inspection is expensive, slow, or cannot see the defect at all: additive builds running for days, welds and molded parts whose internal quality only CT or sectioning would reveal, unattended lights-out machining where a broken tool ruins hours of production, and regulated industries needing per-part traceability. It pays best when scrap cost per event is high, or when catching drift mid-run saves an entire batch. Do not deploy it on an unstable process, since sensors on an out-of-control process just document the variation without fixing it. Expect a real validation campaign correlating signals with destructive or CT ground truth before you grant the system any accept/reject authority. For cheap parts with fast, easy end-of-line checks, conventional inspection plus SPC is usually the better investment.
Key numbersSampling rates from 1 kHz for spindle load and cavity pressure up to 100 kHz for melt-pool photodiodes and 1 MHz for acoustic emission · detection latency milliseconds for tool breakage, seconds to minutes for process drift · melt-pool monitoring generates roughly 1–10 GB per build hour, so terabytes over a multi-day build · retrofit tool monitoring $5–20k per spindle, cavity-pressure instrumentation $5–15k per tool, OEM melt-pool packages $50–250k · validation typically takes dozens of correlated builds or coupons before the system gets accept/reject authority.
ExamplesCNC tool-wear monitoring, additive melt-pool monitoring, molding cavity pressure, weld monitoring, coating inspection.
VideosCavity Pressure Monitoring in Injection Molding (Kistler) · Measurement Science for Real-Time Control of Powder Bed Fusion, NISTIR 8036 (NIST)
Production data is analyzed to detect variation, drift, and loss of process capability.
Strengths & weaknessesIdentifies problems before they create widespread scrap and supports stable high-volume manufacturing. Only works when measurements are reliable and the underlying process is sufficiently understood.
When to useApply SPC to repetitive production of measurable characteristics once the process is capable. A good rule of thumb is to run control charts where Cpk exceeds about 1.33, so sampling can replace 100% inspection, and to chart only the few characteristics that actually predict function rather than everything measurable. It is standard practice, and often contractually required, in automotive and other high-volume supply chains. Two prerequisites really matter: a gauge R&R showing measurement variation well under tolerance, and enough process understanding to act on out-of-control signals rather than tamper. SPC adds little in job shops with lot sizes of a handful, on immature processes still being debugged, or on safety-critical characteristics with marginal capability. In those cases, use 100% inspection or in-process monitoring instead.
Key numbersCpk 1.33 the usual threshold for replacing 100% inspection with sampling, 1.67 for safety-critical characteristics · Cpk 1.33 corresponds to roughly 63 defective parts per million and Cpk 1.67 to under 1 ppm · gauge R&R under 10% of tolerance acceptable, 10–30% marginal, above 30% unusable · control limits at ±3 sigma, giving about one false alarm per 370 points · subgroups of 3–5 parts sampled every 30 minutes to 2 hours · 20–25 subgroups, or 100–125 measurements, to establish limits.
ExamplesMachining dimensions, molding pressure, coating thickness, battery loading, semiconductor critical dimensions.
VideosStatistical Process Control Guide (Quality-One) · Process or Product Monitoring and Control, e-Handbook of Statistical Methods (NIST/SEMATECH)
Parts or coupons are cut, loaded, sectioned, burned, fatigued, or otherwise destroyed to measure performance.
Strengths & weaknessesProvides direct evidence of strength, toughness, fatigue, bond integrity, microstructure, or failure mode. Consumes product and only samples a fraction of production.
When to useUse destructive testing wherever a property cannot be measured any other way: qualifying a new process, material lot, supplier, or welder; validating heat-treat response with tensile and hardness coupons; establishing fatigue and burst margins during design; and periodic lot-release sampling in safety-critical production. Witness coupons built alongside additive parts are the standard pattern. It is also the ground truth against which every NDT method and in-process monitor must be calibrated. Because it consumes product and only ever samples, it cannot screen individual production parts. Use it to prove the process, define the statistical sampling plan deliberately, and rely on nondestructive inspection and SPC to catch part-to-part variation. If a one-off failed part must keep its evidence intact, scan it by CT before any sectioning.
