Material and optical properties of low-temperature NH₃-free PECVD SiN_x layers for photonic applications

The aim of this study was to optimise the deposition recipe of SiN_x films prepared by NH₃-free PECVD in order to obtain the best properties and the lowest propagation losses possible for photonic applications. The study was carried out using the Oxford Plasma Technology 100 PECVD system, employing the deposition recipe suggested by the manufacturer as a starting point with a SiH₄:N₂ chemistry (1.8:1000 sccm), a radio frequency generator (RF) power of 60 W, and chamber pressure of 650 mTorr.

We generated different central composite designs (CCD) with Minitab to study the effect of the SiH₄ flow and the RF power on the material and optical properties of the films following a similar approach to the one described by Tien et al [19], in order to have enough information to estimate how the properties of the layers would change if these deposition parameters were altered in any direction. Once we had a general overview, we focused on 9 experiments that included the factors and levels detailed in section (a) of table 1. The relations estimated from these experiments were confirmed afterwards by conducting tests that included additional factor combinations within the experimental region.

Similarly, we considered the effect of the N₂ flow on the structural properties of the SiN_x layers and its interaction with the SiH₄ flow using a experimental design that included 8 experiments with the factor variations defined in section (b) of table 1.

We also studied the effect of the chamber pressure on the properties of the SiN_x films by varying this parameter from 650 to 1000 mTorr using the deposition conditions described in section (c) of table 1.

Finally, we deposited SiN_x films on substrates with SiO₂ layers that were grown either by PECVD or thermal oxidation in order to study the effect of the SiO₂ layers on the propagation losses of SiN_x strip waveguides.

All the films were deposited at a substrate temperature of 350 °C using the recipes defined in the previous section. The recipes were deposited on bare Si samples ($2\times 2$ cm) targeting thicknesses close to 300 nm for deposition rate, ellipsometry, SEM, AFM and FTIR characterisation. Once the deposition rates were known, films with a thickness of 300 nm were deposited on samples with SiO₂/Si stacks ($3\times 4$ cm) to fabricate strip waveguides for propagation loss measurements. The SiO₂/Si stacks had a layer of either PECVD SiO₂ or thermal SiO₂ with a thickness greater than 2 μm to minimise the losses produced by the leakage of optical modes to the Si substrate.

The layout fabricated on the SiO₂/Si stacks included two sets of waveguides with lengths up to 1.7 cm to measure the optical losses of the SiN_x layers at 1310 and 1550 nm. The width of the strip waveguides was optimised to provide single-mode propagation for both TE and TM polarisations. The waveguides intended for measurements at 1310 nm had a width of 900 nm, while the waveguides for 1550 nm had a width of 1200 nm. In both cases, the designs included grating couplers optimised to have high coupling efficiency (CE) for TE polarisation at each targeted wavelength. The grating couplers for 1310 nm had a theoretical CE ^∼ 58% with a period of 1 μm and filling factor of 50%, while the ones for 1550 nm had a theoretical CE ^∼ 40% with a period of 1.2 μm and 50% filling factor.

Both waveguides and grating coupler were patterned on the SiO₂/Si stack samples using electron beam lithography with a 450 nm thick resist (ZEP520A) and conventional inductive coupled plasma (ICP) etch using a SF₆:CHF₃ chemistry. The waveguides were finally covered with a 530 nm PECVD SiO₂ layer deposited at 350 °C as a top cladding.

The material and optical properties of the SiN_x layers were systematically evaluated using different characterisation techniques. High-resolution scanning electron microscopy (SEM) was used to take cross-sectional images of each deposited layer in order to estimate their thickness, deposition rate and material structure. Atomic force microscopy (AFM) was used to characterise quantitatively the surface roughness of the layers by measuring their root-mean-square roughness (RMS) on a $2\times 2$ μm area. Ellipsometry carried out with an M2000DI ellipsometer was used to determine the thickness uniformity and refractive index of the layers at 632.8 nm. This refractive index (n) was used to estimate the N/Si ratio using equation (1) as proposed by Bustarret et al [16] and Mackel et al [23] using ${{n}_{\text{Si}}}=3.85$ for silicon and ${{n}_{\text{SiN}}}=2.00$ for stoichiometric Si₃N₄.

$$\begin{eqnarray}&&\frac{\text{N}}{\text{Si}}=\left(\frac{4}{3}\right)\,\left(\frac{{{n}_{\text{Si}}}-n}{n-2{{n}_{\text{SiN}}}+{{n}_{\text{Si}}}}\right)\end{eqnarray} \tag{ 1 }$$

