Addressing the complaint of compression damage to sinks and CPU/IHS alleged to be caused by IC Diamond particle contamination
Introduction
After a full investigation by Innovation Cooling it was determined to be impossible to recreate the asserted damage as presented by any particle by compression whatsoever under a worst case model.
Detailed examination of the damage was performed and a profile of the particle was developed that allowed an approximate matching model pieces to be manufactured
The investigation encompassed impressing assorted forms of copper and found that cold rolled, forging or stamping processes hardened the IHS and sinks to a degree that the load required to recreate the asserted damage exceeds the Intel max pressure load specification of 75 lbs.
With applied loads directly on the modeled pieces we were unable to do more than a nearly negligible scuffing impression of surface irregularities with zero embossing imprint into the solid copper itself.
Here are the two complaints the first here on the IHS
The second
to normalize the pictures we took the first picture and auto rotated it 90 Deg clockwise with the gimp preset without any scaling or other changes
Then we pasted it alongside the second image for comparison.
Then marked top and bottom points on each.
Then using the gimp measuring tool measured center point to center point.
The one on the left measures 112 pixels at an angle of 68.87 degrees.
The one on the right measures 112 pixels at an angle of 68.67 degrees.
While both look exactly the same length in scaling/ratioing the pictures, I estimated the one on the left to be approximately 2mm in length and the one on the right to be 3mm.
The image the second image was then scaled +50% and then pasted into the composite to match the 3mm dimension and the top turkey leg width dimension at the maximum point is close to within a couple pixels to identical in width to the dumbbell upper section.
Ratioing the width on the 2mm long dimension I got an average width of .043mm (0.016 in.)
Initial Observations:
After normalizing the two images they then were compared for similarities to model the type of material and conditions that could have caused it. The top left has a kind of dumbbell shape wider at the ends and thinner in the middle the second has a more turkey leg appearance. At the 3mm size the Upper Turkey leg area matches the dumbbell within a couple of pixels. The lower part of the turkey leg was problematic to duplicate as we could custom file the approximate shape clamping the piece to do so was non viable . Matching the particle exactly would increase the test piece surface area some, perhaps a square mm but for the sake of convenience and argument a 3mm straight piece will suffice.
The pictures are a straight forward example of embossing, pressure creating an imprint in the copper. Both shapes have a defined edge and defined flat bottom indicating a hole of some depth. To model the particle to make that sort of imprint you have to mentally extrude the profile and then locate similar particles. It would have a matching outline profile with straight uniform edges and a flat regular bottom. The particle while looking the same are very different, the dumbbell has softer rounded edges perhaps indicating it was exposed to some kind wear while the turkey with it's sharper points looks more like a raw piece with no abrasive action.
The depth of the hole was bothersome as it being it's most distinguishing feature and hard to estimate is it .005? or .010? Along with the flat bottom detail suggests a fairly regular edge in contrast to the outlined shape. All in all a fair ballpark hole depth number might be in the .005 to .008 range with a particle depth and perhaps a actual particle depth size of 2X that in the .010-.020 range. .
Particles selection:
It's not an over sized diamond due to the shape as diamond comes in the form you see below so was the first elimination.
Selected test Particles:
We then embarked on a particle hunt. Immediately, all 'naturally' occurring particles were eliminated because 'natural' particles can be round, disk-shaped, exist as chips or flake-like but would not exhibit 'dumbbell' or 'turkey leg' shapes as these are highly unstable configurations.. The other deciding factor was that on our preliminary particle testing
hard does not mean tough Quartz with an MOHS of 7 would start breaking up at 10 lbs pressure and continue breaking through full range of pressures. At the 70 pound range, granite would just disintegrate etc. Try taking your wife's diamond and hitting it with a hammer or squeezing it in a vice and then posting the pictures, The results would be interesting
As the damage represents classic pressure embossing, the particle selected had to be flat to mimic the flat bottom hole expressed in the pictures.
The shape that caused this damage does not resemble anything one would find in machine shop chaff from milling, drill press grinding etc. In addition, neither aluminum nor copper chaff would not be hard enough to emboss an impression so the selection would have to be elsewhere. While it was easy to draw mental picture of the particles it is less so to to imagine a source for a man-made configuration of that type.