Key numbersASTM E8 tensile coupons use a 12.5 mm diameter gauge section, run 1–5 minutes each, and cost $50–300 at a commercial lab · fatigue testing needs 10^6–10^7 cycles at 10–100 Hz, so hours to weeks per specimen, and 10–30 specimens for one S-N curve · metallographic sections $150–500 per mount · lot-release sampling typically 1–3 coupons per heat, lot, or build plate · universal testing machines $20–150k, fatigue rigs $100–500k.
ExamplesTensile testing, fatigue testing, burst testing, metallography, peel testing.
VideosTensile Testing Part 1, Job Knowledge 69 (TWI) · Tensile Testing Fundamentals and Standards (ZwickRoell)
No processes match the current filters.
Try clearing a facet or broadening the search.
Glossary
Terms that show up in the process explorer and are not obvious from outside the trade. Numbers are typical values, not specifications.
| Term | What it means |
|---|---|
| Anisotropy | Properties that depend on direction. Cast metal is roughly the same in every direction; rolled, forged, laid-up, and 3D-printed material is not, and the weak direction usually governs the design. Etching uses the word differently, for an etch that cuts downward much faster than sideways. |
| Annealing | Heating metal and cooling it slowly to soften it and relieve stress. It undoes work hardening, so a part that has been drawn or bent hard can be annealed and worked again. Deep-drawn cans and drawn tube go through several anneal steps between passes. |
| Autoclave | A pressure vessel that cures composite parts under heat and pressure at the same time, typically 6–7 bar and 120–180 °C. The pressure squeezes voids out of the laminate, which is why autoclave parts hit under 1% voids and out-of-autoclave parts usually do not. Aerospace-scale autoclaves are among the most expensive tools in a composites shop. |
| Axisymmetric | Shaped so that any cross-section through the centerline looks the same, like a pipe, a wheel, or a bottle. Spinning, ring rolling, filament winding, and rotational molding only make axisymmetric parts, which is the first question to ask before considering any of them. |
| Billet | A solid semi-finished block or bar of metal that a later process works down into a part. Forging, extrusion, and machining all start from billet. "Machined from billet" means cut from solid rather than cast or forged, which is strong and accurate but wastes most of the material. |
| Blank | The flat piece cut out before forming starts. Sheet-metal parts begin as a blank that is then drawn, bent, or stamped, and blank size and shape are worked out backwards from the finished part. Blanking is the cutting step that produces it. |
| Bore | An internal cylindrical hole, and the act of enlarging one accurately. Boring is how a hole gets to a tight diameter and true position after drilling, since a drill wanders. Engine cylinders, bearing housings, and hydraulic bodies are all quoted by bore. |
| Burr | The ragged edge of displaced material left where a cutting tool exits. Burrs cut hands, jam assemblies, and hold contamination, so deburring is its own cost line on most machined parts. Waterjet and laser cutting leave much less of it than sawing or milling. |
| Case depth | How far below the surface a hardening treatment reaches. Nitriding gives 0.1–0.5 mm, carburizing 0.5–2 mm, and induction or flame hardening 1–6 mm. The core stays soft and tough underneath, so the part takes contact loads at the surface without becoming brittle. |
| Curing | A chemical reaction that sets a resin or adhesive permanently, as opposed to melting and freezing. Cured material cannot be remelted, which is what separates thermosets from thermoplastics. Cure schedules are quoted as time at temperature, and undercuring shows up later as low strength rather than as an obvious defect. |
| Die | Two unrelated meanings. In forming and casting, a die is the hardened tool that gives a part its shape: stamping dies, forging dies, die-casting dies. In semiconductors, a die is one finished chip after the wafer has been cut apart. |
| Draft | Taper on a part's walls so it can leave a mold or die without dragging. Hot-forged surfaces typically carry 3–7°. Molding, forging, and permanent-mold casting all need it; lost-foam casting does not, because the pattern is vaporized rather than pulled. |
| Ductility | How much a material stretches before it breaks, usually quoted as percent elongation. Ductile metals can be bent, drawn, and deep-drawn; brittle ones crack instead. It is why cast iron is cut and not bent, and why a material's forming limit matters more than its strength in sheet work. |
| Extrusion | Pushing material through a shaped opening so it comes out with that cross-section, the way toothpaste takes the shape of the tube's nozzle. Aluminum window frames, plastic pipe, and pasta are all extruded. It makes constant cross-sections cheaply and cannot make anything that varies along its length. |
