You are on page 1of 5

Materials and Design 41 (2012) 177–181

Contents lists available at SciVerse ScienceDirect

Materials and Design


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

Technical Report

A comparative study on material selection for micro-electromechanical systems


Aditya Chauhan b, Rahul Vaish a,⇑
a
School of Engineering, Indian Institute of Technology Mandi, Mandi 175 001, India
b
Galgotias College of Engineering and Technology, Greater Noida 201 306, India

a r t i c l e i n f o a b s t r a c t

Article history: Material selection is an important task because large numbers of competitive materials are available for
Received 17 February 2012 various technological applications. Numerous material selection techniques are reported whereas most of
Accepted 20 April 2012 the techniques are knowledge based and require performance indices for material selection. In this con-
Available online 8 May 2012
text, Ashby approach is one of the efficient methods which rely on performance indices of materials in
specific application. However Multiple attribute decision making (MADM) approaches do not require
exact physical relation for material selection. We have investigated micro-electromechanical system’s
(MEMS) material selection using MADM approaches and compared their results with that of Ashby
approach. Almost similar materials ranking indicates that MADM approaches are also efficient and ease
to apply without any prior mathematical calculation for materials properties-application relation.
Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction (c) polymers such as polyimmide and (d) ceramics such as carbides
and diamond.
Last decade has witnessed large scale commercialisation of var- This growing set of materials has further increased the scope of
ious new and innovative technologies. Micro-electro-mechanical MEMS applications. However, now the designers are faced with the
systems (MEMSs) are one of the key areas of research and techno- problem of selecting the optimum material for specific MEMS
logical application which has been speedily expanded in last dec- application. To overcome such a problem various multiple criteria
ade. MEMS refer to microscopic devices that have a characteristic decision making (MCDM) and optimisation techniques are re-
length of less than 1 mm but more than 100 nm and combine elec- ported and found to be efficient in various decision making appli-
trical and mechanical components. MEMS research will be contin- cations [2,3]. The Ashby [4] approach is one of the most popular
ued in the near future due to the highly untapped potential that techniques used for material selection problems. MEMS materials
still exists in this field [1]. MEMS technologies have been success- selection has been studied using Ashby approach which is based
fully transformed into the variety of applications including sensors, on multi-objective decision making (MODM) approach [5–9]. This
actuators, power generating devices and bio-medical services. method is efficient in screening compatible materials from large
However, one of the main hurdles that lie in the path of further database of materials. However, MADM techniques are not em-
development of MEMS technology is the fact that the materials ployed in selecting MEMS materials. Various MADM methods have
are very limited as the set of requirements are very narrow. There been reported and found to be promising [10]. Through this paper,
are three basic requirements that the material must possess to be an attempt has been made to compare the results of these tech-
able to be used in MEMS, these are: (a) compatibility with semi- niques (Ashby approach and MADM methods) when are applied
conductor fabrication technology, (b) desirable electromechanical to two case studies including selection of MEMS materials for the
properties and (c) low values of residual stresses. For this reason application of (a) vibratory gyroscope and (b) capacitive micro-
the early attempts at MEMS technology were highly restricted to machined ultrasonic transducer (CMUT).
the silicon family. However, with the development of advanced
fabrication techniques such as LIGA, stereo lithography and laser 2. Materials and methods
micromachining the number of materials that can be used for
MEMS devices have grown considerably and now include four 2.1. Materials
main categories of materials which are: (a) metals such as nickel
and aluminium, (b) non-metals such as silicon and germanium, The success of MEMS technology largely depends upon the
materials performance. The key materials used in MEMS devices
include Si-based materials, various conductors and insulators.
⇑ Corresponding author. Tel.: +91 1905 237908; fax: +91 1905 237945. Si-based materials are generally used for mechanical component.
E-mail address: rahul@iitmandi.ac.in (R. Vaish). The basic MEMS mechanical components are diaphragms, beams,

0261-3069/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.matdes.2012.04.037
178 A. Chauhan, R. Vaish / Materials and Design 41 (2012) 177–181

Table 1
Important MEMS materials and their properties.