So, in selecting a particle model for the tests, a paper staple chosen. Approximate (MOHS of 5, copper MOHS 3). The staples were cut and then filed to 2 (.08 in) and 3mm (.122in.) sizes (2.03mm and 3.11) width is .53mm (.021 inches) thickness is .041mm (.017 inches)
In selecting a staple a material A staple was chosen because a material was required which would be tough enough to handle high stress without breaking, cracking or deforming/losing shape and a staple is also of the approximate dimensions of the alleged particle damage.
Your typical steel staple is pretty tough having been work hardened like the forged hammer you would use to bang a piece of copper into shape
Below are the test pieces utilized in our tests - 3mm above the date 2mm below the date
A test model was devised to investigate exactly what it would take to create those marks. Below is a picture of the general set up
Test set up and conditions
1) A 150lb. Capacity digital scale shipping scale to measure pressure
2) 1.35mm thick cold rolled copper sheet to model the IHS
3) Anvil to apply even pressure 5/16th drill stock ( we used stock vs an actual drill as the drill twist would not allow it to chuck up level)
4) Drill press to apply & control pressure.
5) 2 and 3mm paper staple particle for testing impressions.
The digital scale was calibrated against a separate Push/Pull spring gage on digital scale center and was found to be in agreement within approx 1 pound. The center of the scale was located, marked, and placed directly under the anvil. A 5in. sq piece of copper sheet was used as a base to uniformly spread the load to minimize any hysteresis that might be introduced into the measurement.
For the purposes on the test, 70 lbs was selected as a max point. In the four years Innovation Cooling has been doing public pressure testing, the lowest pressure readingt hat has been measured is 28 PSI and only 2 in the low 70 lb range and on a bell curve the mean is the 55 PSI range.
Applied pressure was held for 10 seconds to allow the copper to flow as copper is incompressible.
Calculated test pressures vs copper pressure thresholds
70 lbs. 2mm area is 0.00168 70/ 0.00168 =41,667 PSI.
70 lbs .3mm area is 0.002562 70/0.002562 =27,322 PSI
Any cold-rolled or cold-forged copper is a hardened copper vs cast copper will have vastly different yield and tensile strengths. Whereas you may be able to make an appreciable embossing mark on pure cast copper at 20,000 psi, on a cold, form-rolled piece the range is much higher in a general range of 50-67,000 PSI. However as we were unable to make any appreciable/measurable embossing marks at 41,667, PSI we will fairly assume we are at some higher PSI number than our indicated test pressures of 27k and 42k PSI.
Cold-forging/stamping sheet copper will further harden the copper and some sources indicate yield strengths to 80,000 PSI including to as high as 180,000 PSI This is important to note as the IHS had the lightest impression which is of all indicative of of the higher, cold working pressures (harder).
A range of 10 heat sinks also tested for hardness and all had been through some kind of cold working process forging, stamping, squeezing or drawing etc. and exhibited the equivalent hardness as seen on the modeled IHS.
For example Innovation Cooling uses Cold-rolled (hardened) copper bar stock to make our heat sinks.
The test set up provided enough control to imprint to a tenth of a pound of designated pressure with a high degree of repeatability between test samples.
The test sample below is with drill stock anvil @ 70 Psi of a 2mm and 3mm length with human hair (.003 in.) and a penny for scale. Viewed from a normal perspective, they appear as minor scuffs on the surface without any depth.
Close up view detail.
Couple of things to note here – the 2mm marks have better definition as they have 42k PSI vs the 27k pressure load of the 3mm marks.
The 3mm made absolutely no embossing imprint into the metal surface and only slightly marred the roughened surface features.
The 2mm demonstrates only a marginal embossing print and when compared to the measured .003 in. human hair or either particle an easy ballpark estimation of well under .001 can be made. And When Measuring with a microscope from the top edge to the bottom mark from the highest point we could find, the depth measured .00025 in. Notice that scratch follow through on the imprint and other surface features remain intact.
To then model our particles on an actual IHS we applied the max load of 70 PSI.
At this point we encountered an immediate difference with the 3mm and were unable to make an identifiable mark @ the 27K PSI after repeated tries.
