1.2.1- Optimisation of the deposition parameters " Window-substrate distance and total pressure
Since very little absorption is observed in the gas phase, there is a need to increase the window substrate distance to a high value, typically 5cm. In that case, and with pressures ranging around 50mbar, the absorption by N2O of the incident radiation is
about 50% (figure 4.3).
The deposition rates are extremely low for pressures below lOOmbar. In contrast, at pressures exceeding 150 mbar, powders were formed as a consequence of secondary gas phase reactions being stimulated, leading to radical recombination and polysiloxane formation [Kamaratos]. Those films were opalescent and very scratch sensitive. Since this is detrimental to the film quality, intermediate pressures are
• Light intensity
Figure 4.4 shows the variation of the deposition rate at 300°C with input light power (40W input power corresponds to ~ 20mW/cm^). It clearly shows that thermal deposition of Si0 2 is insignificant within the operating conditions. Since a constant
precursor flow is used in this experiment, an attenuation of the growth rate increase is observed while increasing the lamp power, consequence of the limitation of the diffusion of photoproducts and reactants to the substrate surface.
40 50
10 20 30
Input Lamp Power (W)
Figure 4.4 : Evolution o f the deposition rate fo r silicon dioxide samples with light exposure (Substrate temperature is 300°C)
• SiH/j/NoQ precursor ratio R
Figure 4.5 shows the variation of the deposition rate with R, the SiH^/N2 0 ratio.
For R<1%, the growth rate is limited by the silane concentration. For R>2% nucléation reactions are promoted in the gas phase, limiting film deposition. The decrease of the growth rate can mainly be attributed to a decrease in the SiHg species concentration which are able to react with O species, since some are reacting with NO species. High values of R may lead to the deposition on the window of materials which have higher absorption to the incoming radiation (e.g., SiO^Ny or SiO^Hy), thus further reducing the growth rate. Figure 4.6 shows the refractive index (n)
evolution as a function of R. Optimised values of 1-2% for R lead to films presenting the 1.462 theoretical value for n in Si0 2 . For higher R values, higher values of n are
observed, reflecting an increase in either H or N incorporation.
• • e
9-0 2 4 6 8 10 12
R% = [silane]/[Nitrous oxide]
Figure 4.5 : Evolution o f the deposition rate V5'. precursor mixture ratio (P= lOOmbar, T=300°C) 2.0 ^1.462 % 1.4 ■ 1.2 1.0 0 2 4 6 8 10 12 R% (Silane/Nitrous Oxide)
* Substrate temperature
Figure 4.7 shows an Arrhenius plot of the growth rate. An activation energy of 0.25eV is extrapolated from those data. Purely pyrolytic reactions give activation energies in the range 0.42eV to 0.95eV [Cobianu] [Vasilyeva] [Taft]. The very low value obtained gives evidence of the reduced importance of the substrate temperature and therefore of a less dominant role of thermally induced surface reactions. The modified chemical pathways induced by the photons clearly enable lower substrate temperatures to be used. No significant variation of the refractive index was observed for temperatures varying from 200 to 400°C.
o I - T em perature (“C) 500 400 300 200 100 10 1 100 1.2 1.8 lOOO/T (K 'l)
Figure 4.7 : Arrhenius plot fo r the growth o f Si02from SiH^ and N2O mixtures.
1.2.1- FTIR spectroscopy
The structural properties of the films were characterised by their vibrational spectra using FTIR spectrometry. Figure 4.8 shows a FTIR spectrum of an oxide film photo-deposited at a substrate temperature of 300° C. The spectrum consists of three characteristic peaks at 1065cm"^ (Si-O-Si stretching), 800cm" ^ (O-Si-0 bending), and 480cm"^ (Si-O-Si rocking) [Boyd-1982] [Lucovsky]. The full width half maximum (FWHM) for the dominant IR mode around 1065cm" ^ has a value of 70cm" characteristic of a stoichiometric Si0 2 » and remains essentially the same for
all the oxides studied. The small peak centred around 885cm" ^ can be attributed either to Si-H, Si-OH, Si2 0 g or to non bridging oxygen [Boyd, 1982-b]. It is most
likely the reflection of a few percent of hydrogen in the film, due to the incomplete desorption of H species at low temperatures. However, the presence of Si-H and Si- OH bonds in concentrations higher than 5% would have led to the presence of other peaks centred around 2270cm" ^ and 3620cm" which were not observed.
The absorption spectra of figures 4.9-a and 4.9-b show that the quality of the film is not significantly affected by either lamp power or precursor mixture ratio used. These parameter however affect the deposition rate more appreciably (e.g., 6Â/min
Si-O S i-O -S i X 01 a - C J - - Si-O C0-- 1200 l O O O B O O B O O Wavenumbep (cm—1)
Figure 4.8 : FTIR spectrum o f a 400A Si02fHm deposited at 300°C
0 w CD u c o 4J t1 e o c o L f- 5 w 10 W 30 W 45 W 1200 1100 1000 900 10% 8% 6% 4% 2% m u c a 4> -P I fi H o C o L h X 0.5% 1200 1100 1000 900 W o v o n u m b o p ( c m - i ) W o v o n u m b B P ( c m - 1 )
Figure 4.9: Sets o f FTIR spectra o f the 1065cm~^ peak fo r various a-lamp power input, b-precursor mixture ratio R.
1.2.3- Film properties
Film properties were investigated on p-type silicon samples, exhibiting a resistivity p = 10 ^ .c m . For each sample, the value Ey of the electrical breakdown field was estimated as an average of the measured values on N devices (N>10). Typical Ey values were measured around 5 MV/cm, although lower values have been obtained for samples deposited at elevated pressures, or at temperatures below 200“C. Values
very low Ey value, consequently strongly reducing the averaged values. This was assumed to be due to particle contamination of the substrates in the laboratory, which is not a clean room facility. This contamination is further affecting the measurements on the very thin films studied. Figure 4.10 gives the evolution of the breakdown field when the precursor mixture ratio R is increased from 1 to 10%.
I
> 1i
10 8 6 4 2 0 6 0 2 4 8 10 12 SiH^/N^O ratio (%)Figure 4.10 : Evolution o f E^, with precursor ratio R
As can be seen, although high R ratios were the cause of high N incorporation, which was reflected by high refractive indices, the breakdown field does not appear to be altered, since no particular decrease of the electrical properties would be expected from SiO^Ny layers (y « x).
Capacitance-Voltage (CV) measurements also revealed that the typical surface state charges were ranging around lO^^-lO^^ cm"^, value which appears to be high comparatively with other conventionally deposited Si0 2 films (lO^^cm"^). This may
be caused by the absence of native oxide cleaning prior to deposition, thus resulting in a high number of interface traps between the silicon substrate and the native oxide, as well as between the native oxide and the deposited layer.
Etch rates measurements were finally performed on the silicon dioxide layers, using a buffered hydrofluoric acid solution, diluted in water at a concentration of Imol.l"^. Typical values of 25Â/s were obtained, to be compared with 16Â/s, characteristic etch rate for thermally grown silicon dioxide [Sze]. This is a much lower value than typically reported for CVD grown oxides (e.g., 5 0 A / s in HE, 0.25mol.l"^ by Adams et a l using PECVD [Adams-1983], and «80Â/s in HE, Smol.l"^ for films deposited at 300°C by Boyer et a l with laser CVD [Boyer]).