3. Challenges and trends of the growth processes of SiC thin films

The main motivation to use thin films is simple: easy growth on a wide variety of substrates. In the case of the SiC, the use of thin films includes other motivations: (a) the cost of SiC bulk substrates is still high; (b) the defect density is relatively high and (c) the area of the substrate available is still small (Fraga, 2011d).

Nowadays, the SiC devices are categorized into two groups: one uses SiC bulks and the other uses SiC thin films grown on Si or insulator/Si substrates. In general, comparative studies show that devices based on SiC bulk substrates exhibit better performance than those thin "%()z/! zc..%/\_zDLLHd^z\$!\*\_z0\$!z"%./0z\$((!\*#!z%/z0+z#.+3\*z%z0\$%\*z"%()/z3%0\$z,.+,!.¥ ties as good as the bulk substrate. In addition, it is necessary to achieve high film growth rate as well as thickness uniformity and homogeneity when deposited on large-area Si wafers, which are important factors to reduce costs. For this, the influence of SiC growth process parameters, such as gases flow rate, substrate temperature, pressure and doping, have been evaluated and optimized.

+z%((1/0.0!z0\$!z+),.%/+\*z)+\*#zz\* z/,100!.%\*#\_z3!z3%((z %/1//z0\$!z.!/1(0/z+¥ 0%\*! z%\*z+1.z3+.'/z+\*z%z"%()/z#.+3\*z5zz\* zz)#\*!0.+\*z/,100!.%\*#^z!#. ¥ ing the throughput of the method, we have found high PECVD SiC film growth rate, from 24 to 36 nm/min, depending on SiH4/CH4 flow rate used (Fraga, 2007) and from 4.0 to 7.0 nm/min for films deposited by RF magnetron sputtering of a SiC target under different Ar/N2 mixtures (Fraga, 2008). In order to evaluate the uniformity, the following tests were performed: (a) before the SiC film deposition, three small steps were created by placing strips on different points of the 2 inch p-type Si wafers, (b) after the deposition, each wafer 3/z10z%\*z "+1.0\$z,0\$z!-1(/\_z cdz 0\$!z 0\$%'\*!//!/z \* z .!/%/0%2%0%!/z+"z 0\$!z,0\$/z3!.!z)!/¥ ured by profilometry and four-points probe, respectively. It was found: thicknesses between 432 and 470 nm and resistivities between 12 and 16 .cm for the pieces of PECVD SiC film, 3\$!.!/z "+.z 0\$!z /,100!.! z "%()/z !03!!\*z 0\$%'\*!//!/z FFJz \* z FKHz \*)z \* z .!/%/0%2%0%!/z !¥ 03!!\*zC^Ez\* zC^EHz
^)^z\$!/!z.!/1(0/z%\* %0!z0\$0z0\$!z0\$.+1#\$,10z+"zz%/z+\*/% !.¥ ably greater than of RF magnetron sputtering. However, both methods present problems in 0!.)/z+"z"%()z1\*%"+.)%05^z 0z%/z\*+0!3+.0\$5z0\$0z+1.z0!/0/z3!.!z,!."+.)! z3%0\$zEz%\*\$z%z3¥ fers whereas the semiconductor industries have been using up to 12 inch wafers. It is likely that the problems of uniformity are more significant in substrates with larger dimensions.

Applications of SiC-Based Thin Films in Electronic and MEMS Devices

http://dx.doi.org/10.5772/50998

321

The residual stress control of SiC thin films is another issue, which can also be added to the challenges. It has been demonstrated that SiC film stress can affect the sensitivity, precision and functionality of thin-film based devices, thus, in some applications, is important to have

SiC film deposition process Young's modulus (GPa) Reference

APCVD 450 Zorman, 1995 LPCVD 396 Fu, 2004 PECVD 88 to 153 El Khakani, 1994 PECVD 56 Flannery, 1998 PECVD 196 Cros, 1997 RF triode sputtering 231 El Khakani, 1994 RF magnetron sputtering 363 Singh, 2012

Co-sputtering 245 to 377 Medeiros, 2012

Initially, the research efforts on semiconductor and dielectric thin films were focused on their electrical properties in order to satisfy the demand in microelectronic devices industry. In general, studies on mechanical properties were limited to internal stress measurements (Tsuchiya, 2008). With the advent of microelectromechanical system (MEMS) technology, the thin films started to be used as mechanical structures, which made fundamental the '\*+3(! #!z+"z0\$!%.z)!\$\*%(z\* z!(!0.+)!\$\*%(z,.+,!.0%!/^z\$!z\$%#\$z+1\*#j/zc+.z!(/¥

Table 2. Young's modulus reported in the literature for SiC films grown on Si substrates.

a low residual stress.