FTIR spectroscopy was carried out using a Varian 600 FTIR spectrometer to determine the compositional structure of the layers including their bond concentration, atom concentration and hydrogen content. The different bond concentrations were obtained from the FTIR spectra after baseline correction with the method described by Yin and Smith [17] using the proportionality factors $K\left(\text{N}-\text{H}\right)=8.2\times {{10}^{16}}$ cm⁻¹ and $K\left(\text{Si}-\text{H}\right)=5.9\times {{10}^{16}}$ cm⁻¹ proposed by Lanford and Rand [15], and the proportionality factor $K\left(\text{Si}-\text{N}\right)=2.4\times {{10}^{16}}$ cm⁻¹ given by Bustarret et al [16]. The hydrogen and nitrogen atom concentrations ($\left[\text{H}\right]$ and $\left[\text{N}\right]$) were then estimated using expressions (2) and (3) [16, 17].

$$\begin{eqnarray}&&\left[\text{H}\right]=\left[\text{N}-\text{H}\right]+\left[\text{Si}-\text{H}\right]\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray}&&\left[\text{N}\right]=\left(\left[\text{N}-\text{H}\right]+\left[\text{Si}-\text{N}\right]\right)/3\end{eqnarray} \tag{ 3 }$$

Where $\left[\text{N}-\text{H}\right]$, $\left[\text{Si}-\text{H}\right]$ and $\left[\text{Si}-\text{N}\right]$ are the concentrations of N-H, Si-H and Si-N bonds, respectively.

The Si atom and Si–Si bond concentrations ($\left[\text{Si}\right]$ and $\left[\text{Si--Si}\right]$) were calculated separately using the N/Si ratio (x) previously estimated from the refractive index using equations (4) and (5) proposed by Mäckel and Lüdemann [23].

$$\begin{eqnarray}&&\left[\text{Si}\right]=\left[\text{N}\right]/x\end{eqnarray} \tag{ 4 }$$

$$\begin{eqnarray}&&\left[\text{Si--Si}\right]=2\left[\text{Si}\right]-1/2\left(\left[\text{Si}-\text{N}\right]+\left[\text{Si}-\text{H}\right]\right)\end{eqnarray} \tag{ 5 }$$

Finally, the cutback method was used to measure the propagation losses of each layer using the Agilent 8163B and 8164B Lightwave systems to obtain the spectral response of the fabricated waveguides at 1550 and 1310 nm respectively.

The structure of the deposited films is important for propagation losses because grains, pores, defects and rough interfaces are a direct cause of scattering losses [24]. Figure 1 contains SEM images that show the microstructure of some of the deposited SiN_x films and their interface with the Si substrate. Image (a) shows a SiN_x layer with excess nitrogen (N-rich), image (b) a layer close to stoichiometry (Si₃N₄) and image (c) a layer with excess silicon (Si-rich). All the films adhere well to the substrate, are free of cracks and have an amorphous structure with no significant changes along the growth direction.

Zoom In Zoom Out Reset image size

The N-rich and Si-rich samples are porous and have a microstructure with columnar growth that resembles the classic appearance of the Zone I type films detailed by Kaiser [24]. This structure can be attributed to the statistical roughening and self-shadowing processes that occur due to the low surface mobility and high sticking coefficient of the species when the surface temperature is low compared to the melting temperature of the deposited materials [24, 25].

These observations are in line with a previous study in which Smith demonstrated that standard PECVD SiN_x layers appear to be completely homogeneous compared to NH₃-free PECVD layers [25]. However, we have proved that the lower density regions between the columns tend to disappear when the chemical composition of the samples moves towards stoichiometry leading to more homogeneous, smooth and dense films as seen in figure 1. This improved microstructure is desirable to achieve light propagation with low losses.

The majority of the physical and optical properties of the SiN_x layers are highly dependent on their thickness and morphology. Examining the behaviour of their deposition rates is important because it affects the nucleation processes that determine the morphology of the deposited layers and it also allows a greater control over the film thickness. Our study suggests that all of the parameters: SiH₄ flow, N₂ flow, RF power and chamber pressure, affect directly the deposition rate of the NH₃-free SiN_x layers without interaction between them as illustrated in figure 2. In this case, the deposition rate increases when either the SiH₄ flow, RF power or chamber pressure are increased during deposition, while it decreases when the N₂ flow rises. Still, the deposition rate is more strongly affected by the RF power, which means that the deposition reaction is mostly plasma power limited.

Zoom In Zoom Out Reset image size

The deposition rate increases by 30% when the SiH₄ flow is doubled. A similar trend in the deposition rate of NH₃-free PECVD films with varying SiH₄ flow has been observed in the literature and can be explained by the higher proportion of Si radicals available that are incorporated into the films in the form of high density Si atoms when the SiH₄ fraction is higher [11, 18]. We believe that the deposition conditions allow the complete dissociation of the SiH₄ and that the process is limited by the N₂ concentration in such a way that an increased number of Si species can be incorporated if there are still N species that can be consumed. Hence, this tendency is only observed because the amount of available Si species is smaller than the amount of N species. Otherwise, it would be expected to see a saturation of the deposition rate at a maximum value due to the depletion of N reactants as Karouta observed in PECVD SiN_x films deposited with higher SiH₄ fractions [18].

