Quick answer

PTM7950 is a phase-change pad that softens at 45 °C and resists pump-out, which suits bare dies and laptops. Paste still wins for quick remounts and flat cold plates. Both need pressure: PTM7950 wants 30 to 40 psi, and ordinary cooler mounts rarely deliver that.

Contents

At a glance

Class
PC Hardware

Provenance

  • Nov 16, 2026Published

TL;DR · PTM7950 vs thermal paste

  • PTM7950 is the right answer for a bare die that keeps pumping paste out, and for anything you do not want to open again. It holds its bond line through thermal cycling and Honeywell's sheet claims no pump-out, bleed-out or dry-out.
  • Paste is the right answer for a flat cold plate you might remove, for a quick remount, and for anything where you want to adjust mounting pressure by feel. Kryonaut is silicone-based, does not cure, and is easy to clean off.
  • Pressure decides the result. Honeywell wants 30 to 40 psi for PTM7950, with 10 psi as the point where anything measurable starts happening. A cooler mount that clamps by hand may not get there.
  • The conductivity labels are not comparable. PTM7950 is rated 8.5 W/m·K; Thermal Grizzly has not published a W/m·K figure since 2020. Buy on measured thermal resistance, not on the number on the tube.

Thermal paste is a suspension of thermally conductive particles in a carrier fluid. PTM7950 is a phase-change material: a thin, firm sheet that is solid at room temperature and turns into a viscous, conformal layer once it passes about 45 °C. They are used in the same place, and they fail in different ways, which is the entire decision.

I have re-pasted enough coolers to have an opinion, but the numbers below are Honeywell’s, Thermal Grizzly’s and Igor’s Lab’s, not my own bench results. What follows is what each material is specified to do, where the published figures are not comparable, and the two conditions that decide which one you should buy.

What each material actually is

PTM7950 is silicone-free and supplied as a sheet in a 0.2 mm to 0.5 mm thickness range (Honeywell TDS). Its selling point is a property list rather than a single number: no pump-out, no bleed-out, no dry-out. That is a direct claim about the failure modes that kill paste on hot, small dies.

Kryonaut is a silicone-based paste with no curing step. Thermal Grizzly publishes its viscosity at 130 to 170 Pa·s, density at 3.7 g/cm³, operating range from −250 °C to +350 °C, and thermal resistance at 0.0032 K/W (Kryonaut datasheet). It is a conventional product, well understood, and it behaves the way paste is supposed to behave.

The mechanical difference is what matters. Paste is applied as a blob or a line and spreads under mounting pressure until it fills the microscopic gaps between two surfaces. The pad is already a sheet of a known thickness, and it thins as it heats and is compressed. A pad is a manufactured bond line; a paste is a bond line you create on the day.

The conductivity numbers are not comparable

This is where most comparisons go wrong, so it is worth being precise.

Honeywell rates PTM7950 at 8.5 W/m·K with a thermal impedance of 0.04 °C·cm²/W, both by ASTM D5470 at 125 W and 40 psi with no shim. Thermal Grizzly does not publish a W/m·K figure at all. Its product page states that the company “has not published general W/(m·K) ratings since the end of 2020” (Thermal Grizzly). That is not evasion. W/(m·K) is a bulk property, and a paste in service is a layer a few tens of micrometres thick where contact resistance dominates.

Igor’s Lab measured PTM7950 with ASTM D5470-17 on their TIMA5 tester and reported 6.374 W/(m·K), with thermal resistance falling from 0.41913 cm²K/W to 0.03427 cm²K/W as the bond line was compressed (Igor’s Lab). They also state that ordinary pastes “cannot achieve much more than 4 to 5 W/(m-K) under the usual conditions on a GPU or CPU.”

Two things follow. First, the measured 6.374 is lower than Honeywell’s 8.5 because it is an effective value that includes the bond line and the contacts, not the bulk material. You can see that in the datasheet’s own numbers: 0.04 °C·cm²/W across a 25 µm layer works out to about 6.25 W/(m·K) if you divide thickness by resistance, which is the same neighbourhood Igor measured. Second, the comparison that matters is 6.4 against 4 to 5, and it is a real gap, not a marketing one — but it is a gap in effective conductivity under pressure, not in the number printed on the box.