Material Elastic modulus E (GPa) Failure strength (GPa) Thermal conductivity K (W/cm/C) Coefficient of thermal expansion a (106/C) Density (kg/m3)
3H–SiC 400 7 3.5 3.3 3200
SCSi(1 1 0) 168 7 1.57 2.33 2300
SCSi(1 0 0) 130 3.4 1.57 2.33 2300
Diamond 800 8.5 6.9 1 3500
Si3N4 250 6.4 0.19 0.8 3100
Aluminium 70 0.17 2.36 25 2700
Copper 120 0.25 3.98 16.6 8960
Poly-Si 159 1.65 0.34 2.8 230
SiO2 70 1 0.001 0.55 2500
Polyimmide 8 0.04 0.001 20 1420
Titanium 110 0.5 0.2 8.5 4510
Nickle 185 0.4 0.899 13 8910
Tungsten 410 0.7 1.78 4.5 19,300
PVDF 2.3 0.05 0.002 140 1780

cantilevers and disks. A variety of major issues concerning materi- in their approach on the following basis: (a) the normalisation
als for the application of MEMS technology have been covered by technique used in TOPSIS is vector while that in VIKOR is linear,
Spearing [11,12]. An extensive survey of thin film mechanical (b) the technique used in TOPSIS tries to evaluate the alternative
properties has been done by Sharpe et al. [13]. Spearing [12] has with the maximum distance from the negative ideal solution while
extended research on other mechanical properties of thin film VIKOR tries to give the alternative closest to the positive ideal solu-
MEMS materials. A list of potential MEMS materials is mentioned tion, (c) VIKOR is a compromise approach and hence, allows differ-
in Table 1. ent users to apply their own judgement to the evaluation process
and (d) VIKOR (extended) can be used to evaluate the fitness of
the final output thereby evaluating feasibility of the final outcome.
2.2. Material selection methods Other methods (which are common in materials selection) are
questionnaire method, artificial neural network and Multiple Attri-
MCDM techniques have been used in various decision making butes Decision Making Approaches (AHP, SAW, WPM, Grey rela-
problems in diverse technological applications. MCDM methods tional analysis and ELECTRE, etc.). These methods are explained
can be categorised into multi-objective decision making (MODM) in detail for materials selection viewpoint [2]. One of the simplest
and MADM approaches. Ashby has proposed systematic material and efficient methods is Weighted Product Method (WPM) [16].
selection procedure using MODM approach. MODM approaches This method is computationally simple and do not require normal-
are useful in screening large number of alternatives. Material selec- isation process. Each alternative is compared with the others by
tion is a problem of optimising various conflicting objectives. multiplying a number of attributes. Each attribute is raised to the
According to Ashby approach, objectives can be clubbed together power equivalent to the relative weight. It is very similar to
to form suitable performance indices. These indices form a basis weighted sum model (WSM).
of comparison for different materials and thus, the material with
the highest performance index is the most suitable candidate for
3. Case studies
a given application. Ashby divided performance index P of any
engineering component as P = f(F, G, M), where ‘f’ denotes function
For the comparison of our results, we here study material
of functional (F), geometric (G) and materials (M) parameters,
selection for two MEMS applications. Pratap and Kumar [5] have
respectively. These parameters are independent of each other
reported MEMS materials selection using Ashby approach. Materi-
and their collective output determines the overall performance of
als indices are evaluated for MEMS gyroscope and capacitive
the component. If there are more than one performance indices,
micromachined ultrasonic transducer (CMUT) [5]. We have further
material selection is compromised solution. Compromised solu-
studied the MEMS material selection using TOPSIS and VIKOR
tions (Pareto-optimal) are the alternatives that have the best
method and compared with that of results obtained from Ashby
trade-off between objectives and are not dominated by any other
approach [5].
alternative in the solution space. Various multi-objective evolu-
tionary algorithms (MOEAs) are extensively investigated for
Pareto-optimal solution in multi-objective decision making prob-
lems. Selection of a best material from all Pareto-optimal solutions
is not possible using Ashby approach. On the other hand MADM
methods can be used to rank finite number of materials. Two
popular methods (technique for order preference by similarity to
ideal solution (TOPSIS) [14] and VlseKriterijumska Optimisacija I
Kompromisno Resenje (VIKOR)) [15] are employed in present study.
TOPSIS methodology is based on relative distance of an alternative
material from ideal and negative ideal solution. The best material
is the one which is close to ideal solution and far from nadir solu-
tion. The ideal solution is the collection of ideal scores in all attri-
butes whereas nadir solution is the combination of the least
performance values. Usually Euclidean distance is used for measur-
ing distances between the alternative materials. The VIKOR meth-
od is a compromise approach MADM model. Even though both
TOPSIS and VIKOR are essentially the same techniques both differ Fig. 1. Schematic sketch for the gyroscope.
A. Chauhan, R. Vaish / Materials and Design 41 (2012) 177–181 179

