7. Results for H/C/O/S Systems

 

7.1. Introduction

 

This chapter builds on the work presented in the previous two chapters and presents the results of MWCVD diamond growth experiments using H₂S additions to 51%CH₄/49%CO₂ (x=0‑0.5%) gas mixtures.  Thus, the effect of lowering T_sub and the use of O containing gas mixtures on the growth of S-doped CVD diamond films can be investigated.  OES and MBMS diagnostics have been used to investigate the gas-phase chemistry of such systems.  The results are correlated with those of SENKIN computer simulations.  The degree of S‑incorporation into the deposited films is discussed with reference to the parameter F_s (as introduced in Section 6.8).

 

7.2. Diamond Deposition From H₂S/51%CH₄/49%CO₂ Gas Mixtures

 

MWCVD experiments were undertaken in which films were deposited using xH₂S/51%CH₄/49%CO₂ (x=0-0.5%) gas mixtures with applied microwave power of 1 kW and pressure of 40 Torr.  Deposition was undertaken at substrate temperatures of 620 and 900°C (as discussed in Section 3.2.9).  Diamond films were analysed by SEM, LRS, four-point probe and XPS, as discussed below.

 

7.2.1. Film Crystallinity

 

Deposited films were examined by SEM in order to investigate the effect on film crystallinity and facet size of H₂S addition to a 51%CH₄/49%CO₂ gas mixture.  Figure 7.1 presents SEM images of diamond films grown at substrate temperatures of (a)-(c) 620°C and (d)-(f) 900°C, at various H₂S additions.  The film grown at the lower T_sub, using an H₂S addition of 100 ppm (Fig. 7.1(a)) exhibits a continuous coating of predominantly triangular facets ~5 mm in size.  Increasing H₂S input levels to 2000 ppm decreases the facet size to < 2.5 mm (Fig. 7.1(b)), while 5000 ppm H₂S additions produce a film primarily composed of small (< 1 mm) facets (Fig. 7.1(c)) with the occasional larger crystallite.  In contrast, there is little change in the surface morphology of films grown at T_sub = 900°C for H₂S additions of 100, 2000 and 5000 ppm, as illustrated by Figures 7.1(d), (e) and (f), respectively.  Here, all three films are continuous and display crystalline facets ~5 mm in size. 

 

 

Figure 7.1. SEM micrographs for films grown using 51%CH₄/49%CO₂ gas mixtures and T_sub of 620°C with H₂S additions of (a) 100, (b) 2000 and (c) 5000 ppm, and T_sub = 900°C with H₂S additions of (d) 100, (e) 2000 and (f) 5000 ppm in a MW plasma reactor.  Conditions: total gas flow 200 sccm, growth time 8 h, pressure 20 Torr, 1 kW applied microwave power.

 

7.2.2. Film Growth Rate

 

Figure 7.2 shows that the growth rates of films grown at T_sub = 900°C are approximately one order of magnitude greater than for those grown at 620°C.  However, despite this difference, the observed growth rate trend, i.e. a fall of ~66% when H₂S addition is increased from 0 to 5000 ppm, is similar for both temperatures. 

Figure 7.2. Film growth rate (measured by cross-sectional SEM), versus H₂S addition for films grown in H₂S/51%CH₄/49%CO₂ gas mixtures at two substrate temperatures.  Key: (■) T_sub = 620°C, (о) T_sub = 900°C, with other process conditions as for Fig.7.1.  Both data sets show a ~66% reduction in growth rate over the range of H₂S addition shown.

 

7.2.3. LRS Analysis of Film Quality

 

Raman spectra of diamond films grown at an H₂S addition of 5000 ppm at (a) 620°C and (b) 900°C are presented in Figures 7.3(a) and (b).  Both spectra can be fitted using a quadratic base line, two Gaussian peaks centred at ~1332 cm^-1 (sharp diamond Raman line and a broader D band), a peak at ~1530 cm^-1 (the G band) and a broad band at ~1450 cm^-1 (polyacetylene [[1]] or a diamond precursor [[2]]).  However, a number of additional features are evident in the spectrum of the sample grown at the lower T_sub, including the presence of peaks at ~1172 cm^-1 (nanophase diamond [[3]] or polyacetylene present in grain boundaries [1]) and 876, 967 and 1088 cm^-1.  These additional peaks may correspond with nanoscale sp³ phases resulting from the small sized crystallites grown at this low T_sub.

 

Figure 7.3. Laser Raman spectra of films deposited using 5000 ppm H₂S addition to 51%CH₄/49%CO₂ gas mixtures for (a) T_sub = 620°C and (b) T_sub = 900°C.  Each spectrum includes a fitted quadratic baseline and Gaussian line shape curve fits.

