~EMOTE MICROWAVE-ENHANCED
CHEMICAL VAPOR DEPOSITION
OF SILICON-NITROGEN (SiXNy ) THIN FILMo/
A Thesis Presented to
The Faculty of the College of Engineering and Technology
Ohio University
In Partial Fulfillment
of the Requirements for the Degree
Master of Science
by
Gary M. Gladysz/
August 1991 ACKNOWLEDGEMENTS
I often wondered whether or not I would finish my thesis, but now that I have (I hope), there are a few people I would like to thank for their part in my actually finishing. Besides myself, Ed and Kathy Gladysz are probably the happiest that I am finishing my thesis. Because, without their guidance, their energy, their love, their financial assistance, and their food, I never would have come this far. I'm sure, after us three, Daniel Gulino and the entire Department of Chemical Engineering are the next happiest to see me finish. without Dan's help, encouragement and unending wisdom and knowledge, I would still be here next year (not uncommon for Chemical
Engineers at O.U.). Of course, the engineering technicians, Jim McKnight and ilarvey Strausbaugh who turned my foggy, vague, and unclear ideas into apparatus which turned out data that makes me appear somewhat intelligent, should, and will not go unappreciated. I also need to thank Madan Gopal, Sudarshan Neogi, and John Wodarcyk for their input, their intelligence, and most of all, their willingness to come to all my parties. Lastly, I would like to thank NASA for partial support of this project through grant NCC 3-194 (and hopefully for hiring me in the near future). i
TABLB OP COB'J.'DlTS
List of Figures ...... iii List of Tables ...... vi
I. Introduction...... 1
II. Literature Review . 4
2.1 Low Earth Orbit . 4
2.2 Materials...... 4 2.3 Remote Chemical Vapor Deposition of Silicon
Nitride ...... 7
III. Experimental Procedure . .. . 11
3 . 1 Vacuum System ...... 11
3 . 2 Instrumentation. . .. . 11 3.3 Deposition System...... 12 3.4 Gas Flow System. .... • .•... 13 3.5 Reactants and Substrates 16 3.6 Procedure . 16
3.6.1 Preparation •.. 16
3.6.2 Achieving Vacuum. 18
3.6.3 Deposition ...... •. 19
3.7 Experimental Design ..•.. 19
IV. Analysis...... 21 4.1 Film Thickness (Deposition Rate) 21 4.2 IR Analysis ...... 23 ii
V. Discussion . · . . · · ··· · · · ·· 35 5.1 Profilometer · ··· · ·· · ·· 35 5.1.1 Film Thickness Versus Temperature 35
5.1.2 Film Thickness Versus Pure Nitrogen stream Flowrate ·· · · · ···· 37 5.1.3 Film Thickness Versus Nitrogen Flowrate Through the Reactant (TMS) · . ···· 37 5.1.4 Downstream Effects ··· · ···· 38 5.2 IR Discussion · · ····· · · · 39 VI. Conclusions · .. · ·· · ·· · · ·· 43 VII. Recommendations ···· ·· ··· ·· 45 VIII. References · . · ·· ··· . · · 47 iii
LIST OF FIGURES
Figure 3.1: Schematic of RMECVD Apparatus · ·· · ·· 14 Figure 3.2: Schematic of the Quartz Reactor ·· · 15 Figure 3 .3: Dimensions of the Quartz Reactor · 18 Figure 4.1: Sample of the Profilometer Data ··· 22 Figure 4.2: The IR Spectra of the thin film produced in
run #1. The substrate was 18 em downstream from the plasma field · · · · ·· · · 25 Figure 4 .3: The IR Spe~tra of the thin film produced in
·run #1. The substrate was 22 em downstream from the plasma field ·· · ·· · · · 26 Figure 4.4: The IR Spectra of the thin film produced in
run #2. The substrate was 18 em downstream from the plasma field · · · ·· · ·· 27 Figure 4.5: The IR Spectra of the thin film produced in
run #3. The substrate was 18 em downstream from the plasma field ·· · ··· · · 28 Figure 4.6: The IR Spectra of the thin film produced in
run #3. The substrate was 22 em downstream from the plasma field · ·· · ·· · · 29 Figure 4.7: The IR Spectra of the thin film produced in
run #5. The substrate was 18 cm downstream from the plasma field ·· · · · · · · 30 Figure 4.8: The IR Spectra of the thin film produced in iv
run #5. The substrate was 22 cm downstream
from the plasma field ...... 31
Figure 4.9 The IR Spectra of the thin film produced in
run #6. The substrate was 18 cm downstream
from the plasma field ...... 32
Figure 4.10: The IR Spectra of the thin film produced in
run #6. The substrate was 22 em downstream
from the plasma field ...... 33
Figure 4.11: The IR Spectra of the thin film produced in
run #7. The substrate was 18 em downstream
from the plasma field ....•... .. 34
Figure 5.1: Film thickness versus temperature for runs
#1-4. The substrates were 18 em downstream
from the plasma field ...... 35
Figure 5.2: Film thickness versus temperature for films
deposited 22 em downstream from the plasma
field .....•.•.•....•. .. 36
Figure 5.3: Film thickness versus pure nitrogen stream
flowrate 18 cm downstream from the plasma
field. The temperature was 300°C and the
nitrogen flowrate through TMS was 106
secm ...... •..... 37
Figure 5.4: Film thickness versus the flowrate of nitrogen
through TMS 18 cm downstream from the plasma
field. The temperature was 300°C and the
flowrate of the pure nitrogen stream was 106 v
secm ...... 38
Figure 5.5 The IR spectra of the thin film deposited in
Run #1. The substrate was 18 em downstream
from the plasma field •...... •. 41
Figure 5.6 The IR spectra of a thin film deposited by
MECVD using TMS at 25°C • . . . . . • . 42 vi
LIST OF TABLES
Table 2.1: Reaction Efficiency of Metals in LEO 5
Table 2.2: Reaction Efficiency of Polymers and
Composites in LEO . 6
Table 2.3: Weight 'Loss of Silicon Based Coatings on Epoxy Resin Substrates deposited at 25°C · ·· · 8 Table 2.4: Physical Properties of TMS ··· · 10 Table 3.1: Deposition Parameters . .. . . ··· · 20 Table 4.1: Profilometry Information . ... · · · 22 Table 4.2: Infrared Peak locations of Silicon
Compounds em"? 23 1
:I • :II1'1'RODUCT:IOII
The purpose of this research is to protectively coat materials used in space applications. The space systems operating in low Earth orbit (LEO), which extends 200 to 700 kilometers in altitude, are exposed primarily to atomic oxygen (AO). The occurrence of AO is due to the dissociation of diatomic oxygen (02) by UV radiation. Other constituents of LEO include UV radiation, electrons, ions, and micrometeoroids. Composites are now, and will be in the future, a major component in space systems. They were used in the Galileo Spacecraft, are being used in the Space Shuttle Program, and will be extensively used in Space station Freedom. The most likely to be used are an organic matrix/graphite fiber composite because of their high strength, stiffness, low mass, and low coefficient of thermal expansion.