Pressure is the deciding variable

Honeywell is explicit that pressure drives performance. Its datasheet states that 30 to 40 psi is needed, and that “a minimum pressure of 10 psi is needed to observe a significant reduction in bondline thickness.” Below that, you are not getting the material the datasheet describes.

That is a harder requirement than it sounds. Most CPU coolers use spring-loaded screws that bottom out against a standoff, and the actual pressure depends on the spring rate and how far it compresses. Most mounts are designed to work with paste, which needs far less pressure to spread. If your cooler is a stock boxed unit with plastic push-pins, or a laptop heatsink held by four small screws into a thin chassis, you may never reach the pressure Honeywell specifies.

There is no published number from Thermal Grizzly for Kryonaut’s required pressure, which is the point. Paste is forgiving of a mediocre mount. PTM7950 is not. If you cannot control or verify the mounting pressure, paste is the lower-risk choice, and the CPU overheating guide covers how to tell whether the mount is the problem in the first place.

Pump-out, drying, and the long haul

Pump-out is the mechanism that makes a good paste go bad. The die heats and expands, the cold plate heats and expands at a different rate, and the paste is squeezed out of the interface over hundreds or thousands of cycles. Thermal Grizzly’s own guidance acknowledges the failure: replace the paste if temperatures deteriorate or if the cooler has been removed.

A phase-change material is designed around that cycle instead of fighting it. It liquefies at 45 °C, flows into the gaps, and re-solidifies as it cools, so it re-conforms on every thermal cycle rather than being progressively displaced. Honeywell’s reliability data is aimed squarely at this: the sheet lists passing a 150 °C bake for 1,000 hours, a double-85 (85 °C / 85% humidity) test for 1,000 hours, and 1,000 temperature cycles between −55 °C and +125 °C, with thermal impedance staying at or below 0.045 °C·cm²/W throughout (TDS).

Igor’s Lab’s real-world result is consistent with that. On a Manli RTX 4080 Gallardo, temperatures settled at 66 to 67 °C after burn-in. That is a single card and a single test, not a universal claim, and I would not quote it as a temperature win over paste on your specific card. It is evidence that the material performs in a real mount.

The trade-off is handling. Igor describes the pad as firm and brittle, notes that the protective films can crack it if you are careless, and recommends the pre-assembled versions with adhesive strips for exactly that reason (page 3). Paste does not have that problem. It has the opposite problem: it is easy to apply badly and easy to apply too much.

Where each one belongs

Buy PTM7950 for a bare die with a small contact area that runs hot and cycles constantly: a graphics card, a laptop, a handheld. Buy it for a machine you do not want to open again, and for any mount where you have already seen paste pump out once. Buy it when you can torque the cooler down properly.

I may earn a commission if you buy through these affiliate links. As an Amazon Associate I earn from qualifying purchases.

Search AmazonLink checked 2026-11-16 Search NeweggLink checked 2026-11-16

Buy paste for a desktop CPU under a tower cooler with a proper spring mount, for a cold plate you expect to remove within a year, and for any situation where you want to adjust the mount and try again. Kryonaut’s published properties are unremarkable in a good way: it does not cure, it cleans up with isopropyl alcohol, and you can re-seat a cooler without destroying the interface.

A word on buying the pad: the counterfeit and repackaged market for PTM7950 is real, and the material’s performance depends on it being the actual Honeywell product in the stated thickness. Buy a pre-cut sheet from a seller who states the thickness and the source, not a loose square of unknown provenance.

What I would check before buying

Four checks, and the first two decide most of the answer.

  1. Can you apply real pressure? Spring screws with standoffs, a proper backplate, and a cooler that is not held by push-pins. If the answer is no, paste is the safer choice regardless of the conductivity numbers.
  2. Has this mount already pumped out once? If you have re-pasted the same card twice and the temperatures fell again both times, that is the signature case for a phase-change pad.
  3. Is the die bare or under a heat spreader? A bare GPU die with a small contact patch is the strongest case for PTM7950. A large, flat CPU heat spreader is the strongest case for paste.
  4. What does your sensor actually say? Measure before you change anything. The CPU temperature check covers the tools, and the GPU hotspot delta census covers the edge-versus-hotspot split that tells you whether the problem is the interface or the mount.

