Preprint typeset in JINST style - HYPER VERSION

Thermal and tensile strength testing ofthermally-conductive adhesives and carbon foam

Maxwell Chertok∗, Minmin Fu, Michael Irving, Christian Neher, Mengyao Shi, KirkTolfa, Mani Tripathi, Yasmeen Vinson, Ruby Wang, Gayle Zheng

University of California, DavisOne Shields Ave., Davis, CA 95616, USA

∗ E-mail: chertok@physics.ucdavis.edu

ABSTRACT: Future collider detectors, including silicon tracking detectors planned for the HighLuminosity LHC, will require components and mechanical structures providing unprecedentedstrength-to-mass ratios, thermal conductivity, and radiation tolerance. This paper studies carbonfoam used in conjunction with thermally conductive epoxy and thermally conductive tape for suchapplications. Thermal conductivity and tensile strength measurements of aluminum-carbon foam-adhesive stacks are reported. Initial radiation damage tests are also presented. These results caninform bonding method choices for future tracking detectors.

KEYWORDS: Thermal conductivity; Tensile strength; Silicon tracking detectors.

arX

iv:1

608.

0827

9v2

[ph

ysic

s.in

s-de

t] 1

Sep

201

6

Contents

1. Introduction 1

2. Design of Thermal Testing Apparatus 22.1 Temperature Measurement 2

3. Sample Preparation 43.1 Epoxy Samples 43.2 Thermally Conductive Tape Samples 4

4. Thermal Testing Procedure 5

5. Thermal Resistance Results 65.1 Epoxy Results 65.2 Thermally Conductive Tape Results 7

6. Tensile Tester & Results 8

7. Radiation Damage Testing 9

8. Discussion 10

9. Acknowledgments 10

1. Introduction

Carbon foam, thermally conductive tape, and thermally conductive epoxies are materials well-suited for particle detector mechanical structure applications due to their good thermal conduc-tivity, structural characteristics, and radiation hardness. The thermal properties of adhesives andcarbon foam are well known separately. Thermal conductivity k (see Table 1): 3M 8805 tape:k = 0.6 W/m ·K [1]; 3M boron nitride loaded epoxy TC-2810: k = 0.8− 1.4 W/m ·K [2]; 10%dense carbon foam: k = 26 W/m ·K. To create accurate heat dissipation models of components forpotential use in future detectors, experimental data are needed to determine how different combina-tions of materials affect resulting collective thermal and mechanical properties. This paper presentsmeasurements of thermal and mechanical properties of combinations of such materials. We alsoreport first radiation hardness results for these structures.

In general, when two or more materials are combined in a single sample such that heat flowssequentially through each material, the thermal properties of the composite are readily calculable.In the case of the combination of carbon foam and epoxy however, the two components do not

– 1 –

remain separated. Instead, the epoxy is observed to wick into the carbon foam to such a degreethat this method of prediction is inadequate. Although the thermal conductivities of the carbonfoam and epoxy are known separately, it is difficult to predict the thermal conductivity of thecombination since heat propagates through the epoxy and carbon foam structure both sequentiallyand in parallel. This motivates the necessity of a device to determine the thermal performanceexperimentally.

Thermally conductive tape is an alternative to epoxy and could potentially be used for bondingelements of detector modules or support structures. Upon application, thermally conductive taperemains as a solid layer and does not wick into the carbon foam, which results in a weaker bond.However, this adhesive is desirable for its ease of use and good sample-to-sample repeatability ascompared to epoxy and other adhesives. We present an alternative to such adhesives in anotherpaper [3].

Characterizing bond mechanical strength is important to ensure strength and durability of themodules and mechanical structures following assembly and after substantial radiation exposure. Apneumatic tensile tester was designed and used to measure the ultimate tensile strength of epoxyand tape bonded carbon foam aluminum stacks.

2. Design of Thermal Testing Apparatus

To measure the thermal resistance and conductivity of a sample, we must determine the temperaturedrop ∆T across the sample when a known amount of power (heat) P flows through it. Table 1defines these quantities for future reference.

Quantity Symbol Formula

Thermal Resistance R ∆T/PThermal Conductance C 1/RThermal Conductivity k Cd/A

Table 1: Definitions of resistance, conductance, and conductivity. A is the cross sectional area ofthe sample [m2], and d is the thickness of the sample [m].