One drawback of these composite is that they are rapidly attacked by AD. For Space station Freedom to be cost effective, it must have an operational life of at least 30 years. Unprotected epoxy/graphite trusses would last only a fraction of that time.
At the elevated velocities through which these space systems move, AO can chemically react with many materials which are relatively unreactive in the surface environment. 2
When AO reacts with a material, the material experiences degradation in the form of surface roughening and mass (weight) loss, which impairs its mechanical, electrical, and optical properties. In order to extend the life of space systems in the LEO's harsh environment, a protective barrier coating must be applied to the components. The barrier must be strong and flexible, and have a strong interfacial bond between it and the substrate. It must be abrasion resistant, which comes from a strong interfacial bond within the barrier, and be stable in a UV environment. Most importantly, it must protect the material from AO attack.
) A silicon nitride (Si3N4 thin film is one protective barrier that meets the above criteria. Presently, Si3N4 is being used widely in the microelectronics field as a passivation layer protecting silicon integrated circuits from oxidation and moisture damage. Because of its protective properties, silicon nitride is now finding uses in space applications. It provides excellent protection from AO attack.
In order to apply a protective coating to a material, a suitable deposition technique must be found. Chemical vapor deposition (CVO) is widely used in the engineering, electronics, and optics fields for applying thin films. To ensure good adhesion to a substrate, the partial pressure of water vapor must be minimized within the reactor vessel; thus 3 very low pressures are needed. Once low pressure is achieved, the reactants are allowed to enter the vacuum chamber. Once in the vacuum chamber, activation energy is needed for the reactants to deposit. Normally, this activation energy is provided by heat.
Microwave-enhanced chemical vapor deposition (MECVD) is a way of incorporating microwaves to provide the activation energy needed to deposit thin films. In MECVD, the substrate is located in the plasma field produced by the microwaves dissociating the reactant gases. It has been shown to produce silicon nitride (SiNx) films at substrate temperatures as low as 25°C. One drawback of MECVD is that the substrate is in contact with the plasma and is thus continuously bombarded by ions, electrons, and photons produced by the microwave radiation.
Downstream or remote MECVD is a way of utilizing microwave radiation without SUbjecting the substrate to the damaging plasma field. The substrate is located at some distance from the plasma field and the excited gas is then directed to the substrate. A higher substrate temperature will be needed in remote MECVD since the gas is only excited and not dissociated, as in the plasma field. Remote MECVD has been shown to produce SiNx films at temperatures as low as 200°C with the appropriate chemical precursors. 4
II. LITBRATURE REVIEW
2.1 Low Earth Orbit Environment
The atomic oxygen (AO) found in LEO (200-700 km in altitude) 1 is produced when UV radiation dissociates 2 ° - Although the density of AO is low (on the order of 109
3 19 atoms/cm compared to 10 molecules °2/c m3 at sea level and room temperature) and has relatively low kinetic energy, spacecraft at this.orbit travel at 8 kIn/sec, which gives AO sufficient energy to cause reactions with the surface of the spacecraft_ The relative translational energy of this system is 5 eV. 1, 2
The harmful effect of the LEO atmosphere was not realized until the space shuttle missions began. These flights gave evidence which showed surface degradation after only tens of hours in LEO _ The degradation was attributed to AO.
However, protection from AO is not the only barrier that must be overcome in LEO. For example, severe UV and charged particle radiation, intense thermal cycling, and micrometeoroids are also problems which must be tackled. All of these must be taken into consideration when designing structures which will be in LEO for long durations. 3
2.2 Materials
Several types of materials have been tested to see how 5 well they could hold up in LEO. Metals, polymers, composites, and protective coatings were all tested on Space Shuttle flights and/or on the Long-Duration Exposure Facility (LDEF).
LDEF was placed in LEO by a space shuttle flight in April
1984 and was expected to be retrieved 11 to 18 months later.
It carried materials to be tested in the LEO environment.
