Thermal Grease Shootout
Thermal grease, commonly known as thermal goo (also known as thermal paste, thermal compound etc) is a fluid substance which helps to increase thermal conductivity between two surfaces.
If two surfaces made perfect contact there won’t be any need for thermal compounds. But as we don’t live in a perfect world, the contact between two surfaces is almost always not perfect. There are little crevices, surface topology variation etc that do not allow for a perfect contact. To overcome this, material is applied between the two surfaces to improve their contact and hence heat transfer.
In the world of computers thermal compounds are used between chips (CPUs, GPUs, memory, motherboard chipsets, voltage regulation circuitry etc) and heat sinks to improve the former’s heat dissipation.
All electronic circuits generate heat. High performance parts generate more. Over clocked high performance parts generate even more heat. This heat has to go somewhere. The best way to get rid of it is to transfer it to another object so that the chip remains cool (we want to protect the chip). The other substance must have a couple of properties
- 1. It must be able to transfer heat quickly (conductance –how much heat is removed in unit of time within a given temperature difference) and efficiently (thermal conductivity).
2. It must be able to withstand high temperatures without itself getting damaged in the process
Metals are usually very good at conducting heat. Thus they are usually employed as heat sinks. There is no point keeping the heat trapped in the heat sink. It will just start heating the chip as soon as the thermal balanced is reversed. To remove heat, heat sinks are either connected to fins directly, or via heat pipes. A fan usually helps to dissipate heat from the fins. This heat is thus taken away from the heat sink and the chip.
Thermal compounds help to improve the contact between the chips and the heat sink. A good thermal compound thus needs to have similar properties as the heat sinks.
EFFICIENCY OF THERMAL COMPOUNDS
The efficiency of thermal compounds depends on the type(s) of metals incorporated into them. The efficiency as noted above is measured as thermal conductivity. This is the ability of an element to conduct heat. It is measured in Watts per Kelvin per meter. For smaller contacts it can me measured in Watts per Kelvin per centimeter. (W/cmK). The higher the thermal conductivity the higher the ability of the metal to remove heat from the chips. The following is a table of metals found in thermal compounds with their thermal conductivity values
THE ANATOMY OF A THERMAL COMPOUND
All thermal compounds (except those that are in liquid form) are made up of two basic components:
1. The Vehicle
This is the medium in which the actual material(s) that are meant to conduct heat are mixed in. In most situations this is silicone grease. Sometimes synthetic oils can be used in place of silicone grease (e.g. Arctic Silver’s AS 5). Though they are employed as a vehicle, they are usually responsible for the majority of heat transfer from the chips.
2. The Conductors
This is the actual component that conducts heat away from the chips. A thermal compound may be based on any one of the following.
a. Ceramics A ceramic powder is suspended in silicone grease. Commonly used ceramics are
- i. Beryllium Oxide
ii. Aluminum Nitride
iii. Aluminum Oxide
iv. Zinc Oxide
v. Silicon Dioxide
Examples of Ceramic based thermal compounds include Thermalright Chill Factor II and III.
b. Metals As the name suggests metal particles are suspended in the vehicle. The common types of metals are Aluminum and Silver. Arctic Silver series is a good example of a metal based thermal compound (though not entirely metal based, see below). These are better than ceramic based compounds but might conduct electricity (or have a certain level of capacitance). Thus a level of care has to be taken during their use. They might not be suitable for certain circuits e.g. those involved with power regulation.
There are other conductors like Carbon particles (Diamond dust) and liquid metal conductors. The latter is very expensive and the former is usually not commonly used.
In most cases the most important factor that determines the conductance is not the use of the conductors itself but the vehicle in which the conductors are suspended.
PROPERTIES OF A GOOD THERMAL COMPOUND
How good (or bad) a thermal compound is, is usually determined by its ability to conduct heat. However that is not the only quality that one should look for in a thermal compound. Most of what makes or breaks a thermal compound is dependent on the characteristics of the vehicle (the fluid component) used. A good thermal compound, apart from conducting heat efficiently, must also have the following qualities
- 1. Ability to fill gaps between the two surfaces
2. Ability to adhere to the surfaces
3. Ability to resist drying and flaking
4. Ability to perform consistently
5. Ability to not conduct electricity
6. Ability to resist oxidation
7. Ability to spread easily on application
Almost all thermal compounds sold today adhere to these properties. The only two that are variable are #5 and #7. Arctic Silver 5 (AS5) is known to have a certain level of capacitance, while Ceramique is thicker and relatively more difficult to spread as compared to other compounds (coincidently both come from the same manufacturer –Arctic Silver). There are compounds that are easier to spread than others. This is dependent on the viscosity of the vehicle used
APPLYING THERMAL COMPOUNDS
There are many ways in which to apply thermal compounds. The exact technique is usually determined by personal preferences. Ideally the technique should be determined by:
- 1. The compound being used
2. The material on which the thermal paste is being applied (i.e. Heat Sink)
1. The Compound used
Compounds that spread easily can be applied by either using a 1/3 the size of a pea gob of thermal paste at the center of the processor heat spreader or a line drawn across the heat sink spreader (running across the cores)
Compounds that are of high viscosity (thicker compounds) like Arctic Silver’s Ceramique are better applied by making an “X” at the center of the heat spreader.
2. The Heat Sink Used
a. The Standard Intel Design Coolers
These are the ones with a central copper core. For these a small gob of thermal compound (1/3 the size of a pea) in the central is the way to go
b. The Square Heat Sink Bases
For these either use an “X” pattern (thick compounds), or a pellet sized gob/ line pattern (thin compounds).
c. Direct Touch Heat Pipe Coolers
These are usually budget implementations. The problem with these coolers is the presence of crevices between the heat pipes. The best idea is to apply the thermal paste onto the heat sink base and spread it out. Spreading out might leave air bubbles behind, but in any case any area of non-contact (i.e. the crevices) will have air trapped in it. It is a good idea to use thicker (higher viscosity agents) on these coolers as they tend to stay in their place. The following pictures show how the line method works for these cooler.
