Thermal Conductivity of Ni-coated MWCNT Reinforced 70Sn-30Bi AlloyMd Muktadir Billah and Quanfang Chen*

MEMS and Nanomaterials Lab, University of Central Florida, Orlando, FL 32816, USAQuanfang.Chen@ucf.edu

Abstract

The thermal conductivity of thermal interface material (TIM) is essential since the heat dissipation of electronic chips in an integrated circuit is achieved through TIM which forms the bottleneck of the heat conduction. In the current study, the effect of MWCNTs reinforcement on thermal conductivity of a typical TIM material in 70Sn-30Bi alloy has been investigated. To prevent CNT aggregation and to enhance the interfacial bonding between CNTs and the metal, CNTs were coated with nickel by electroless deposition prior to the material synthesis. The metal coating also allows avoiding the separation of CNTs from the molten metal due to buoyancy effect. The dispersion of metal-coated CNTs were assisted by sonication. The thermal conductivity of 3 wt. % Ni/MWCNTs reinforced 70Sn-30Bi alloy was found to be more than 170 percent greater than that of the base alloy.

1. IntroductionThermal management has emerged as a critical challenge to advanced electronic systems, including advanced integrated circuits (ICs), high power electronics such as high power amplifiers and phased-array radars, high-energy lasers, and high-power microwave systems [1]. Owing to the ever increasing need for higher performance, the input power of the advanced electronic systems has been greatly increased to achieve the desired operational power density. As a result the resistive or Joule heating can produce a heat flux of up to 1,000 W/cm2 [1]. If this large heat flux cannot be dissipated instantly and continuously, the accumulated heat would quickly raise the device’s substrate temperature thus degrading the system’s performance and eventually damaging the electronic system [1]. Therefore, efficient thermal management is necessary to insure the continuous removal of heat away from the electronic devices.

Fig. 1 Sketch of heat removal challenge in power electronics (arrows show needed heat dissipation, not in scale).

Details of a typical thermal management system for electronics are shown in Fig. 1 that it normally includes a heat sink and a heat spreader which is bonded to the electronic chip by a

1

thermal interface material (TIM). Since the heat sink and the heat spreader are mostly premade with thermally conductive materials such as copper or aluminum, the function of a TIM material is more as an adhesive or solder to bond the heat spreader and the electronic device together. Since the heat generated by the electronic device needs to be dissipated continuously “downwards” to the heat sink as depicted by the red arrows depicted in Fig. 1, the thermal resistance (inverse of the thermal conductance) between the electronic device and the heat sink is critical to the cooling efficiency. The total thermal resistance (equation 1) between the device and the heat sink includes: 1) contact thermal resistance Rc1 which is formed mainly by micro/nano cavities (right of Figure 1) between the electronic device (silicon die, e.g.) and the TIM, 2) the thermal resistance (hi/Ki) formed by the TIM which is proportional to the thickness (hi) and inversely proportional to the thermal conductivity (Ki), 3) the contact thermal resistance Rc2 which is formed between the TIM and the heat spreader due to the micro/nano voids formed and 4) the thermal resistance formed by the heat spreader (hs/Ks). Although contact resistances can be reduced by elaborate machining techniques such as polishing and degassing, the thermal resistance of the TIM and the heat spreader are largely dependent on the ratio of thickness over the thermal conductivities. Since the heat spreader material normally has a large thermal conductivity (385W/m-K of Cu, e.g.), which normally leads to a small temperature drop (<10C, [2]). However, the thermal conductivities of available TIMs are much smaller [3], in a range of 1~20W/m-K (the upper end goes to low melting alloys). It has indicated that a 50m thick TIM (Sn-Ag alloy with the best thermal conductivity) can produce a temperature drop of more than 30C on an electronic system [2]. Therefore, the efficiency of a thermal management (cooling) system (Figure 1) is largely limited by the thermal conductivity of the TIM material.

