Characterisation and optimisation of PECVD SiNx as an antireflection coating and passivation layer for silicon solar cells

Related Articles Growth and band alignment of Bi2Se3 topological insulator on H-terminated Si(111) van der Waals surface Appl. Phys. Lett. 102, 074106 (2013) Effect of microwave plasma treatment on silicon dioxide films grown by atomic layer deposition at low temperature J. Appl. Phys. 113, 064102 (2013) Strain engineering in epitaxial La1−xSr1+xMnO4 thin films J. Appl. Phys. 113, 053906 (2013) Synthesis and characterization of MoS2/Ti composite coatings on Ti6Al4V prepared by laser cladding AIP Advances 3, 022106 (2013) Molecular beam epitaxy of high structural quality Bi2Se3 on lattice matched InP(111) substrates Appl. Phys. Lett. 102, 041914 (2013)


I. INTRODUCTION
An antireflection coating (ARC) is an important feature of crystalline silicon solar cells.Ideally, this layer should not only maximize optical transmission but simultaneously suppress surface recombination.Amorphous hydrogenated silicon nitride a-SiN x :H (hereafter referred to as SiN x ) synthesised by low-temperature PECVD has become the state-of-the-art ARC layer for c-Si solar cells to fulfil these two requirements. 1,2 0][11][12][13][14][15] Details of the deposition processes and silicon substrates employed in these studies are included in Table I. Figure 1 summarizes the results by plotting the upper limit to the  e S eff,UL is cited at excess carrier density n = 10 15 cm −3 and recalculated using the latest Auger model. 16G.1.The upper limit to the effective surface recombination velocity S eff,UL as a function of refractive index n at 632 nm for SiN x -passivated FZ p-Si substrates.SiN x is deposited by a variety of deposition techniques and gas mixtures as summarised in Table I.The dash-line highlights the opposing trend presented in Ref. 15.
effective surface recombination velocity S eff,UL as a function of the refractive index n.The presented S eff,UL and n are at an excess carrier density of n = 10 15 cm −3 and a wavelength of 632 nm, respectively.As can be seen, irrespective of deposition techniques and reactant gas mixtures employed in Refs.10, 12, and 13 an apparent trend is observed: S eff,UL decreases as n increases.More complicated trends can be observed for the results in Refs.9 and 11, whereby S eff,UL first decreases with increasing n and saturates for n above 2.3.The results presented in Refs.9-13 make it appear that the optimum surface passivation is acquired by Si-rich SiN x of high n.Since high-n SiN x is well-known to be highly absorbing of short wavelength light, this trend implies there is a trade-off between optical transmission and surface passivation.However, two previous studies suggest that the trade-off can be circumvented: Schmidt et al. 14 achieved a low S eff,UL (15.2 cm/s on FZ 1.0-• cm p-Si) using a stoichiometric SiN x (n = 1.9) by including N 2 with SiH 4 and NH 3 , and our recent work 15  to obtain the film in Ref. 15 is by increasing the deposition pressure in a microwave/radio-frequency (μW/RF) PECVD reactor.These latter results are also included in Figure 1, showing that as the deposition pressure increases, both S eff,UL and n decrease.This trend is opposite to the previously assumed trend, providing the opportunity to obtain low absorption and low recombination.The two works 14,15 show that remarkably low surface recombination is achievable by lowly absorbing SiN x and imply that the trade-off between optical transmission and surface passivation can be circumvented.
In this paper, we first review the SiN x deposition system employed in this work and the SiN x growth mechanisms proposed in the literature.These are presented in Section II.We then conduct two experiments to characterise and optimise SiN x as an ARC and passivation layer for solar cells.Section III presents the first experiment, which is a central composition experiment (CCE) that examines six deposition parameters.After reporting on the dependence of SiN x properties (i.e.deposition rate, structural, optical and electronic properties) on the deposition parameters, we state the optimized deposition conditions that attain low absorption and low recombination.On the basis of the SiN x growth mechanism described in Section II and of our experimental results, we then discuss how the deposition parameters affect the film properties.Further, we perform correlative studies to understand the relationships between various SiN x properties.Section IV presents the second experiment in which the gas-flow rate is varied and the other depositions conditions are set to their optimal values.It demonstrates that a relatively constant and low S eff,UL (< 10 cm/s) on low-resistivity (≤1.1 cm) p-and n-Si can be achieved by a single SiN x layer within a broad range of n = 1.85-4.07.

