How to control shrinkage in metal injection molding parts

Every metal injection molding program lives or dies on one number. That number is the MIM shrinkage factor, the percentage a part contracts between the mold cavity and the finished sintered part. For stainless steel feedstock it usually lands between 17 and 20 percent. Miss it by half a point and a 25 mm boss moves 0.13 mm, which is enough to break an assembly.

I have been on the tooling side of MIM for about two decades, and I still treat MIM shrinkage as a variable to be managed rather than a constant to be looked up. What follows is the working method we use: what moves the number, what tolerance it can hold, and how we prove the factor before a production tool gets cut.

Why the mold cavity is always bigger than the part

The process sequence explains the arithmetic. Feedstock, metal powder bound in a polymer, fills a cavity at room temperature and produces a green part at essentially cavity size. Debinding strips the binder. Sintering then closes the remaining porosity and the part contracts in every direction at once. The cavity is therefore cut oversize by the shrinkage factor:

cavity dimension = final dimension / (1 - shrinkage)

For an 18 percent factor and a 25 mm target, the cavity is cut at about 30.5 mm. That is a large, deliberate oversize, and it is the reason metal injection molding tooling is quoted with a shrinkage assumption written down on the drawing rather than held in someone's head.

Those bands are the honest expectation for as-sintered parts. A supplier who offers +/-0.05 mm across a full drawing in MIM either has not run the part yet or quietly plans to machine it afterwards. The mill paperwork travels with the factor, which is why we keep the certificate and the dimensional report on the same shelf as our stainless steel 316L data.

One habit saves arguments later. Measure the green part before it ever sees a furnace. If the green dimensions are already off, the problem is the tool or the molding window, and no sintering profile will rescue it.

Five things that actually move MIM shrinkage

When a dimension drifts, it drifted for one of these reasons.

Powder loading comes first. Feedstock typically carries 60 to 65 percent metal powder by volume. Push the loading up and shrinkage falls, because there is less binder to remove. Push it too far and the melt gets stiff, short shots appear, and the molder spends the savings on scrap. The trade is that simple, and it is why loading sits where it sits rather than at either extreme.

The binder system is second. Wax-based, PEG-based, and POM-based systems debind differently and leave different pore structures behind, so the same powder can sinter to a different final size under a different binder chemistry. Changing binders mid-program is a tool-change event, not a paperwork event.

Sintering profile is the widest lever on MIM shrinkage. Peak temperature, hold time, atmosphere, and dew point all shift densification. A 30 degree change in peak temperature on 316L moves a 25 mm dimension by roughly 0.05 to 0.1 mm. Furnace-to-furnace variation is real, so a factor proven in one furnace is not automatically valid in the next one.

Geometry is fourth. Gate location and freeze-off decide how evenly the cavity packs, and uneven packing becomes uneven density, which becomes uneven contraction. I will come back to that in a moment.

Support and fixturing during sintering is fifth, and it is the most underrated of the five. A part that sags on a setter plate at 1,350 degrees C will not shrink the way the flow analysis predicted, no matter how good the analysis was.

Feedstock control: where MIM shrinkage starts

Lot-to-lot feedstock variation runs around +/-0.15 percent on MIM shrinkage, which sounds small until it stacks on top of profile variation. Across a twelve month program that is the difference between a comfortable tolerance and a recurring deviation report.

We control it with incoming checks on every lot: solids loading by thermal analysis, melt flow rate, and powder particle size distribution. Typical MIM powder sits at a D50 between 8 and 15 microns. A shift toward coarse powder changes packing and changes the sintering response. None of this needs exotic equipment, but somebody has to run the tests and record the results against the lot number. When I audit a supplier, that log is one of the first things I ask to see.

Debinding: the step buyers forget

Debinding sits between molding and sintering, and it decides how uniform the part is when densification begins. A slow thermal cycle removes binder gradually and leaves an open pore channel structure. A rushed cycle traps binder in the core, and the trapped material has to escape later at higher temperature, when the metal skeleton has already started to stiffen.

The visible result is blistering, internal voids, or a carbon residue that shifts the final dimensions. None of those look like a shrinkage problem on the shop floor, and all of them end up as one in the inspection report. We treat the debinding curve as part of the shrinkage recipe, not as a warm-up step.

Where MIM shrinkage stops being isotropic

Tooling is usually scaled by a single factor in all three axes, and for a simple part that works fine. For a complicated one it does not.

