Arthur Sherman - Chemical Vapor Deposition for Microelectronics (779637), страница 26
Текст из файла (страница 26)
The substrate temperature was kept relatively low, 230°C, the excitation frequency was 13.56 MHz, and pressures were 500 to 700 mTorr.Deposition rates were found to be on the order of 500 to 600 A/min,much faster than for pure tungsten. Apparently, adding silicon to the glowdischarge enhances deposition rates of tungsten silicide, even for small percentage additions (about 1%). The ratio of W to Si could be adjusted by controlling the amount of SiH 4 added to the WF6 , and the film resistivity dependedon this ratio, as shown in Figure 18.142Chemical Vapor Deposition for MicroelectronicsCONTENTW0f(ncm))(tJ.1 )( 10-30.990.450.15x'-,.-----1)(1It----_)(_)(_1 X 10-56008001000ANNEALING TEMP.
(OC)Figure 18: Resistivity versus annealing temperature for different compositionWSi x films annealed in N 2 for 60 minutes. I8For films close to stoichiometric, WSi 2 , the resistivity even after annealingis '""100 JlSl-cm. Since these films were amorphous, both before and after annealing, the reduction in resistivity must have been due to outdiffusion of hydrogenand fluorine. The difference, then, between the observed resistivity (""-'100J,lr2-cm) and the expected value ('""50 /lSl-cm) may be due to lack of propercrystallization.5.6.2 MolybdenumAttempts to deposit PECVD molybdenum have not been successfu I inproducing pure low-resistivity films when MoF6 + H 2 are used as the reactants.
16Plasma-Enhanced CVD143When the H2 /MoF6 ratio was below 5: 1, the films were porous. Ratios above7:1 yielded stable films. Analysis indicated a fluorine content in these films ashigh as 15%. Why such a high concentration of fluorine is found here whentungsten films deposited under similar conditions have less than 1%, is notclear.
The measured resistivity of the PECVD molybdenum film (3000 A)was >10,000 pQ-cm.A more successful approach to such PECVD of molybdenum films wasreported 19 using MoCl s and H2 • In this case, a sublimator operating at 90°to 100°C was needed to vaporize the solid MoCl s . Again, a parallel-plate, coldwall plasma reactor was used for the depositions. Temperatures ranged from170° to 430°C, and the pressure was 1 Torr. The as-deposited films turned outto be amorphous with small amounts of chlorine incorporated. Annealing athigh temperatures reduced the resistivity by more than one order of magnitude,as shown in Figure 19, although it is still too high. Also, the resistivity of theas-deposited films decreased as substrate temperature increased.
The improvedresistivity after heat treatment probably reflects a crystallization of the originally amorphous material, as well as the diffusion of chlorine.Euc:~ 10- 4>I-enenwa: 10- so5001,000ANNEALING TEMPERATURE (GelFigure 19: Resistivity of PECVD molybdenum versus annealing temperature. 19Molybdenum disilicide was deposited in the same experimental apparatus 19by adding SiH 4 to the reactant gas mixture. It was possible to vary the MoSistoichiometry by adjusting the flow rates of Si H4 and H2 • At the same time:the film resistivity also varied, as shown in Figure 20.
Unfortunately, the bestresistivity achieved was '""JaDO pQ-cm when a typical thin film value for MoSi 2would be '""J100 pQ-cm. The high resistivity may have, again, been due to chlorine trapped in the film. However, it is not clear why the silicide film would differ so radically from the molybdenum film just discussed.144Chemical Vapor Deposition for MicroelectronicsE 10- 2c:U-~>~C/)la10 em'lminenwa::DCEPO. TEMP. 400 C10- 4RF POWER 100 Wo10203040H2 CARRIER (cm~min)Figure 20: Resistivity of MoSi x as a function of SiH 4 and H 2 flow rates.195.6.3 TantalumUsing plasma enhancement, there is an alternative to the thermal CVDthin film TaSi 2 reviewed in the previous chapter. 2o Experiments were carriedout in a vertical cold-wall tubular reactor with a single wafer placed normalto the flow. Plasma excitation was provided by a coil wrapped around thequartz tube and positioned above the wafer. The reactive gases flowed from topto bottom.