 

In order to gain a measure of film quality (i.e. the ratio of sp³:sp² carbon bonding) the quality factor, Q (as discussed in Section 2.3) is presented in Figure 3(c).  Somewhat surprisingly, Q for films grown at the lower T_sub is generally slightly higher than those deposited at 900°C.  Although the scatter is considerable it is apparent that, in both cases there is a slight fall in film quality with increased H₂S addition for both sets of data. 

Figure 7.4. Plot of the diamond film quality factor, Q, vs. H₂S addition.  Key: (■) T_sub = 620°C, (о) T_sub = 900°C, with other process conditions as for Fig.7.1.  The lines are linear least squares fits to the two data sets (as labelled).

 

7.2.4. Four-point probe Film Resistivity Measurements

 

There is a general fall in resistivities with increased H₂S addition, for the films grown at both high and low T_sub (Figure 7.4).  The scatter seen for samples grown with low H₂S additions and at high T_sub is due to the measurements being made at the limit of the apparatus capabilities.  There is also a large offset of ~3 orders of magnitude between resistivity measurements of high and low T_sub grown films.  Apart from this offset, the trend in film resistivities for H₂S additions over 1000 ppm is similar for both datasets. Both show a ~100-fold reduction in resistivity when H₂S additions are increased from 1000 to 5000 ppm.

Figure 7.5. Plot of four point probe resistivity vs. H₂S addition for films deposited using H₂S/51%CH₄/49%CO₂ gas mixtures.  Key: (■) T_sub = 620°C, (о) T_sub = 900°C, with other process conditions as for Fig.7.1.

 

7.2.5. XPS measurements of Film S/C ratio

 

XPS analysis has afforded S/C ratios of high temperature films grown with input gas mixture H₂S concentrations of 3500 ppm (S/C ~0.07%) and 5000 ppm (S/C ~0.08%).  We note that XPS yields an average signal over the whole area of the substrate (1 cm²), and so the position of the S (i.e. grain boundaries versus crystallites) cannot be determined.  S/C ratios for all other deposited films were below the detection limit of the apparatus (~0.05%).

 

7.3. OES of H₂S/51%CH₄/49%CO₂ Microwave Plasmas

 

An optical emission spectrum of a 1%H₂S/51%CH₄/49%CO₂ MW plasma is presented in Figure 7.6.  Emissions due to C₂ Swan bands are clearly present at ~470, 516 and 560 nm.  Also evident are emissions of the 3^rd Positive and 5B bands (260-370 nm) and Angstrom bands (~451, 484, 520 and 561 nm) of CO.  Emissions due to CH (431 nm) and H (483 nm) are also seen. 

 

Figure 7.6. Plot of an optical emission spectrum of a 1%H₂S/51%CH₄/49%CO₂ MW plasma.  Conditions: applied microwave power 1 kW, pressure 40 Torr

 

Emission spectra taken of plasmas containing 0% and 1%H₂S were almost identical.  Therefore, six emission wavelengths were chosen and their relative intensities were plotted vs. H₂S addition, as presented in Figure 7.7.  It is clear that the intensities of the CO and H emission features stay relatively constant or even increase slightly over the range of H₂S additions, while those of C₂ and CH are seen to reduce by ~18 and 20%, respectively, over the same range.  This suggests that increased H₂S addition to the 51%CH₄/49%CO₂ gas mixture causes a reduction in hydrocarbon concentrations.

 

Figure 7.7. Optical emission peak intensity vs. H₂S addition to a 51%CH₄/49%CO₂ gas mixture.  Emission features are as labelled and percentages given indicate the drop in intensity of the indicated feature over the range of H₂S addition shown.  Conditions: applied microwave power 1 kW, pressure 40 Torr.

 

7.4. Computer Simulations of H₂S/51%CH₄/49%CO₂ Gas Mixtures

 

SENKIN computer simulations of the gas phase chemistry of H₂S/51%CH₄/49%CO₂ gas mixtures were carried out using a combination of the GRI-mech 3.0, C-S linking and Leeds combined methane combustion mechanisms (see Sections 4.5.1 to 3).  Figure 7.8 shows SENKIN predicted mole fractions, X, versus H₂S addition to a 51%CH₄/49%CO₂ gas mixture at (a) 20 and (b) 40 Torr.