However, because of the Space Shuttle Challenger accident in
1986, LDEF was not retrieved until 1990.4
During Space Shuttle flights, over 20 metals have been tested for their tolerance to AO. The results show that all except silver, copper, and osmium were unaffected by the LEO environment. silver formed a non-conductive heavy oxide layer which eventually led to mass loss due to flaking or scaling. Both copper and osmium formed oxides which have relatively high vapor pressures. Below is a table listing the reaction efficiency of the three metals.
Table 2.1 Reaction Efficiency of Metals in LEO 2
Material Reaction Efficiency(cm3/atom)
Copper (various forms) 0.9-1.7 x 10-24 Silver (various forms) Heavily Attacked Osmium 0.026 6
Organic materials are generally more reactive than even the most reactive metal. Space Shuttle flights showed that perfluorinated polymers performed better than organic polymers. Table 2.2 lists some of the polymers tested.
Table 2.2 Reaction Efficiency of Polymers and Composites 2,4,5,6
Material Reaction efficiency cm3/atom
Kapton 3 x 10-2 4 Mylar 3.4 Tedlar 3.2 Polyethylene 3.7 Polyslfone 2.4 Epoxy 1.7 Polystyrene 1.7 Polybenzimidazole 1.5 Polyester Heavily Attacked Polyester with Antioxidents Heavily Attacked Perfluorinated Polymers: Teflon, TFE < 0.05 Teflon, FEP < 0.05 25% Polysiloxane/45% Polyamide 0.3 Graphite/Epoxy: 1034C 2.1 5208jT300 2.6 Silicones: RTV-560 0.2 DC-1104 0.2 T-650 0.2 DC1-2577 0.2
Composites have many uses in today' s and tomorrow's space applications. They have many applications in the Space 7
Shuttles and also the Galileo Spacecraft. 7 A graphite/epoxy composite will be used as structural elements for Space station Freedom and solar arrays. Metal matrix composites returned from the LDEF showed magnesium MMC's were attacked by AO while aluminum MMC's were not. Table 2.2 lists some of the organic composites and their reaction efficiency in LEO.
2.3 Remote Microwave-Enhanced Chemical Vapor Deposition
(RMECVD> Silicon Nitride Thin Films
silicon nitride, as well as other silicon based thin films, are being examined for their application as protective coatings for space systems, as well as electrical and optical applications. Table 2.3 shows the weight loss of an epoxy resin substrate with various silicon-based protective coatings exposed to atomic oxygen. All of these thin films reduce the mass (weight) loss sUbstantially.
Silicon nitride satisfies the requirements that a good thin film should exhibit for space applications. Presently, however, the largest use for silicon nitride thin films is silicon integrated circuits. It provides a passivation layer which prevents oxygen, sodium, and moisture damage.
There are many different ways of depositing silicon nitride thin films. Chemical vapor deposition (CVO) is one technique, but one which 8 requires temperatures between 750-1500oC. 1, 8 Many of the materials which need to be protected (composites and polymers, for example) would be destroyed at such temperatures.
Table 2.3 Weight Loss of Silicon based Coatings on Epoxy resin substrate Deposited at 250C 9.
Sample Fsilane Gas/Flow Thickness Weight loss description (seem) (seem) (nm) (mg. cm- 2 (coating hr-1 ) type)
Control 5.06 (Untreated) ------a-Si:H 10 Ar/45 50 0.04
P-SiN 25 NH3/25 50 0.03
P-SiON 25 NH3/25: N2O/2 50 0.18
P-Si02 25 N2O/25 50 0.09 Plasma generated ~n a M~erowave chamber: Pressure:100mTorr 02flow:200 seem Microwave power:200W
Currently, microwave-enhanced chemical vapor deposition
(MECVD) is widely used to deposit silicon nitride thin films. silicon Nitride has been deposited at temperatures as low as
2SoC using this technique. 10 One drawback of MECVD is that the substrate is located directly in the plasma field. While in contact with the plasma, the substrate is constantly bombarded with ions, electrons, and photons. 11, 12
One of the newest techniques for depositing thin films is downstream or remote MECVD (RMECVD). This utilizes microwaves to provide excitation of one or more of the reactants. However, the deposition takes place away from the plasma field.
Because the reactants are only excited, not dissociated 9 outside of the plasma field, additional activation energy must be supplied to deposit a film. As mentioned earlier, MECVD requires no external heat. In RMECVD, silicon nitride films were deposited at temperatures as low as 200°C, well below the melting point of many polymers of space interest.
Dzioba, Meikle, and streater were able to deposit silicon nitride films on Si, InP, and InGaAs. 11 The substrates were located 60 em downstream from their 2.45 GHz plasma field.
They were able to deposit 5 nm thick films using silane (SiH4 ) and nitrogen at temperatures between 250-400oC.
Tsu, Livsky, and Mantini deposited silicon nitride films using silane or disilane combined with either nitrogen or ammonia. 12 Their silicon substrate was located 10 em downstream from the plasma field. Films were deposited at temperatures as low as 200°C.
The biggest advantage of RMECVD over MECVD is that in the former, the substrate is located away from the plasma in the microwave chamber. In the glowing region, electrons, ions, and photons can interact with the surface and cause defects in the substrate, which might weaken it. These particles could also sputter away the surface of the substrate or film.
Both of the above examples can cause absorption sites and make it easier for AO to react with the substrate. It is also found that RMECVD SUbstantially reduces the amount of hydrogen bonding. Thin films produced by MECVD do not perform as well as gate insulators in metal-insulator-semiconductors (MIS) or 10 metal-oxide-semiconductors (MOS) device structures. The reason for this is because of the hydrogen bonds which are produced in larger quantities when using this technique. Lastly, the radiation in the plasma zone leads to high interface state density in MISFET's when the film is deposited. Hence, not only is RMECVD applicable to protecting substrates in LEO from AO, it also will find uses in microelectronics technology.