Upon examining the IHS its appearance was more to the look of a forged piece indicating a further hardening beyond what you would experience with a simple cold rolled/worked piece.
So at this point , the 2mm piece was tested the several times and these are the two best partial prints we could manage.
2mm scuff with combined close-up view produces zero surface penetration @ 70lbs.
View of the second partial print w/ human hair for scale @ 70 lbs..
It was estimated that to fully recreate the alleged particle damage one would most likely have to be in the 200 lb range (+100k PSI more or less) or enough to plasticize the copper to make a workable impression.
This is 300% three times the maximum recommended Intel board load and over 4 times the Intel nominal sink loading of 40-45 lbs.
The general concern here would be of fracturing the MB at these dangerously elevated pressures.
General discussion and limitations of test
The primary condition of the test setup deals with a more than worst case scenario beyond what you could possibly experience in real life as the test comprises a tool steel anvil pressed flat and level against a test piece and no where else. In any event it does the job of bringing understanding as to the limits and level of difficulty involved in embossing/damaging at the nominal pressure loads hardened copper.
There were several other limiting factors that would support the fact that the compression damage as presented could not be possibly be the result of any contamination of any kind but these factors are beyond the scope of this effort.
A key point here requires the understanding and emphasis of the design considerations of the IHS. The IHS's primary role is to protect the CPU both thermally and mechanically. For the thermal aspect, besides the use of conductive copper in heat transfer, close mechanical tolerances are required for CPU contact to the IHS. Intel has a specification requirement of // .002 or Flat and Parallel to within two thousandths of an inch across the surface of the entire IHS.
This kind of tolerance is the hallmark of a forged part and whether actually forged or manufactured with a forge-like process is immaterial, but what is most important i in the employment of this process is that it is the one chosen when it is desirable to make the structurally hardest copper part possible. Mechanically, you require a piece that will contact the CPU, not flex under load or bend under stress; you need a piece that remains stable during the assembly processes and is resistant to cosmetic dings and scratches etc.
With a typical cold rolled hardened piece, as you would see in a heat sink, to make an impression of a .005 depth is likely in the approximate pressure range of 140 -150lbs., or 70-85K PSI.
For a forged piece like the IHS a conservative number of 200 lbs + would have to be applied to recreate a + .005 depth.
When one fully understands the hardness of the copper used in heat sinks and an IHS it is easy to understand the extreme pressure required to emboss them. One can only conclude that as the required pressures to recreate the purported particle/compression damage so far exceed the typical operating range that no particle damage can be assumed and the marks were made by some alternative means.*This being the case there was no particle contamination possible.
Innovation Cooling takes vigorous exception to the suggestion of particle contamination of our product IC Diamond from either our suppliers, contract mixer, or in house processing. Each stage is held to the highest level of standardized industry practice. The diamond utilized in our compound comes from the USA's largest supplier and is of the same quality and is in fact exactly the same material used in lapping high end optical equipment. To imply that large random over sized diamonds or contamination of any kind are part of the mix is an indicator of ignorance of the standardized screening process that have been in place for the last hundred years. This would be like finding silver nuggets in your silver compound or 100 grit particles on your 400 grit sandpaper. It it is extremely unlikely to happen.
Our supplier conforms to standard mil specifications in manufacturing and processing.
Our contact mixer QC's all material for size, shape, consistency etc. to mil spec. upon receipt. We pay for that service. Then mixes the material in a clean room environment on million dollar mixing machines with the material heated and under vacuum.
The compound is received in sealed containers that are then tapped in a similar manner as you would a beer keg. Material is then pumped through a line directly through a valve into a medical syringe and immediately capped
Conclusion
Under a worst case scenario with extreme pressures we were unable to recreate whatsoever the alleged damage IC Diamond was purported to have caused.
Cold-worked, hardened copper sharply limits any significant damage to above standardized maximum pressures
The marks were not caused by particle contamination and were made by some other means and as such no particle contamination can be assumed.
Replacement offer
After Investing the time and effort into this exercise an interest has grown to acquire the actual pieces and we would like to buy the components for full replacement value for a more detailed examination and will of course pay for all shipping charges.