If the temperatures are already fine, neither product will help. Repasting a healthy cooler is how people end up with a worse result than they started with. The thermal throttling fix guide is the place to start when the machine is actually throttling, because airflow, fan curves and mount pressure all come before the material choice.

Frequently asked questions

Is PTM7950 better than thermal paste?
It is better at one specific job: holding its performance over months of thermal cycling without pumping out or drying. Honeywell rates it at 8.5 W/m·K and describes it as silicone-free with no pump-out, bleed-out or dry-out issues. Paste is easier to apply, easier to remove, and cheaper. Neither is universally better, and the deciding variable is usually mounting pressure.
Does PTM7950 need a burn-in period?
It needs to pass through its phase-change temperature to work as intended. Honeywell’s material changes state at 45 °C, so a few minutes of real load is what turns the pad into a thin, conformal layer. Idle temperatures before that first heat cycle are not a useful measurement.
How much mounting pressure does PTM7950 need?
Honeywell specifies 30 to 40 psi to reach the intended bond line, and states that a minimum of 10 psi is needed to see any significant reduction in bond-line thickness. That is more than most cooler mounts apply by hand. If your cooler uses spring screws, tighten them evenly in a cross pattern until they bottom out, and check the contact pattern when you next remove it.
Can PTM7950 be reused after removing the cooler?
No. Once the pad has melted and conformed it is a one-time interface. Removing the cooler tears it, and it cannot be repositioned like a paste. Honeywell’s own handling guidance and Igor’s Lab both point to the pre-assembled versions with adhesive strips for the same reason: the material is thin, firm and can crack if you handle it wrong.
Why does Thermal Grizzly not publish a W/m·K rating?
Because the company stopped publishing general conductivity figures at the end of 2020. Its Kryonaut product page says so directly, and it publishes viscosity, density, thermal resistance and operating range instead. That is not a dodge; W/(m·K) is a bulk material property that does not describe how a paste behaves in a thin, high-pressure layer.
How long does thermal paste last before it needs replacing?
There is no universal service life. Thermal Grizzly’s own guidance is to replace the paste if temperatures deteriorate or if the cooler has been removed, and Kryonaut’s datasheet states it does not cure. In practice a good paste lasts years on a desktop. The reason people re-paste is usually pump-out on a bare die, not drying.
Is PTM7950 good for a laptop or a GPU?
Those are its strongest cases. A bare die with a small contact area and a lot of thermal cycling is exactly where pump-out ruins paste, and that is where a phase-change pad holds its thickness. Igor’s Lab measured a Manli RTX 4080 Gallardo settling at 66 to 67 °C after burn-in with PTM7950. Laptops benefit for the same reason, plus the pad does not migrate.
What thickness of PTM7950 should I buy?
Honeywell lists the series across a 0.2 mm to 0.5 mm thickness range, and the correct choice depends on the gap your cold plate leaves. Too thick and the thermal resistance climbs; too thin and it will not bridge. For a standard desktop CPU or GPU cold plate, the thinnest version the seller offers in a pre-cut sheet is usually the right starting point.

Sources and corrections

Last updated
Methodology
See our methodology for research and review standards. It draws on 7 cited sources, listed below, each checked against the original page on the date above. The material properties here come from Honeywell’s PTM7950 technical data sheet and Solstice’s phase-change selection chart, both of which state their test methods. Thermal Grizzly’s figures come from its own Kryonaut datasheet and product page. The measured conductivity and thermal resistance figures are Igor’s Lab’s, taken with ASTM D5470-17 on their TIMA5 tester, and I name the exact values rather than rounding them into a marketing claim. Where I do arithmetic on two published numbers I say so. I have not measured either interface material myself, and I quote no temperature win that a cited source does not contain. Every source was opened live on 2026-11-16.
Corrections
Spotted an error? Email a correction with the affected line and what you found. The draft includes this page’s address. Verified changes are recorded here. Contact details.

About the author

Written by Lowell K. Wood IV, who builds and runs TechFuelHQ from St. Louis, Missouri.