2.4 Tetramethylsilane (TMS)
Tetramethylsilane is commonly used in MECVD to produce silicon nitride films and is replacing silane because silane is very toxic and flammable. It structure is one silicon atom with 4 methyl groups around it ( Si(CH3) 4 ) . Table 2.4 lists some physical properties of TMS.
Table 2. 4P hlYSl.CaI Propert'l.es 0 f TMS Formula Boiling Melting Refractive Density Weight Point °c Point °c Index (g/cc) (20°C) I 88.23 I 26-28 I -99 I 1.3580 I 0.648 I 11 III. EXPERIMENTAL PROCEDURE
3.1 VACUUM SYSTEM
Figure 3.1 shows a diagram of the apparatus used in this experiment. The following is a listing of the components of the vacuum system.
Mechanical Fore Pump: Used initially to reduce the pressure in the system to 10-2 Torr. It also is used to back the diffusion pump, which will be discussed later.
Diffusion Pump: Manufactured by Bendix Inc., it has a normal pumping speed of 8.2 Torr-liters/sec. at 7 x 10-3
Torr. It maintains steady state operation below 10-3 Torr with a backing pump. The diffusion pump has a plateau speed of 4300 liters/second.
stainless steel Bell Jar: It has an approximate volume of .12m3 (4ft3 ). It is equipped with six 2.75 inch
Conflat flange feedthrough ports.
Liquid Nitrogen Baffle Trap: It is located between the diffusion pump and gate valve and used to condense vaporized pump oil from the diffusion pump.
3.2 INSTRUMENTATION
Pressure Sensors: There are two types of pressure sensors. The first is an ionization gauge. It is 12 functional at pressures below 10-5 Torr. The second type are thermocouple pressure ports. There are two of these: one located in the fore pump line and the other on the bell jar.
Rotameters: There are two ALLBORG rotameters with a range from 106 to 1999 seem.
Pressure gauge: A Premier pressure gauge in the nitrogen line just after the exit from the tank. It can measure pressures from 0 to 32 psig.
Pressure readout: A digital readout is used in conjunction with the ionization pressure device. It has a working range of below 10-4 torr. There are two analog pressure readouts used with the thermocouple ports, good for pressures between 10-3 to 1.0 torr.
Temperature readout: Monitors the temperature of the thermocouple connected to the substrate heater.
3.3 DEPOSITION SYSTEM
Quartz Reaction Chamber: A quartz tube closed at one end with the open end fitting to a Conflat flange port on the belljar. It has dimensions of 1.5 inches diameter by
10 inches length.
Quartz feed line: There are two of these. They are
1/8 inch diameter and supply the reactor with the gases needed for the reaction. 13
Substrate Heater: It is made of Chromalox and has dimensions of 7 inches by 1/2 inch by 1/4 inch. It has a maximum temperature of 700°C.
Microwave Source: A Sharp Model RA52 microwave oven was used. It emits microwaves at a frequency of 2.45 GHz and has a power of 500 watts.
3.4 GAS FLOW SYSTEM
The gas flow system can be seen in Figure 3.1.
Nitrogen coming from the high pressure tank is split into two streams. The first stream, the pure nitrogen stream, comes to a valve followed by a rotameter. Just before entering the belljar, there is a Swagelok check valve. The stream continues through the stainless steel feedthrough and into a Tygon tube. This is connected to a quartz tube, which is placed in the reactor. The second stream (the reactant stream) follows a similar path. First encountering a valve and then a rotameter, it bubbles through the reactant TMS. The saturated nitrogen next passes through a separate Swagelok check valve while entering the belljar. Once in the belljar, the reactant stream follows a Tygon tube into a quartz tube, which is already positioned in the reactor.
The system is made up entirely of Tygon tUbing, with a few exceptions. These exceptions are as follows: The VACUUM CHAMSER
f..~J4w? __ RE:ACTANT NITROGEN -' ------
VENT TO HOOD t
Figure 3.1 Schematic of RMECVD Appartus
~ ~ 15 quartz tubes, which enter the reactor, stainless steel tUbing, which make up the feedthroughs and the tUbing which enters and leaves the flask. stainless steel tubing was used near the reactant because of the possibility of reaction with the Tygon tUbing.
MICR~VE 8eLL..3~~ C~"1BER L,.;U'\I-L~I FLA"GE FEEDTHR0UC3H PORT ALUt...rt'llUt-t C::JLLJlCtfl!