The literature has shown that devices based on thin films, grown on the same conditions, often did not exhibit similar performance (Fraga, 2012). Identify and overcome the causes of the non-reproducibility is another challenge. The synthesis of high-quality SiC films with reproducible properties is fundamental for the advancement of the SiC thin film device 0!\$\*+(+#5^z 0z%/z'\*+3\*z0\$0z0\$!z,.+,!.0%!/z+"z0\$%\*z"%()/z.!z/0.+\*#(5z !,!\* !\*0z+\*z0\$!z !,+¥ /%0%+\*z+\* %0%+\*/\_z%^!^\_z0\$!z.!,!0(!z\* z,.!%/!z+\*0.+(z+"z !,+/%0%+\*z+\* %0%+\*/z%/z%),+.¥ tant to ensure the reproducibility film. The electrical resistivity and thickness measurements of films, deposited under same conditions, are the most used parameters to evaluate their reproducibility.

40!\*/%2!z.!/!.\$z\$/z!!\*z +\*!z+\*z#.+30\$z+"z%z"%()/z0z(+3z+.z\$%#\$z0!),!.01.!z,.+¥ ess aiming to produce high quality films. Three methods have been most frequently used: \_zz\* z/,100!.%\*#^z(!zDz+),.!/z0\$!/!z)!0\$+ /^z\$!z)+/0z.%0%(z%//1!z"¥ ing CVD SiC films for device applications is the high temperature necessary to assure the surface reactions and good deposition rate (Ong, 2006). The low temperature deposition is 2!.5z%),+.0\*0z".+)z0\$!z,+%\*0z+"z2%!3z+"z !2%!z%\*0!#.0%+\*^z\$!z/5\*0\$!/%/z+"z(+3w0!),!.¥ 01.!z !,+/%0! z\$%#\$(5w+\* 10%2!z%z "%()/z\$/z!!\*zz#+(z+"z)\*5z.!/!.\$z#.+1,/z "+¥ cused on the development of TFTs, solar cells and heterojunction bipolar transistors (Cheng, 1997). This has encouraged more studies on the optimization of low-temperature methods as PECVD and the different sputtering processes. PECVD SiC films have been deposited at a .!(0%2!(5z(+3z0!),!.01.!zc(!//z0\$\*zFCC[d^z\$!z"%()/z,.!/!\*0zz#++ z \$!/%+\*\_z\$%#\$z !,+¥ sition rate and good uniformity. Sputtering process presents poor sidewall coverage due to 0\$!z/%#\*%"%\*0z %/0\*!z!03!!\*z0\$!z0.#!0z\* z0\$!z/1/0.0!^z-!/% !/\_z0\$!z !,+/%0%+\*z%/z".!¥ quently made at room temperature and usually has a low deposition rate (Ong, 2006).


+z%((1/0.0!z0\$!z+),.%/+\*z)+\*#zz\* z/,100!.%\*#\_z3!z3%((z %/1//z0\$!z.!/1(0/z+¥ 0%\*! z%\*z+1.z3+.'/z+\*z%z"%()/z#.+3\*z5zz\* zz)#\*!0.+\*z/,100!.%\*#^z!#. ¥ ing the throughput of the method, we have found high PECVD SiC film growth rate, from 24 to 36 nm/min, depending on SiH4/CH4 flow rate used (Fraga, 2007) and from 4.0 to 7.0 nm/min for films deposited by RF magnetron sputtering of a SiC target under different Ar/N2 mixtures (Fraga, 2008). In order to evaluate the uniformity, the following tests were performed: (a) before the SiC film deposition, three small steps were created by placing strips on different points of the 2 inch p-type Si wafers, (b) after the deposition, each wafer 3/z10z%\*z "+1.0\$z,0\$z!-1(/\_z cdz 0\$!z 0\$%'\*!//!/z \* z .!/%/0%2%0%!/z+"z 0\$!z,0\$/z3!.!z)!/¥ ured by profilometry and four-points probe, respectively. It was found: thicknesses between 432 and 470 nm and resistivities between 12 and 16 .cm for the pieces of PECVD SiC film, 3\$!.!/z "+.z 0\$!z /,100!.! z "%()/z !03!!\*z 0\$%'\*!//!/z FFJz \* z FKHz \*)z \* z .!/%/0%2%0%!/z !¥ 03!!\*zC^Ez\* zC^EHz
^)^z\$!/!z.!/1(0/z%\* %0!z0\$0z0\$!z0\$.+1#\$,10z+"zz%/z+\*/% !.¥ ably greater than of RF magnetron sputtering. However, both methods present problems in 0!.)/z+"z"%()z1\*%"+.)%05^z 0z%/z\*+0!3+.0\$5z0\$0z+1.z0!/0/z3!.!z,!."+.)! z3%0\$zEz%\*\$z%z3¥ fers whereas the semiconductor industries have been using up to 12 inch wafers. It is likely that the problems of uniformity are more significant in substrates with larger dimensions.