| Fixture | Purpose-built hardware that holds a workpiece in a known position while it is machined, welded, or measured. A jig is the same idea but also guides the tool. Fixtures are a large part of NRE and are usually what makes a second machine slower to bring up than expected. |
| Flash | Thin excess material squeezed into the gap where two halves of a mold or die meet. It is normal in forging, die casting, and injection molding, and it has to be trimmed off afterwards, which is a separate operation with its own tooling. Flashless processes exist and cost more per part. |
| Gauge | Sheet or wire thickness on an old numbering scale where a bigger number means thinner. It varies by material and by country, so gauge numbers are ambiguous and drawings should carry millimeters. Also used generally for a measuring instrument. |
| Grain flow | The direction of a metal's internal structure after it has been deformed rather than melted or cut. Forging, ring rolling, and thread rolling bend the grain to follow the part's shape, which is why they beat castings and machined parts on fatigue strength. |
| Green | A part that has been pressed, printed, or molded but not yet sintered, fired, or cured. Pressed metal powder sits at 55–70% of theoretical density while green, and it is often machined in that state because it cuts easily. Foundry "green sand" is a separate usage, meaning moist clay-bonded molding sand. |
| Grit | Abrasive particle size, where a bigger number means finer. Rough grinding runs 40–80 grit and finishing 400 grit and above. Each step should remove the scratches left by the one before it, which is why polishing sequences skip no more than one grade at a time. |
| Hardness (HRC, HV) | Resistance to indentation. HRC is the Rockwell C scale used for steel: 30–45 HRC for a tough core, 55–62 HRC for a hardened case, and above roughly 45–50 HRC cutting tools stop working and the part has to be ground. HV is Vickers, used for thin layers and coatings, with nitrided surfaces at 700–1,200 HV. |
| Heat-affected zone (HAZ) | The band of base metal beside a weld that got hot enough to change structure without melting. It is often the weakest part of the joint. Arc welding leaves 1–5 mm of it, laser welding under 1 mm, and friction stir welding avoids melting altogether. |
| Hot isostatic pressing (HIP) | A furnace cycle at 100–200 MPa of argon pressure and 900–1,250 °C that closes internal voids, taking castings and printed parts above 99.9% density. Hold times run 2–4 hours inside an 8–12 hour cycle. It is usually bought from a toll processor rather than run in house. |
| Kerf | The width of material a cutting process consumes. Laser cutting takes 0.1–0.5 mm, waterjet about 1 mm, and plasma or oxy-fuel 1.5–5 mm. Kerf sets how tightly parts can be nested, and in wafer dicing it decides how much silicon area the saw streets cost. |
| Layup | Placing fiber cloth or tape into a mold ply by ply, by hand or by machine, before it is cured. The stacking sequence and the fiber angles are the design, so layup is where a composite part's properties are decided rather than a step that follows the design. |
| Mandrel | A shaped former that a part is built up on or held by from the inside. Filament winding and rotational forming use one to define the internal shape, and tube bending uses one inside the tube to stop the wall collapsing. Getting the mandrel back out afterwards is a design constraint of its own, which is why some are collapsible or dissolvable. |
| Micron (µm) | One millionth of a meter, and the unit most manufacturing tolerances and surface finishes are quoted in. A human hair is roughly 70 µm across, ordinary machining holds tens of microns, and precision grinding holds single digits. Also called a micrometer. |
| Near-net shape | A part that leaves its primary process close enough to final dimensions that only light finishing is needed. Investment casting, closed-die forging, and metal injection molding are near-net routes. The saving shows up in machining time and material, not in the primary process itself. |
| NRE | Non-recurring engineering: the one-time cost of getting a part into production, covering tooling, fixtures, programming, process development, and qualification. NRE is what sets the breakeven volume between a low-tooling process and a high-tooling one. |
| Peening | Hammering a surface with shot, or pulsing it with a laser, to leave it in compression. Cracks struggle to start or grow in a compressed surface, so peening buys fatigue life on springs, gears, and landing gear without changing the alloy or the geometry. |
| Porosity | Voids inside a part. Castings trap gas and shrinkage voids, press-and-sinter powder parts are 7–15% porous unless densified further, and welds go porous when shielding is lost. Porosity costs fatigue life, pressure tightness, and weldability well before it costs static strength. |