Currently available TIMs include [3-12]: 1) Thermal greases such as Shin-Etsu series products and Bergquist TIC-7500.The thermal conductivity of thermal greases is less than 10 W/m-K. 2) Phase changed thermoplastic adhesives (epoxy) such as Power Strate series products and Bergquist Hi-Flow series products. These materials are typically melted in the 50-80C range and the thermal conductivity is less than 5 W/m-K after adding a variety of fillers to enhance their thermal conductivity. 3) Phase changed low melting alloys such as Thermax HF-601 10-BT. The thermal conductivities of these low melting temperature alloys are in the range of 10-20 W/m-K depending on the alloys used and their compositions. 4) Thermal conductive elastomers such as Chomerics therm-A-form T642. The thermal conductivity is less than 3 W/m-K. In summary, low melting alloys (<170C) have relatively greater thermal conductivities among available TIM technologies but their thermal conductivities are not sufficient. Therefore, there is a need to increase the thermal conductivity of low melting alloys. Since Sn-Bi alloys are more environmentally friendly than conventional solders, have a melting point as low as 138C and relatively high thermal conductivities (K~19 W/m-K), the question becomes whether the thermal conductivity of a Sn-Bi alloy can be increased significantly to create an advanced TIM material. On the other hand, carbon nanotubes (CNTs) have long been regarded as ideal one-dimensional building blocks for developing advanced thermal conductive materials due to their nanometer size and the exceptional thermal conductive (about 3000 W/m-k for a multi-walled carbon nanotube (MWCNT) [13-14]). Therefore a quest has been made to if the relatively low cost MWCNTs can be used to significantly enhance the thermal conductivity of SnBi alloy. However, due to the difficulties in mixing and dispersion owing to the large difference in material properties (density and wetting) MWCNTs cannot be added directly into SnBi alloy by conventional cost effective material fabrication processes like the melting and casting.

2

In this paper, authors have studied effects of nickel coated MWCNTs reinforcement in 70Sn-30Bi composite on its thermal conductivity. MWCNTs were coated with Ni by an electroless method to retain the separation of MWCNTs while the coated nickel is used to obtain the desired density and the wetting with SnBi alloy. The MWCNT-SnBi composite was synthesized using sonication assisted melting and casting route. The measured thermal conductivity of the Ni/MWCNTs reinforced 70Sn-30Bi composite was found to be 170% greater than that of the 70Sn-30Bi solder alloy. The details of material synthesis and characterization as well as the discussion have been provided.

2. Experimental2.1 Materials preparation and characterization

MWCNTs having a diameter of 30-50 nm and axial length 10–20 µm were purchased from the CheapTubes Inc. CNTs were first ultrasonically dispersed in an aqueous solution of HNO3 (70%) at 80C for 2 h for the purpose of purification and functionalization. After that CNTs were washed with deionized (DI) water thoroughly. The acid treatment helps to get rid of impurities [15-18] as well as to modify surface properties for better wetting/dispersity in the electrolyte due to resulting in high-density activation sites [17]. The activation process is important to make CNTs hydrophilic and improve adherence of metallic particles to CNTs surface.

MWCNTs were coated with nickel by electroless method in a two-step process. First MWCNTs were activated by ultrasonically dispersing in the electroless plating solution under ultra-violate (UV) exposure for 20 minutes. Then a NaBH4 solution was added to form the NiO clusters as the catalytic sites and ultra-sonicated for further 10 minutes [19]. After rinsing with DI water the MWCNTs were then coated with nickel by dispersing in the plating solution for 20 minutes at 50C. Finally, the Ni-coated CNTs were washed with DI water, filtered and dried in a vacuum desiccator. Table 1 shows the plating bath composition. Ammonia solution was used to adjust the pH to 8.5. Zeiss ULTRA-55 Field Emission Gun Scanning Electron Microscope (FEG-SEM) and the Energy Dispersive Spectroscopy (EDS) were used to study the coating morphology.

Ni-coated MWCNTs were added in SnBi alloy using sonication assisted melting and casting route with the experimental setup sketched in Fig. 2. The setup includes a heating unit, a crucible, an ultrasonic probe, a control unit, and an inert gas supply. Pure tin and bismuth metals were weighed and melted in a small crucible in argon gas atmosphere. Weighted Ni-coated MWCNTs were added at 100C above the melting point of the alloy and the high temperature was maintained during sonication process to keep the viscosity sufficiently high [20-21]. A 20 kHz, 600 W ultrasonic Sonicator (Misonix) with a titanium probe was used for the sonication. Following the sonication, the temperature was maintained for a further 10 minutes to allow potentially trapped bubbles get escaped. The sonication process forms transient cavitation and acoustic streaming that results in homogenous dispersion of MWCNTs [22-24]. Finally, the SnBi alloy with dispersed Ni-coated MWCNTs was cooled down to room temperature.

