THERMAL CONDUCTIVITY AND CONTACT RESISTANCE

2y ago
105 Views
2 Downloads
355.76 KB
8 Pages
Last View : 15d ago
Last Download : 2m ago
Upload by : Rosemary Rios
Transcription

Proceedings of IPACK2007ASME InterPACK '07July 8-12, 2007, Vancouver, British Columbia, CANADAIPACK2007-33026THERMAL CONDUCTIVITY AND CONTACT RESISTANCEMEASUREMENTS FOR ADHESIVESPeter Teertstra*Microelectronics Heat Transfer LaboratoryDepartment of Mechanical EngineeringUniversity of WaterlooWaterloo, Ontario, CanadaABSTRACTThermal adhesives that contain large concentrations ofhigh thermal conductivity filler materials, such as ceramics ormetals, are widely used by the electronics industries in a varietyof applications. The thermal properties of these materials, suchas the thermal contact resistance across a bonded joint and thethermal conductivity of the bulk material, are critical to theselection of the “best” material. A method is presented for themeasurement of these thermal properties using a steady-state,guarded heat flux meter test apparatus based on the welldocumented and familiar ASTM test standard D-5470. Fivedifferent adhesive materials are tested and a linear fit of theresulting resistance versus thickness data are used to determinethe bulk thermal conductivity and contact resistance values.Four of the five materials tested had conductivity values of lessthan 1 W/mK, and the data demonstrates that a small butsignificant thermal contact resistance exists between theadhesive and the substrate for each of the adhesives.Greek Symbols TBLTkareabondline thicknessthermalconductivitybcjW mKheat transfer rateRthermal resistanceoTxtemperatureodistance* Assistant Professor, Member ASMECbulkcontactjointINTRODUCTIONAdhesive materials are widely used in the manufacture ofelectronic components and systems for a variety ofapplications, including the attachment of heat sinks orspreaders, die attach, underfill and encapsulation. Theseadhesives typically include large concentrations of fillermaterials intended to enhance the relatively poor thermalconductivity of the base material. Currently available thermaladhesives utilize a variety of filler materials, such as metallicsilver or aluminum, ceramics, oxides, or carbon black, invarious binders comprised of epoxy (two-part or heatactivated), RTV silicone, acrylics, etc.Given the manypossible combinations of the various ingredients, it is notsurprising that a significant number of thermally-enhancedadhesive materials are currently available to the electronicspackaging industry.The goal for the users of thermal adhesives is theminimization of the thermal resistance to heat conductionacross a joint formed when two surfaces are bonded. This isachieved through reductions in both the thermal resistanceassociated with conduction through the bulk adhesive materialand the thermal contact resistance between the bondingm2mmQoSubscriptsa adhesiveNOMENCLATUREAtemperature differenceWCWCm1Copyright 2007 by ASME