| Prepreg | Fiber cloth or tape pre-impregnated with partly cured resin. It buys a repeatable 55–65% fiber volume fraction and under 1% voids, at $40–150/kg plus freezer storage at -18 °C and a 10–30 day out-life once thawed. |
| Prismatic | Made of flats, pockets, holes, and slots rather than curved surfaces, like a valve body or a machined bracket. Prismatic parts suit 3-axis milling; sculptured surfaces need 5-axis work or a different process, so the distinction decides which machine quotes the job. |
| Quenching | Cooling hot steel fast, in water, oil, or gas, to lock in a hard structure that slow cooling would not produce. Fast quenching gives hardness and also distortion and cracking risk, so quench severity is a trade-off rather than a setting to maximize. Quenched steel is almost always tempered afterwards. |
| Ra | Arithmetic average surface roughness, in micrometers; lower is smoother. Sand castings run Ra 6.3–25 µm, die castings 0.8–3.2 µm, and ground surfaces 0.1–0.4 µm. Optical surfaces are quoted in nanometers RMS instead. |
| Residual stress | Stress locked into a part by processing, present with no external load applied. Quenching, welding, and heavy machining leave tensile stress that distorts parts and encourages cracking. Shot peening, laser peening, and thread rolling add compressive surface stress deliberately, because it resists fatigue. |
| Runner, gate, and sprue | The channels that carry molten material from the machine into the mold cavity. The sprue is the entry, runners distribute, and the gate is the narrow final opening. They solidify too and are cut off as scrap, so their size and placement drive both cycle time and yield. |
| Sintering | Heating a powder compact below its melting point so the particles bond and the part densifies. It is the consolidation step in powder metallurgy, metal injection molding, binder jetting, and ceramics. The part shrinks as it densifies, by 15–20% linear in metal injection molding, so tooling is cut oversize to compensate. |
| Spinneret | A plate pierced with many fine holes that molten polymer is pumped through to form filaments, the way a showerhead splits a stream. Hole count and shape set the fiber denier and cross-section, and the spinneret is the precision part of an otherwise brute-force line. |
| Springback | Elastic recovery when the forming load comes off, so the part relaxes away from the die shape. Tooling and press-brake programs overbend to compensate, and the leftover variation is one reason bend angles are quoted at about ±1° rather than tighter. |
| Stamping | Pressing sheet metal between a punch and a die to cut or form it, one hit per station, at hundreds of parts per minute on a progressive line. Tooling is expensive and the per-part cost is very low, which is the whole reason car bodies are made this way. |
| Tempering | Reheating quenched steel to a moderate temperature to trade some hardness back for toughness. Untempered quenched steel is hard and brittle enough to be useless for most parts. The tempering temperature is the dial that sets the final balance. |
| Thermoplastic and thermoset | Thermoplastics melt and resolidify repeatedly, so they can be injection molded, welded, and reground. Thermosets cure once into a permanent network and then char rather than melt. It decides whether a plastic part can be recycled, heat-staked, or repaired. |
| Tolerance | The allowed range on a dimension, written as ±0.1 mm or as a fit class. Tolerance drives cost more sharply than size does, because each tightening step can add a process: cast, then machine, then grind. Tolerance should be specified from what the assembly needs, not from what looks tidy. |
| Undercut | A feature that traps the tool or the mold, so the part cannot simply be pulled out or reached in a straight line. Molds solve it with side actions and lifters, which add cost and cycle time; machining solves it with special tooling or another setup. Designing undercuts out is usually cheaper than tooling around them. |
| Work hardening | Metal getting harder and less ductile as it is deformed cold. It is useful when it strengthens a finished part and a nuisance when it stops a part being formed further or makes it difficult to machine. Austenitic stainless steel work hardens fast, which is why it punishes a dull tool. |
| Workpiece | The material being worked on, as distinct from the tool doing the work. Useful because most process descriptions turn on which of the two moves, spins, or is held: in turning the workpiece rotates, in milling the tool does. |
How to choose a manufacturing process
Picking a process means answering two separate questions: can it make the part, and do the economics work? Keep in mind that a finished component usually combines several processes. An aircraft bracket might begin as a forged billet, then be CNC-machined, heat-treated, shot-peened, anodized, dimensionally inspected, and ultrasonically tested.