Table 2 Now the TOPSIS and VIKOR approach are applied to the given
MEMS materials ranking based on Ashby, TOPSIS and VIKOR approaches for vibratory data. The desirable properties for the selection of material for a
gyroscope applications.
vibratory gyroscope are: a lower value of Young’s modulus and
Materials Area (sq. units) Ashby ranks TOPSIS ranks VIKOR ranks coefficient of thermal expansion while a higher value of failure
ScSi(1 0 0) 83892.47662 1 5 3 strength and thermal conductivity are required. These are the
ScSi(1 1 0) 80028.43982 2 2 2 properties which form the positive ideal solution and vice versa
Si3N4 22793.96725 3 6 5 is done for negative ideal solution. The ranks of the materials using
Diamond 21472.58846 4 3 4
3H–SiC 6589.380338 5 1 1
TOPSIS and VIKOR methods are tabulated in Table 2.
Poly-Si 3336.88908 6 8 7
SiO2 1806.620134 7 9 10
3.2. Capacitive micromachined ultrasonic transducer (CMUT)
Copper 391.6963383 8 4 6
Aluminium 350.810943 9 7 8
Titanium 194.9262995 10 10 12 CMUTs consist of a fixed electrode on a surface and a membrane
Tungsten 167.3507988 11 12 9 electrode that can vibrate. It works on piezoelectricity and variable
PVDF 78.60438347 12 14 14 capacitance. Ultrasonic waves generate by the vibration of the pie-
Nickle 62.99471832 13 11 11
Polyimmide 36.61140402 14 13 13
zoelectric membrane. The application of CMUT’s ranges from sim-
ple detection devices to integrated actuators in control systems to
non-destructive testing and bio-medical applications. Fig. 2 shows
3.1. Vibratory gyroscope an illustration of typical CMUT cross-section.
The two indices used to measure the performance of a CMUT
A gyroscope is used to sense and measure angular motion using are: (M4 = k/a); a higher value of this index essentially means that
vibrating mechanical elements. It works on coriolis component of the materials chosen for the vibrating membrane are thermally
applied force to sense angular rates. A vibratory gyroscope has more stable. This is done because change in operational tempera-
one or more proof masses that are excited by means of a feed-back ture causes dimensional change which causes a shift in the funda-
control device. The amplitude of excitation is carefully controlled mental frequency. Thus, thermal stability becomes an important
and frequency of vibration is set to natural resonant frequency. criterion while selecting material for the membrane. The second
3=2
When the proof mass is made to rotate about an axis perpendicular performance index is, M 5 ¼ rf =ðE  qÞ1=2 . This is composite index
to the plane of vibration, the coriolis component of the force sets created by combining two separate indices [5] (E/q)1/2 and (r3=2f =E)

the proof mass into vibration along the axis perpendicular to the which are responsible for maximising flexural rigidity and sensitiv-
excitation direction but in the plane of vibration. This secondary ity under a specified pressure, respectively. In order to rank studied
vibration is dependent on the angular rate of rotation and thus, materials, we combine the two indices to from a third index
is employed to find out the angular rate. Schematic sketch of the (M6 = M4  M5) similar to M3 (in vibratory gyroscope). Thus, an
gyroscope is shown in Fig. 1. overall higher value of M6 indicates better performance of a given
The two indices selected for performance measure are: (M1 = rf/ alternative. The relative rank of materials based on the composite
E) [5] (maximisation of this index means that the component will index is mentioned in Table 3.
possess a higher cut-off frequency thereby increasing its upper In order to study material selection using MADM approaches,
limit of functionality) and (M2 = k/Ea2) (maximisation of this index TOPSIS and VIKOR methods are applied to the given data. The
means that the component will have a higher deflection for a given desirable properties for the selection of material for CMUT devices
frequency) thereby improving its sensitivity. Hence desirable are: lower value of Young’s modulus, coefficient of thermal expan-
materials require larger values of M1 and M2. In order to give quan- sion and density while higher value of failure strength and thermal
titative ranking to the materials, we combine the two indices to conductivity are desirable. Ranks obtained from MADM
from a third index (M3 = M1  M2). Thus, an overall higher value approaches (TOPSIS and VIKOR) are mentioned in Table 3.
of M3 indicates better performance of a given alternative. M3 can
be interpreted as area under the plot of M1 vs M2. This method is 4. Results and discussions
similar to weighted product method (MADM approach) [16] when
we assume unit weights for both the indices (M1 and M2). The It is clear from Tables 2 and 3 that ScSi (1 0 0) and diamond are
relative rank of materials based on the composite index (M3) is the best materials (based on Ashby approach) for the application
mentioned in Table 2. of vibratory gyroscope and CMUT applications respectively

Table 3
MEMS materials ranking based on Ashby, TOPSIS and VIKOR approaches for CMUT
applications.