surfaces and the adhesive layer. The selection of the “best”adhesive for a particular application often involves manydifferent considerations, not the least of which is the eventualthermal performance of the resulting joint. Although themanufacturers and vendors of thermal adhesives usuallyprovide values for bulk thermal conductivity, often little detailis provided in regards to the test method or the impact ofthermal contact resistance on the overall joint resistance. As aresult, it may be advantageous to perform measurements of thethermal properties to validate the vendors data and provideadditional insight into the behavior of these adhesive-bondedsystems.Thermal joint resistance and thermal conductivitymeasurements for thermally-enhanced adhesive materials arepresented by a number of different researchers in the openliterature. Bolger [1] performed steady state thermal jointresistance measurements for single and multiple layers ofepoxy tape adhesives with diamond, silver, aluminum andalumina fillers at high pressures between polished aluminumsurfaces. Using these data, Bolger [1] calculated the effectivevalue of thermal conductivity k for each of the adhesives anddeveloped a correlation to predict the volume fraction of thefiller required for a particular k value.Mirmira, Marotta and Fletcher [2] performed thermalcontact resistance measurements using a steady-state test for avariety of adhesive materials, including epoxies, cements andsilicone. One thickness was considered in each case and resultsare presented in terms of the overall joint resistance only.Kilik et al. [3] presents thermal conductivity data for avariety of copper and aluminum filled epoxy adhesives frommeasurements performed using a transient test method basedon lumped capacitance analysis. This procedure provides muchfaster results than a steady-state test but does not give any datafor thermal resistance across the adhesive / surface contact.Campbell, Smith and Dietz [4], Hasselman et al. [5] andBrowne [6] all present results for thermal properties of filledadhesive materials based on measurements performed using alaser-flash method. This method is well described by theASTM Standard E-1461-01 “Standard Test Method forThermal Diffusivity by the Flash Method” [7], which containsa description of the use of the laser flash method to measurethermal diffusivity of homogeneous, isotropic, solid materials.The standard recommends caution be exercised when derivingthermal conductivity from thermal diffusivity measurements. Inparticular, erroneous results can occur when the laser flashmethod / equipment are used to predict specific heat formixtures where the components have significantly differentthermal diffusivities. In the case of the researchers citedearlier, the standard test method was modified to includesubstrate materials to support the adhesive layer and facilitatemeasurement of thermal contact resistance, while bulk thermalconductivity was calculated from a laser-flash measurement ofa cured pellet of the adhesive. The results of Hasselman et al.[5] suggest that the contact resistance between the adhesive andthe bonded surfaces is at least as significant as the bulkFig.1 Thermal interface material test apparatusconduction resistance. It should be noted that the cost of alaser flash test apparatus places it beyond reach of manyresearchers.In this study a method will be presented for determiningthe bulk thermal conductivity and thermal contact resistance atthe bonded surfaces for adhesive materials using thermal jointresistance measurements. These tests will be performed with amodified guarded heat flux meter test apparatus, as per theASTM Standard D-5470 [8]. This type of test equipment isoften available in most thermal test laboratories and the methoddescribed in this study is easily adaptable to work with most ofthese systems.APPARATUS AND METHODOLOGYThermal Interface Test ApparatusAll measurements were performed using a thermalinterface material (TIM) test apparatus, with its design basedon the guarded heat flux meter device recommended by theASTM standard [8] with a number of modifications. The testcolumn in the apparatus shown in Fig.1 is comprised of twocalibrated electrolytic iron flux meters with a 25 x 25 mm crosssectional area. Temperatures are measured at 10 mm intervalsalong each of the flux meters using five 1 mm dia. x 25 mm2Copyright 2007 by ASME