Engineering factors
| Factor | Why it matters |
|---|---|
| Material compatibility | Melting point, ductility, conductivity, chemistry, cure behavior, brittleness, and oxidation eliminate many processes immediately. |
| Geometry | Internal passages, undercuts, wall thickness, aspect ratio, symmetry, draft, and tool access govern manufacturability. |
| Size and mass | Press force, chamber size, furnace size, machine travel, lifting, and handling create hard limits. |
| Tolerance | Every process has a native accuracy range; forcing tighter tolerances usually requires secondary machining or inspection. |
| Surface finish | Surface condition affects appearance, friction, fatigue, sealing, cleanliness, optics, and coating adhesion. |
| Mechanical properties | Grain flow, porosity, fiber orientation, residual stress, heat treatment, and anisotropy may matter more than nominal material grade. |
| Production rate | Cycle time, cooling, curing, line speed, deposition rate, and setup time determine practical throughput. |
| Secondary operations | Deburring, machining, heat treatment, coating, joining, cleaning, and inspection may dominate total cost. |
Economic and strategic factors
| Factor | Why it matters |
|---|---|
| Tooling and NRE | Dies, molds, fixtures, programming, process development, qualification coupons, and validation determine upfront risk. |
| Breakeven volume | Low-tooling processes are cheaper at low volume; high-tooling processes are usually cheaper at scale. |
| Yield and scrap | First-pass yield can matter more than nominal machine rate, especially in batteries, semiconductors, composites, optics, and high-value alloys. |
| Scrap recoverability | Metal chips may retain recycling value; cured thermosets, contaminated powders, mixed materials, and batteries may not. |
| Installed-base scarcity | Large presses, HIP vessels, autoclaves, large-envelope 5-axis machines, and precision optical equipment may have only a handful of qualified suppliers. |
| Capacity utilization | Continuous and capital-intensive processes become uneconomic when underutilized. |
| Qualification burden | Aerospace, medical, nuclear, automotive, and defense customers may require years of testing, source approval, audits, and process lock-down. |
| Switching cost | Requalifying a new supplier or process may cost far more than any apparent unit-price savings. |
| Inspection burden | A process that makes parts cheaply but requires 100% CT or extensive destructive testing may not be economical. |
| Process maturity | Laboratory feasibility does not guarantee stable yield, repeatability, maintenance, production uptime, or qualified equipment. |
| Supply-chain concentration | Feedstocks, specialty tooling, chemicals, machine vendors, and operators can create hidden single points of failure. |
| Environmental & permitting | Solvents, plating baths, powders, explosives, radiation, emissions, and hazardous gases can determine site feasibility and launch time. |
Comparing processes quantitatively
Descriptions alone don't let you compare processes. For each process, it helps to record:
| Field | What to record |
|---|---|
| Typical native tolerance | Before secondary machining. |
| Typical surface roughness | Usually Ra in micrometers or microinches. |
| Practical size range | Smallest and largest typical component. |
| Typical production volume | Prototype, low, medium, high, or continuous. |
| Tooling and NRE | Rough order of magnitude. |
| Machine capex | Rough order of magnitude. |
| Cycle time / throughput | Seconds, minutes, hours, meters per minute, wafers per hour, or tons per year. |
| Material utilization | Yield and primary scrap modes. |
| Common secondary ops | Machining, heat treatment, coating, cleaning, joining, inspection. |
| Main defect modes | Porosity, cracks, warpage, burrs, voids, contamination, delamination, inclusions. |
| Qualification difficulty | Low, moderate, high, or extreme. |
| Supplier scarcity | Commodity, specialized, scarce, or globally constrained. |
Volume categories
Tooling-cost bands
Treat these as rough categories rather than quotes. Part size, material, geography, supplier availability, qualification, and production rate can shift actual costs by orders of magnitude.