2.2 Measurement of thermal conductivity of MWCNT/SnBi composite

3

The thermal conductivity was measured using the one-dimensional thermal conduction principle with the same setup used for Cu/CNT composites [25]. The one-dimensional steady state thermal conduction state was obtained in the laboratory within an insulating system covered with Styrofoam (Fig. 3). The setup consists of a resistive heater and two thin film RTDs (Resistance Temperature Detectors). As the heating source, a solid piece of copper wrapped around two parallel 120 Ω resistors for a measured resistance of 59.2 Ω was used.

Table 1 Ni Plating Bath Composition.

Fig.2 Ultrasonic assisted melting.

RTDs are temperature sensors based on resistance change linearly versus the temperature. The RTDs were Omega F3105 Class A. RTDs were mounted on a quartz plate at 8.5 mm distance. The sample for thermal conductivity measurement was placed making contact with the heating source and a heat sink. A solid piece of copper was used as the heat sink with water being pumped through it. The mounted RTDs were then used to measure the temperature of the hot and the cold ends.

Fig. 3 Schematic top view of thermal Fig. 4 Wheatstone bridge circuit with RTDs.

Conductivity measurement setup.

The thermal conductivity can be derived based on one-dimensional thermal conduction equation 2. The length used for thermal conductivity measurement, L = 8.5 mm, A is the cross-sectional area (width X thickness) of the samples, P is the power can be adjusted by using current I based on equation 3 in which R= 59.2 Ω; T1 and T2 are temperatures of the hot end and the cold end respectively, measured using the RTDs. A thermocouple was also used for verifying the

4

Chemicals Concentration (g/L)

NiSO4.6H2O 35

NaH2PO2 35

C6H5Na3O7 18

NH3 solution For pH adjustment

measured temperatures. The temperature induced change in resistances of RTDs were measured using a Wheatstone bridge circuit as shown in Fig. 4.

K= PLA(T1−T 2)

(2)

P=I 2 R (3)

NI USB 6008 DAQ was used as a very stable excitation power source applying voltage VEX

across the bridge. The RTD and bridge output voltage VO is an indication of the RTD’s resistance related by equation (4). Three high-precision resistors having a very low-temperature coefficient were used in the Wheatstone bridge. The DAQ was plugged into a laptop with LabVIEW 2011. Thus, the resistance of the RTD was measured. Two Wheatstone bridges were used for the two RTDs. The resistance of the RTDs was linearly proportional to their temperatures. Initially, the temperature measurements using the RTDs were calibrated. Calibration was also done for the total heat loss of the setup. Here heat transfer includes the three categories: (1) thermal conduction through the insulating setup; (2) convection to the environment; and (3) radiation to the environment. The total power, P = Pconduction + Pconvection + Pradiation. This finally leads to equation (4) to measure Csetup [25]. As composite sample of thermal conductivity K is placed in the setup, K is therefore measured from equation (6) [25].

V O = V EX (R3

R3+ RRTD) – (

R2

R1+R2)

(4)P=C setup (T 1−T2) (5)

P=¿) . (T 1 - T 2) (6)

Csetup was measured by measuring the temperature difference between the hot and the cold end using equation (3) before placing the sample for thermal conductivity measurement. Current was measured using a multimeter. Csetup is the total equivalent resultant thermal conductance, was measured for different current values. For lower ampere values, as the temperature gradient was small, the Csetup was found to be dependent on the temperature gradient (T1 - T2). However, for large temperature gradient, Csetup reached a constant value. Therefore all the thermal conductivity measurements were carried out in stable temperature gradient range. Prior to conducting the tests, the setup was calibrated by measuring the thermal conductivity of pure copper. For every temperature measurement, 2 hours of thermal relaxation was used to reach the thermal equilibrium.

3. Results and Discussion

Figure 5 shows the calibrated results of RTDs. RTDs were calibrated between 0C and 100C to find the resistance – temperature coefficients from the linear relationship between the resultant resistance and the temperature.

5

Fig. 5 Relationship of temperature vs. resistance of (a) RTD 1 and (b) RTD 2.

The resistance – temperature coefficients were found to be 0.3865 ohm/C and 0.3785 ohm/C for RTD 1 and 2 respectively. These values were used in LabVIEW to measure temperatures. Csetup

was measured by varying the passing current through resistors which results in a different temperature gradient across the hot and the cold end. Csetup was found to be constant 0.01385 W/ K for a temperature gradient above 25C (Fig. 6). The thermal conductivity of all samples was measured at a temperature gradient of above 25C.