ceramic RTDs and the test sample is placed between the lappedupper and lower surfaces. A heater block with four embeddedcartridge heaters is positioned at the bottom of the test columnwhile the temperature of the cold plate at the top of the columnis regulated using a glycol-water solution from a constanttemperature bath. The contact pressure at the interface betweenthe flux meters is measured using a load cell and adjusted usinga linear actuator. The test column and loading mechanism arecontained with a vacuum chamber capable of pressures lessthan 10-4 atm. All measurements are performed using aKeithley 2700 data acquisition system controlled with Labviewprogram running on desktop PC. A full description of theapparatus and details of its construction and operation arepresented by Culham et al. [9].The uniformity of the pressure distribution across thecontacting surfaces of the heat flux meters was verified usingPressurex[10] Ultralow (25 – 85 psi) pressure indicating film.The procedure used for measurement of the joint resistancebetween the flux meter surfaces contacting the sample is asfollows. The sample is placed between the flux meters andsmall preliminary load is applied to align the test column. Avacuum is drawn in the test chamber in order to reduce heatlosses due to convection, and the data acquisition software isstarted. Once the desired contact pressure and sampletemperature are achieved and a steady state condition isreached, the temperatures of the RTDs are recorded. A sampleof the RTD temperatures plotted vs. position in the flux meteris shown in Fig.2.The total heat flow rate through each of the flux meters iscalculated byQ k (T ) AdTdx80oTemperature ( C)704020060120Position (mm)Lower (heater) flux meterUpper (cold plate) flux meterFig.2. Heat flux meter temperature distributionAdhesive Test Sample PreparationTo avoid damage to the precision joint surfaces of the heatflux meters, the measurements were performed using testsamples consisting of two aluminum substrates bonded togetherusing a uniform thickness layer of adhesive. A set of forty 25 x25 x 6.43 mm aluminum Al-2024 T-351 blocks were machinedfrom bar stock and the faces were lapped to achieve a flat,smooth surface. An average surface roughness Ra 0.3 µ mand an average out-of-flatness of 2 µ m over the 25 mm widthof the sample were measured using a Mitutoyo SJ-400 surfaceprofilometer. The co-planarity of the blocks was verified usinga high precision micrometer ( 1 µm ) to measure the thicknessof the blocks at the center and the corners. The maximumvariation between these thickness measurements was 10 µ mwith an average difference of 6 µ m .The test samples were assembled using a pair of blocks, asshown in Fig. 3. Steel shims were placed at the corners of thejoint to set the thickness of the adhesive layer and kapton tapewas used to assist in aligning the blocks and preventing theadhesive from running out of the joint. Once the surfaces werecoated with adhesive and the blocks were put together, a smallclamp was used to bring the surfaces of the blocks into contactwith the shims and hold the sample until the adhesive hadcured. The manufacturer’s recommendations regarding curingtemperature and duration were followed in each case.Four different test samples were prepared for each of theadhesives examined in the study, three using 0.008”, 0.013”and 0.017” shim stock and one with no shims. The bondlinethickness (BLT) was calculated by subtracting the thickness ofeach of the blocks measured at the center using a micrometer( 1 µm) from the total thickness of the test sample after theadhesive had cured. The micrometer was also used to measuredthe corner thicknesses of the completed samples to verify that a(1)A 6.25 10 4 m 2 is the cross sectional area, and d T d x isthe temperature gradient, calculated using a linear least squaresfit of the data. The total heat flow rate calculated for the lowerand upper flux meters varied by less than 5% for all cases.The thermal joint resistance is determined using TQ T joint5030where k (T ) is the thermal conductivity of the calibrated fluxmeter material, correlated with respect to average temperature,R joint 60(2)where Q is the mean value from the upper and lower heat fluxmeters and T is the temperature difference at the joint,calculated based on an extrapolation of the least squares fit ofthe data to the contacting surfaces of the flux meter, as shownin Fig.2.The uncertainty in the joint resistance measurements of thecurrent test apparatus is 2.0 % , as reported by Savija[11]based on the accumulated uncertainties in the heat flow rateand joint temperature measurements.3Copyright 2007 by ASME

Table 1 Adhesive descriptions and bondline thicknessesMaterialDescriptionAHigh thermalconductivityRTV Silicone (1-part)0.0100.1960.3280.439BAluminum-filled 2-partepoxy putty0.0890.2040.3360.419CAluminum-filled 2-partepoxy bonding resin0.1060.2140.3820.451DSilver-filled 2-part epoxy0.1140.1930.3150.398ESilver-filled thermoplastic0.1230.2490.3650.450Fig.3 Test sample assemblyuniform bondline thickness had been achieved. The averagevariation in bondline thickness for all of the samples,determined by the maximum difference in the thicknessesmeasured at the corners divided by the center thickness, was10%. As a result, an uncertainty of 5 % is assumed for allbondline thickness measurements.The five different adhesives used in this study wereselected to represent a cross-section of the various types of thecurrently available materials. The adhesives, identified asmaterials A, B, C, D and E, are described in Table 1 along withthe bondline thickness measurements for all of the samples.BLT(mm)Fig.4. Thermal resistance network for adhesive samplejoint, Ra , it is necessary to reduce the effects of thermalcontact resistance between the aluminum block and the fluxmeter, Rc , and the conductive resistance through the blocks,Test ProcedureAll thermal resistance measurements were performed at amean joint temperature of 50 oC with a load of 150 N applied tothe test column. Vacuum conditions were maintained in the testchamber during the measurements to reduce convective heatloss from the flux meters.The joint resistance measured for each of the test samples,R j , can be modeled as a series combination of five resistances,Rb , using the following relationshipR a R j 2 R c 2 Rb(3)This approach assumes that the surface finishes at the lowerand upper block / flux meter interfaces and the blockthicknesses are similar. A low viscosity, carbon black filledthermal interface material was used between the sample and theas shown in Fig. 4.In order to determine the resistance across the adhesive4Copyright 2007 by ASME