Why the continuous / discrete distinction matters
A startup running a continuous extrusion or rolling line is a very different business from one running CNC machines, injection-molding presses, or additive-manufacturing systems. Continuous processes are usually driven by utilization, throughput, energy cost, and commodity-scale capital intensity. Discrete manufacturing is more often driven by tooling, cycle time, part complexity, labor, yield, and qualification.
Core takeaway
Don't compare processes only by whether they can produce the same CAD shape. Two parts with identical geometry may behave very differently depending on whether they were cast, forged, machined, molded, printed, sintered, welded, or heat-treated. The process determines grain structure, porosity, anisotropy, residual stress, surface condition, defect population, inspectability, qualification burden, and eventual unit economics.
Key questions for engineering decisions
- Can the process make the geometry in the required material?
- Can it achieve the necessary properties, tolerance, and finish?
- Can defects be detected and controlled?
- Can it scale to the required volume?
Key questions for investment and business analysis
- What is the real yield?
- What is the qualification burden?
- How scarce is qualified capacity?
- What fraction of cost comes from secondary operations and inspection?
- How much utilization is needed to earn an acceptable return on equipment?
- Is the process genuinely differentiated, or merely technically feasible?
- Does the company control a bottleneck, or is it dependent on scarce upstream equipment, feedstocks, or suppliers?
A manufacturing advantage is rarely just a faster machine. Durable advantages usually come from some combination of process know-how, qualification, yield, automation, proprietary tooling, operating data that's hard to replicate, scarce installed capacity, and integration across multiple manufacturing steps.
Head-to-head: the usual short list
These are the high-level process families that most often compete for the same part. Treat the figures as rules of thumb rather than quotes. The tables after it settle the choices inside a family: which casting route, which sheet-forming route, and how to join metal.
| Process | Volume sweet spot | Tooling cost | Per-part cost | Geometry & tolerance | Pick it when |
|---|---|---|---|---|---|
| CNC machining | 1–5,000 | Minimal (fixtures only) | High, roughly flat with volume | Complex 3D; ±0.01–0.05 mm | Geometry is still changing, tolerances are tight, or volume cannot justify tooling. |
| Injection molding | 5,000–millions | $20k–$200k molds | Cents at scale | Complex thermoplastic; ±0.05–0.1 mm | A moldable thermoplastic design is frozen and lifetime volume clears roughly 5,000–10,000 units. |
| Casting | 1s (sand) to 100k+ (die) | $500–$20k patterns to several $M die tools | Low–moderate, plus machining | Complex 3D with cores; ±0.1 mm (die) to ±1 mm (sand) | The shape is too complex or too large to machine economically, and cast properties are good enough. |
| Sheet-metal fab | 1 to a few tens of thousands/yr | Minimal (standard punches & brakes) | Moderate | Cut & bent sheet; ±0.25 mm on bends | You need enclosures, brackets, or chassis quickly, at volumes too low to pay back stamping dies. |
| Stamping | Tens of thousands/yr and up | $20k–$300k progressive dies | Cents at dozens of strokes/min | Formed sheet; excellent repeatability | A stable sheet-metal design runs tens of thousands per year and up. |
| Forging | 10,000+ | $10k–$100k die sets | Moderate, plus finish machining | Near-net solid shapes; draft and flash, no internal cavities | Fatigue- and impact-critical parts need grain-flow strength that castings cannot match. |
| Extrusion | Hundreds of kg to continuous | Low (aluminum dies a few $k) | Low per meter | Constant cross-section only | The part is a long constant profile (aluminum especially) that would waste material if machined. |
| Additive | 1–500 | None | High; post-processing can cost more than the print | Internal channels & lattices; ±0.1–0.2 mm | The geometry cannot be machined or molded at all, or lead time matters more than unit cost. |
Choosing a casting route
Once casting is the answer, the next question is which casting process. The routes below differ mainly in tooling cost, minimum wall, and finish, so annual volume decides most cases. Aluminum has the most options; iron and steel rule out the permanent-mold and die-casting rows.
