An advanced semiconductor material selection for switching devices in electric vehicles using three multiple attribute decision making methods

1.   Introduction

Electric vehicles (EV) have become a very prominent choice for pollution-less transportation. All the environmental threats arisen by the green house gases emitted by conventional vehicles and the economic concerns of different countries over the unavailability of petroleum resources has propelled the search for an alternative fuel which can replace the fossil fuels used in vehicles. This has kick-started the use of electric energy for the propulsion of a vehicle. Electrical vehicle generates less pollution due to the absence of any form of emissions but the generation of electricity for EV's can indirectly contribute to pollution from power plants. This can be taken care by using renewable energy for generation which can make EV a zero emission vehicle.

Power Electronics has been vital in boosting the research on Electric Vehicles (EV's). A brief description about the various power electronics devices and their uses in EV's are carried out by ^[1]. The latest development of power semiconductor devices in EV's done by ^[2] highlights the impact of semiconductor switches in the success of EV's and also discusses predominantly used switches likes IGBT's and power MOSFET's. Unlike other applications, inside an EV power semiconductor device are subjected to face high load fluctuations due to the unpredictability in drive conditions. Hence, the design of power switching devices invites special care. There are many semiconductor materials available at material level for device manufacturing. Selection of a proper material can considerably improve the performance. The paper intends to do the selection based on many material properties along with cost constraint, which is a novel approach. Various parameters affecting the performance of semiconductor devices in EV's are discussed.

Almost all of the power electronics converter systems in automotive applications use silicon based power semiconductor switches. Silicon has already reached its fundamental theoretical limits. This leads to exploring new semiconductor materials which can be used in the fabrication of semiconductor switches. Some semiconductor materials which can replace the use of silicon in power devices in near future are GaAs, 3C-SiC, 4H-SiC, 6H-SiC, GaN and Diamond etc. The performance of these particular semiconductors hugely rely on their material properties. Some semiconductors may excel in certain properties but may lag behind when it comes to some other properties. Hence, the selection of a proper semiconductor material to make power electronics switching devices out of it requires considerable attention. In this paper, three Multiple Attribute Decision Making (MADM) methods namely Analytical Hierarchy Process (AHP) method, Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) method and PROMETHEE method are used to identify the best possible material, which can be used for the fabrication of semiconductor switches to be used in an EV. Si, GaAs, 3C-SiC, 4H-SiC, 6H-SiC, GaN and Diamond are the materials chosen as primary candidates for switch fabrication. These materials are compared with each other based on their material properties and the most suitable material is chosen using above MADM methods. In this paper, a brief description about different semiconductor materials are presented in Section 2. The material properties which determines the performance of semiconductor switches are discussed in Section 3. The three MADM methods used to determine the ranking of semiconductor materials are explored in Section 4 which is followed by the conclusion in Section 5.

2.   Semiconductor materials

There are many different types of semiconductor materials that can be used within electronic devices. Each has its own advantages, disadvantages and areas where it can be used to offer the optimum performance. Some commonly used semiconductor devices are semiconductor devices are Germanium, Silicon, Gallium arsenide. Silicon carbide, Gallium Nitride, Gallium phosphide, Cadmium sulphide, Lead sulphide etc. This paper focuses mainly on Silicon, Gallium Arsenide, 3 different grades of Silicon Carbide, Gallium Nitride and Diamond. A study about the effect of the material properties of all these semiconductors on the performance of a power electronics switching device if they are used to manufacture the same device is done in detail. A brief description about these semiconductor materials are given below.

2.1.   Silicon

Silicon-based power devices have long dominated the power electronics and power systems applications. The need for faster devices with high voltage and high switching frequency capability is growing, especially for advanced power conversion. Applications where devices are required to operate at higher than $150^\circ C$ junction temperature, high voltages, high switching frequencies, and high power densities are growing especially for electric vehicle applications. Silicon-based devices are not able to meet these stringent requirements without costly cooling systems, large number of devices in series and parallel, and costly active or passive snubbers. The latter adds to the overall size and weight, a mostly undesirable feature of silicon-based power converters.