show that devices based on SiC bulk substrates exhibit better performance than those thin "%()z/! zc..%/\_zDLLHd^z\$!\*\_z0\$!z"%./0z\$((!\*#!z%/z0+z#.+3\*z%z0\$%\*z"%()/z3%0\$z,.+,!.¥ ties as good as the bulk substrate. In addition, it is necessary to achieve high film growth rate as well as thickness uniformity and homogeneity when deposited on large-area Si wafers, which are important factors to reduce costs. For this, the influence of SiC growth process parameters, such as gases flow rate, substrate temperature, pressure and doping, have been

The literature has shown that devices based on thin films, grown on the same conditions, often did not exhibit similar performance (Fraga, 2012). Identify and overcome the causes of the non-reproducibility is another challenge. The synthesis of high-quality SiC films with reproducible properties is fundamental for the advancement of the SiC thin film device 0!\$\*+(+#5^z 0z%/z'\*+3\*z0\$0z0\$!z,.+,!.0%!/z+"z0\$%\*z"%()/z.!z/0.+\*#(5z !,!\* !\*0z+\*z0\$!z !,+¥ /%0%+\*z+\* %0%+\*/\_z%^!^\_z0\$!z.!,!0(!z\* z,.!%/!z+\*0.+(z+"z !,+/%0%+\*z+\* %0%+\*/z%/z%),+.¥ tant to ensure the reproducibility film. The electrical resistivity and thickness measurements of films, deposited under same conditions, are the most used parameters to evaluate their

40!\*/%2!z.!/!.\$z\$/z!!\*z +\*!z+\*z#.+30\$z+"z%z"%()/z0z(+3z+.z\$%#\$z0!),!.01.!z,.+¥ ess aiming to produce high quality films. Three methods have been most frequently used: \_zz\* z/,100!.%\*#^z(!zDz+),.!/z0\$!/!z)!0\$+ /^z\$!z)+/0z.%0%(z%//1!z"¥ ing CVD SiC films for device applications is the high temperature necessary to assure the surface reactions and good deposition rate (Ong, 2006). The low temperature deposition is 2!.5z%),+.0\*0z".+)z0\$!z,+%\*0z+"z2%!3z+"z !2%!z%\*0!#.0%+\*^z\$!z/5\*0\$!/%/z+"z(+3w0!),!.¥ 01.!z !,+/%0! z\$%#\$(5w+\* 10%2!z%z "%()/z\$/z!!\*zz#+(z+"z)\*5z.!/!.\$z#.+1,/z "+¥ cused on the development of TFTs, solar cells and heterojunction bipolar transistors (Cheng, 1997). This has encouraged more studies on the optimization of low-temperature methods as PECVD and the different sputtering processes. PECVD SiC films have been deposited at a .!(0%2!(5z(+3z0!),!.01.!zc(!//z0\$\*zFCC[d^z\$!z"%()/z,.!/!\*0zz#++ z \$!/%+\*\_z\$%#\$z !,+¥ sition rate and good uniformity. Sputtering process presents poor sidewall coverage due to

quently made at room temperature and usually has a low deposition rate (Ong, 2006).

Cost Fair Fair Fair Uniformity Fair Fair Fair Substrate versatility Good Very good Very good Stress control Poor Very poor Good Throughput Varies Very good Fair

CVD PECVD Sputtering

!/% !/\_z0\$!z !,+/%0%+\*z%/z".!¥

0\$!z/%#\*%"%\*0z %/0\*!z!03!!\*z0\$!z0.#!0z\* z0\$!z/1/0.0!^z-

Table 1. Comparison among the main methods used to grown SiC films.

evaluated and optimized.

320 Physics and Technology of Silicon Carbide Devices

reproducibility.

The residual stress control of SiC thin films is another issue, which can also be added to the challenges. It has been demonstrated that SiC film stress can affect the sensitivity, precision and functionality of thin-film based devices, thus, in some applications, is important to have a low residual stress.