Fig. 6 Calibration curve for the Csetup. Fig. 7 Thermal conductivity of the composites.

The samples were placed over the hot and the cold ends and the thermal conductivities were calculated by using equation 6, and the results are shown in Fig. 7. It indicates that for 0.5 wt. % Ni/MWCNTs addition, the resultant thermal conductivity of 70Sn-30Bi was increased by 23.35 %. Figure 7 indicates that the resultant thermal conductivities are almost linearly proportional to the addition of MWCNTs. For 3 wt. % addition of Ni/MWCNT, the resultant thermal conductivity was found to be more than 170% of that of the 70Sn-30Bi base alloy.

6

Fig. 8 (a) Ni-coated MWCNTs and (b) Dispersion of CNTs in the matrix.

Utilizing carbon nanotubes to increase material’s thermal conductivity and other material properties including the mechanical strength has always been a challenge due to difficulties in obtaining good separation/dispersion of CNTs and poor CNT-matrix interfacial bonding. The sonication in electrolyte has proven efficient to completely separate MWCNTs and the subsequent nickel deposition onto MWCNTs will retain the separation, as shown in Fig.8a. Since nickel has a similar density to that of SnBi alloy, the addition of Ni-MWCNT into the molten SnBi alloy avoid the separation of MWCNTs due to buoyancy effect that under the assistance of sonication the MWCNTs can be dispersed well within the SnBi alloy, verified by Fig. 8b after solidification. In addition, the coated nickel on MWCNTs forms good interfacial bonding between MWCNTs and the SnBi matrix 9Fig. 8b) that is important for thermal conduction and mechanical strength (significantly increased, will be published in another paper). The coating results in improved interfacial bonding that attributes to increased thermal conductivity [26-28], since good interfacial bonding increases thermal conduction by phonons. In the Sn-Bi matrix, MWCNTs are found to be dispersed homogenously (Fig. 8b) which is very importance to enhance thermal conductivity since CNT-CNT contact resistance is very high. Therefore the nickel coating results in better dispersion and better interface bonding that lead to the enhanced thermal conductivity.

4. Conclusion

MWCNTs have been successfully added into 70Sn-30Bi alloy via melting route assisted with sonication. MWCNTs were precoated with nickel by electroless deposition that not only prevent aggregation of MWCNTs but also avoid separation of MWCNTs from the molten alloy due to buoyancy effect. The resultant thermal conductivity of 3 wt. % Ni/MWCNTs reinforced 70Sn-30Bi was found to be more than 170% greater than that of the pure 70Sn-30Bi alloy. The enhanced thermal conductivity was attributed to the good dispersion of metal coated MWCNTs and the good interfacial bonding with the matrix resulted from the Ni coating.

Acknowledgements This work was financially supported by the National Science Foundation of USA.

7

REFERENCES

1. Ryan J. McGlen , Roshan Jachuck and Song Lin, Integrated thermal management techniques for high power electronic devices, Applied Thermal Engineering, Volume 24, Issues 8-9, June 2004, Pages 1143-1156

2. Jeffrey P. Calame, Robert E. Myers, Frank N. Wood, and Steven C. Binari, Simulations of Direct-Die-Attached Microchannel Coolers for the Thermal Management of GaN-on-SiC Microwave Amplifiers, IEEE TRANSACTIONS ON COMPONENTS AND PACKAGING TECHNOLOGIES, VOL. 28, NO. 4, DECEMBER 2005

3. Farhad Sarvar, David C. Whalley and Paul P. Conway, Thermal Interface Materials - A Review of the State of the Art, 2006 Electronics System integration Technology conference, Dresden, Germany.

4. Carl Zweben, Ultrahigh-thermal – conductivity packaging materials, 21th IEEE SEMI THERM Symposium, September 2005

5. K Zhang, Y Chai, M F Yuen, D G W Xiao and P C H Chan, Nanotechnology 19 (2008) 215706

6. K Zhang, Y Chai, M F Yuen, D G W Xiao and P C H Chan, Carbon nanotube thermal interface material for high-brightness light-emitting-diode cooling, Nanotechnology 19 (2008) 215706 (1-8)

7. Lee, T. et al., "High Thermal Efficiency Carbon Nanotube-Resin Matrix for Thermal Interfacematerials", Proc. 55th Electronic components and technology conference, 2005, pp. 55-59.