2.2.   Silicon carbide

SiC is a wide band gap semiconductor material which is mostly used for the production of power switching devices. These devices are lighter, smaller, more compact and more efficient which make them ideal for high voltage power electronic applications. SiC is suitable for elevated temperature applications due to its high thermal conductivity and wider band gap ^[3]. SiC-based power devices have higher breakdown voltages (5 to 30 times higher than those of Si) because of their higher electric breakdown field. SiC devices are thinner, and they have lower on resistances ^[4].

2.3.   Gallium nitride

Unique materials properties of GaN-based semiconductors have stimulated a great deal of interest in research and development in materials growth and opto-electronic, and electronic devices using this semiconductor system. The major advantages of nitride-based devices that make them extremely promising for high power high temperature applications are high electron mobility and saturation velocity, high sheet carrier concentration at hetero-junction interfaces, high breakdown field, and low thermal impedance when grown over SiC substrates. The chemical inertness of nitrides is a key property to provide high reliability ^[5]. GaN and SiC diodes have similar switching properties, but as the temperature increases, the switching performance of the GaN diode is better than that of the SiC diode.

2.4.   Diamond

Diamond shows the best theoretical performance, exceeding every other WBG semiconductor by a factor of several times in every category. However, its processing problems have not yet been solved. After several years of research, there are still processing issues with SiC because of the high temperatures required; diamond is a harder material and needs even higher temperatures for processing, and not as much research has been done on its processing yet. The literature has reported the use of diamond in sensors ^[6] and field emission devices ^[7].

3.   Properties of semiconductor materials

The properties of semiconductor materials which affect the performance of devices are discussed below. The properties of various semiconductor materials namely Si, GaAS, 3C-SiC, 4H-SiC, 6H-SiC, GaN and Diamond at 300K is shown in Table 1 ^[8,9].

3.1.   Bandgap (BG)

The minimum energy difference between the top of the valence band and the bottom of the conduction band is called as Bandgap. It is a major factor determining the conductivity of a material. A semiconductor has a relatively small non-zero band gap. Some semiconductors are classified as "wide-bandgap" semiconductors because of their wider bandgap ^[10]. Higher energy gap devices are able to operate at higher frequencies and temperatures ^[11]. Also wide-bandgap semiconductors can withstand more radiation without losing their electrical characteristics. Thirdly, the larger bandgap results in a lower intrinsic carrier concentration and higher operating junction temperature ^[2].

3.2.   Electron mobility (EM)

Electron Mobility is a parameter which decides how quickly an electron can move through a semiconductor. The conductivity of a semiconductor is proportional to the product of mobility and carrier concentration. Higher mobility will result in better conductivity of the semiconductor. Mobility is a very important parameter which effects the performance in semiconductors. Higher electron mobility and electron saturation velocity allow for higher frequency of operation. The best material for high-frequency power switching applications should have a large mobility.

3.3.   Hole mobility (HM)

Hole mobility is the analogous quantity of electron mobility for holes. The highest electron and hole mobilities at room temperature of any wide-bandgap semiconductor is clearly an immensely attractive property. Very few semiconductors have an electron to hole mobility ratio below three and also have an electron mobility above 1000 $cm^2$ /V.s which is considered as a desirable parameter ^[5]..

3.4.   Breakdown or critical field (BF)

A material's electric field strength is the greatest determinant of its voltage blocking capability. For many devices, a semiconductor material with a high electric breakdown field is desirable. High breakdown electric field enables the fabrication of very high-voltage, high-power devices. High doping levels can be achieved with a high electric breakdown field, which ultimately helps to form thinner device layers at the same breakdown voltage levels. The breakdown field in SiC is about 8 times higher than in silicon. This is important for high-voltage power switching transistors.