Materials Area (sq. units) Ashby ranks TOPSIS ranks VIKOR ranks
Diamond 138.3577915 1 4 4
ScSi(1 1 0) 27.1817485 2 2 2
3H–SiC 23.50722353 3 1 1
ScSi(1 0 0) 10.45999965 4 3 3
Si3N4 5.914089343 5 5 5
Poly-Si 1.822172463 6 8 6
Tungsten 0.11150366 7 13 14
Copper 0.039133215 8 7 8
Aluminium 0.020607231 9 6 7
Nickle 0.018449592 10 12 12
Titanium 0.015991405 11 11 11
SiO2 0.005884688 12 9 9
Polyimmide 5.08131E06 13 10 10
PVDF 3.37978E06 14 14 13
Fig. 2. Cross-sectional sketch of CMUT devices.
180 A. Chauhan, R. Vaish / Materials and Design 41 (2012) 177–181

similar components could be different and hence different indices


are supposed to be used in it. Ashby approach is a highly reliable
technique and its results are highly accurate. Originally the Ashby
approach was extended to the MEMS materials by Srikar and Spear-
ing [11]. Spearing has extended the performance indices to MEMS
devices by approximating various MEMS component as beams,
plates, diaphragms, disks and springs. Additionally various issues
regarding application of materials in MEMS application have been
covered by Spearing [12], which included detailed analysis of effects
of scaling, processing, fabrication, material characteristics and de-
sign issues. Thus, a comprehensive approach was established to
evaluate the material suitability for a given MEMS application using
the well established Ashby approach.
On the other hand MADM techniques like TOPSIS and VIKOR
can be used without prior knowledge of physical relations of mate-
rials properties for their specific applications [17]. It offers the fol-
lowing advantages when used: (a) it can be used to evaluate the
ranks of alternatives regardless of the number of attributes associ-
ated with it, (b) the technique can be applied regardless of the
Fig. 3. TOPSIS and VIKOR materials ranks plotted against Ashby materials ranks for
interdependence of the attributes, (c) no functional relationship
vibratory gyroscope applications.
needs to be established or used, (d) the user can apply weights
as necessary to prioritise different attributes as per the given de-
understudy. 3H–SiC, ScSi (1 1 0) and diamond are found to be best sign constraints, (e) less computational time required, and (f)
materials using TOPSIS and VIKOR methods. The results from two non-numeric attributes and weights can also be considered using
MADM approaches (TOPSIS and VIKOR) are in good agreement fuzzy techniques. However material selection requires more
and quite similar to that of given by Ashby approach. Significantly sophisticated methods since materials properties vary with fabri-
high linear correlation coefficients are observed between the ranks cation methods and scale of study.
obtained from Ashby-TOPSIS (+0.82), Ashby-VIKOR (+0.88) and Even though it is an open question for the researchers to com-
TOPSIS–VIKOR (+0.94) in case of vibratory gyroscope application pare, differentiate, evaluate and critically analyse various MADM
(Fig. 3). Similarly for CMUT application, linear correlation coeffi- and MODM methods for material selection and screening proce-
cients are Ashby-TOPSIS (+0.81), Ashby-VIKOR (+0.80) and dures, during the study we come across an additional phase of
TOPSIS–VIKOR (+0.98) as shown in Fig. 4. It indicates that the evaluation. Mostly it is the case that the various performance indi-
MADM techniques deliver the very similar result as that of Ashby ces associated with the overall performance of component may be
approach. Results of MADM methods can be further improved by at conflict with each other. As is the case in the second case study
considering subjective weights (for all the properties under study) of CMUT the composite index M 5 ¼ r3=2 f =ðE  qÞ
1=2
is made up of
based on expert opinion. In the present study equal weights are as- two separate indices which are (E/q)1/2 and (r3=2 f =E) [5]. As is
signed for all properties under study in both the case studies. How- observed one tends to maximise Young’s modulus while other tries
ever, Ashby technique is an efficient method and directly correlates to minimise it to increase the overall performance of the same
the effect of various physical parameters on the overall performance component. It then becomes additional problem to evaluate or
of a device. However, the formulation of indices can be a cumber- rank the materials based on these conflicting indices. Also, addi-
some and difficult task based on the fact that each application is tionally when the number of such indices increases to a large num-
inherently different from another application and also, exact func- ber it becomes difficult to apply visual techniques to select or
tional relation between the parameters must be known and require screen the materials. In such cases additional techniques may be
expert decision. The constraints imposed on different category of required to come to conclusive results. Pratap and Kumar [5] has
reported two separate techniques to combine different indices to
generate a composite index. First is the value function approach
in which a normalised performance index Pnorm is evaluated as
Pnorm = a1M1 + a2M2 + a3M3 where ai (i = 1,2,3) are the weights to
be decided by the designer (Ra = 1) and individual indices are
normalised as Mnorm = Mi/Mmax. However, this method has inher-
ent drawback of masking the effect of weaker index. Therefore,
WPM has been used to couple two indices to provide a more justi-
fied approach of evaluating a composite index. In the original paper
visual and graphical techniques have been used to screen the vital
materials, and the final selection is a qualitative approach. In order
to have a quantitative approach we have extended the WPM to cre-
ate a single performance index which may be directly used for
ranking. It brings us to the discovery that a hybrid technique
may be evolved which combines the best features of MADM and
MODM methods for efficient analysis of material performances
for designated applications. However, it requires extensive re-
search. It is never ending challenge for researcher because of fast
development of materials and their applications. Further compari-
son studies are required to examine consistency. A sensitivity anal-
Fig. 4. TOPSIS and VIKOR materials ranks plotted against Ashby materials ranks for ysis is essential for all MADM techniques to find out which method
CMUT applications. is more appropriate for a specific problem.
A. Chauhan, R. Vaish / Materials and Design 41 (2012) 177–181 181