8. H. Huang, C.H. Liu, Y. Wu, S.S. Fan, Aligned Carbon Nanotube Composite Films for Thermal Management, Advanced Materials, Vol.17 (2005), pp.1652-1656.

9. J. Xu, T.S. Fisher, “Enhanced thermal contact conductance using carbon nanotube arrays”, Proc 9th Intersociety Conference on Thermal and thermomechanical Phenomena in Electronic Systems,Vol.2, June. 2004, pp.549-555.

10. Baratunde A. Cola, Xianfan Xu, and Timothy S. Fisher, Increased real contact in thermal interfaces: A carbon nanotube/foil material, APPLIED PHYSICS LETTERS( 90), 093513(2007)

11. Francesco Toschi, Emanuela Tamburri, Valeria Guglielmotti, Preparation and thermal characterization of Carbon Nanotubes-based composites for applications in electronics packaging, Second International Conference on Quantum, Nano and Micro Technologies,2008-8-4

8

12. Ngo, Q. et al., "Thermal Interface Properties of Cu filled Vertically Aligned Carbon Nanofiber Arrays ", Nano Letters, Vol. 4, No. 12, 2004, pp. 2403-2407.

13. S. Berber, Y-K. Kwon, and D. Tomanek, Unusually high thermal conductivity of carbon

nanotubes, Physical Review Letters 84 No. 20 (2000) 4613-4616.

14. P. Kim, L. Shi, A. Majumdar, and P.L. McEuen, Thermal transport measurements of

individual multi-walled nanotubes, Physical Review Letters, 87 (2001) 215502,1-4

15. Li Q, Fan S, Han W, Sun C, Liang WJ. Coating of carbon nanotube with nickel by electroless plating method. Appl. Phys. 1997; 36: 501–503.

16. Peng YT, Hu YZ, Wang H. Tribological behaviors of surfactant-functionalized carbon nanotubes as lubricant additive in water. Tribol. Lett. 2007; 25: 247-253.

17. Zhao B. , Yadian B. L., Li ZJ, Liu P, Zhang YF. Improvement on wettability between carbon nanotubes and Sn. Surface Engineering 2009; 25: 31-35.

18. Jagannatham M, Sankaran S, Prathap H. Electroless nickel plating of arc discharge synthesized carbon nanotubes for metal matrix composites. Applied Surface Science 2015; 324: 475–481.

19. Yan T, Li L, Wang L. A simple nickel activation process for electroless nickel-phosphorus plating on carbon fiber. BioResources 2013; 8: 340-349.

20. Giridhar V, Arunraj RS, Dhisondhar R. Ultrasonic nano-dispersion technique of aluminum alloy and carbon nano-tubes (CNT) for automotive parts applications. Int. J. Adv. Res. Technol. 2013; 1: 2321-0869.

21. Yang Y, Lan J, Li X. Study on bulk aluminum matrix nano-composite fabricated by ultrasonic dispersion of nano-sized sic particles in molten aluminum alloy. Mater Sci Eng A 2004; 380: 378–383.

22. Akio K, Atsushi O, Toshiro K, Hiroyuki T. Fabrication process of metal produced by vortex method. J. Japan. Inst. Light Met. 1999; 49: 149–154.

23. Mussert KM, Vellinga WP, Bakker A, Zwaag SVD. A nano-indentation study on the mechanical behavior of the matrix material in an AA6061-Al2O3 MMC. J. Mater. Sci. 2002; 37: 789–794.

24. Duralcan Metal Matrix Composites, May–June Data Report Package, Dural Aluminum Composites Corporation, San Diego, CA 92121.

25. Chai G, Chen Q. Characterization Study of the Thermal Conductivity of Carbon Nanotube Copper Nanocomposites. Journal of Composite Materials 2010; 44.

26. Kittel C. Introduction to Solid State Physics, 6th Edition, 1986; John Wiley & Sons, New York.

27. Susumu A, Morinobu E, Tomoyuki S, Atsushi K. Fabrication of nickel multiwalled carbon nanotube composite films with excellent thermal conductivity by an

9

electrodeposition technique. Electrochemical and Solid-State Letters 2006; 9 (8): 131-133.

28. Ngo Q. Cruden BA, Cassell AM, Sims G, Meyyappan M, Li J, Yang CY. Thermal interface properties of cu-filled vertically aligned carbon nanofiber arrays. Nano Letters 2004; 4(12): 2403-2407.

10