3.5.   Saturation velocity (SV)

In a semiconductor, the maximum velocity a charge carrier (generally an electron) can attain is called as saturation velocity. A high saturation velocity is advantageous for the performance of FET's operating at high frequencies. The high frequency switching capability of a semiconductor material is directly proportional to its saturation velocity. Higher saturation velocity allows charge in the depletion region of a device to be removed faster; therefore, the reverse recovery current of WBG semiconductor based diodes is smaller, and the reverse recovery time is shorter ^[10].

3.6.   Thermal conductivity (TC)

Thermal conductivity is the property of a material's ability to conduct heat. Higher thermal conductivity means that the material is superior in conducting heat more efficiently. Higher thermal conductivity combined with wide bandgap and high critical field give WBG semiconductors an advantage when high power is a key desirable device feature. Generally, maximizing the thermal conductivity of the semiconductor material and the associated package material is desirable ^[5]. Very high thermal conductivity reduces the thermal resistance of the device die and also allows more efficient heat transfer from heat sink and yields a lower junction temperature. Higher thermal conductivity means that heat generated in a device can be more easily be transmitted to the case, heat sink, and then to the ambient.

3.7.   Maximal operation temperature (MOT)

There is a great need for electronic circuits and devices to work at high temperatures. The need for electronics to work at high operating temperatures is the result of two primary situations, either the need for greater power delivery or a higher temperature environment. For high-temperature operation of power semiconductor devices their thermal stability in terms of high-temperature performance and possibility of thermal runaway are very important considerations ^[12]. The importance of higher operating temperatures was discussed by ^[13] and found out that Si-based power devices has a limited operating capability up to 200$^\circ$C whereas WBG devices are capable of operating at temperatures as high as 600$^\circ$C. Thus, the converter has a higher reliability. The switching frequency of the devices is also limited because of the heat generated by the devices, primarily the switching losses. The increased operating temperature will also reduce the weight, volume, cost, and complexity of thermal management systems. The operation temperatures for various semiconductor materials were discussed by ^[14].

3.8.   Coefficient of thermal expansion (CTE)

Every material has a CTE measured in parts per million per degree Celsius/kelvin (ppm/$^\circ$C or ppm/k). It is the rate at which a material expands in response to heat. This is an important property because two joined materials that do not expand identically when heated will inflict stress on each other until a break occurs. Semiconductor materials that have a closer CTE match to available electrically insulating ceramics can more easily be adapted for high power and wider temperature excursion applications. It is this phenomenon that is the basis of nearly every single mechanical failure type in power modules. Semiconductor materials that have a closer CTE match to available electrically insulating (thermally conducting) ceramics can more easily be adapted for higher power and wider temperature excursion applications than ^[5].

3.9.   Dielectric constant (DC)

It is desirable that the dielectric constant of the semiconductor to be low. The dielectric constant is the measure of the capacitive loading of the device and a low DC produces reduced device impedances ^[15]. But there is a contradiction that larger the DC, smaller will be the device ^[16]. So the selection of materials depending on the DC parameter is a trade-off depending on the application.

3.10.   Cost (CO)

Cost of semiconductor is often expressed in terms of cost per wafer. Cost per wafer is perhaps the most widely used cost metric in the semiconductor industry. It is used to compare different equipment, processes, materials, Fab operating costs etc to obtain one indicator of operating cost. Cost per wafer can also be used as a benchmarking metric. Cost per wafer at the fab level can be simply obtained using the total manufacturing cost upon the total number of wafers produced. Cost per wafer at the equipment level is typically computed using the cost of equipment depreciation, cost of labor, maintenance and materials cost, cost of energy and other facilities as well as building depreciation costs.

