You are on page 1of 6

This article appeared in a journal published by Elsevier.

The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elseviers archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright

Author's personal copy

Materials and Design 31 (2010) 113117

Contents lists available at ScienceDirect

Materials and Design


journal homepage: www.elsevier.com/locate/matdes

Material selection for microelectronic heat sinks: An application of the Ashby approach
G. Prashant Reddy a, Navneet Gupta b,*
a b

Chemical Engineering Group, Birla Institute of Technology and Science (BITS) Pilani, Rajasthan, India Electrical and Electronics Engineering Group, BITS Pilani 333 031, Rajasthan, India

a r t i c l e

i n f o

a b s t r a c t
This paper focuses on optimal materials selection for microelectronic heat sinks to maximize the thermal, mechanical and electronic response based on electro-thermal heat transfer analysis using the Ashby approach. In this work, material indices have been developed for a number of properties of heat sinks supported by materials selection tables/graphs. It is found that aluminum based alloys/metals perform better than other available materials for microelectronic heat sinks. 2009 Elsevier Ltd. All rights reserved.

Article history: Received 13 May 2009 Accepted 2 July 2009 Available online 14 July 2009 Keywords: Heat sink Ashby approach Heat transfer Materials selection Microelectromechanical systems

1. Introduction Heat sinks are the most common and cost-effective hardware employed for the thermal management of microelectronic circuits and microelectromechanical systems (MEMS) devices. Heat sinks nd wide applications in microelectronics and have become almost essential to modern integrated circuits like microprocessors, digital signal processors (DSP), graphics processing units, and more [1]. In common use, it is a metal object brought into contact with an electronic components hot surfacethough in most cases, a thin thermal interface material mediates between the two surfaces. Microprocessors and power handling semiconductors are examples of electronics that need a heat sink to reduce their temperature through increased thermal mass and heat dissipation (primarily by conduction and convection and to a lesser extent by radiation). Materials selection for engineering design needs a clear understanding of the functional requirements for each individual component and various important criteria/factors need to be considered. The selection of materials for microelectromechanical systems (MEMS) is complicated by the highly integrated multifunctional roles of the components. The conventional set of MEMS materials like silicon compounds, metals and alloys, ceramics/glasses, polymers and composites [1,2] although compatible with existing micromachining techniques, are not an optimal choice for the

maximum performance of devices. The growing interest in developing microelectronic heat sinks on various electronic devices presents an opportunity to expand the present set of MEMS materials to improve the functionality of such devices by optimal material selection. The existence of several techniques, mathematical and physical models to integrate large MEMS material sets into microsystem design has provided an impetus in adopting a rational approach for material selection in electronic/MEMS component design and fabrication [210]. As a step towards such an approach, we focus on the material selection for heat sinks which are widely used in microelectronic circuits for optimum thermal management [11]. The key performance indices for microelectronic heat sinks are thermal conductivity (k), electrical resistivity (qe), thermal expansion (a) and Youngs Modulus (E) [5]. This paper discusses a strategy for selecting suitable materials for heat sinks based on electrothermal heat transfer analysis compatible with Ashby approach in order to improve the device performance. The paper is organized as follows: Section 2 discusses the materials and their properties used in MEMS devices. Section 3 reviews the Ashby methodology. Section 4 is devoted to heat sinks and equations involved in heat transfer, while Section 5 involves application of Ashby method for material selection in microelectronic heat sinks. We discuss our results and conclusions in Section 6. 2. Materials and properties for MEMS devices

* Corresponding author. Tel.: +91 9772976336; fax: +91 1596244183. E-mail addresses: prashantrg.bits@gmail.com (G. Prashant Reddy), dr.guptanavneet@ gmail.com (N. Gupta). 0261-3069/$ - see front matter 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.matdes.2009.07.013

The present day technologies involving processing techniques such as bulk-micromachining, surface micromachining and soft