Weld lines and jetting create local density differences. Thin walls cool faster than thick sections, so they pack less completely. Long, unsupported features sag sideways under their own weight while the binder is gone and the metal skeleton is still weak. A part lying flat on a setter plate also behaves differently at the bottom face than at the free top surface, because friction and thermal contact differ.

The practical result is that axis-to-axis shrinkage can vary by 0.3 to 0.5 percent on a difficult part, and flatness on a 60 mm plate can drift 0.1 to 0.15 mm without fixture support. We handle it with ceramic setters, sometimes with light weights on top, and we measure the part on all three axes instead of trusting a single number from a single direction.

Wall thickness sets the boundary. Below about 0.5 mm the fill becomes erratic and shrinkage control gets expensive. Above about 6 mm you start fighting internal voids, which is a density problem that shows up as a shrinkage problem. For a part outside that window, investment casting is often the more forgiving route.

What tolerance MIM shrinkage can really hold

As-sintered, we quote +/-0.3 percent of the dimension as the default, with a practical floor around +/-0.05 mm on small features. That is the honest envelope for MIM shrinkage. On a 10 mm dimension it works out to +/-0.03 mm, so the floor governs. On a 100 mm dimension it is +/-0.3 mm, so the percentage governs. Buyers who understand which of the two rules applies to a given feature stop arguing about tolerances that were never available in the first place.

Below that band you have three options, and all three cost money. Sizing or coining squeezes selected dimensions after sintering and can hold +/-0.05 mm on those features only. CNC machining after sintering gives you a true precision fit at the cost of giving back the MIM advantage on that feature. Designing the critical interface as a separate, more tightly controlled component is the third option, and it is often the cheapest of the three.

The design rule I give buyers is short. Mark four or five dimensions as critical, and let the rest of the drawing carry the general band. A MIM drawing with every dimension toleranced at +/-0.05 mm is not a tight drawing, it is an unfinished drawing.

How we prove a MIM shrinkage factor on a new tool

Nobody knows the exact factor for a new part before the first furnace run. What a competent supplier knows is how to converge on it quickly and cheaply.

1. Start from the feedstock supplier's published factor for the grade.
2. Cut a pilot tool, single cavity wherever the geometry allows.
3. Mold a first article run of 30 to 50 green parts, and measure them green before sintering.
4. Sinter that run under the production profile and measure again.
5. Calculate the real factor per axis, not per part.
6. Adjust the cavity, or tune the sintering profile if the error is uniform and small.

On a well understood grade that sequence takes two to three weeks. On titanium or a new low-alloy grade it takes longer, because the process window is narrower and the first runs usually expose something the simulation missed. We re-run the whole sequence whenever a new furnace comes online or the load pattern on a furnace changes.

A program where the assumption was wrong

We molded a small instrument housing for a medical device customer in 17-4PH. The starting assumption was 17.5 percent. First article came back at 18.4 percent, nearly a full point high, and the 30 mm overall length landed 0.2 mm short of nominal with a flatness deviation at the gate end.

Two causes combined. The feedstock lot ran slightly finer than the qualification lot, and the gate froze earlier than the flow analysis predicted, so the far end packed light. We re-cut the cavity for 18.3 percent and lengthened the gate land by 0.3 mm. The second run held within +/-0.25 percent across three lots, and the program went to production with a documented factor and a Cpk above 1.33 on the critical fit.

Cost of that lesson was four weeks and one cavity revision. Cost of shipping the first run would have been a recall.

Six mistakes that cause most MIM shrinkage problems

• Tolerancing the part as if it were machined, with no allowance for the as-sintered band.
• Switching feedstock supplier on price without a shrinkage trial on the actual part.
• Approving a first article measured green, or measured on one axis only.
• Assuming a factor proven in one furnace transfers to another furnace.
• Ignoring setter design, then blaming the tool for flatness.
• Changing the sintering profile inside the PPAP window without re-measuring dimensions.

What belongs on the drawing

Four items. The material grade with its acceptance method. The critical dimensions, marked as critical. The default as-sintered band of +/-0.3 percent, stated rather than implied. And the sintering profile together with the proven shrinkage factor as part of the qualification package, so the next engineer who touches the program is not starting from zero.

Shrinkage in MIM is not a defect to be eliminated. It is a designed-in contract between the tool, the feedstock, and the furnace. Programs that treat it as engineering data from day one run quietly for years. Programs that treat it as a surprise spend those years in deviation meetings.

Our how we work page lists the inputs we need to quote a MIM part with a realistic shrinkage assumption, and our FAQ answers the tolerance and material questions buyers raise most often. Our quality page shows how the factor and the dimensional report are documented at PPAP.