l"LI~t-< I ~ t- Flgure 3.2 Schematic of the Quartz Reactor 16 3.5 Reactants and Substrates Nitrogen was used both as a carrier gas and reactant. As a carrier gas, it was bubbled through TMS (tetramethylsilane) and the saturated nitrogen entered the reactor. TMS was chosen because of its low boiling point (26-28°C). The TMS was 99.9+ % pure. The substrate on which the thin films were deposited were <111> silicon substrates manufactured by Monsanto. They had a minimum and maximum thicknesses of .014 and .015 inches respectively. 3.6 PROCEDURE 3.6.1 Preparation The deposition of SiNx by RMECVD in this study was done on <111> silicon wafers. Before deposition, the silicon substrate was cleaned with acetone and then fastened to the substrate heater by means of a screw clamp. One substrate was placed on each side of the clamp. Once the substrates were secure, the equipment needed was be loaded into the quartz reactor. Needed in the reactor was the following: the substrate heater, the pure nitrogen line, and the reactant line (See Figure 3.2). Both of the gas lines were put into the reactor at a fixed 17 distance. These distances were measured from the open end of the reactor. The pure nitrogen line was 15.5 cm. into the reactor while the reactant line exits 7.5 cm. into the reactor. The position of the substrate heater will be discussed in detail later in this section. Once the reactor was set up internally, the belljar was sealed and the mechanical fore pump started. The cooling water for the diffusion pump and the liquid nitrogen condenser was turned on and the external parts of the reactor were put in place. These consisted of an aluminum collar and the microwave source. The collar fit over the Conflat flange and extended down the quartz reactor so approximently 3 inches of the quartz was protruding. The exposed quartz reactor was then placed in the microwave chamber. The microwave source was a commercial microwave oven; therefore, a hole was cut on one side of it to accommodate the reactor. Because of the wall, the reactor extended only 2.5 inches into the source. The aluminum collar was used to absorb any microwaves which might be emitted from the hole. This was also the point where the plasma field ends. with the background mentioned in the previous paragraph, an accurate description of the heater and substrates placement can be given (see Figure 3.3). The heater was placed so that the left edge of the clamp was 3.5 cm. from the conflat flange. Therefore, the middle of 18 the substrate on the right was approximately 18 cm. downstream from the plasma. The clamp was 2.54 cm, this will put the substrate to the left at approximately 22 cm downstream. -3.5c:rn I 12 em I b c:m -----i ] , I I ,I I I I I I I I ... _~ .... I SUBSTRATE CLAMP HEATER ------I , I I I :I:!*!I!:!:!:I~!*!I!I!I n 7 em I Figure 3.3 Dimensions of the Quartz Reactor 3.6.2 ACHIEVING A VACUUM Once all of the preliminary activity was completed, 19 the belljar was sealed and the mechanical fore pump started. When the pressure reached 10-2 torr, the diffusion pump was switched on. When the pressure of the system reached 3 x 10-6 torr, the gas lines were started. 3.6.3 DEPOSITION The TMS was placed in a 100 ml Erlenmeyer flask which was inserted in the appropriate nitrogen line. The gas cylinder valves were opened and the regulator adjusted to bring the delivery pressure to 10 psig. The pure nitrogen stream was turned on and introduced into the reactor. This is an attempt to remove the oxygen from the system which could react and contaminate the SiNx film. Once this stream flowed for 2 minutes, the reactant stream was allowed to enter the reactor. The flow of each stream was adjusted by separate rotameters. The microwave source was turned on and the timer set for 2 minutes. After the 2 minutes, both streams were turned off along with the diffusion pump. The cooling water remained on for 45 minutes after the diffusion pump was turned off. After this time the belljar was vented and the appartus removed. 3.7 EXPERIMENTAL DESIGN Many preliminary trials were done before the ranges of 20 flowrates and temperatures were chosen. The range was chosen keeping in mind that relatively low temperatures were needed while maintaining significant deposition rates. As far as the flowrate of the two nitrogen streams is concerned, the lower the flowrate, the higher the deposition rate. Table 3.1 gives the parameters for several runs. Table 3.1 Deposition Parameters Run # Temperature Nitrogen Nitrogen (OC) Flowrate Flowrate (seem) Through TMS (seem) 1 400 106 106 2 350 106 106 3 300 106 106 4 250 106 106 5 300 274 106 7 300 370 106 8 300 466 106 9 300 106 274 10 300 106 370 11 300 106 466 21 IV. ANALYSIS 4.1 Film Thickness (Deposition Rate) The film thicknesses were determined using a Dektak IIA Profilometer. This instrument has a diamond stylus which scans by moving linearly across the surface to give a profile of the path. In order to get a film thickness measurement, part of the substrate was masked off during deposition; this was done by putting the substrate under the clamp. Figure 4.1 shows an example of a typical surface profile. A calculation which can be made from the film thickness is the deposition rate. Since all of the samples had run times of 2 minutes, a deposition time per minute can be determined by dividing the film thickness by 2. Table 4.1 gives information on film thickness. The no film (NF) and too thin (TT) notation in· Table 4.1 are not the same. The TT means that there is visible evidence of a film (a change of the reflective property of the sUbstrate) but the interpretation of the profile data could not separate the film from the curvature of the silicon wafer. The NF notation simply means that there in no visible or profile evidence of a film. -'-'-'---- _._._------_... _---- -_.