Author's personal copy

114

G. Prashant Reddy, N. Gupta / Materials and Design 31 (2010) 113117

lithography, have made it possible to introduce, shape, and integrate a large number of engineering materials into MEMS elements [2,3]. These materials are traditionally grouped into four classes: metals and alloys, glasses and ceramic, polymers and elastomers, and composites. The properties of materials commonly required in mechanical design are the Youngs Modulus (E), density (q), Poissons ratio (m), fracture strength (rF), yield strength (rY), fracture toughness (KIC), coefcient of thermal expansion (a), specic heat capacity (Cp), loss coefcient (g), and residual stress (rR) [3]. Using the method of Ashby, the designers have to consider the above material properties to optimize device performance and reliability in microsystems design. Certain electrical properties like resistivity and conductivity are also considered for electrical components in such designs. Compared to the properties of macroscale (bulk) structures, the properties of microscale structures can potentially be functions of the length scale as well as the details of processing techniques employed. However, it is possible to quantitatively relate micromechanical properties to bulk properties in many cases by focusing on structures with minimum feature sizes greater than 1 lm [3]. Some of the properties whose physical origins scale down to the atomic level and which can be included in micromechanical (>1 lm) and bulk structures include the Youngs Modulus, density, Poissons ratio, coefcient of linear expansion, and the specic heat. Sharpe [8] have tabulated initial design values based on an extensive survey of such measurements whose values are listed in Table 1 along with nominal bulk values tabulated by Ashby and Jones [12]. Thus, 0.8Ebulk < El < Ebulk can be concluded where the subscript l indicates microscale. Therefore, for the initial stages of micromechanical design, bulk values of these properties can be used. Based on this, Table 2 summarizes the initial design values for various material properties.

3. Materials selection: the Ashby method The Ashby method sets out the basic procedure for selection, establishing the link between the material and function as shown in Fig. 1. A material has several attributes like density, strength, Youngs Modulus, cost, resistance to corrosion, etc. [5]. A design demands a certain prole of these and the following tasks have been suggested: (i) identifying the desired attribute prole and (ii) comparing it with those of real engineering materials to nd the best match. The selection strategy of materials involves four main steps, comprising of translation, screening, ranking and supporting information [5,13]. These steps as illustrated in Fig. 2, can be interlinked to select the best suitable materials in microsystem design.

Material Material Classes

Function

Shape Material Attributes Material Indices Process

Fig. 1. Basic process linkage for material selection.

All Materials

Translate Design Requirements (express as function, constraints,

Table 1 Comparison of bulk and microscale properties. Materials Aluminum Copper Gold Nickel NiFe alloy Diamond-like carbon Poly Si Single crystal Si SiC Silicon nitride Silicon oxide Ebulk (GPa) Ref. [12] 69 124 82 214 130234 7001000 130180 130180 430445 280310 5080 El (GPa) Ref. [8] 70 120 70 180 120 800 160 125180 400 250 70

objectives and free variables)


rF,bulk (MPa)
Ref. [12] 200 400 220 400 4002000 800010,000 20004000 20004000 400010,000 50008000 8001100

rF,l (MPa) Ref. [8]


150 350 300 500 1600 8000 12003000 >1000 6000 1000

Screen Using Constraints (eliminate materials that do not satisfy Design Requirements)

Rank Using Objective (find the screened materials that satisfy design requirements)

Table 2 Recommended initial design values of material properties. Property Density, q (kg m3) Youngs Modulus, E (GPa) Poissons ratio, m () Fracture strength, rF (MPa) Linear expansion coefcient, a (K1) Specic heat, Cp (J kg1 K1) Intrinsic loss coefcient, gi Recommendation

Seek supporting information (research the family history of top-ranked candidates)

ql = qbulk (approx)
0.8Ebulk 6 El 6 Ebulk 0.25 rF,l = rF,bulk (approx) al = abulk (approx) Cp,l = Cp,bulk 102 < gi (polymers) 105 < gi (metals) 107 < gi < 104 (ceramics) 1 GPa < rR < 1 GPa

Final Material Choice

Residual stress, rR (MPa)

Fig. 2. Steps involved in materials selection.

Author's personal copy

G. Prashant Reddy, N. Gupta / Materials and Design 31 (2010) 113117

115

Following Ashby approach [5], the design process can be divided into four stages. In the rst stage, the designer requires range of values for many classes of materials in order to arrive at the device concept. Bulk values can typically be used to make this choice. In the next stages, the concept is translated into a class of structures and the properties of a few materials are required to a greater precision for feasibility analysis. In these stages, bulk values can be used for many properties within the bounds listed in Table 2, although measurements are typically necessary to evaluate residual stresses and intrinsic loss coefcients. In the nal stages of design, the designer requires the properties of very few materials to very high precision to evaluate the performance and reliability of the device in detail. In the Ashby approach to material selection [4,5] a function is sought to describe the performance (p) of the element under consideration. In general, this function has the form

fan often aids a heat sink by providing increased airow over the heat sinkthus maintaining a larger temperature gradient by replacing the warmed air more quickly than passive convection achieves alonethis is known as a forced air system [15]. Heat sinks thus form an important constituent in MEMS devices like micro heat exchangers and micro pumps apart from traditional microelectronic circuit applications as cooling devices. Meticulous material selection keeping in mind various design aspects and governing equations of heat transfer concerning heat sinks should optimize the device performance and Ashby methodology is one such approach.