II. SIN X DEPOSITION AND GROWTH MECHANISMS
The SiN x films are deposited in a static laboratory-scale μW/RF PECVD system (Roth & Rau AG, system AK400).Figure 2 depicts a cross-section of the deposition system, which consists of two gas inlets: a gas shower at the top for injecting NH 3 or/and Ar, and a gas ring at the bottom for injecting SiH 4 .The plasma is excited around quartz tubes by means of two continuous-wave magnetron microwave generators with a frequency of 2.45 GHz, introducing up to 2000 W of microwave power into the process chamber.The RF plasma is excited near the graphite substrate by means of an RF generator with a frequency of 13.56 MHz, creating a bias voltage of up to 300 V between the substrate and the ground.
Here we adapt the SiN x growth model proposed by Smith et al., 17 Kessels et al. 18 and Oever et al. 19 to discuss possible chemical reaction schemes for SiN x growth in this work.For the gas mixture of N 2 and SiH 4 , Smith et al. 17 and Kessels et al. 18 proposed that an a-Si:H-like surface layer is created by SiH a radicals and the a-Si:H-like surface is simultaneously reacted with N radicals, leading to the formation of SiN x .Oever et al. 19 studied the plasma chemistry for the gas mixture of NH 3 , SiH 4 and Ar and refined the growth model, concluding that SiN x is formed by the direct surface reaction between the a-Si:H-like layer and NH b radicals.We adapt the abovementioned models and refine a simple reaction scheme for SiN x growth in this work: Gas phase dissociation: -NH 3 is introduced into the system through the shower-head which forms part of the ceiling of the deposition chamber above the μW quartz tubes and is dissociated: -SiH 4 is fed from the downstream gas ring that is between the quartz tube and RF-biased graphite substrate and is dissociated: (ii) Formation/insertion: An a-Si:H-like layer is formed on the silicon wafer by SiH a radicals and simultaneously inserted with NH b radicals.(iii) Film densification by cross-linking at elevated substrate temperature to break Si-H and N-H and form Si-N: Concurrently, the excessive incorporation of NH b radicals breaks and replaces the existing Si-H bonds (at elevated substrate temperature), forming Si-N: or otherwise, the excessive incorporation of SiH a radicals breaks and replaces the existing N-H bonds, forming Si-N:

III. EXPERIMENT 1: CENTRAL COMPOSITION EXPERIMENT
For the optimization of six deposition parameters, we performed the CCE that consisted of 28 different deposition conditions (parameter sets) and four additional replications of the baseline.Each set of experiments involved varying one deposition parameter while keeping the other parameters constant and at their baseline value.The relationships between SiN x properties and individual deposition parameters are described and discussed in this section.

A. Experimental details
Table II presents the baseline and range of the deposition parameters varied in this work.It is common to have an offset between the set-point of the substrate table temperature and the actual silicon sample temperature, 9, 11 so all deposition temperatures reported here refer to the actual wafer temperature measured by an external calibrated thermocouple, and not to the reactor's set-point temperature.
The lifetime samples in this experiment were p-type {100} FZ-Si wafers with a resistivity of 0.85 • cm and a thickness of 300 μm.All samples were etched in tetramethylammonium hydroxide (TMAH) at ∼85 • C to remove saw damage.The p-Si samples were cleaned by the RCA procedure and diffused with phosphorus to getter iron and other metallic impurities. 20The phosphorus glass was then removed in HF acid and the phosphorus-doped silicon layer was removed by etching in a 1:10 HF:HNO 3 solution.Next, all wafers were cleaned by the RCA procedure, dipped in HF to remove the native oxide, and then coated with SiN x on both surfaces by two sequential depositions.The effective carrier lifetime τ eff of the samples was measured using a Sinton Instruments WCT-120 operated in either transient or generalized quasi-steady-state mode, as described elsewhere. 21he S eff,UL ( n) can be calculated according to Where W is the Si substrate thickness and τ bulk,intrinsic is the Si intrinsic bulk lifetime parameterized by Richter et al. 16 Reflectance and C-V measurements were performed on double-side-polished n-type {100} FZ-Si with a resistivity of 1.0 • cm and a thickness of 290 μm.Metal-insulator-semiconductor (MIS) test structures were fabricated for C-V measurements.In this work, the front metal contact was formed by evaporating aluminum through a shadow mask to create circular dots of diameter ∼700 μm and thickness ∼100 nm.The rear contact was formed with a GaIn eutectic.More details on the fabrication of reflectance and C-V measurements samples, as well as the characterisation of the wavelength-dependent refractive index n(λ) and extinction coefficient k(λ), and the interface defect density D it and effective insulator charge Q eff associated with SiN x , can be found in Ref. 15.
The FTIR transmission spectra were measured on the same samples as used for reflectance and C-V measurements, using FTIR spectrometer (Bruker Vertex 80V) with a resolution of 6 cm −1 .The measurement showed three distinctive absorption peaks associated with Si-N, Si-H and N-H vibrational modes about 850, 2220 and 3340 cm −1 , respectively. 22The bond density [A-B], defined as the number of bonds per unit volume, can be determined by 23 where α(ω) is the absorption coefficient at wavenumber ω, k A-B is the proportionality constant in cm −2 .In this work, k A-B for Si-N, Si-H and N-H is taken from Ref. 22 having the values of 2 × 10 19 , 2 × 10 20 and 1.2 × 10 20 , respectively.