| Process | Tooling | Economic volume | Minimum wall | Tolerance & finish | Pick it when |
|---|---|---|---|---|---|
| Sand casting | Patterns and coreboxes $500–20k, 2–6 weeks | One-off to a few thousand a year | About 4 mm | ±1 mm on small features; Ra 6.3–25 µm | The part is large, heavy, or made of iron or steel, and you can machine the functional surfaces afterward. |
| Investment casting | Wax dies $5k–50k, 8–16 weeks | Tens to tens of thousands | About 1.5 mm | ±0.13 mm over the first 25 mm; Ra 1.6–3.2 µm | The part is under about 50 kg in a hard-to-machine alloy, and fine detail saves you machining time. |
| Permanent mold | Molds $40k–200k, 8–16 weeks | 1,000–100,000 a year | About 3 mm | ±0.25–0.5 mm; Ra 3.2–6.3 µm | You run aluminum or magnesium in the middle volume band and want heat-treatable parts without paying for a die-cast cell. |
| Low-pressure die | Dies $150k–800k | Tens of thousands a year | About 3 mm | ±0.4 mm; Ra 3.2–6.3 µm | A structural aluminum part needs low porosity and full solution treatment, and a 3–10 minute cycle is acceptable. |
| High-pressure die | Dies from tens of thousands to several million, 12–20 weeks | Tens of thousands a year and up | 1–3 mm | ±0.1–0.25 mm over 25 mm; Ra 0.8–3.2 µm | Walls are thin, volume is high, and you do not need to weld or solution treat the casting. |
| Lost foam | Pattern tooling $50k–250k, 8–16 weeks | Roughly 10,000 a year and up | About 3 mm | ±0.25 mm per 25 mm; Ra 6–12 µm | The casting would otherwise need many sand cores, and volume covers the foam tooling and the process debugging. |
| Centrifugal | Molds $5k–50k per size, good for thousands of pours | One-off to steady production | 5–125 mm wall | OD ±0.5–2 mm; 3–6 mm of bore machining stock | The part is a hollow round (pipe, liner, ring) and you can machine the bore, where the light inclusions collect. |
Sheet-metal forming routes
Sheet parts can be made half a dozen ways, with tooling costs spread over three orders of magnitude. Annual volume settles most of it; the rest comes down to whether the shape is prismatic, round, or a long constant profile.
| Process | Tooling | Economic volume | Stock & size | Tolerance | Pick it when |
|---|---|---|---|---|---|
| Press brake | Standard punches and dies, near zero | 1 to a few tens of thousands a year | 0.5–12 mm, bends to 3–4 m | ±0.25 mm on flanges, about ±1° on bends | You need enclosures, brackets, or chassis quickly and any fab shop can turn them around the same day. |
| Stamping | Progressive dies $20k–300k, 8–20 weeks | Tens of thousands a year and up | 0.2–6 mm sheet | ±0.05–0.25 mm, to ±0.025 mm fine blanked | The design is frozen and running 20–100 strokes per minute turns the die cost into cents per part. |
| Roll forming | Roll sets $15k–100k per profile, 8–16 weeks | Tens of thousands of linear meters a year | 0.3–6 mm, profiles to about 600 mm wide | ±0.2–0.5 mm on the section, ±1 mm cut length | The part is a constant cross-section bought by the meter, or longer than a press brake can bend. |
| Stretch forming | Form dies $5k–50k in steel, kirksite, or epoxy | Tens to a few thousand pieces | Meter-scale panels on tables to 15 m | ±0.5–1.5 mm on contour | You need a wrinkle-free skin with shallow compound curvature and can trim 50–150 mm of gripped edge per side. |
| Hydroforming | Tooling $50k–500k, 12–24 weeks | Tens of thousands a year | Tube expansion 20–50% in steel, less in aluminum | ±0.5–1 mm | One seamless hollow member can replace a welded assembly of stampings and stiffness per kilogram matters. |
| Metal spinning | Mandrels $500–5,000 in wood, aluminum, or steel | 1 to roughly 5,000–10,000 a year | Blanks 50 mm to 5 m across, 0.5–6 mm | ±0.25–1 mm, Ra 0.4–3.2 µm | The shape is axisymmetric (cone, dome, tank head) and a deep-draw die would cost fifty times the mandrel. |
| Incremental forming | Forming tool and backing plate, a few hundred dollars | 1 to about 100 pieces | 0.5–3 mm sheet, wall angle 60–70° maximum | ±0.5–1 mm | You need a handful of panels, die cost would dominate, and minutes to hours per part is acceptable. |
Joining metals
Most assemblies need a joining method, and the choice sets distortion, inspectability, and whether the joint can ever come apart. Arc welding is the default for steel structure. The other options matter when the parts are thin, dissimilar, or cannot take the heat.