5. Ashby approach to heat sink materials We now utilize the Ashby method for various mechanical properties such as thermal expansion (a), Youngs Modulus (E), thermal conductivity (k), and electrical properties such as resistivity (qe), conductivity by determining best suitable materials for optimizing the performance of microelectronic heat sinks. Case-I: To prevent electrical coupling and stray capacitance between a microchip and heat sink, (Fig. 3) the heat sink must be a good insulator, meaning a high resistivity, qe > 1019 lX cm. But at the same time it must also have the highest possible thermal conductivity (k) to drain away heat as fast as possible from the chip. The translation step is summarized as follows:
Function Constraints Heat sink (1) Material must have qe > 1019 lX cm (2) All dimensions specied Maximize thermal conductivity (k) Choice of material

p f fF; G; Mg

where F, G and M express the functional requirements, geometric parameters and material indices, respectively [2,14]. In many cases the variables in Eq. (1) can be separated to give

p f1 fFgf2 fGgf3 fMg

Eq. (2) permits great simplication. For all F and G, the performance can be optimized by optimizing the appropriate material indices. This optimization can conveniently be performed using graphs with axes corresponding to different material properties or material indices. 4. Heat sink design and performance A heat sink is an environment or an object that absorbs and dissipates heat from another object using thermal contact (either direct or radiant) [15]. Heat sinks function by efciently transferring thermal energy (heat) from an object at a relatively high temperature to a second object at a lower temperature with a much greater heat capacity. This rapid transfer of thermal energy quickly brings the rst object into thermal equilibrium with the second, lowering the temperature of the rst object, fullling the heat sinks role as a cooling device [12,16]. The most common design of a heat sink is a metal device with many ns, as shown in Fig. 3. The high thermal conductivity of the metal combined with its large surface area due to the ns results in the rapid transfer of thermal energy to the surrounding, cooler material. This cools the heat sink and whatever it is in direct thermal contact with. Use of uids (for example coolants in refrigeration) and thermal interface material (in cooling electronic devices) ensures good transfer of thermal energy to the heat sink. Similarly a fan may improve the transfer of thermal energy from the heat sink to the air by moving cooler air between the ns [15]. A heat sink usually consists of a base with one or more at surfaces and an array of comb or n-like protrusions to increase the heat sinks surface area contacting the air, and thus increasing the heat dissipation rate. While a heat sink is a static object, a

Objective Free variables

Mathematically, p = f{qe, k}. We have to maximize both k and qe for optimum heat sink performance. These steps can be implemented by using the kqe graph developed by Sharpe [8]. From the graph shown in Fig. 4, we nd that Aluminum Nitride (AlN) or Alumina (Al2O3) satises the constraints and our objective to maximize thermal conductivity and resistivity is achieved with these materials. Case-II: Thermal stress is the stress that appears in a body when it is heated or cooled but prevented from expanding or contracting. It depends on the expansion coefcient (a) of the material and on its modulus, E. A development of the theory of expansion leads to the relation [5]

a cG qC p =3E

where cG is the Gruneisens constant and its value ranges between about 0.4 and 4, but for most solids it is near 1. Since qCp is almost constant, we can infer that a is proportional to 1/E. Also the moduli of materials depend upon their melting point as:

100kT 3X

where k is the Boltzmanns constant and X the volume-per-atom in the structure. Substituting Eq. (4) and qCp = 3 106 J/m3 K [5] in Eq. (3), we get

a cG =100T

We can make the following observations based on these equations: (i) As temperature of the material used in heat sink increases (T), the value of thermal expansion a also decreases.

Fig. 3. Common design of a heat sink in a metal device with many ns.

Author's personal copy

116

G. Prashant Reddy, N. Gupta / Materials and Design 31 (2010) 113117

Fig. 4. Graph showing the contour of thermal conductivity (k) vs. electrical resistivity (qe) for different class of materials.

(ii) For high values of Youngs Modulus, the value of a must be signicant (i.e. neither too high nor too low). (iii) Despite satisfying the above mechanical properties, the material should have high electrical resistivity. Mathematically, the performance parameter of the heat sink is as follows:

p f fa; E; Tg

We must look to maximize the Youngs Modulus and look for materials whose convective heat transfer coefcient (h) increases with increase in temperature (T). From the a vs. E graph shown in Fig. 5 [5], we can conclude that Al, AlN, Al2O3 and to some extent even Cu and Zn alloys can be best possible materials for microelectronic heat sinks. For Al, the value of convective heat transfer coefcient increases with increase in temperature (T) [17].