4.   Multiple attribute decision making (MADM) methods

Decision making is one type of optimization method which can be defined as the study of identifying and choosing alternatives based on the values and preferences of the decision maker. Making a decision implies that there are alternative choices to be considered, and in such a case we want not only to identify as many of these alternatives as possible but to choose the one that best fits with our goals, objectives, desires, values, and so on ^[17]. According to ^[18], decision making should start with the identification of the decision maker(s) and stakeholder(s) in the decision, reducing the possible disagreement about problem definition, requirements, goals and criteria. The procedures of MADM can be summarized in five main steps as follows ^[19]:

● Step 1: Define the nature of the problem;

● Step 2: Construct a hierarchy system for its evaluation (Figure 1);

● Step 3: Select the appropriate evaluation model;

● Step 4: Obtain the relative weights and performance score of each attribute with respect to each alternative;

● Step 5: Determine the best alternative according to the synthetic utility values;

● Step 6: Outrank the alternatives referring to their utility values.

For finding out the best semiconductor material for making a switching device with superior performance for an EV, some important methods, known as Multiple Attribute Decision Making (MADM) methods have been used. MADM methods is used to solve problems which involves selection from a finite number of alternatives ^[20,21]. Using MADM methods we can arrive at a solution, i.e., best alternative or rank of entire alternatives, if we have the information about the attributes of various alternatives. An MADM method is carried out with the help of a decision table which has the information about alternatives (here Si, GaAS, 3C-SiC, 4H-SiC, 6H-SiC, GaN and Diamond), attributes (here SCM, BG, EM, HM, BF, SV, TC, MOT, CTE, DC and CO), weight or relative importance of attributes and measures of performance of alternatives. The alternatives and attributes are available from Table 1 and the weights has to be found out. Once the weights are obtained then Analytical Hierarchy Process (AHP) method, Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) method and PROMETHEE method are used to obtain the ranks of all materials. Based on these ranks, the best material has been chosen for fabrication of semiconductor switches used in EV's.

4.1.   Determination of weights

The procedure for determination of weights of all alternatives are as follows:

$1.$ Obtaining the normalized semiconductor materials properties:

All the elements in the decision table must be normalized to the same units in order to carry out the MADM techniques. The properties of semiconductors which was shown in Table 1 is normalized using the below relations,

For a beneficial attribute

For a non-beneficial attribute

where, $w_{(jmax)}$ is the maximum value of attribute in the alternatives for a beneficial parameter and $w_{(jmin)}$ is the minimum value of attribute in the alternatives for a non-beneficial parameter. A beneficial attribute is the one where the maximum value is considered as the better performance (eg: BG, EM etc.) and a non-beneficial attribute is the one where the minimum value is considered as the better performance (eg: DC, CO etc.).

Similarly all the attributes for different alternatives are normalized using the above relations. Now normalization of CTE invites special attention as it is not normalized depending upon least or higher values of attributes. Since every material should have a similar CTE to Si or SiC as possible (Si is considered as the substrate for all materials here) ^[22], the difference between the CTE's of different alternatives with Si is found out and this measure of closeness to the CTE of Si is later used to assign values or weights for all alternatives ranging from 0 to 1 with the help of a machinability index table ^[20] as shown in Table 2.

The final normalized semiconductor materials properties are shown in Table 3. This table is also considered to be a 7$\times 10$ matrix, $A_{1}$ for future computations.

$2.$ Determining the relative importance of each attributes with respect to the objective:

Here, a pair-wise comparison matrix using a scale of relative importance is constructed. When an attribute is compared with itself it is assigned a value of 1, which means the diagonal elements of the pair-wise comparison matrix will be always 1. Numbers 3, 5, 7 etc correspond to verbal judgments 'moderate importance', 'strong importance', 'very strong importance ' etc ^[23]. Now considering N attributes, we will have a square matrix $A_2$ of order N$\times$N. Here, $a_{ij}$ value denotes the comparative importance of attribute i with respect to attribute j. Also, $a_{ij}$ = 1 if i = j and $a_{ji}$ = 1/$a_{ij}$. The pair-wise comparison matrix (also known as the relative importance matrix) $A_2$ for the semiconductor materials parameters are shown in Table 4.