Dependence of film properties on deposition parameters
Figure 3 plots the dependence of the deposition rate, bond densities, optical and electronic properties of plasma SiN x on the various deposition parameters.These dependencies are now described.
Deposition rate-For throughput considerations, the deposition rate by PECVD should be as high as possible.Figure 3(a) shows the effect of deposition parameters on the deposition rate.The deposition time for all conditions was 3 min, and the resulting film thickness ranged from 50 to 200 nm.We found that increasing the total gas flow within the tested range causes a threefold increase in the deposition rate.The factor increase is ∼2 for the decrease of NH 3 /SiH 4 gas flow ratio and smaller (∼1.2) for the variation of other parameters.Optical properties-Figure 3(c) depicts the responses of n at 632 nm and k at 360 nm to the deposition parameters.As we can see, n decreases strongly with increasing NH 3 /SiH 4 gas flow ratio, μW plasma power, pressure and total gas flow, whereas it is hardly affected by the variation of deposition temperature and RF bias voltage.Moreover, Figure 4(a) and 4(b) show the dispersion relation for both the refractive index n and the extinction coefficient k.The results are consistent with those reported in other studies, [24][25][26] showing a decrease of n and k with increasing wavelength.As expected, SiN x of low n (≤2.0) exhibits low absorption of short-wavelength light and almost negligible absorption at wavelengths above 360 nm, implying it is better suited as the ARC in solar cells.Indeed, compared to the SiN x of n = 2.5 and an equivalent passivation, the optical simulation of our previous work 15 found that the SiN x of n = 1.9 would enhance the photogenerated current density by more 0.66 mA/cm −2 or 1.40 mA/cm −2 for solar cells encapsulated in glass and ethylene-vinyl acetate (EVA) or operating in air, respectively.This enhancement is due to the film's low k rather than it having the optimum n.In addition, we find that, irrespective of the variation of deposition parameters, an increase of k at 360 nm is generally accompanied by an increase of n at 632 nm, as made clearer with Figure 4(c).
Surface passivation-Figure 3(d) plots S eff,UL of the as-deposited SiN x .It indicates that S eff,UL depends strongly on deposition temperature, pressure and gas flow ratio, and slightly on total gas flow, μW plasma power and RF bias voltage.Whereas S eff,UL increases with increasing NH 3 /SiH 4 gas flow ratio, S eff,UL decreases significantly as pressure increases.An optimum deposition temperature is observed at 300 • C, leading to a minimum S eff,UL .
Figure 5 shows the injection-dependent effective lifetime τ eff ( n) for SiN x -passivated p-type silicon samples.We briefly comment on how τ eff ( n) is affected by temperature and pressure since passivation is the most sensitive to these parameters.For reference, the Auger limit parameterized by Richter et al. is also plotted. 16We note that none of the τ eff ( n) curves cross over except for SiN x deposited at 405 o C, which has a maximum τ eff at n = 5 × 10 15 cm −3 .This means that we can meaningfully compare τ eff (and hence S eff,UL as presented in Figure 2(d)) at a particular n, such as 1 × 10 15 cm −3 .
Figure 5(a) shows that the shape of the τ eff ( n) curves diverges from the shape of the intrinsic lifetime curve as temperature increases.More specifically, there is an increasing injection dependence at low n, whereby τ eff increases with n, consistent with the silicon surface being in inversion due to positive insulator charges. 27,28 ith C-V measurements, we confirm this speculation.As evident in the inset of Figure 5(a), Q eff increases with an increase of deposition temperature.In contrast to the temperature variation, Figure 5(b) shows that there is no change to the injection dependence of τ eff ( n) when pressure is varied, which correlates to a relatively constant Q eff .The dependence of Q eff on deposition temperature but not pressure is consistent with the insulator charge originating from thermal-assisted charge injection from the silicon substrate. 29inally, the optimum deposition condition that achieves an SiN x with low absorption and low recombination is concluded to be 300 • C, 0.2 mbar, NH 3 /SiH 4 /Ar gas flow at 20/20/20 sccm, μW plasma power at 500 W and RF bias voltage at 150 V.Note that even though the film deposited at 0.4 mbar provides lower S eff,UL than that at 0.2 mbar, we select an optimum deposition pressure of 0.2 mbar instead of 0.4 mbar because the spatial uniformity of the film was poor when deposited at 0.4 mbar.These optimum deposition conditions are utilized to examine the trade-off between optical transmission and surface passivation in Section IV.