A gra ph ite res istive heater perm itted su bstrate tem peratu res of850°C. Excitation frequencies of 600 kHz or 3.5 M Hz were used to createthe plasma. The source for Ta was TaCl s, and for Si was SiH 2 CI 2 . As before,the TaCl s vapor was produced in a sublimator using H 2 as the carrier gas. Thechoice of SiH 2 CI 2 rather than SiH 4 was made to minimize gas phase (homogeneous) nucleation. Deposition pressures were in the 1 to 2 Torr range.These authors found that it was possible to deposit amorphous films whoseTa concentration ranged from 10 to 80 mol % by changing the reactive gasmix.
Another feature of the films was that under certain conditions they contained substantial quantities of chlorine and hydrogen. Also, they did notadhere to either silicon or silicon dioxide after annealing (argon atmospherefor 1 hour at 900°C). When the substrates were dry etched in an Hel plasmafor 2 minutes, they adhered to the substrate even after annealing. Since thisetch removed about 50 A, it appears that the native oxide on the silicon and/orsome other surface impurities on both the silicon and silicon dioxide werecausing the lack of adhesion.Plasma-Enhanced CVD145It was found that the stoichiometry could be adjusted simply by changingthe SiCI 2 H2 flow rate, as shown in Figure 21. For these curves, the substratetemperature was 580°C, the H2 flow was 400 seem, 2the partial pressure ofTaCl s was 30 mTorr, and the RF power was 10 watts/em at 3.5 MHz.30nmrnlnSi•20<1J-+JItSs..s:::0 10.,.+J.,.V')0c...•Q)c00515...seem20Figure 21: Silicon and tantalum deposition rates versus SiCI 2 H2 flow rates.2°For the initially amorphous films, annealing converted the film to a polycrystalline one and reduced the resistivity from 180 pn-cm (as deposited) to60 J-Ln-cm.
By increasing the substrate temperature to 650°C, the initial filmswere polycrystalline. By reducing the TaCl s flow (PTaCl s ""'20 mTorr), increasing the H2 flow to 800 seem, and lowering the RF frequency to 600 kHz at2 watts/em 2, the authors were able to deposit films with initial resistivitiesof 70 pn-cm. Annealing reduced this value to 50 pr2-cm. The interface between the TaSi 2 and an underlying layer of polysilicon was smooth, as shownin Figure 22. Also for these conditions no chlorine and only a trace of carbonwas detected.146Chemical Vapor Deposition for MicroelectronicsFigure 22: TEM cross section through TaSi polycide structure.
20As shown in the previous chapter, similar high-quality TaSi 2 polycrystalline films can be prepared by thermal CVD at the same or lower depositiontemperature.5.6.4. TitaniumRecent studies have described the PECVD of titanium disilicide, TiSi 2 ,in a hot-wall tubular reactor. 21 ,22 This silicide is of particular interest since ithas the lowest resistivity of those we have been considering-14 J,LQ-cm.
Although no studies of thermal CVD of TiSi 2 films have been reported in theliterature, it is c1aimed 21 that for the SiH 4 /TiCI 4 reaction, deposition temperatures above 600°C are required, and that this leads to films with larger thandesirable grain size. Plasma excitation obviously can lower the deposition temperature, and amorphous TiSi 2 films can be deposited. Post deposition anneal can give the desired crystal structure.Depositions were done at 50 kHz with SiH 4 • TiCI 4 and Ar as a diluent.At power levels of about 100 watts, the deposition rates were 60 to 80 A/min.Anneals of the as-deposited films were done at 600° to 650°C.
Comparisonof the time behavior of resistivity after annealing between TiSi 2 and WSi 2films is shown in Figure 23. Significant is the lower temperature anneal possible for the TiSi 2 film and its substantially better resistivity (2:1). As longas the anneal time at 650°C was greater than 5 minutes, it was found that TiCI 4flow (50 to 90 sccm) and deposition temperature (380° to 460°C) had verylittle effect on the final sheet resistance.It was found that after annealing the film composition was always stoichiometric, regardless of as-deposited composition, as long as sufficient siliconwas available. Some chlorine was detected in the as-deposited films (aboutPlasma-Enhanced CVD147ANNEAL (OC)o}A90010001100:~~g}oW Six(BRORS, et.
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