Figure 7.8. Plots of SENKIN predicted species mole fraction, X, versus H₂S addition to a 51%CH₄/49%CO₂ gas mixture for a pressure of (a) 20 and (b) 40 Torr.  Data for species not containing S are presented as thick lines to aid clarity. Conditions: simulation temperature 2000 K, reaction time 5 s.

 

It is clear from Figure 7.8 that the species trends are similar for the two pressures presented.  In both cases H₂S addition leads to a rise in predicted mole fraction of CS, [CS].  A rise is also seen in [H₂S], [S], [CS₂] and [S₂], while [C₂H₂], [H₂O], [CO₂], [CH₄] and [CH₃] all fall with increased H₂S addition.  At both pressures [SO] and [SO₂] are predicted to be ~10^-7 and 10^-11, respectively.  Both species mole fractions are below the estimated detection limit of the MBMS apparatus of ~10^-6.  [H₂] and [CO] are predicted to be 0.5 for both pressures.  Increasing the simulation pressure from 20 to 40 Torr has the effect of decreasing the absolute values of [S₂] while causing an increase in [CS₂].

 

7.5. Molecular Beam Mass Spectrometry Studies of H₂S/51%CH₄/49%CO₂ Microwave Plasmas

 

All MBMS measurements were made under the same conditions as the deposition runs, except that the chamber pressure was reduced from 40 to 20 Torr.  This was because the presence of the probe made the plasma unstable at pressures >20 Torr.  Measurements were made of H₂S/51%CH₄/49%CO₂ MW activated gas mixtures as a function of H₂S addition (0‑1%).

 

Figure 7.9 shows combined plots of MBMS measured species counts and SENKIN predicted species mole fractions, against added H₂S.  The respective vertical scales in each plot have been arranged so as to aid comparison.  The plots of C₂H₂, CH₄ and CH₃ show good agreement between the measured and predicted species trends, with H₂S addition resulting in a decrease in species mole fraction and measured counts.  Similarly, experiment and simulation are in agreement in the case of H₂S, CS and CS₂, with species mole fraction and measured counts increasing upon increased H₂S addition.  There is, again, good agreement between MBMS and SENKIN results in the case of SO.  However, the predicted mole fraction, X, of SO is ~10^-7, which is approximately 100 times below the estimated detection limit of the MBMS apparatus.  This suggests that SO is present within the microwave plasma at concentrations much greater than predicted by SENKIN.  A similar conclusion can be drawn from the SENKIN and MBMS results for SO₂.  Here, the scatter in SO₂ measured species counts data suggests SO₂ concentrations are near the MBMS detection limit (X ~10^-5), whereas the predicted SO₂ mole fractions are far below this, i.e. ~10^‑11.  On the other hand, the MBMS measured counts of S and S₂ appear to be at or below the MBMS detection limit, suggesting that the predicted mole fractions of S (~10^‑4) and S₂ (~10^‑5) may be overestimates. 

 

Figure 7.9. Plots of MBMS measured species counts (right hand scale) and SENKIN predicted species mole fractions (left hand scale) vs. H₂S addition to a 51%CH₄/49%CO₂ gas mixture, for the following species: (i) C₂H₂, (ii) CH₄, (iii) CH₃, (iv) H₂S, (v) CS, (vi) CS₂, (vii) SO, (viii) SO₂, (ix) S and (x) S₂.  Conditions are as given in Fig.7.1, except that the pressure was reduced to 20 Torr to improve plasma stability in the presence of the sampling probe.  SENKIN computer simulation results are for a H₂S/51%CH₄/49%CO₂ gas mixture at a temperature of 2000 K and a pressure of 20 Torr.  Key: о MBMS species counts, (·) SENKIN computer simulation mole fractions.  The two data sets in each panel have been vertically scaled to illustrate the degree of consistency between the experimental and modelled trends.

 

7.6. Discussion

 

The results presented in the preceding sections will be discussed and compared with those obtained using H₂S/1%CH₄/H₂ gas mixtures (Chapter 6) in the following section.  H₂S addition to 51%CH₄/49%CO₂ MW plasmas during diamond CVD is seen to have a considerable impact upon the growth rates, quality and resistivities of the resulting films.  S addition also induces changes in film morphology, particularly in the case of low temperature deposition, e.g. T_sub = 620°C.  Reduction of T_sub from 900 to 620°C results in a ~10-fold drop in film growth rate.  Such reduction in growth rates upon lowering of T_sub are consistent with previous observations (see Section 5.2.1 and Refs. [4] and [5]) but the increased substrate-plasma separation used during low T_sub deposition in the present studies may also contribute to the observed variation.  However, the fact that we observe very similar downward trends in film growth rate for increased H₂S additions at both deposition temperatures suggests that the trend is primarily due to gas phase reaction instigated by H₂S, rather than gas-surface interactions.