Discussion on the effect of deposition parameters on film properties
We now discuss how each process parameter affects the deposition rate and chemical bond density in relation to the growth mechanism described in Section II.
Precursor gas flow ratio is commonly varied to alter SiN x film properties as it directly tailors the partial pressure of the resultant radicals.As the NH 3 /SiH 4 increases, the proportion of NH b radicals in the reactor increases while the proportion of SiH a radicals decreases.We therefore observe that 1) deposition rate decreases, 2) [Si-H] decreases and [N-H] increases and 3) n and k decreases.
The deposition temperature hardly affects reaction steps (i) and (ii) but causes significant change in film densification (step iii).According to Figure 3(b), both [Si-H] and [N-H] decrease as the deposition temperature increases.This behaviour is attributable to the film densification through reaction (c).As temperature increases, the film is densified through the restructuring of Si-H and N-H to form Si-N. Film densification by cross-linking at elevated temperature would also result in a decrease in film thickness, as evident in Figure 3(a).
The influence of deposition pressure and μW plasma power on the hydrogen bond densities can also be elucidated.Changing the deposition pressure and plasma power has a strong effect on the gas phase dissociation rate and the shape of the electron distribution function. 30,31 s pressure decreases, μW plasma becomes less confined near the quartz tubes and approaches the downstream SiH 4 gas injection region. 11The decrease of pressure therefore enhances the dissociation of SiH 4 but suppresses the dissociation of NH 3 , leading to an increase in [Si-H] and a decrease in [N-H].Moreover, the excessive incorporation of SiH a radicals (at low pressure) further reduces [N-H] through reaction (e).
Next, note that NH 3 in our reactor is injected from the top shower head and is mainly dissociated by the μW plasma power through reaction (a).This means the increase of μW plasma power enhances more dissociation of NH 3 than of SiH 4 .We therefore observe that [N-H] increases with the increase of μW plasma power.Again, the excessive flux of impinging NH b radicals (at high μW plasma power) further breaks Si-H bonds through reaction (d).These conclusions are consistent with Figure 3(b), which shows that [N-H] increases as [Si-H] decreases when either pressure or μW plasma power is increased.
Plasma power is primarily dominated by μW plasma rather than by RF plasma in a μW/RF PECVD reactor because the ion density in a μW plasma is more than one order of magnitude higher than in an RF plasma. 31The main purpose of the RF plasma in our system is to create a bias voltage between the plasma and the substrate, thereby enhancing the flux of radicals onto the substrate surface.Based on the discussion above, we are not surprised to see in Figure 3(b) that an increase in the RF bias voltage is accompanied by a slight increase in [Si-H], [N-H] and the deposition rate.
Without sufficient information on the relative amount of gas phase consumption of a precursor, also called depletion, we cannot postulate how an increase of total gas flow causes an increase of [Si-H] and a decrease of [N-H].Nevertheless, the strong increase of deposition rate as well as the increase of [Si-H] with increasing total gas flow further confirms that SiN x deposition rate in our system is primarily limited by the supply of SiH 4 precursor gas or consequently by the incorporation of SiH a radicals.