- .- ... -- ~.. 1 fl r ()fiIe f o r f\ u rl r\Jo. ~3 18 em Downst re orn 300() 2700 ~ 7400 -- (n E ";1 o L oo~ L ~ 1800-- 01 c ~ 1500 en ~ i 200 c -Y. .~ 900 .L ~--- GOl) 300 o ,'--..,l c>:cf=tv'>.or--ca:~J • I, edC , I ! I,I,I, II () 200 400 600 800 1000 1 2()() -1400 1 (lOO 18()C) 20()O Sc on Dist a nee (tT1 ic rons ) L . ------ Figure 4.1 Sample Profilometer Data rv rv 23 , f TahIe 4. 1 Prof'l1. ornet ry l.n ormat'loon Run # Film Deposition Film Deposition Thickness Rate 18 em Thickness Rate 18 em DS* DS* 22 cm DS* 22 em DS* (run) (run/min) (run) (nm/rnin) 1 3,097 1,549 47 24 2 44 22 lost --- 3 224 112 22 11 4 34 17 TT --- 5 86 43 TT --- 6 178 89 23 12 7 34 17 TT --- 8 27 14 NF --- 9 TT --- NF --- 10 TT --- NF --- TT - Film too thin to get a thickness NF - No film deposited DS* - Downstream 4.2 IR ANALYSIS The infrared analysis of the composition of the thin film was performed on a Perkin Elmer Model 1600 Fourier Transform Infrared Spectrometer. Table 4.2 lists the peaks which commonly appear in silicon nitride thin films. The Figures 4-2 to 4-11 are the spectra taken for some of the thin films. 24 Table 4.2 Infrared Peaks of Silicon Compounds in cm- 1 Group Name Main Peak Secondary Peak Other Peaks 5 w SiCH3 1265 1408" 787-746 5 w Si (CH3 ) z 1257 1408" 847-826 5 w Si (CH3 ) 3 1250 1408" 847-819 775-740w SiO 1120-10005 SiOSi 10525 800" SiCHz 1060-10405 SiCHzCHzSi 1160-1136" 1075-1040" SiN 860-8105 s: strong w: weak v 100 CJ ~ :fj .....a C/] ~ cd J-4 E-4 50- 1500 1000 Wavenutnbers (ern-I) RR1_1 Res= 4.000000 07/19/91 12:20 Figure 4.2 The IR spectra of the thin film produced in run #1. The substrate was 18 cm downstream from the plasma field. (\J U1 120 - ~ 100 Q .s ...... ,.J 8 ~ cd J...4 ~ 00- 60 1500 1000 Wavenurnbe.·s (cln -1) I~R1_2 Res= 4.000000 07/19/91 12:05 Figure 4.3 The IR spectra of the thin film produced in run #1. The substrate was 22 cm l\J downstream from the plasma field. 0'\ 140 - (1) c:J ~ ~ ..... S 120 ~ ('j J-f f-4 100 - 1500 1000 Wavellurnbers (ern-I) RR2_1 Res= 4.000000 07/19/91 12:12 Figure 4.4 The IR spectra ,of the thin film produced in run #2. The substrate was 18 cm downstream from the plasma field. rv ....J 110 - Q) 100 CJ Q ~ ...... , .~ I~IIJIIII.~~ aen r' r I Q cd ..... E--.. 90 _. 1500 1000 Wavenurnbers (crn-1) RR3_1 Res= 4.000000 07/19/91 12:14 Figure 4.5 The IR spectra of the thin film produced in run #3. The substrate was 18 em downstream from the plasma field. f\) 00 J 10- Q) c:J ~ +J:l ...... a f1) ~ cd ~ 100- 1500 1000 Waventllubers (cln-l) I~R3_2 Res= 4.000000 07/19/91 12:22 Figure 4.6 The IR spectra of the thin film produced in run #3. The substrate was 22 em downstream from the plasma field. tv \0 105- QJ (.) ~ 100 ~ ...... -+J S acd ~ E-4 95 - 90 1500 1000 Waven rr rrr bers (ern -1 ) RR5_1 Res= 4.000000 07/t9/f)1 12:23 Figure 4.7 The IR spectra of the thin film produced in run #5. The substrate was 18 cm downstream from the plasma field. w o -.1 Q) (J ~ cd ..0 ~ o rn ..0 -< J 5()0 1000 WaveIllllubers (cn"l-l) R5_2 Res= 4.000000 07/ 1 ~)/~)1 1 2 :2 () Figure 4.8 The IR spectra of the thin film produced in run #5. The substrate was 22 crn w downstream from the plasma field. ~ f\.) w 18 em was substrate 12:29 The #6. run 07/19/91 in (ern-I) produced 1000 field. film 4.000000 thin > s plasma the Re the Wavenurnbers of from spectra IR The downstream 1500 4.9 I~6_1 00- 100 Figure .... cd ~ en s ~ (,) cv E--4 :f1 ...... "*.~k. ,~.:;,Jln ·n1____,.,_"".v""",,, , ,>,\~~~iL•..·='·~_·""'.2;'-'j1_u·•• . ], -.1 - Q) (.) ~ co ..0 .... o en ..0~ -.2 1500 1000 Wavenurnbers (ern-I) RO_2 Res= 4.000000 07/19/91 12:3() Figure 4.10 The IR spectra of the thin film produced in run #6. The substrate was 22 ern w downstream from the plasma field. w 120- Q) (.) ~ .s .~-+J (I]8 ~ cd ~S-4 100- 90 .J , J 1500 10(lO Wavenllrnbel·s (clu-l) I~es= 07/1~}/9112:25 I~R7_1 4.000000 Figure 4.11 The IR spectra of the thin film produced in run #7. The substrate was 18 em downstream from the plasma field. VJ ~ 35 v. DISCUSSION 5.1 Profilometer Data 5.1.1 Film Thickness Versus Temperature In general, as temperature increased, film thickness increased. Figure 5.1 shows the relationship of film thickness to temperature for films deposited at 18 cm downstream from the plasma. This graph shows an extremely large increase between 350 40000 ,.,...... and 400°C. Another feature 00 E o ~ JOOOO * of this graph shows that the (f) CJ') c deposition rate at 300°C is <:»-« 00 20000 (f) greater than the rate at Q) c ~ 350°C. This does not follow .2 .= 10000 the general trend. o...... "..r'TTT"~rT'TT,..,..,.-n-TTTT'TTTT"TT'T'CI,..,.,..,"'T"'I'T"'I~rTT"T'"rT'T'T'1* Visually, the film 2 deposited at 400°C appears Figure 5. 1 Fi lm thickness versus temperature for runs #1-4. The substrates were 18 to be milky white in color cm downstream from the plasma field. this indicates a non- stoiciometric film. On the left side of the same SUbstrate, there is a thinner deposit of film. It has a thickness of 100 nm, as compared to 3097 nm, This difference in film thickness occurs over a distance less than one centimeter. There can be several explanations for such an occurrence. One 36 explanation is that there may be a variation in substrate thickness. This would create a temperature gradient across the substrate surface therefore creating a variable deposition rate. Another explanation could be the possibility of flow patterns developing in the quartz reactor. This would lead to concentration gradients across the reactor. The concentration gradient is the most probable cause of this. The film deposited at 500 300°C appeared to be the most uniform of the four temperatures utilized in this C/) experiment. For this reason, C/) ~200 * .x: the rest of the experiments .