5. Conclusions [3] Opricovic S, Tzeng GH. Compromise solution by MCDM methods: a


comparative analysis of VIKOR and TOPSIS. Eur J Oper Res 2004;156:
445–55.
MADM approaches (TOPSIS and VIKOR) are employed in MEMS [4] Ashby MF. Multi-objective optimization in material design and selection. Acta
materials selection. Their results are in close agreement with that Mater 2000;48:359–69.
[5] Pratap R, Kumar AA. Material selection for MEMS devices. Ind J Pure Appl Phys
of Ashby approach. Materials are selected for MEMS vibratory
2007;45:358–67.
gyroscope and capacitive micro-machined ultrasonic transducer [6] Guisbiers G. Wautelet, materials selection for micro-electromechanical
devices. Good concurrence (significantly high linear-correlation) systems. Mater Des 2007;28:246–8.
[7] Qian J, Zhao YP. Materials selection in mechanical design for microsensors and
between results of Ashby approach and MADM techniques like
microactuators. Mater Des 2002;23:619–25.
TOPSIS and VIKOR indicates that MADM techniques can be used [8] Parate O, Gupta N. Material selection for electrostatic microactuators using
for material selection without prior knowledge of exact physical/ Ashby approach. Mater Des 2011;32:1577–81.
mathematical relations of materials properties for their specific [9] Reddy GP, Gupta N. Material selection for microelectronic heat sinks: an
application of the Ashby approach. Mater Des 2010;31:113–7.
application. However more work needs to be extended for examine [10] Xuebin L. Study of multi-objective optimization and multi-attribute decision-
robustness of MADM techniques in material selection. making for economic and environmental power dispatch. Electric Pow Syst Res
2009;79:789–95.
[11] Srikar VT, Spearing SM. Materials selection in micromechanical design: an
Acknowledgements application of the Ashby approach. J Microelectromech Syst 2003;12:
3–10.
[12] Spearing SM. Materials issues in microelectromechanical systems (MEMS).
One of the authors (RV) acknowledges support from the Indian Acta Mater 2000;48:179–96.
National Science Academy (INSA), New Delhi, through a grant by [13] Sharpe WN, Bagdahn J, Jackson K, Coles G. Tensile testing of MEMS materials-
the Department of Science and Technology, (DST), New Delhi, un- recent progress. J Mater Sci 2003;38:4075–9.
[14] Milani AS, Shanian A, Madoliat R, Nemes JA. The effect of normalisation norms
der INSPIRE faculty award-2011 (ENG-01). in multiple attribute decision making models: a case study in gear material
selection. Struct Multidiscip Optim 2005;29:312–8.
[15] Opricovic S, Tzeng GH. Extended VIKOR method in comparison with
References outranking methods. Euro J Oper Res 2007;178:514–29.
[16] Athawale VM, Chakraborty S. A comparative study on the ranking performance
[1] Judy JW. Microelectromechanical systems (MEMS): fabrication, design and of some multi-criteria decision-making methods for industrial robot selection.
applications. Smart Mater Struct 2001;10:1115–34. Int J Ind Eng Comput 2011;2:831–50.
[2] Jahan A, Ismail MY, Sapuan SM, Mustapha F. Material screening and choosing [17] Chauhan A, Vaish R. Magnetic material selection using multiple attribute
methods – a review. Mater Des 2010;31:696–705. decision making approach. Mater Des 2012;36:1–5.

You might also like