How Do You Select a Thermal Interface Material (TIM)?

Selecting a thermal interface material starts with three things you already know about your assembly: the size and variation of the gap, the pressure the stack-up can apply, and whether silicone is acceptable near your optics, contacts or sensors. Thermal conductivity is the fourth question, not the first. A 12 W/m·K pad that cannot conform to a varying gap will run hotter in service than a 3 W/m·K pad that fills it completely. This guide walks the decision in the order a design engineer meets it.

Is your bottleneck the material or the interface?

Before changing material, split the thermal path and measure where the resistance actually sits. The path runs die junction → package internals → case or lid → thermal interface material → heat sink or cold plate → ambient or coolant. Only one of those links is the TIM.

The reason this step matters is that the interface is invisible in a datasheet. Two mating surfaces that look flat are microscopically rough, and the gaps between them are filled with air — roughly 0.026 W/m·K, against about 401 W/m·K for copper [Industry reference]. A thermal interface material earns its place by displacing that air. If your measured bottleneck is the cold plate flow rate or the package itself, a higher-conductivity TIM will not fix it.

Assembly pressure is the variable most often left out of this analysis. A material characterised on a test fixture at high clamping pressure behaves differently under a low-pressure cold-plate mount, because it no longer wets the surface the same way. Where possible, measure interface thermal resistance under your real assembly conditions rather than relying on a headline conductivity figure.

What gap and what pressure? Matching the form factor

Form factor is decided by gap range and by how much compression force the assembly can tolerate — not by conductivity. Pads suit repeatable, defined gaps; dispensed gap fillers and gels suit gaps that vary across a board; phase change materials suit thin, high-pressure bondlines under a heat sink.

The table below maps Henkel’s BERGQUIST and LOCTITE thermal interface range against those two axes. Thickness tells you the gap the material is built for; Shore hardness tells you how much stress it will impose on the component while it fills that gap.

[Source: Henkel, “lt-6885 Materials for Telecom and Datacom Optoelectronics”; “Material Solutions for AI Data Center”; “lt-8442 Cloud/Hyperscale Datacenter Switches, Routers and Servers”.] Values are for selection comparison. Confirm against the current datasheet revision for your grade before design-in.

Two patterns are worth reading out of that table. First, hardness and conductivity do not move together: TGP HC5000 is softer than TGP 7000ULM but carries less conductivity, so a fragile component may be better served by the softer, lower-conductivity grade. Second, the dispensed materials cover gaps a die-cut pad cannot — if your gap varies across the board because of substrate warpage or stack-up tolerance, a gap filler or gel removes the thickness decision entirely.

When do you have to avoid silicone?

Avoid silicone where volatile siloxane can reach an optical surface, an electrical contact, or a sensor. In optical modules, cameras and photonic assemblies, migrated siloxane deposits on lenses and windows and raises insertion loss over time. On relay and connector contacts, it can form an insulating film under arcing.

Outside those cases, silicone is usually the better engineering choice: it stays compliant across a wide temperature range and dominates the high-conductivity end of the range. The decision is therefore not “silicone is bad” but “is there a contamination-sensitive surface in this enclosure?”

Three routes exist when the answer is yes. A fully silicone-free pad such as BERGQUIST GAP PAD TGP 3004SF removes the chemistry altogether. A non-silicone phase change material such as LOCTITE TCP 4000 D or LOCTITE TCF 4000 PXF does the same in a thin, high-pressure bondline. Where silicone is tolerable but outgassing must be controlled, a low-outgassing grade such as BERGQUIST GAP FILLER TGF 3500LVO is a middle path.

Note the trade-off in the table: the silicone-free grades sit at 3.0 and 4.0 W/m·K, while the silicone gap fillers reach 14.5 W/m·K. Removing silicone currently costs conductivity, so the contamination risk has to be real before you pay for it.

What fails first: pump-out or compression set?

Different form factors age in different ways, and the failure mode determines what your qualification test must look for. Greases and some pastes pump out; pads take a compression set; phase change and cured gels are designed to resist both.

Pump-out is the gradual migration of material out of the interface under repeated thermal cycling, leaving the centre of the contact area starved. It is most likely where the bondline is thin, the assembly pressure is low, and the duty cycle swings often — which describes a large part of modern liquid-cooled hardware. Henkel positions its phase change materials directly against this: they apply with the convenience of a paste but, in the manufacturer’s words, do not suffer the pump-out and long-term thermal degradation seen with grease [Source: Henkel, “lt-8442 Cloud/Hyperscale Datacenter Switches, Routers and Servers”]. LOCTITE TCF 4000 PXF is specified as showing no pump-out, dry-out or pull-out.