Correlations between SiN x properties
a. Optical properties versus structural properties.Since an increase in refractive index n is believed to correlate to a shift in the chemical composition toward being more Si rich, 32 we plot n as a function of [Si-H]/[N-H] in Figure 6 for all deposition conditions.We can then express the results by the linear relationship where the uncertainty represents the 95% confidence interval.This empirical relation indicates that, over a wide range of deposition conditions, the optical properties primarily depend on the ratio of b. S eff,UL versus bond density.[Si-H] is regularly considered as a measure of surface passivation quality owing to the hypothesis that higher [Si-H] implies a higher probability that hydrogen terminates the Si dangling bonds at the SiN x -Si interface. 9, 10Lauinger et al. found a clear increase in τ eff with increasing n of SiN x for p-type silicon wafers, where an increase of n corresponds to an increase of [Si-H] and a simultaneous decrease of [N-H].This was observed both for remote PECVD SiN x precursor gases (NH 3 /SiH 4 ) 9 as well as direct PECVD SiN x precursor gases (N 2 /H 2 /SiH 4 ). 10 Figure 7 plots S eff,UL against [Si-H] for the films in our work.It shows that when the NH 3 /SiH 4 gas flow ratio or RF bias voltage is varied, surface passivation improves as n increases, corresponding to an increase in [Si-H].These results are consistent with Refs. 9 and 10.However, contrary to the trend, when other deposition parameters are varied, no consistent correlation can be established between S eff,UL and [Si-H].As evident in Figure 7, S eff,UL keeps increasing with an increase in [Si-H] when pressure and μW plasma power are varied.More complicated behaviour is induced by varying deposition temperature whereby S eff,UL exhibits a minimum at [Si-H] = 7 × 10 21 cm −3 .Clearly, there is no universal dependence of S eff,UL on [Si-H].
Similarly, we do not find a universal correlation between S eff,UL and [Si-N] for our as-deposited films.The blue squares in Figure 8 indicate that [Si-N] has little influence on the as-deposited SiN x passivation.Figure 8 also plots S eff,UL against [Si-N] for the same samples after a rapid-thermal anneal (RTA) at 800 • C for 5 seconds.][35] Higher [Si-N] improves the thermal stability of SiN x passivation, attributing to lower effusion of H 2 molecules from the denser SiN x layer into the ambient during short annealing step.Figure 9(a), we find that S eff,UL depends strongly on D it , irrespective of the varied process parameters.As D it increases over three orders of magnitude, S eff,UL increases by two orders of magnitude.However, S eff,UL shows different relationships with Q eff when the SiN x is altered by varying different deposition parameters.As evident in Figure 9(b), S eff,UL is almost constant as Q eff increases from 3 × 10 9 cm −2 to 3 × 10 11 cm −2 , and then exhibits two orders of magnitude variation when Q eff = 4 × 10 11 cm −2 .Figure 9 shows that S eff at SiN x -passivated silicon surface in this work depends primarily on interface defect density rather than charge.

IV. EXPERIMENT 2: CIRCUMVENTING THE TRADE-OFF BETWEEN OPTICAL TRANSMISSION AND SURFACE PASSIVATION
Inspired by the vastly different responses of n and S eff to various deposition conditions, we now examine the challenge mentioned in Section I of developing a single SiN x layer that circumvents the trade-off between optical transimission and surface passivation.
In relation to this goal, we note that Lauinger et al. 9 achieved a saturation of τ eff but by very Si-rich SiN x with n above 2.3, and Hoex et al. 38 presented a relatively constant S eff,UL within a refractive index range of 1.9-2.4,however, the S eff,UL associated with their SiN x -passivated p-type 8.4-• cm FZ Si substrates is relatively high (50-70 cm/s).Indeed, within a broad range of n (1.85-4.07)associated with SiN x , a constant low S eff,UL on low resistivity (≤1.1 • cm) Si substrates has not been achieved.

A. Experimental details
The optimized deposition conditions presented in Section III B were applied on three types of low resistivity FZ Si samples: 0.85 • cm p-type, 0.45 • cm n-type, and 1.10 • cm n-type.The wafers received the same TMAH silicon etch, RCA clean, and double-side SiN x deposition that were applied to the lifetime samples in Section III.Note that no gettering step was applied on n-Si samples, since n-Si has a lower sensitivity to metal contaminants than p-Si. 39The S eff,UL at n = 10 15 cm −3 and n at 632 nm were characterized in accordance with the procedures described in Section III A. In this experiment, the NH 3 /SiH 4 gas flow ratio was then varied to obtain a broad range of n, ranging from 1.83 (sub-stoichiometric) to 4.07 (close to a-Si:H).