 

A large offset was observed between the measured film resistivities of samples grown at high (900°C) and low (620°C) T_sub.  We suggest that the lower resistivity of the low temperature grown samples is probably due to H-induced surface conduction [[6]-[7][8][9]].  This effect is reported to be annealed out of films at temperatures above ~700°C in vacuum and ~190°C in air [8].  The resistivity of such annealed films is several orders of magnitude greater than the H-terminated samples, pre-annealing.  We therefore suspect that the growing diamond surface is (at least partially) H‑terminated at T_sub = 900°C.  However, at the end of a deposition run the plasma is shut off and the film is left to cool down in a 100%CO₂ atmosphere.  During cooling from 900-700°C, we envisage that the H‑induced surface conductivity is annealed out, thereby resulting in the highly resistive film observed.  T_sub = 620°C is below the temperature for such annealing.  In this scenario, therefore, films grown at this lower substrate temperature would retain their H‑induced surface conductivity intact when cooled in a CO₂ atmosphere.  Previously films have been deposited at T_sub = 900°C using H₂S/1%CH₄/H₂ gas mixtures (see Chapter 6).  After growth, these films were cooled in a H₂ atmosphere and the measured film resistivities (Section 6.2.4) were comparable with those obtained for the low temperature deposited films in the present work.  This observation lends further credence to the hypothesis outlined above.

 

Increased H₂S additions (>1000 ppm) appear to cause a reduction in film resistivities, regardless of T_sub.  It is important to note that the reduction in film resistivities observed between high T_sub grown films with 0 and 5000 ppm H₂S additions is much greater than the ~3-fold reduction seen for H₂S doping of 1%CH₄/H₂ gas mixtures, using similar process conditions (Section 6.2.4).  However, it remains unclear whether this reduction is a result of S-incorporation into the diamond films or due to an increase in graphitic phases within the film (as indicated by a parallel reduction in the quality factor).

 

OES measurements of C₂ and CH emission features suggest that increased H₂S additions caused a drop in gas phase hydrocarbon concentrations.  This conclusion is reinforced by the observed trends in MBMS measured counts and SENKIN predicted mole fractions for C₂H₂, CH₄, and CH₃.  The methyl radical is generally believed to be the major diamond growth precursor in most low pressure CVD reactors [[10],[11]].  It therefore seems reasonable to associate the measured drop in CH₃ MBMS counts with the observed drop in film growth rates upon increased H₂S addition.

 

A reaction scheme for the chemistry of H₂S addition to 51%CH₄/49%CO₂ gas mixtures is presented in Table 7.1.  It should be noted that all except Reaction 4 are composite conversions composed of many elementary reactions.  Reaction 1 is the dominant reaction involving S containing species.  This is shown by the large SENKIN predicted mole fractions and MBMS measured counts of CS₂ and CS.  Investigations of 50%CH₄/50%CO₂ gas mixtures (Section 5.7.3) have concluded that the majority of the input carbon is transformed to CO via Reaction 2.  Any excess CH₄ not destroyed by this reaction is able to undergo further conversions to produce C₂H₂ and CH₃ via Reactions 3 and 4, respectively.  However, Reaction 1 is able to compete with these reactions, with the result that increased H₂S additions lead to a reduction in the CH₄ available for production of C₂H₂, CH₃ and H₂O.  This is analogous to the situation seen for H₂S addition to 1%CH₄/H₂ gas mixtures (Section 6.8).  As stated above, the reduction in CH₃ concentrations caused by increased H₂S addition leads to a reduction in deposited film growth rates.

 

a)      Reaction taken from Section 5.7.3.

b)      Reaction taken from Table 6.1 in Section 6.8.

 

Table 7.1. Proposed reaction scheme for the gas phase chemistry of H₂S/51%CH₄/49%CO₂ diamond CVD gas mixtures at 2000 K and ~20 Torr.  DG_reac values were calculated using species Gibbs free energies of formation obtained from Reference [12] and were also used to calculate reaction equilibrium constants, K_eq.

 

In addition, we recall that there is a large (~2 mm) separation between the visible edge of the plasma and the substrate in the case of films grown at low T_sub, whereas the growth surface of samples deposited at high T_sub are in direct contact with the visible edge of the plasma.  The larger substrate-plasma separation for low T_sub grown films clearly causes a further reduction in the CH₃ flux reaching the substrate surface, as a result of radical recombination reactions (e.g. involving CH₃ and H radicals).  Arguably of even greater importance, is the reduced mobility of species across the growing film surface at low T_sub [5,[13],[14]].  Such diffusion is considered necessary for the growth of uniform diamond CVD films, as carbon containing species move to step edges allowing ordered crystal growth.  These two effects are likely to be responsible for the much greater change in deposited film surface morphology observed upon H₂S addition at lowered T_sub, compared to growth at the higher T_sub. 