Cooler Master Hyper 212 plus Cooler
A straight line of thermal compound applied across the cooler’s base
The compound is spread using a credit card (one that is not in use!)
Thermal compound on the processor after the cooler was installed and eventually removed to see the result
REMOVING THERMAL COMPOUNDS
Removing thermal compounds is much easier than its application. Either use one of the commercially available thermal compound removers (e.g. those from Arctic Silver) or use an alcohol swab (easily available from your chemist). Once the surface is clean don’t forget to reapply thermal compound before reassembly!
THE MAGIC OF A LITTLE GOB
Think it is not possible for a little teeny (smaller than a BB pellet sized) gob of thermal paste to spread over a processor heat spreader? Watch the video for proof
Remember that the use of a thermal paste is a compromise solution. In an ideal world the metal to chip contact would be perfect thus negating the use of thermal goo. You need just enough to ensure good contact between two surfaces.
THE SHOOT OUT
The shoot out involves the following participants
*The exact nature of the cooler master thermal compound is unknown. This is the standard compound that comes bundled with its Hyper 212 Plus cooler.
Only two products (both from Arctic Silver) have recommended curing times (the time before the thermal compound is able to achieve maximal thermal conductance). For Ceramique the time recommended is 25 hours of use and for AS5 it is about 200 hours of use.
ARCTIC SILVER AS5 & ELECTRICAL CIRCUITS
The only product in the shoot out which can not be used near electrical traces, pins and leads is AS5. AS5 can be considered to be a mixed compound as it composed mostly (>99%) of Aluminum and also consists of little amounts of Zinc and Aluminum oxide among other ceramics. These are suspended in a synthetic oil rather than silicone. Though it does not conduct, it is capacitive and can cause problems if it bridges two electrical contacts.
Not all tests are conducted in an ideal environment and testing thermal compounds has a lot of subjectivity and other strings attached to it.
For starters each thermal compound was tested three times and under similar ambient temperatures using the same system. This can produce standardized results within these parameters, but will not be applicable to different ambient temperatures and other systems in use.
It is important to understand that the shape of heat spreaders on processors and heat sink base can be variable even within the same product line. No heat spreader or heat sink lapping was performed during the test. The products were tested as they were provided by the manufacturer.
The pressure applied on the heat spreader can be variable even when using the same test bed due to human factor
The method of thermal paste application can also impact performance. Though the method used for each paste was the one that was personally tested to be the best, it might not be so.
The products themselves were shipped from various parts of the world and might have been exposed to higher storage temperatures than the manufacturers recommend and thus affect their performance.
Finally the way the test was conducted, it did not measure the efficiency of the thermal paste. It measured the temperature of the processor and the degree to which it varies between various compounds. As all other variables (test bed) were similar, the differences in results should only reflect the ability of the thermal paste to transfer heat. Ideally thermal conductance of each compound should be measure to really test its caliber. This is beyond the scope of this article and the means of the reviewer as well.
Finally as the test was done over a period of a month, the longevity factor (ability to perform consistently and not degrade over time) was not ascertained.
Though it can not really be counted as a limitation, but personal bias is a factor that can easily tilt results. I have been using Arctic Silver AS5 and Noctua NT-H1 for a very long time. I have used The Chill Factor a couple of times. This was my first encounter with the rest of the products reviewed here. As all thermal pastes come in identifiable packaging I could not blind myself to them. I could have asked another investigator to apply thermal paste, but that would have added another form of bias. In the end I had enough faith in my (moral) self to conduct the tests myself.
The thermal compound was only applied between the Processor and the cooler.
The test was run 5 times per thermal compound. The temperature of the hottest core during each run was averaged over the 5 runs. Each thermal paste was tested thrice, thus producing 15 discrete results for every thermal compound.
The results are presented relative to the best performer. The difference between this and the rest are shown graphically. The base line for the best performer is taken as “0” and all other results are relative to this.
The results should come as no surprise as far as the winner is concerned. The best performer is Arctic Silver’s AS5. Despite its disadvantage, it is still the best performer amongst the pack.
The rest of the products fall with in a maximum of 2°C of AS5, which is not really much. For all intents and purposes any one of these compounds can be used with a fair degree of confidence. They all do what they are designed to do –improve heat transfer between the processor (Integrated Circuit) and the heat sink.
This was perhaps the most difficult review I have done. The amount of products at hand, the number of applications of the compounds and the variables needed to be kept under control all amounted to a higher level of difficulty. Test parameters were kept under control as much as possible (especially ambient temperatures – I had to cool down or at times wait for the room to become hot enough so that the temperature for all tests remained the same). Only one investigator (I, myself) applied the thermal paste and installed the cooler.
Despite test limitations, the results do show that all thermal compounds are not made equal. They also show that the difference between the performers is not very significant. Even if you don’t have access to the best thermal compound in the world, the ones bundled with budget coolers will also suffice.
If you do, however, have a choice, the better performing thermal compounds would do better with better prepared surfaces (heat sinks and heat spreaders).
For almost 90% of us the type of thermal compounds should not really matter. The differences are not that huge to loose sleep over. For the 10% who must have the squeeze out every centigrade of performance the chart shows who is who of thermal compounds.
This article owes a lot to Olin Coles of Benchmark Reviews. My work pales in comparison to what he has done regarding thermal compound testing. Though my results are my own, I did learn a lot from his articles. Thank you Olin.