B. Results and discussion
The S eff,UL for the three types of samples is plotted against n in Figure 10.While there is no universal relationship between S eff,UL and n, the trend in this experiment is similar to those presented in Refs.9 and 11 (see Figure 1).In these cases, the variation in n was achieved by varying the NH 3 /SiH 4 gas flow ratio.As indicated in Figure 10, S eff,UL first decreases as n increases and then saturates over the range n = 1.85-4.07.This behaviour is seen on all three types of substrates.Note that the S eff,UL for 0.45 • cm n-Si seems typically lower than the S eff,UL for 1.1 • cm n-Si.One possible reason is that the Auger model proposed by Richter et al. 16 underestimates the bulk intrinsic lifetime for the 0.45 • cm n-Si, leading to a lower S eff,UL .Note further that all S eff,UL reported in this section is by the as-deposited SiN x .
In addition to the similar trend, we note that (i) at an equivalent value of n, our SiN x provides a very low S eff,UL , where the lowest S eff,UL is 5.3 cm/s, 0.6 cm/s and 1.8 cm/s for the aforementioned three types of Si samples and (ii) the saturation of S eff,UL in Refs.9 and 11 starts at n = 2.3 whereas it starts at a lower n of 1.85 in this work.Overall, very low S effUL is attained over a range of n where k is also low.The results of Figure 10 suggest that an appropriate deposition of SiN x can eliminate the trade-off between ARC absorption and surface passivation.In short, solar cells that require high surface passivation need not be compromised by light absorption in the SiN x .

V. CONCLUSION
The properties of SiN x films synthesised by a μW/RF PECVD reactor were altered by varying the substrate temperature, pressure, NH 3 /SiH 4 gas flow ratio, total gas flow, μW plasma power and RF bias voltage.After reporting on the dependence of the SiN x film properties on the deposition parameters, we determined optimized deposition conditions that attain low absorption and low recombination.On the basis of the SiN x growth models proposed in the literature and of our experimental results, we discussed how each process parameter affects the deposition rate and chemical bond density.By studying the correlations between the structural, optical and electrical properties, we found that for the SiN x prepared in this work 1) S eff,UL does not correlate universally with the bulk structural and optical properties such as chemical bond densities and refractive index, and 2) S eff,UL depends primarily on the defect density at the SiN x -Si interface rather than the insulator charge.Finally, employing the optimized deposition condition, we achieved a relatively constant and low S eff,UL on low-resistivity (≤1.1 cm) p-and n-type c-Si substrates over a broad range of n = 1.85-4.07.The results demonstrate that the trade-off between optical transmission and surface passivation can be circumvented by a judicious deposition of SiN x .Although we focus on photovoltaic applications, this study may be useful for any device for which it is desirable to maximize light transmission and surface passivation.

FIG. 4 .
FIG. 4. Wavelength-dependent (a) refractive index n and (b) extinction coefficient k for representative SiN x films determined from spectrophotometry measurements.The correlation between n at 632 nm and k at 360 nm is also illustrated in (c).

FIG. 6 .
FIG. 6. Relationship between measured refractive index n at 632 nm and SiN x bond density ratio: [Si-H]/[N-H].The solid line is a linear best fit and the dashed lines are its 95% confidence intervals.

FIG. 7 .
FIG. 7. S eff,UL as a function of [Si-H] for SiN x deposited at a variety of conditions.
FIG.10.Extracted S eff,UL for three types of FZ c-Si substrates as a function of n at 632 nm.In this experiment, only the NH 3 /SiH 4 gas flow ratio was varied to alter the SiN x properties.

TABLE I .
Summary on the details of PECVD processes and silicon substrates employed in references.
a Process variable altered in Refs.10-13 is gas flow ratio.b FZ-float zone; ρ bulk -nominal bulk resistivity; W-silicon substrate width.c Measured temperature on substrate table by thermal coupler.d Temperature setpoint.

TABLE II .
Baseline and tested range of PECVD deposition parameters.
a Ar gas flow is kept constant at 20 sccm.b sccm denotes cubic centimetre per minute at standard temperature and pressure.