2.1 Temperature Measurement

For the thermal measurement, we sandwich the sample between two stainless steel blocks to createa vertical stack. A heater sits on top of the upper block and sends heat down the stack while aPeltier cooler and fan attached to the bottom of the stack draw heat away. The cooler’s purpose isto center the stack at room temperature in order to reduce heat exchange with the environment. SeeFigures 1 and 2 for a schematic of the stack and photograph of the apparatus.

A set of 4 thermistors imbedded within the steel blocks allows us to track the temperature dropacross the sample, 2 in each thermal block. The temperature at each thermistor is calculated usingthe Steinhart-Hart equation [4]

T =1

a+b ln(Re)+ c[ln(Re)]3, (2.1)

– 2 –

where Re is the temperature-variable resistance (Ω) of the thermistor in question and a, b and c arethe Steinhart-Hart constants included in the manufacturer’s specifications for the thermistors.

A 24-bit temperature measurement USB data acquisition device (DAQ) from MeasurementComputing collects the temperature data at 0.5 s intervals from the four thermistors. They arereferred to, from top to bottom, as thermistors A,B,C and D, respectively. By using the locationsof the thermistors and assuming a linear temperature dependence through each of the blocks ,wecalculate the temperature differential across the sample using the following 3 equations:

Ttop = TB −TA −TB

yA − yB(yB − ytop) (2.2)

Tbottom = TC +TC −TD

yC − yD(ybottom − yc) (2.3)

∆T = Ttop −Tbottom (2.4)

TA, TB, TC, and TD are the temperatures (K) measured from each of the four thermistors, yA,yB, yC, and yD are the vertical positions of the four thermistors, while ytop and ybottom denote thepositions of the top thermal block’s interface and the bottom thermal block’s interface respectively.The positions are shown in Fig. 1.

Figure 1: Schematic of stack for thermal testing. Figure not to scale.

– 3 –

Figure 2: UC Davis thermal tester apparatus. The upper and lower thermal blocks are insulated toreduce heat exchange with the environment. The DAQ is read out on a PC, not shown.

3. Sample Preparation

This section describes the method of preparing the samples used in the thermal and mechanicaltesting apparatus. Samples were bonded using either boron nitride loaded epoxy or thermallyconductive tape.

3.1 Epoxy Samples

Epoxy samples consist of two 25mm x 25mm x 2.5mm 6061 polished aluminum coupons sand-wiching a 25mm x 25mm x 4mm piece of carbon foam, bonded with 3M boron nitride loadedTC-2810 epoxy on both interfaces between aluminum and carbon foam.

We use an EFD Nordson robotic dispenser to apply the epoxy to the aluminum coupon in linesat a speed of 10 mm/s. (Figure 3.) The sample is then placed into a jig with a specified depth, anda teflon squeegee is drawn over the sample and edge of the jig. This leaves a consistent layer ofepoxy. (Figure 5) After both aluminum coupons are prepared, the carbon foam is sandwiched inbetween, a 200 g mass is placed on top of the stack, and the sample is left for 24 hours to cure.

3.2 Thermally Conductive Tape Samples

The thermally conductive tapes used were 3M 8805 and 3M 8810. The intrinsic properties of thesetwo tapes are identical, but the 3M 8805 tape has a thickness of 125 µm while the 3M 8810 tapehas a thickness of 250 µm. We apply the tape to two aluminum coupons of the same dimensions,and sandwich the carbon foam piece in between. Following the procedure obtained from 3M, wethen apply a pressure of 50 PSI for 5 seconds to ensure an optimal bond.

– 4 –

Figure 3: Left: EFD Nordson robotic dispenser applying epoxy to aluminum coupon. Right: jigand epoxy sample on aluminum coupon after squeegee procedure.

4. Thermal Testing Procedure

The cured sample is inserted into the thermal tester apparatus, with Arctic Silver 5 thermally con-ductive paste applied to each sample–stainless steel block interface, as shown in Figure 4. A 1kg mass is placed on the weight table (centered directly over the vertical axis passing through thesample stack) and the system is left for 18 hours to settle as the thermally conductive paste spreadsout. Repeated testing of the same sample indicates that variations in the thermal paste’s thicknessresult in no more than 2% variation in measured thermal resistance.