Fig. 5. Thermal expansion (a) vs. Youngs Modulus (E) for different class of materials.

Author's personal copy

G. Prashant Reddy, N. Gupta / Materials and Design 31 (2010) 113117

117

transfer between the chip and heat sink. From standard values, aluminum is the best possible material to satisfy this requirement. We assume that solid A is silicon based material/substrate.

6. Conclusions Material selection for MEMS design of microelectronic heat sinks following the Ashby approach of material indices has been discussed in this paper. While each index is related to a specic performance parameter, a device usually requires several performance parameters, to be optimized for enhanced performance. We nd that consistent results are obtained in all the three cases for different mechanical and electrical properties of the heat sink. Hence aluminum based alloys/metals are very promising materials for microelectronic heat sink. References
[1] Madou MJ. Fundamentals of microfabrication. 2nd ed. Boca Raton (FL): CRC Press; 2002. [2] Srikar VT, Spearing SM. Materials selection in micromechanical design: an application of the Ashby approach. J MEMS 2003;12:319. [3] Senturia SD. Microsystem design. 1st ed. Norwell, MA: Kluwer Academic Publishers; 2001. [4] Spearing SM. Materials issues in microelectromechanical systems (MEMS). Acta Mater 2000;48:17996. [5] Ashby MF. Materials selection in mechanical design. 2nd ed. Oxford (UK): ButterworthHeinemann; 1999. [6] Ashby MF. Multi-objective optimization in material design and selection. Acta Mater 2000;48:35969. [7] Qian J, Zhao YP. Materials selection in mechanical design for microsensors and microactuators. Mater Des 2002;23:61925. [8] Sharpe WN. Mechanical properties of MEMS materials. In: Gad-el-Hak M, editor. The MEMS handbook. Boca Raton (FL): CRC Press; 2001. p. 3.13.25. [9] Shieh J, Huber JE, Fleck NA, Ashby MF. The selection of sensors. Prog Mater Sci 2001;46:461504. [10] Srikar VT, Spearing SM. A critical review of microscale mechanical testing methods used in the design of MEMS. Exper Mech 2003;43:22837. [11] Zeinab SAR. Optimization and thermal performance assessment of pinn heat sinks. J Appl Sci Res 2007;3:22735. [12] Ashby MF, Jones DRH. Engineering materials: an introduction to their properties & applications. 2nd ed. Oxford (UK): ButterworthHeinemann; 1996. [13] Gregory J. Materials selection for mechanical design. Cambridge: Massachusetts Institute of Technology, MIT Press; 2005. [14] Pratap R, Kumar A. Materials selection for MEMS devices. Ind J Pure Appl Phys 2007;45:35967. [15] Seely JH, Chu RC. Heat transfer in microelectronic equipment. 1st ed. New York: Marcel Dekker, Inc.; 1972. [16] Gilleo K. <http://www.et-trends.com/les/Heat_Sinks_Question.pdf>; 2005. [17] Holman JP. Heat transfer. 8th ed. New York: McGraw-Hill; 2001.

Fig. 6. Schematic view of contact between microchip and heat sink.

Case-III: Thermal contact resistance Consider a case wherein 2 solid bars are brought into contact as shown in Fig. 6. Suppose bar A is the microchip and bar B is the heat sink. Let the length of bar A be DxA and that of bar B be DxB performing an energy balance on the two materials, we obtain:

q kA AT 1 T 2A =DxA T 2A T 2B =hC A1 kB AT 2B T 3 =DxB or; q T 1 T 3 =fDxA =kA A hC A1 DxB =kB Ag 6


where the quantity (hCA) is called the thermal contact resistance and hC is called the contact coefcient [17], while kA and kB are coefcients of thermal conductivity for A and B, respectively. T1 is the initial temperature at surface A, T3 is the leaving temperature at B, while T2A and T2B are the temperatures at the point of contact. However, there is no 100% contact between the two and the existence of a uid between the spaces joining the two solids cannot be ignored. Therefore, we must note that there are two principal contributions to the heat transfer at the joint: (i) The solidsolid conduction at the spots of contact. (ii) The conduction through entrapped gases in the void spaces created by the contact. Designating the contact area by Ac and the void area by Av, the equation for heat transfer can be written as:
1

q T 2A T 2B =fLg =2kA Ac Lg =2kB Ac g kf Av T 2A T 2B =Lg T 2A T 2B =hC A1 7

The value of kf is neglected when compared with kA and kB. Thus we have to look for materials which have higher values of hC and higher kB values at elevated temperatures for maximum heat