$3.$ Determining the relative normalized weight:

Once the normalized matrix is formed then the relative normalized weight $w_{j}$ of each attribute has to be found out by first calculating the geometric mean of the i-th row and then normalizing the geometric means of rows in the comparison matrix. This can be done using the following relations:

The final tabulated values $P_j$ and ($w_j$) of all the attributes is shown in Table 5. This is also considered to be a 10$\times matrix,$ A_{3} $ for future computations.

$4.$ Calculate matrices $A_{4}$ and $A_{5}$ such that $A_{4}$ = $A_{2}$$\times$$A_{3}$ and $A_{5}$ = $A_{4}$/$A_{3}$.

$A_{5}$ can be obtained as $A_{5}$ = $A_{4}$/$A_{3}$. For eg., $A_{51}$ can be calculated as,

$5.$ Determine the maximum eigen value $\lambda_{max}$ which is the average of matrix $A_{5}$.

$6.$ Calculate the consistency index CI using below relation. The smaller the value of CI, the smaller will be the deviation from consistency.

$7.$ From the Random Index (RI) values shown in Table 6 ^[20], obtain the RI for the number of attributes used in decision making. Here N = 10 and hence RI = 1.49.

$8.$ Calculate the consistency ratio CR using below equation. A CR of 0.1 or less is considered to be acceptable and indicates the selection of relative importance matrix is properly done.

$9.$ Once the CR value is verified we can make sure that the obtained values for weights are correct and can be chosen for further computations.

4.2.   Analytic Hierarchy Process (AHP) method

Developed by ^[23], the Analytic Hierarchy Process (AHP) Method is one of the most popular technique for decision making problems. The determination of rank of materials using AHP method is as follows:

$1.$We can use now the matrix $A_3$, which is the normalized weight matrix of AHP method (Refer Table 7). It can be noted that the sum of all the weights of attributes is equal to 1.

$2.$Once the CR value is verified, the overall or composite performance scores for the alternatives has to be obtained by multiplying the normalized value of each alternative with its corresponding normalized weight value $w_j$ and summing over the attributes for each alternatives.i.e., the 7$\times 10$ matrix $A_1$ is multiplied with the 10$\times 1$ matrix $A_3$ to get a 7$\times 1$ column matrix $A_6$ (as shown in Table 8) whose values are the composite performance scores of the corresponding alternatives.

$3.$Arranging the score in descending order gives the rank of the semiconductor materials according to which it can be chosen for power electronics devices. Diamond ranks the highest whereas GaN ranks the lowest. The score and rank of all the alternatives and their ranks are shown in Table 9.

4.3.   Technique for order preference by similarity to ideal solution (TOPSIS) method

Developed by ^[24], this method is based on the concept that the chosen alternative should have the shortest Euclidean distance from the ideal solution. TOPSIS gives a solution that is not only closest to the hypothetically best, but that is also farthest from the hypothetically worst. The procedure for the selection of the best alternative from the available alternatives using TOPSIS method is discussed below:

$1.$Determine the objective and identify the pertinent evaluation attributes.

$2.$Based on the information available on attributes prepare a matrix. In this particular case this matrix is nothing but the matrix shown as table properties (i.e., Table 1). Each row of this matrix is allocated to one alternative, and each column to one attribute.

$3.$In the case of subjective attributes we can use the machinability index table ^[20] which was explained earlier (See Table 2) to get an objective value for it.

$4.$Next step is to obtain the normalized decision matrix, $T_{ij}$. This can be obtained using the following relation:

where $m_{ij}$ is an element of the actual table, which is non-normalized and $m_{ij}$ corresponds to the value of the j-th attribute for the i-th alternative and N is the number of alternatives.

Here again normalization of CTE is to be done with special care as mentioned previously. Initially, the CTE attribute values has to be normalized using the method of subjective measure of attribute (Refer Table 2) because the CTE performance is determined by measuring the closeness to the substrate over which it is mounted (here Silicon). This makes it a subjective quantity which should be assigned a value depending on its closeness to the substrate. This value can be directly taken from Table 3 mentioned previously. These values are latter normalized using the TOPSIS method. Similarly, all the attributes of all the alternatives can be normalized using TOPSIS method which is shown in Table 10. It is also considered as a $7\times 10$ matrix $T_{1}$ for future purposes.