RESULTSThe adhesive samples were tested using the sameconditions as the preliminary measurements. In each case atleast two measurements were performed for each sample, andfor each of these tests the sample was removed, the contactingsurfaces were cleaned with acetone, the carbon black thermalinterface material was applied and the sample was repositionedin the test column. The adhesive resistance, Ra , is plottedversus bondline thickness for each of the five adhesivematerials in Figs. 5 – 9, and the Ra data are given in Table 3.The uncertainties in the measured joint resistance and thederived contact and bulk resistance values were combinedusing the method described by Moffat[12]flux meter in all cases because it provided good repeatability inthe results.The values for Rc and Rb were determined empiricallyusing a series of preliminary tests. In the first test, the heat fluxmeters were brought into contact with only a thin layer of thecarbon black interface material present at the interface. Thejoint resistance was measured for this case, which correspondsto a single value of contact resistance, Rc . In the next test, asingle 6.43 mm aluminum block was placed between the fluxmeters with the carbon black TIM on both the upper an lowerblock surfaces. The overall joint resistance measurement in thiscase corresponds to the conduction resistance through a singleblock and contact resistance across two joints, as shown in Eq.(4). In the third test, two blocks were tested with the carbonblack TIM present in all three joints, with the overall jointresistance corresponding to Eq. (5). These tests were eachperformed three times to demonstrate the repeatability of theresults, and the data are presented in Table 2.2 Rc Rb 0.204(4)3 Rc 2 Rb 0.347(5)δ RaRa δ R j Rj 2 δ Rc δ Rb Rc Rb 2 122 (6)which gives an overall uncertainty on the adhesive resistance of 3.5 % . This value, along with the uncertainty on thebondline thickness measurements, are presented as error bars inFigs. 5 – 9.1.501.25Table 2 Contact and conduction resistance test resultsRj( C W)oR j ,avg( C W)1.00ooTest#Ra ( C/W)Descriptiony 3133.9x 0.0264One contactOne blockTwo .34620.35830.3370.0600.750.500.250.2040.000.0E 001.0E-042.0E-043.0E-044.0E-045.0E-04BLT (m)Fig. 5 Thermal joint resistance results for material A0.3471.21.0y 2521.6x 0.0345Ra ( C/W)0.8oValues of Rc 0.061 and Rb 0.082 o C W Rb werecalculated using a simultaneous solution of Eqs. (4) and (5). Itshould be noted that the value for Rc is similar to the measuredjoint resistance for a single contact, as presented in Table 2.Because Rc and Rb are calculated based the measuredjoint resistance using simple, linear mathematical relationships,it is assumed that the uncertainty in each of these quantities willbe similar to that derived for the test apparatus by Savija[11], 2.0 % .0.60.40.20.00.0E 001.0E-042.0E-043.0E-044.0E-045.0E-04BLT (m)Fig. 6 Thermal joint resistance results for material B5Copyright 2007 by ASME

Table 3 Adhesive layer resistance 50.110.12oRa ( C/W)y 1955.6x 0.10250.750.500.250.000.0E 001.0E-042.0E-043.0E-044.0E-045.0E-04BLT (m)Ra (W mK )MaterialFig. 7 Thermal joint resistance results for material C1.00.8oRa ( C/W)y 2246.2x 0.02470.60.40.20.00.0E 001.0E-042.0E-043.0E-044.0E-045.0E-04The heat transfer through the adhesive joint is modeled asa series combination of thermal contact resistances at thesubstrate / adhesive interfaces, Rc , and conduction resistanceBLT (m)Fig. 8 Thermal joint resistance results for material Dthrough the bulk material, RbRa 2 Rc Rb0.14where Rb is calculated based on the familiar expression forone-dimensional heat conduction0.12y 209.97x 0.0336Rb oRa ( C/W)0.100.08BLTk A(7)Bondline thickness, BLT, is expressed in meters. SubstitutingEq.(7) into Eq.(6) gives a linear expression for the adhesivethermal resistance as a function of bondline thickness, whichcan be determined for a particular material using a linear, leastsquares fit of the data in Figs. 5 – 9. The thermal conductivityof the bulk material is calculated using the slope of the line0.060.040.020.000.0E 00(6)1.0E-042.0E-043.0E-044.0E-04 1 1 k A slope 5.0E-04BLT (m)(8)where A 607 10 6 m 2 is the cross sectional area of theadhesive, a value that has been corrected to account for theFig. 9 Thermal joint resistance results for material E6Copyright 2007 by ASME