Figure 4: Sample inserted in apparatus, ready for thermal testing. The stainless steel blocks withembedded thermistors are encased in insulating foam.

After the requisite settle time, data are collected for 80 minutes, adequate to reach steady stateheat flow for this setup. The heater and cooler are adjusted to ensure the sample is maintained nearroom temperature and to maximize the temperature difference across the sample, increasing signal

– 5 –

to noise. In practice, 3K is a sufficiently large temperature drop to provide good results withoutdamaging the heater or Peltier. The data file outputted by the DAQ is run through a python scriptwhich averages the thermal resistance measured at 70, 75, and 80 minutes. We subtract the smallcontributions to the thermal resistance from the aluminum coupons and silver thermal paste, so ourresults apply to the adhesive and carbon foam only.

5. Thermal Resistance Results

Samples made using the epoxy and tape at a variety of thicknesses were tested using the thermaltester. The results are presented in this section.

5.1 Epoxy Results

A series of samples were made using epoxy layers of 50 µm, 125 µm, and 225 µm thickness oneach side. In Figure 5, the mean thermal resistance of samples made at each total thickness isplotted, with the error bar indicating one standard deviation for the samples made at that thickness.Three samples made at a later date using a fresh tube of epoxy showed approximately 100% higherresistance and are plotted separately in red. Finally, further testing with a third sample of epoxyshowed consistent results with the higher value (red) for the thinnest layer, and with the blue plottedvalues for the thicker layers in Figure 5.

0 100 200 300 400 500

Total Epoxy Thickness (µm)

0.4

0.6

0.8

1.0

1.2

1.4

Therm

al R

esi

stance

(K

/W)

3M TC-2810 Thermally Conductive Epoxy

New Epoxy Tube

Old Epoxy Tube

Figure 5: Thermal resistance of epoxy–carbon foam samples versus thickness.

As seen in Figure 6, SEM photos of sectioned samples after testing indicate an uneven epoxylayer and inconsistent wicking into the carbon foam. As controls, we tested slabs of epoxy withoutcarbon foam, and these showed consistent sample to sample thermal resistance. Furthermore, testswith only carbon foam and no epoxy also showed consistent results (albeit with high resistance).

– 6 –

Figure 6: SEM photo of epoxy to carbon foam interface.

5.2 Thermally Conductive Tape Results

A series of samples were made with 250µm, 375µm, and 500µm total thickness of 3M 8800 seriesthermally conductive tape. Tests showed a linear correlation between sample resistance and tapethickness, as shown in Figure 7. By fitting a line of the form T R(x) = ax+ b, we can isolate the

0 100 200 300 400 500 600

Total Tape Thickness (µm)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

Therm

al R

esi

stance

(K

/W)

3M 8800 Series Thermally Conductive Tape

TR(x) = 0.0033x+ 0.2218

Data

Figure 7: Thermal resistance of tape samples at varying thicknesses.

thermal conductivity of the tape from the carbon foam. This gives T R(x) = 0.0033x+ 0.2218,and k = 0.47 W/m ·K for the thermal tape, 22% lower than the manufacturer specification of 0.6

– 7 –

W/m ·K [1]. The second term of the fitted equation gives the thermal conductivity of the carbonfoam, kcarbon =Ccarbon ·0.004m/(0.0254m)2 = 27.96 W/m ·K, in agreement with the specificationof 26 W/m ·K.

6. Tensile Tester & Results

We have designed and built an apparatus (photograph in Figure 8) to measure ultimate tensilestrength (UTS) for these bonded samples. The tensile tester consists of four pneumatically actuatedpistons bolted to the corners of a square base plate. The pistons push on four rods attached to thecorners of an upper plate. The sample under test is attached to two aluminum blocks via highstrength epoxy. Hooks are screwed into these blocks, and the hooks are then attached to eye boltsin the center of the upper and lower plates. The system pressure is gradually increased, and the flowof air is stopped at the moment of breakage. The pressure at the moment of breakage is recordedusing a pressure sensor attached to a manifold.