 

Table 7.1 also sets out two possible routes to the production of SO (Reactions 6 and 7).  The first involves a composite reaction between CO₂ and H₂S producing CO, H₂ and SO, while in the second reaction H₂O and H₂S react to form SO and hydrogen.  At 2000 K, the Gibbs free energy for both reactions, DG_reac, is greater than zero, indicating that, at equilibrium, the reverse of the reactions as written are spontaneous (i.e. DG_reac < 0).

 

However, as stated previously, the MBMS measurements suggest that the SENKIN calculations underestimate the true SO and SO₂ concentrations.  This might reflect an overestimation of the forward rate constant of one or more of the elementary reactions of which Reaction 1 is comprised (Table 4.2 in Section 4.5.2), which would result in a reduced concentration of H₂S available to participate in Reactions 6 and 7, and thus a lower predicted SO mole fraction.  This, in turn, would cause an underestimation of the predicted mole fraction of SO₂ produced via Reaction 8.  Alternatively, or additionally, it may reflect inadequacies in the kinetic data for S/O and H containing species reactions that we have taken from the Leeds combined methane and SO_x combustion mechanism (as discussed in Section 4.5.3).  The fact that some 70% of these rate constants are either estimations or extrapolations from lower temperatures [[15]] serves to underline the inadequacy of the database of reactions involving H/C/O/S containing species.

 

Table 7.2 presents a comparison between SENKIN species mole fractions for 1%H₂S addition to both 1%CH₄/H₂ and 51%CH₄/49%CO₂ gas mixtures.  [H₂], [C₂H₂], [H₂S], and [CS] are predicted to be similar for both gas mixtures, whereas [CH₃] and [CH₄] are ~100 times smaller for the 1%H₂S/51%CH₄/49%CO₂ gas mixture.  For a 1%H₂S/1%CH₄/H₂ gas mixture, [S] and [S₂] are predicted to be at, or below, the estimated MBMS detection limit (in accordance with the lack of experimental measurements of these species).  Changing to a 1%H₂S/51%CH₄/49%CO₂ gas mixture has the effect of increasing [S] (~100 fold) and [S₂] (~10 fold).  This has the effect of increasing [S] to a level above the MBMS detection limit, as reflected by the experimental detection of S.

 

 

Table 7.2. SENKIN predicted species mole fractions for a 1%H₂S addition to 1%CH₄/H₂ and 51%CH₄/49%CO₂ gas mixtures.  Conditions: reaction time 5 s, Pressure 20 Torr, temperature as shown for each gas mixture.

 

The S incorporation parameter, F_s = [CH₃]´[CS], is plotted in Figure 7.10 as a function of H₂S addition to both 1%CH₄/H₂ and 51%CH₄/49%CO₂ gas mixtures.  It should be noted that the SENKIN product for the 51%CH₄/49%CO₂ gas mixture has been scaled vertically by a factor of five to fit onto the plot.  XPS measured S/C ratios for films deposited using these gas mixtures (at T_sub = 900°C) are indicated also.  The figure illustrates the point that, for a given H₂S addition, 1%CH₄/H₂ gas mixtures are found to yield films with higher S/C ratios than 51%CH₄/49%CO₂ mixtures.  Such an observation is consistent with the smaller SENKIN predicted values of [CH₃]´[CS] calculated for the latter gas mixture.  Indeed, this parameter is seen to maximise in the case of the H₂S/51%CH₄/49%CO₂ gas mixtures, at ~0.42% added H₂S.  This is midway between the only two H₂S addition values (0.35 and 0.5%) at which the S content was measurable by XPS. 

Figure 7.10. Plots of F_s vs. H₂S addition to (a) 1%CH₄/H₂ (T_gas = 1630 K) and (b) 51%CH₄/49%CO₂ (T_gas = 2000 K) gas mixtures obtained from the SENKIN calculations.  Conditions: pressure 40 Torr, reaction time 5 s.  The numbers present the S/C ratio (as measured by XPS) for films deposited at the H₂S additions indicated.  Note that the values of F_s for the H₂S/51%CH₄/49%CO₂ gas mixture have been scaled up by a factor of 5.

 

7.7. References