| Process | What it joins | Heat & distortion | Speed | Equipment | Pick it when |
|---|---|---|---|---|---|
| Arc welding | Steel and most alloys, thin sheet through heavy plate, with filler | 0.5–3 kJ/mm, 1–5 mm heat-affected zone, ±1–3 mm before machining | 100–500 mm/min; 1–8 kg/hr deposited, to 45 kg/hr submerged | $500–5k power source, $100k and up for a robot cell | You are fabricating structural steel and want one process that covers field repair through robotic production. |
| Resistance spot | Lap joints in 0.5–3 mm sheet per layer, mostly steel; needs two-sided access | Confined to a 4–7 mm nugget, so panels stay flat | 0.2–1 second per spot; 3,000–5,000 spots in a car body | Robot gun cells; electrode tips redressed every 2,000–10,000 welds | You are lap-joining sheet at automotive rates and can reach both sides of the flange. |
| Laser welding | Precision joints, roughly 1 mm of penetration per kW in steel | 0.2–2 mm weld width, heat-affected zone under 1 mm | 1–10 m/min, to 20 m/min on thin sheet | $150k–1m per cell | Fixturing can hold a 0.1 mm gap and volume justifies the cell, as on battery tabs or hairpin stators. |
| Electron beam | 25–50 mm in a single pass, including titanium and other reactive metals | Lowest of the fusion routes; depth-to-width 20:1 to 50:1 | 0.5–5 m/min, plus 1–15 minutes of pump-down per cycle | $500k to several million, vacuum chamber and radiation controls | The weld has to be deep and narrow in one pass, or the metal needs vacuum to stay clean. |
| Friction stir | 1–25 mm aluminum plate and extrusion, including 2xxx and 7xxx | No melting; flat enough to machine afterward | 0.5–3 m/min, with 5–50 kN of axial force to react | $250k–2m, or roughly $100k for a head on an existing machining center | The seam is long and straight in aluminum, especially alloys that crack when fusion welded. |
| Brazing | Thin sections and dissimilar pairs, through capillary gaps of 0.02–0.10 mm | Whole assembly reaches 590–1,200 °C, but the base metal never melts | 1–4 hour furnace cycle, hundreds of joints per load | $5k–50k in fixtures and furnace tooling | One heat has to make many joints, and service temperature stays well below the filler's melting range. |
| Adhesive bonding | Dissimilar, thin, or heat-sensitive materials; 20–40 MPa lap shear over a 0.1–0.5 mm bond line | None from the process, but service tops out at 80–120 °C for common epoxies | About 24 hours at room temperature, or 1 hour at 120–180 °C | Adhesive $20–200/kg, fixtures usually under $5k | You want the joint sealed as well as loaded, and you can keep the adhesive in shear rather than peel. |
| Mechanical fastening | Anything you can drill or form a hole in, including field assembly | None, but every hole carries a stress concentration around 3 | Immediate; no cure or cool-down | $0.05–5 per fastener commercial, $1–20 in aerospace grades | The joint has to come apart for service, or assembly happens in the field without process control. |