Compression set is the pad equivalent: after long compression at temperature the material does not fully recover, contact pressure relaxes, and interface resistance creeps up. This is why the ultra-low modulus grades matter — they reach the required contact with less stress, and have less stored stress to relax.

The practical consequence is that measuring initial thermal resistance is not a qualification. Thermal cycling to the duty profile the product will actually see, then re-measuring, is what separates a material that holds from one that does not.

Can one material handle both heat and EMI?

Yes, but you pay for it in conductivity. An EM-absorbing pad has to carry absorber loading that displaces thermal filler, so its conductivity sits below a purely thermal pad of the same thickness.

BERGQUIST GAP PAD TGP EMI4000 illustrates the trade cleanly: 4.0 W/m·K with electromagnetic absorption to 77 GHz, in a 0.75 – 2.0 mm silicone-free pad. Against the 12 W/m·K of a purely thermal grade at similar thickness, roughly two thirds of the conductivity is given up to gain the absorption. In a radar or millimetre-wave module where the alternative is a separate absorber layer plus a separate TIM — and therefore an extra interface — that trade is usually worth making. In a module with no EMI requirement, it never is.

Where do carbon-based and boron nitride TIMs fit?

They are filler technologies, not a separate product category, and they mostly explain why a given grade behaves the way it does. Carbon fillers such as graphite and graphene conduct exceptionally well in-plane but far less through thickness, which makes graphite sheet good at spreading heat sideways across a phone or a battery module and poor at moving it from a die into a heat sink. They are also electrically conductive, so they need insulation where they sit near live parts.

Hexagonal boron nitride takes the opposite position: lower peak conductivity than carbon, but electrically insulating, chemically inert and stable at high temperature. That combination is why BN-filled grades appear in power electronics, where the interface has to conduct heat while withstanding voltage.

For a selection decision, the useful summary is short. If you need lateral heat spreading, look at carbon. If you need to conduct heat through an interface that must also insulate, look at BN-filled silicone grades. Neither changes the gap range, the assembly pressure or the silicone decision — they only change which grades are candidates.

A pre-specification checklist

• Split the thermal path and confirm the interface is genuinely the bottleneck before changing material.
• Record the gap as a range, not a nominal value. Include substrate warpage and stack-up tolerance.
• Record the assembly pressure your mounting can actually deliver, and characterise candidates at that pressure.
• Decide the silicone question early. It removes or restores whole branches of the range.
• Check whether the joint must be reworkable — it rules some grades in and others out.
• Qualify with thermal cycling, not with initial resistance alone, and know which failure mode you are looking for.

Frequently Asked Questions

What is the difference between thermal conductivity and thermal impedance?
Thermal conductivity (W/m·K) is a property of the material itself. Thermal impedance is the resistance of the actual joint, and it includes the bond-line thickness plus the contact resistance at both mating surfaces. A high-conductivity pad that cannot conform to your gap can still deliver worse impedance than a softer, lower-conductivity one.

How do I choose between a thermal pad and a dispensed gap filler?
Pads suit gaps that are known, repeatable and available as a stocked thickness, and they keep manual assembly simple. Dispensed gap fillers suit gaps that vary across the board, large contact areas, and automated lines where volume is metered per part. The variation in the gap, not its nominal value, is usually what decides.

When is silicone-free worth the loss in conductivity?
Where volatile siloxane can reach an optical surface, an electrical contact or a sensor. In this range the trade is explicit: the silicone-free grades sit at 3.0 and 4.0 W/m·K while the silicone gap fillers reach 14.5 W/m·K. If nothing in the assembly is siloxane-sensitive, silicone is usually the better engineering choice.

What should a TIM qualification test look for?
The failure mode that matches your form factor. Dispensed and grease-like materials are qualified against pump-out, so cycle them thermally and look for a starved centre contact area. Cured pads are qualified against compression set, so check whether the material still recovers enough thickness to fill the gap after cycling. Initial thermal resistance alone tells you neither.

Related reading on this site: Three Misconceptions About Heatsink Selection covers the same interface-resistance blind spot from the heat sink side, and Dow TC-3120 thermal gel for data centres looks at one dispensed gel in a data-centre context. For the wider assembly picture, see our Consumer Electronics solutions and the full Thermal Interface Materials range.

If you have the gap range, the assembly pressure and the duty cycle to hand but are not sure which form factor they point to, discuss your application with an engineer — send us those three numbers and we will narrow the candidates with you.