$5.$Decide the weights of the different attributes. Weights determined earlier can be used for this purpose. We can recall the matrix $A_3$ which is the normalized weight matrix of AHP method (Refer Table 7).

$6.$Find out the weighted normalized matrix $V_{ij}$. It can be obtained by the multiplication of each elements of the column of the matrix $T_{ij}$ with its associated weight $w_{j}$. The elements of the weighted normalized matrix $V_{ij}$ can be expressed as,

All the equivalent weighted normalized elements of the $7\times 10$ matrix $V_{1}$ can be determined using the above equation. The complete weighted normalized matrix $V_{1}$ is shown in Table 11.

$7.$Obtain the positive ideal and the negative ideal solutions, i.e., the best and worst solutions. It can be determined as,

and

${V_{j}}^{+}$ means the positive ideal or best value of the particular attribute among the values of the attributes of different alternatives. Where higher values are desirable i.e., beneficial attributes (here BG, EM etc up-to CTE), ${V_{j}}^{+}$ indicates the higher value of the attribute. ${V_{j}}^{+}$ indicates the lower value of the attribute if the attributes are non-beneficial i.e., where least values are desirable (here DC and CO).

${V_{j}}^{-}$ means the negative ideal or worst value of the particular attribute among the values of the attributes of different alternatives. Where higher values are desirable i.e., beneficial attributes (here BG, EM etc up-to CTE), ${V_{j}}^{-}$ indicates the lower value of the attribute. ${V_{j}}^{-}$ indicates the higher value of the attribute if the attributes are non-beneficial i.e., where least values are desirable (here DC and CO).

The positive ideal value ${V_{j}}^{+}$ and the negative ideal value ${V_{j}}^{-}$ of all the attributes are shown in Table 12 and Table 13.

$8.$Determine separation measures. The separation of each alternatives from the ideal value can be obtained using the following equations.

where i varies from 1 to N i.e., 1 to 7 here.

The complete ${S_{i}}^{+}$ and ${S_{i}}^{-}$ values of all the alternatives are shown in Table 14.

$9.$Obtain the relative closeness of a particular alternative to the ideal solution. It is denoted as $P_{i}$ which can be determined as,

$10.$The complete composite performance score $P_{i}$ of all the alternatives and arranging the score in descending order gives the rank of the semiconductor materials according to which it can be chosen for power electronics devices. The score and rank of all the alternatives and their ranks are shown in Table 15.

The alternative having the highest $P_{i}$ value is ranked one and so on. Hence we can decide on the most preferred and least preferred alternatives for a particular problem (in this case, selection of semiconductor materials) accordingly. We can see 4H-SiC is the most preferable choice for selection of semiconductor material whereas GaN is the least preferred.

4.4.   PROMETHEE method

The PROMETHEE method was first proposed by ^[25] and ^[26] and later improvised ^[27]. The values of attributes need not necessarily be normalized or transformed into a common dimensionless scale in this method. Instead it is assumed higher score value means a better performance. The weights of the criteria has to be determined separately in this method, which has been already carried out. The procedure to use PROMETHEE method to determine the rank of semiconductor materials to be used in switching devices in EV's is as follows:

$1.$Determine the individual alternative comparison matrix for all the attributes.

A 7$\times 7$ matrix A$_N$ is formed comprising of zeros and ones. For a beneficial attribute, an alternative having an higher value is assigned 1 when compared with an alternative having lower value and assigned 0 when compared with an alternative having higher value. For a non-beneficial attribute, an alternative having an higher value is assigned 0 when compared with an alternative having lower value and assigned 1 when compared with an alternative having higher value. An alternative compared with itself is assigned 1 which makes all the diagonal elements of the 7$\times 7$ matrix A$_N$ as 1. If both alternatives has the same value it is assigned as 0 when comparing. The A$_{1}$ matrix for the attribute band-gap (BG) is shown in Table 16. Similarly the alternative comparison matrix for the nine other attributes have been found out.