Figure 8: Front view of tensile tester showing pneumatic pistons, frame, connecting hooks andpressure readout.

The results of testing a series of samples described above are shown in Table 2. As seen inFigures 9 and 10, the broken epoxy samples contain sizable pieces of carbon foam broken andlifted along with the aluminum coupon. This indicates that the carbon foam structure breaks beforethe epoxy interface in certain areas. The 3M 8800 tape tends to stay fully attached to the aluminumcoupon, and only lifts traces of the carbon foam surface when broken. No correlation between UTSand previous heating of the samples (from the thermal testing described above) was observed.

– 8 –

Sample Type Layer ThicknessSample 1(kPa)

Sample 2(kPa)

Sample 3(kPa)

Sample 4(kPa)

3M 8805 Tape 125µm 219 243 249 2763M 8810 Tape 250µm 332 372 233 2553M TC-2810 Epoxy 225µm 1234 742 >1273

Table 2: Ultimate tensile strength (UTS) of tape and epoxy samples.

Figure 9: Epoxy interface after tensile failure.

Figure 10: Tape interface after tensile failure.

7. Radiation Damage Testing

One of our samples, made by bonding aluminum to carbon foam using 3M TC-2810 epoxy, wassubjected to 1014 neutrons/cm2 neutron fluence at the McClellan Nuclear Research Center. Thethermal resistance measured before and after irradiation were 1.328W/K and 1.312W/K respec-tively. This is roughly a 1% change, well within the typical error of measurement. Thus weconclude that neutron radiation has no effect on the thermal conductivity of the substances in oursamples. UTS for this sample was determined as described in the previous section, and results areconsistent with those in Table 2.

– 9 –

8. Discussion

These results indicate that although 3M TC-2810 epoxy has higher thermal conductivity, its ther-mal properties when used to bond carbon foam are inconsistent from sample to sample, even whensamples were made using identical conditions. This is particularly apparent for the thinnest 50 µmepoxy layers. It is likely that differences in secondary properties such as temperature, humidity,and viscosity influence wicking into the carbon foam, and thus greatly affect the thermal propertiesof the samples at this thickness. We also note that the data sheet provided by 3M indicates a ∼ 40%batch to batch uncertainty in thermal conductivity [2]. More consistent results were obtained forepoxy layers of 125 µm and above. Samples made with thermally conductive tape showed consis-tent sample to sample results with a highly linear relationship between tape thickness and thermalresistance.

The epoxy samples showed an up-to 66% difference in UTS between samples made using thesame well-controlled procedure. Despite this inconsistency, even the weakest of the epoxy sampleswas still over twice as strong as any tape sample tested. However, the UTS of these samples weremuch lower compared to those made with epoxy, due to the fact that the tape cannot wick intothe foam structure like the epoxy does. A modest increase in UTS is observed by doubling thethickness of the thermally conductive tape at the expense of nearly doubling thermal resistance(since the adhesive is the most resistive part of the sample).

Finally, radiation damage testing of epoxy–foam samples performed with neutrons to a fluenceof 1014 n/cm2 showed no change in thermal or mechanical properties. Tests with charged particlesand at higher fluences are necessary to ensure suitability at harsh radiation environments, such asat High Luminosity-LHC.

9. Acknowledgments

This work at the University of California, Davis was supported by U.S. Department of Energy grantDE-FG02-91ER40674 and by U.S. CMS R&D funds via Fermilab.

References

[1] 3M Electronics Materials Solutions Division, 3M Thermally Conductive Adhesive Transfer Tapes 8805,8810, 8815, 8820, July, 2015.

[2] 3M Electronics Materials Solutions Division, 3M Thermally Conductive Epoxy Adhesive TC-2810,August, 2014.

[3] M. Chertok, M. Fu, M. Irving, C. Neher, M. Tripathi, R. Wang, and G. Zheng, “Reactive Bonding Filmfor Bonding Carbon Foam Through Metal Extrusion,” arXiv:1606.07677[physics.ins-det].

[4] J. S. Steinhart and S. R. Hart, “Calibration curves for thermistors,” Deep Sea Research andOceanographic Abstracts 15 no.~4, (1968) 497 – 503.

– 10 –