reduction in area due to the shims in the corners. The thermalcontact resistance is determined using the y-intercept of thecorrelation, corresponding to the value of the joint resistancewhen the thickness of the adhesive layer goes to zero, i.e.Rb 0 in Eq. (6)Rc Ra y intercept 22W/mK. Material E, the silver-filled thermoplastic, had a bulkthermal conductivity of approximately 10 times the value of allof the other materials. However, if one were to compare theresults on the basis of the total adhesive resistance, the firstodata point for material A, Rc 0.052 C W (RTV silicone,no shim) is similar to the no-shim data for material E,(8)Rc 0.058 o C W . When no shim was used to prepare thesample, the RTV had a significantly smaller bondline thicknessvalue, 0.01 mm vs. 0.123 mmfor the silver-filledthermoplastic. This result provides an excellent reminder thatall of these adhesive materials should be applied in as thin alayer as possible, regardless of the bulk thermal conductivityvalue.Using the least squares equations presented with the data inFigs. 5 – 9, values for the thermal conductivity and thermalcontact resistances are calculated, as presented for each of theadhesive materials in Table 4.Table 4 Thermal conductivity and contact resistance results( C W)Materialk (W m K RcoACKNOWLEDGMENTSThe author acknowledges Prof. J. Richard Culham for hissupport of this research and the use of the test equipment in theMicroelectronics Heat Transfer Laboratory. The contributionof Dr. Ahmed Zaghlol from R-Theta Thermal Solutions Inc. isalso greatly appreciated.REFERENCES[1] Bolger, J. C., “Prediction and Measurement ofThermal Conductivity of Diamond Filled Adhesives,”Proceedings of the 42nd Electronic Components andTechnology Conference, pp. 219 – 229, San Diego,CA, May 18 – 20, 1992.[2] Mirmira, S. R., Marotta, E. E. And Fletcher, L. S.,“Thermal Contact Conductance of Adhesives forMicroelectronic Systems,” AIAA Journal ofThermophysics and Heat Transfer, Vol. 11, No. 2, pp.141 – 145, 1997.[3] Kilik, R., Davies, R. and Darwish, S. M. H., “ThermalConductivity of Adhesive Filled with Metal Powders,”International Journal of Adhesion and Adhesives, Vol.9, No. 4, pp. 219 – 223, 1989.[4] Campbell, R. C., Smith, S. E. and Dietz, R. L.,“Measurement of Adhesive Bondline EffectiveThermal Conductivity using the Laser Flash Method,”Proceedings of the 15th IEEE Semiconductor ThermalMeasurement and Management Symposium, pp. 83 –97, San Diego, CA, Mar. 9 – 11, 1999.[5] Hasselman, D. P. H., Donaldson, K. Y., Barlow, F. D.,Elshabini, A. A., Schiroky, G. H., Yaskoff, J. P. andDietz, R. L., “Interfacial Thermal Resistance andTemperature Dependence of Three Adhesives forElectronic Packaging,” IEEE Transactions onComponents and Packaging Technologies, Vol. 23,No. 4, pp. 633 – 637, 2000.[6] Browne, J., “Evaluation of an Advanced ThermalTransfer Adhesive for Thermal ManagementApplications,” Proceedings of the 16th IEEEDISCUSSIONA method has been proposed and demonstrated for themeasurement of the thermal conductivity and thermal contactresistance across joints bonded with thermally enhancedadhesive materials. The thermal joint resistance data measuredusing the steady-state test procedure is well fit by a linearfunction of the bondline thickness, which suggests that thethermal conductivity of the bulk material is a constant valuethat is not affected by the thickness of the adhesive layer.Although the contact resistance between the adhesive layerand the substrate is significantly smaller than bulk conductionresistance for most cases, it is not negligible. The contactresistance is of similar magnitude for most of the adhesivesexamined in this work, suggesting that Rc may be as much afunction of the surface finish, substrate material or contactpressure as the properties of the adhesive material itself.Material C had a significantly larger contact resistancecomponent that the rest of the adhesives, perhaps due to itsliquid hardener component that caused it to have a much lowerviscosity than the other materials during mixing andapplication. This may have lead to settling of the filler materialduring the curing process, resulting in a non-homogeneousmixture near the bonded surfaces and a higher effective contactresistance value. Additional measurements will be required inorder to quantify the relationship between surface finish andthe contact resistance for adhesive materials.Most of the materials examined in this study had similarbulk thermal conductivities, within the range 0.53 – 0.847Copyright 2007 by ASME