Initially, the research efforts on semiconductor and dielectric thin films were focused on their electrical properties in order to satisfy the demand in microelectronic devices industry. In general, studies on mechanical properties were limited to internal stress measurements (Tsuchiya, 2008). With the advent of microelectromechanical system (MEMS) technology, the thin films started to be used as mechanical structures, which made fundamental the '\*+3(! #!z+"z0\$!%.z)!\$\*%(z\* z!(!0.+)!\$\*%(z,.+,!.0%!/^z\$!z\$%#\$z+1\*#j/zc+.z!(/¥ tic) modulus is the key mechanical property to use them as structural layer in MEMS devices. As can be observed in Table 2, crystalline and polycrystalline SiC films grown by atmospheric pressure chemical vapor deposition (APCVD) and LPCVD processes exhibit higher Young's modulus than the amorphous produced at low temperatures by PECVD and sputtering. In addition, it has been observed that the most of the SiC films still exhibit lower Young's modulus than the reported for SiC wafers which is in the range between 330 and 700 GPa depending on the polytype (Zorman and Parro, 2010).

One issue that should be considered is that, although the researches on SiC film growth have been mainly tocused on the deposition of SiC on Si substrates, MEMS applications frequently require the growth of SiC thin films on sacrificial and insulating layers, as for example SiQ2 or SigNy, grown on Si substrates. Thus, to evaluate the influence of substrate type on the properties of SiC films is another important point. Some works have investigated the growth and properties of SiC films on SiO2, SigN4 and poly-Si (Fleischman, 1998; Wu, 1999; Chen, 2000). A known drawback of SiC on insulator layers is the high stress caused by the large lattice and thermal mismatch between SiC and insulator, which become post-deposition annealing necessary to minimize this problem and improve the quality SiC film on insulator. Other particularities of SiC growth on insulator/Si substrates include: the effect of insulator layer thickness on the properties of SiC film and the choice of the suitable insulator for each device application. Chen et al. reported the SiC growth by CVD on the following substrates: thermally oxidized Si substrates with SiO2 thicknesses of 30, 50, 70, and 100 nm, Si substrates with native oxide of 2 nm and 3 µm thick phosphosilicate glass (PSG). They observed that in the thickness range between 30 and 70 nm, the SiO2 serves as a compliant layer which reduces the strain between SiC film and the substrate besides allows the growth of a more oriented SiC film. In relation to choice of the insulator, Si.N. has shown more suitable than SiO2 due to: (i) its higher dielectric constant that can reduce the leakage currents, and (ii) its thermal expansion coetficient is much closer to that of SiC than one of SiO2 thus the stress in SiC film grown on Si.N., will be lower than that on SiO2 (Cheng, 2003). Nevertheless, most of the SiC thin-film MEMS devices reported in the literature use the SiC2 as sacrificial layer and/or substrate electrical isolation (Chang, 2008; Mishra, 2009). One reason for this is the easy formation of SiO2 achieved by thermal oxidation of Si substrates.

In our researches, we have explored the properties of SiC thin films grown on SiO/Si substrates by PECVD and RF magnetron sputtering for the development of strain gauges (Fraga, 2010) and pressure sensors (Fraga, 2011a; Fraga 2011c). Young's modulus was found to be 65 GPa for PECVD a-SiC film and 57 GPa for nitrogen-doped PECVD a-SiC film. These values are near to 56 GPa that was reported for PECVD a-SiC film grown on Si substrates (Flannery, 1998). On the other hand, the Young's modulus values found by us for RF-sputtered SiC and SiCN films on SiO%Si were considerably lower than those reported in the literature for films on Si substrates. We found Young's modulus of 40 GPa and 88 GPa for sputtered a-SiC and SiCN films respectively, whereas other authors have found values above 200 GPa for sputtered SiC film grown on Si (see Table 2) and 117 GPa for SiCN film grown on Si (Sundaram, 2004). For CVD processes, the following Young's modulus was found: 426 ± 100 GPa for SiC grown on SiO%Si by APCVD (Fleischman, 1998) and 426 GPa for SiC grown on Si3N4/Si by LPCVD (Cheng, 2002).

In general, the literature has shown that the substrate type influences the structure, morphology, electrical and mechanical properties of SiC films. However, although there are differences related to the substrate type, the properties, such as high Young's modulus, high chemical resistance and high thermal stability among others, that make the SiC film attractive for a variety of harsh environment MEMS device applications are maintained.