$2.$Multiply the individual alternative comparison matrix of all the attributes with its corresponding weights.

For eg., All the elements of Table 16 is multiplied with the weight obtained for the attribute band-gap (BG) i.e., 0.084 to get Table 17. Similarly all the other individual alternative comparison matrix of each attributes are multiplied with its corresponding weights determined earlier (Refer Table 5).

$3.$Once all the weighted individual alternative comparison matrices are obtained, the final performance matrix A$_{PM}$ of all alternatives has to be determined using the following equations.

where N is the number of attributes (here BG, EM, ....., CO.) i.e., 10 and i, j varies from 1 to 7. It can be noted that all the diagonal elements of the final performance matrix A$_{PM}$ is equal to one as sum of all the weights determined is one. The final performance matrix A$_{PM}$ is shown in Table 18.

$4.$Determine the row-wise and column wise matrices S$_{R}$ and S$_{C}$ as follows:

where,

and

The final row-wise and column wise matrices S$_{R}$ and S$_{C}$ is shown in Table 19.

$5.$Determine the final performance scores P$_{i}$ of each alternatives as follows:

where, i = 1 and j varies from 1 to 7.

$6.$Arranging the final performance scores P$_{i}$ in descending order gives the rank of the semiconductor materials according to which it can be chosen for power electronics devices. The score of all the alternatives and their ranks obtained using PROMETHEE method is shown in Table 20.

The alternative having the highest Pi value is ranked one and so on. Hence, we can decide the most preferred and least preferred alternatives for a particular problem (here selection of semiconductor materials) accordingly. We can see 4H-SiC is the most preferable choice whereas 3C-SiC turned out to be the least preferable using PROMETHEE method.

5.   Conclusion

The different material properties of semiconductors like Si, GaAs, 3C-SiC, 6H-SiC, 4H-SiC, GaN and Diamond are studied. Also, their effect on the performance of switching devices are discussed. The ranks of different semiconductor materials, which can be used for the fabrication of high power semiconductor switches, are estimated using three multiple attribute decision making methods, namely Analytic Hierarchy Process (AHP) method, Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) method and PROMETHEE method. The ranks obtained using all the methods are shown in Table 21.

Using AHP method Diamond achieves the highest ranking whereas 4H-SiC secures the best position when TOPSIS and PROMETHEE methods are used. Even though Diamond has better electrical properties, its research and development is still at an infant stage. In EV applications where cost of the vehicle plays a huge role in its success and acceptability, the high cost of diamond makes its use in EV application nearly close to impossible at present. Si, even though being reached its technological limits and having a poor device performance compared to WBG semiconductor materials hasn't lost its significance due to its matured device technology and reduced cost and high availability. In WBG semiconductors, SiC technology is getting more matured than GaN technology due to its better research and commercialization. Also, the reduced thermal conductivity of GaN pulls it out of competition with SiC when it comes to deciding switching devices for EVs. Gallium is not naturally found in a deposit which makes it further complicated in extracting the material. GaAs is fairly brittle which limits the wafer diameter whereas silicon wafers can typically found in higher diameters. This significant decrease in usable area adds again to the cost of a GaAs wafer. Also, the toxicity of arsenic itself invites special handling to prevent adverse effects. In future when the WBG semiconductors technology advances in par with Si technology, then SiC materials are going to be the best contenders for power electronics semiconductor switches. The applied MADM techniques suggests that 4H-SiC will be a better choice when it comes to selection of materials for fabrication of semiconductor switching devices in EV's when compared with its fellow material grades due to its superior performance and reduced cost.

Conflict of interest

The authors declare that there is no conflicts of interest in this paper.