~ ..cr- were performed 100 at this E G: temperature. The film itself 250 300 350 400 450 had a bluish tint over the Temperature COe) Figure 5. 2 The graph fi lm thickness versus entire substrate. Most of teq:>erature for the fi lm deposited 22 cm downstream from the plasma field. the other deposits were rainbow-colored in appearance. This appearance comes from varying thicknesses in the film, reflecting the light at different wavelengths. The film deposited on substrates 22 cm downstream followed the general thickness versus temperature trend. Figure 5.2 illustrates the film thickness versus temperature. These films are much thinner than the other film deposited just 4 cm upstream. 37 5.1.2 Film Thickness Versus Pure Nitrogen Flowrate The variation of thickness with the flow rate of pure nitrogen stream is difficult to interpret. Figure 5.3 shows 2500 .r'" a general decrease in (fJ * §2000 ~ ~ thickness with increasing CIl Ol * ~ 1500 nitrogen flow. This can be ~ CIl explained as follows: The (fJ ~ 1000 .:::L .~ * increase in the flow of ..c: t- 5OO excited nitrogen dilutes the E i.L * available TMS and permits 1 0 2 Nitrogen Stream shorter residence time, Figure 5. 3 Fi lm thickness versus pure causing smaller deposition nitrogen stream flowrate 18 em downstream. The temperature is 300°C and a flowrate of 106 seem of rates. Flow patterns could nitrogen through TMS. account for the scattering of the points. 5.1.3 Film Thickness Versus Nitrogen Flowrate Through the Reactant (TMS) Figure 5.4 shows the relationship between the thickness and flow of nitrogen through TMS at 18 cm downstream. It shows a steady decline as the flowrate increases. There are two reasons for the decrease in deposition rate. First, the increased flowrate of nitrogen would mean 38 an increased outlet velocity. This would allow more TMS to * enter the plasma and deposit on the walls of the reactor. since much of the TMS is consumed, less is available * for downstream deposition. 102 Nitrogen Flowrate Through Reactant (seem) Second, the higher gas Figure 5. 4 Fi lm thickness versus flowrate flowrate leaves shorter of nitrogen through TMS 18 cm downstream. Temperature is 300°C while the pure nitrogen flows at 106 seem. residence time, leading to decreased deposition rates. At 22 cm downstream, all of the samples were too thin or there was no film at all. 5.1.4 Downstream Effects In general, the further downstream a substrate is located, the lower the deposition rate. In the series of experiments, only three of the substrates at 22 cm downstream had significant deposition. They occurred at runs numbered 1, 3, and 7. The rates at 22 cm downstream were smaller than the rates at 18 em, The rates were only 1.55, 9.82 and 12.92% of the upstream deposition rates. This not only is due to the distance downstream, it is also due to the dissipation of reactant along the length of the reactor. 5.2 IR Data 39 The IR analysis gives a qualitative measure of the composition in the film. Figures 4.2 - 4.11 clearly indicate the presence of carbon (C), oxygen (0), hydrogen (H), and nitrogen (N). The strongest peak, located at 1050 cm" indicates the presence of si-o bonds. SiN is the next 1 • strongest at 810 cm- The peak at 1274 is due to SiCH3 which indicates possibly unreacted TMS. These seem to be the only peaks apparent when analyzing the substrates 18 cm downstream from the plasma. The substrates 22 cm downstream, reveal peaks not seen in the others. A shoulder located at 1140 cm- 1 appears on the side of the sio band. This is most probably SiCH2CH2Si (see Table 4.2). There are limitations with the analysis technique used. The main one is the IR Spectrometer's sensitivity to the film thickness. The films had to be over 500 angstroms to get a readable spectrum. The use of Rutherford Backscattering Spectroscopy (RBS) would quantify the composition of the films. Gopal (1991) was able to determine the composition of film made from a TMSjnitrogen MECVD system using RBS. 1O The film had the following composition ratio: si: 1.2, N 2.0, 0 3.0, C :2.0, H 5.0 The IR spectrum of the film deposited in Run #1, 18 cm 40 downstream in this study, and the above film is shown in Figure 5.5 and Figure 5.6. The film from this study has a stronger Si-N band (810 cm") and also a broader si-o band (1120 cm") , Looking at the Si-N peaks, the one from this study is stanger compared to the si-o peaks. This implies a larger amount of the Si-N bonds present when depositing by RMECVD. There are also more impurties (besides Si-O) in the film deposited by MECVD. They are as follows (See Figure 5 . 6): 127 0 cm-1 - S i CH 3 1 1682 cm- - CNH 2 3 60 cm-1 - S iH 1 - • 2966 cm- CH3 None of these peaks are present when viewing a thin film deposited by RMECVD using a similar scale (see Figure 5.5). Although, when the scale is expanded between 400 and 1500 em" small peaks do appear. Therefore, some impuries are present in the RMECVD films, but in much smaller quantities. r-I ~ 600 -4 i 400 4.