Every MIM part shrinks 15% to 20% linearly in the sintering furnace. Holding ±0.3% on the finished part therefore means controlling a shrinkage of about 17% to within a few hundredths of its own size. On a 30 mm feature, a 0.1% change in shrink is 0.03 mm. That is the arithmetic behind every MIM tolerance discussion, and it explains why some features hold ±0.3% easily while others need a secondary operation.
This article explains where shrinkage variation and distortion come from, what we control at each stage to hold ±0.3%, and where that number stops being realistic.
Why MIM parts shrink, and why the number is not fixed
MIM feedstock is metal powder mixed with a binder, with powder loading typically around 58% to 65% by volume. After the binder is removed, sintering at about 1,300 to 1,380 °C for stainless steels closes most of the space the binder occupied. The part densifies to 96% to 98% of theoretical density and shrinks accordingly.
The tool is cut oversize by a shrink factor, typically 1.16 to 1.22. In theory, shrink is the same in every direction. In practice, four things make it vary from part to part and from direction to direction:
- Powder loading varies slightly between feedstock lots. A lower loading means more binder to remove and more shrink.
- Molding density varies with packing pressure and gate position. Regions packed harder shrink less. Flow direction can leave a difference of a few tenths of a percent between shrink along and across the flow.
- Section thickness affects how completely each region densifies. Thick sections often shrink slightly more than thin ones.
- Furnace conditions, mainly temperature uniformity and position on the tray, change final density and therefore final size.
Where distortion comes from
Shrink variation changes size. Distortion changes shape. In the furnace the part is hot enough to creep, and four forces act on it:
| Mechanism | What you see | Typical cause |
|---|---|---|
| Gravity sag | Overhangs droop, long spans bow, tall thin walls lean | Unsupported features at sintering temperature |
| Setter friction | Bottom face shrinks less than the top, giving a slight taper or “foot” | Part dragging on the setter as it shrinks |
| Uneven densification | Warp toward the thicker side | Large changes in wall thickness |
| Released molding stress | Twist or bow that appears only after sintering | Uneven packing, gate too far from thick sections |
How we hold ±0.3%: control at every stage
No single step holds the tolerance. Each stage removes a share of the variation, and the stages compound.
| Stage | What we control | Why it matters |
|---|---|---|
| Design (DFM) | Wall thickness 1 to 6 mm, adjacent sections within about 1.5:1, flat setting face, symmetric mass where possible | Uniform sections shrink uniformly; a flat face resists sag and friction distortion |
| Feedstock | One feedstock lot per order, incoming checks on melt flow and density | Lot-to-lot powder loading changes shrink |
| Tooling | Gate into the thickest section, balanced runners, cavity dimensions from measured shrink data, steel-safe margins on critical features | Even packing gives even shrink; steel-safe allows correction by removing steel |
| Molding | Part weight monitored against a tight band, fixed pack pressure and time, gate-freeze study | Part weight is the best early indicator of molding density |
| Debinding | Ramp rate set by the thickest section, debind weight loss verified | Incomplete debinding leaves carbon and uneven sintering |
| Sintering | Setter type chosen per part (flat ceramic, V-groove, custom), loading map per tray, furnace temperature verified by surveys or test rings | Temperature and position decide final density and size |
| Post-sinter | Sizing, coining or straightening only on features that need it | Brings one or two features to ±0.02 to ±0.05 mm without machining the whole part |
The first adjustment loop
Shrink factors are averages. Your geometry will shrink slightly differently in different directions. So at T1 we measure every critical dimension on parts from every cavity, compare each with CAD, and calculate an effective shrink per feature. The steel is then corrected feature by feature. Because critical features were built steel-safe, the correction removes steel rather than adding it, and one round is usually enough. If you used a rapid prototype tool first, its shrink data goes into the production tool before it is cut, which often removes most of the correction.
Setting the part in the furnace
How the part sits in the furnace is a design decision as much as a process one. A part with a flat face parallel to the parting line sits on a standard flat setter and shrinks evenly. A part with a curved base or a long overhang needs a V-groove or a custom ceramic support that follows its shape, or a tool biased to compensate for a known sag. All of these work. All of them add cost and some variation. The best support is the one the design makes unnecessary.
What ±0.3% realistically means
±0.3% is a capability for well-designed features, not a blanket promise for every dimension on every part. Use this as a planning guide; final numbers depend on geometry.
| Feature type | Typical as-sintered capability | With sizing or secondary operation |
|---|---|---|
| Linear dimension within one wall or rigid section | ±0.3% (not tighter than about ±0.03 to ±0.05 mm on small features) | ±0.02 to ±0.05 mm by coining or sizing |
| Hole position from a datum on the same rigid section | ±0.3% | ±0.02 mm by reaming or machining |
| Dimension across a thin, unsupported span | ±0.5% or wider | Straightening or fixture sizing |
| Flatness of a long, thin plate | About 0.1% to 0.2% of length | Flattening or grinding |
| Hole diameter | ±0.3% to ±0.5% | H7 class by reaming |
| Angles | About ±0.5° | Coining or machining |
The best features for ±0.3% are short, measured within one rigid section, and dimensioned from a datum on the same section. The hardest are distances across thin spans, flatness of large plates, and features on the far side of a thickness change.
Measurement and capability
A tolerance is only real if it is measured consistently. We agree three things with customers before T1:
- The datum scheme, matching how the part locates in your assembly.
- The critical-to-function dimensions that get full measurement per cavity at T1 and capability monitoring in production. These are typically held to Cpk ≥ 1.33.
- The measurement method: CMM program, fixture, and sample size. Non-critical dimensions are covered by a general tolerance and sampling.
The trade-offs
- Tighter than ±0.3% costs money. Every feature brought tighter by coining, sizing or machining adds a step, a fixture and an inspection.
- Thin, open shapes cost yield. Parts that need custom supports load fewer per furnace run and scatter more. Sometimes a rib that adds 2% of mass saves far more in yield.
- Higher sintering temperature is not a cure. It raises density but widens shrink scatter. In 316L it raises the risk of delta ferrite and magnetism.
- Multi-cavity tools add a variable. Each cavity shrinks slightly differently, so very tight features may justify fewer cavities.
Before you put ±0.3% on the drawing
Mark the critical-to-function dimensions and their datum, and leave the rest at ±0.5% or ISO 2768-m. Send the STEP file and drawing with that marking. Within 7 working days you get a free DFM review that flags which features we expect to hold ±0.3% as-sintered, which need a secondary step, and what geometry change would avoid it, together with a tiered quote: prototype tool (4 to 5 weeks) and sintered samples (10 to 15 working days after the tool is ready), with a dimensional report on every critical feature, a pilot lot, and production pricing.
Share a link to your STEP file and drawing through our quote request form, or email leo@lichengmim.com.