SemiconductorThermalMeasurementandManagement Symposium, pp. 83 – 97, San Jose, CA,Mar. 21 – 23, 2000.[7] “Standard Test Method for Thermal Diffusivity by theFlash Method,” ASTM Standard E-1461-01, CopyrightASTM International, Conshohocken, PA, 2007.[8] “Standard Test Method for Thermal TransmissionProperties of Thermally Conductive ElectricalInsulation Materials,” ASTM Standard D-5470-06,Copyright ASTM International, Conshohocken, PA,2007.[9] Culham, J.R., Teertstra, P., Savija, I. and Yovanovich,M.M., “Design, Assembly and Commissioning of aTest Apparatus for Characterizing Thermal InterfaceMaterials,” Eighth Intersociety Conference onThermal and Thermomechanical Phenomena inElectronic Systems, San Diego, CA May 29 - June 1,2002.[10] Sensor Products Inc., 300 Madison Ave., Madison,N.J., 07940.[11] Savija, I., “Method for Determining ThermophysicalProperties of Thermal Interface Materials,” M.A.Sc.Thesis, Department of Mechanical Engineering,University of Waterloo, Waterloo, Ontario, Canada,2002.[12] Moffat, R.J., “Describing the Uncertainties inExperimental Results,” Experimental Thermal andFluid Science, Vol. 1, pp. 3 – 17, 1988.8Copyright 2007 by ASME

earlier, the standard test method was modified to include substrate materials to support the adhesive layer and facilitate measurement of thermal contact resistance, while bulk thermal conductivity was calcula

Related Documents:

Energies 2018, 11, 1879 3 of 14 R3 Thermal resistance of the air space between a panel and the roof surface. R4 Thermal resistance of roof material (tiles or metal sheet). R5 Thermal resistance of the air gap between the roof material and a sarking sheet. R6 Thermal resistance of a gabled roof space. R7 Thermal resistance of the insulation above the ceiling. R8 Thermal resistance of ceiling .

thermal conductivity measurement methods is warranted to be included in this work. Two general classifications of thermal conductivity measurement methods are: (1) steady-state and . [12] and the guidelines given by ASTM E 1225-04 [10]. Although the thermal conductivity of TRISO fuel is unknown, the TFTCMS is designed to measure has an

2013, Fuchs et al. 2015). Since statistical solutions are applied to predict thermal conductivity, machine learning algorithms' usage to thermal conductivity prediction is reasonable. This study generates a machine learning approach to predict thermal conductivity by using drilling data. Drilling parameters are recorded from the surface.

The Atlas Scientific EZO class electrical conductivity circuit, employs a method of scaling resolution. As the conductivity increases the resolution between readings decreases. The EZO class electrical conductivity circuit will output conductivity readings where the first 4 digits are valid and the others are set to 0.

Stator-to-Case Thermal Contact Resistance Developed model for lamination thermal contact resistance to enable estimates of through- stack thermal conductivity for new materials. The model includes both solid and fluid components to calculate the inverse of TC

thermal conductivity. Among the few available theoretical models on the thermal properties of MOFs, the molecular dynamics (MD) simulation method is most widely used.18 20 MD simulations conducted by Huang et al. and Zhang and Jiang reveal that the main thermal resistance in MOF-5 stems from the flexible polymeric Received: September 5, 2015

[l] or physical vapor deposition [2] onto turbine components with an intermediate NiCoCrAlY alloy bond coating to improve adherence and to reduce oxidation. The thermal conductivity of these coatings is sensitive to the deposition technique, microstructure, density, and interface thermal resistance between layers [3].

10 Annual Book of ASTM Standards, Vol 02.02, in the Related Material section (gray pages). 11 Available from Standardization Documents, Order Desk, Bldg. 4, Section D, 700 Robbins Ave., Philadelphia, PA 19111-5094, Attn: NPODS. TABLE 1 Chemical Composition Limits NOTE 1—When single units are shown, these indicate the maximum amounts permitted. NOTE 2—Analysis shall be made for the elements .