> C) ~ ~~ ~ "s m c:I a:s M E-4 200 o r- 4000 3000 2000 1000 WavenuD1ber:9 (cxn-l) RR1_1 Res= 4.000000 08/06/91 11:16 Figure 5.5 The IR Spectra of the film produced in Run #1. The substrate was 18 cm ~ downstream from the plasma field ...-a ~ 80 ~ ~~ ~ ..... 8U] ~ ~ Jo.4 E-4 60 4000 3000 2000 1000 Wavenurnbers (ern-1) T3_1 Res= 4.000000 08/08/91 11:27 10 ~ Figure 5.6 The IR spectra of a thin film deposited by MECVD using TMS at 25°C. l\J 43 VI. Conclusions 1. An RBS study needs to be performed to quantify the composition of these films. After this, it can be determined if TMS is an adequate replacement for silane. 2. Downstream or remote chemical vapor deposition is a feasible way of depositing thin films at temperatures low enough to avoid harming composites used for space systems. 3. The films deposited in run #3 were the most uniform film of the stUdy. The conditions were as follows: Temperature = 300°C Flowrate of pure nitrogen stream = 106 sccm Flowrate of nitrogen through TMS = 106 sccm Studies need to be performed to find out if smaller flowrates would improve the quality of the film. 4. All depositions had nitrogen, oxygen, silicon, hydrogen, and carbon present in the films. The IR spectrum indicates these were bonded in the following way: Sia, SiN, SiCH3 From the results of the IR, it can be concluded that the films deposited are probably silicon oxynitride with some unreacted TMS present 44 5. The set up used in this study is adequate for performing downstream MECVD, although many problems are associated with it. In the present set up, both gas streams enter the plasma zone. The result is a very heavy deposit on the reactor walls. Modifications are suggested in the Recommendation section (Chapter 7). 6. An RBS stUdy needs to be performed to determine if RMECVD reduces the amount of hydrogen in the film. 7. The variation in deposition rate in run #1 on the substrate, which is 18 cm downstream, is most likely due to flow patterns developing in the reactor. 8. The RMECVD films as compared to the MECVD films deposited by Gopal (1991)4 seem to have more Si-N bonds present and also less hydrocarbon impurties. 45 VII. Recommendations 1. Modifications should be made to the existing reactor to improve the system. As it is now, the two feed lines and the substrate heater are not fixed tightly inside the reactor. To improve this, a fixture should be built to hold them more securely inside the reactor. This will allow the same flow patterns to develop in each trial and from that, a model can be developed. 2. Ideally, a new reactor should be designed and built. with a new quartz chamber sized to fit over the substrate heater in the belljar, the entire present quartz reactor can become the plasma zone where only nitrogen is excited. The excited nitrogen will leave from the plasma zone, enter into the quartz reactor, and combine with the nitrogenjTMS stream and deposit. on the substrate. Since the TMS will never be in the plasma, the deposition on the wall of the reactor will be eliminated, thus leaving more TMS to deposit on the substrate. The elimination of TMS in the plasma zone should also provide for a smoother film. The silicon nitride soot formed in the plasma zone travels down the reactor and becomes incorporated into the film. 3. Greater control is needed over the parameters. Better temperature control is needed along with an accurate way of 46 determining the amount of TMS entering the reactor. New rotometers are needed to handle smaller flowrates of gas. A better location is needed for the pressure sensor to determine the pressure inside the reactor. 4. A more sensitive IR Spectrometer is needed. 47 VIII• OPBRB.CBS 1. Nishikawa, Takao, Kalsumi Sonoda, Korchiro Nakanishi, "Effects of Atomic Oxygen On Polymers Used as Surface Materials for Spacecraft", 21s t symposium on Electrial Insulating Materials 1988 2. Leger, L., J. Visentine, B. Santos-Mason, "Selected Materials Issues Associated with Space station", 18 t h International SAMPE Technical Conference, Oct. 7-9, 1986 3. Rutledge, Sharon K., Phillip E. Paulsen, Joyce A. Brady, "Evaluation of Atomic Oxygen Resistant Coatings for Fiberglass-Epoxy Composites In Low Earth Orbit" 34 t h International SAMPE symposium, May 8-11, 1989 4. Tennyson, R.C., G.E. Mabson, w.o. Morison, J. Kleiman, "Composites in Space", Advanced Materials and Processes, April 1991 5. Tennyson, R.C., G.E. Mabson, w.o. Morison, J. Kleiman, "Composites in Space", Avanced Materials and Processes, May 1991 6. English, Lawrence K., "Atomic Oxygen: Achilles' Heel of Man in Space", ME, August 1987, p39 7. Archer, John S., Robert L. Brown, Parker J. Congill, "Development of Graphite Adapter for Galileo Spacecraft", 29 t h National SAMPE symposium, April 3-5, 1984 8. Adams, A. C. , Plasma Deoosited Thin Films, ed. Frank Jensen, CRe Process, Boca Raton, Fla. 1985 9. Klemberg-Sapieha, J.E., M.R. Wertheimer, D.G. Zimcik, "Plasma Deposited Multi Purpose Protective Coatings for Space Applications", ESA Journal 1989, vol 13 10. Gopal, M., "Microwave Enhanced Chemical Vapor Deposition of Silicon Compound Thin Films and Their Characterization" M.S. Thesis Ohio University 1991 11. Dzioba, steven, S.Meikle, R.W. Streater, "Downstream Plasma Induced Deposition of SiNx on Si, InP, and InGaAs", J. Electrochem Soc. p2599 48 12 . Tsu D. V., G. Lucovsky, M.J. Mantini, "Local Atomic structure in Thin Films of Silicon Nitride and Silicon Diimide Produced by Remote Plasma Enhanced Chemical Vapor Deposition", The American Physical Society, 1986, p7069