On the contrary, the deposition rate decreases slightly when the N₂ flow is increased by 1.5 times. This effect is expected, since it is equivalent to reducing the SiH₄ flow and can be explained by the reduction of the proportion of Si species available for deposition. This effect is similar to the one observed by other authors with standard PECVD layers when increasing the NH₃ or N₂ flow [26]. In fact, it can also be observed in the results presented by Mao for NH₃-free layers [11], but the decrease in deposition rate in that case is stronger. We believe that this difference in the strength of the effect is due to the fact that there was an excess of N species in all the recipes that were tested in this study.

The deposition rate doubles for a 5 times increase in RF power. There is no precedent on the obtained trend between the deposition rate and RF power in NH₃-free films. Nevertheless, the behaviour is similar to the one described by several authors in both PECVD SiN_x and SiON films [18, 26, 27]. The deposition rate increases because the higher RF power provides higher plasma energy that results in larger concentrations of active species that raise the incorporation of Si and N atoms into the films making them become thicker in shorter time. This last result is contrary to the trend observed by Mao in NH₃-free films [11], but the discrepancy can be explained by the use of a different reactor and RF powers that are at least one order of magnitude higher than the ones used in this study.

Finally, the deposition rate increases by about 30% when the pressure is increased 1.5 times. As with the RF power, there is no precedent in this result but is similar to the effect observed in PECVD SiN_x [26] and PECVD SiON layers [27]. The higher deposition rates achieved with higher pressures are related to the increase in residence time that occurs on the precursor gases with increasing total pressure. For instance, the high dissociation efficiency of SiH₄ leads to the creation of more Si radicals when the residence time is longer, achieving an effect similar to the one obtained with higher RF powers that increases the deposition rate.

In general, high deposition rates are desirable to speed up the fabrication process and reduce the cost of photonic industrial applications. In this case, we achieved deposition rates as high as 20 nm min⁻¹. This deposition rate is about one order of magnitude higher than the typical rate for LPCVD SiN_x films, but still not higher than the rates observed in standard PECVD SiN_x films. As a result, NH₃-free PECVD SiN_x layers can not be considered as an alternative for improving the deposition time during fabrication. They can only be seen as an alternative for obtaining lower propagation losses without reducing the deposition rate that is already being used with standard PECVD layers.

Thickness uniformity is important in order to achieve consistent dimensions over large substrate areas when fabricating photonic devices for commercial applications. In this study we quantified the thickness uniformity of NH₃-free PECVD SiN_x layers deposited on 6'' wafers in terms of their non-uniformity percentage, which is defined by expression (6) [28]:

$$\begin{eqnarray}&&\text{non}-\text{uniformity}\left(\% \right)=\frac{{{t}_{\text{max}}}-{{t}_{\text{min}}}}{2{{t}_{\text{mean}}}}\end{eqnarray} \tag{ 6 }$$

where t_max is the maximum thickness, t_min the minimum thickness and t_mean the mean thickness measured in a 6'' wafer.

The experiments show that the non-uniformity percentage of the NH₃-free PECVD SiN_x layers has only a clear relationship with the RF power and pressure used during deposition. Figure 3 indicates that the non-uniformity percentage increases with increasing RF power at the same time that it decreases with increasing pressure. This result is consistent with the observation of other authors when depositing SiN_x with the standard PECVD recipe [29, 30].

Zoom In Zoom Out Reset image size

The large thickness non-uniformity observed at higher RF powers is mainly due to the high growth rates achieved under these conditions, as they cause a depletion of precursors that allows intermediate active species to stream over the substrate. This effect can be boosted by the lower stability that the plasma has when the RF power is high as pointed out by Wei et al [30]. Nevertheless, the thickness non-uniformity of the layers can be reduced by increasing the pressure regardless of the RF power because there is a better gas distribution uniformity during deposition at high pressures. We observed that the non-uniformity drastically reduces when the pressure is increased to 950 mTorr and therefore we opted to use this value for our fabrication process. In fact, we were able to reduce the overall wafer non-uniformity of the layers deposited at 650 mTorr from 15% to values as low as 3% for nearly stoichiometric layers and 1% for Si-rich layers as illustrated in figure 4.

Zoom In Zoom Out Reset image size

In most cases, the thickness non-uniformity of the NH₃-free PECVD layers is slightly higher than that of standard PECVD SiN_x films which is typically  <2.5% [29]. Yet, we have demonstrated that NH₃-free PECVD layers with relatively low non-uniformity can be achieved when the chamber pressure is high. For similar deposition parameters, the non-uniformity of the NH₃-free PECVD layers is comparable to that of standard PECVD films.

Our study suggests that, even though the rest of the optical and structural properties of the NH₃-free SiN_x layers seem altered by the deposition conditions, the observed changes only reflect the effect that the deposition parameters have on the N/Si ratio of the films. In fact, our experiments revealed that the only structural parameter that can be modified directly with the deposition conditions is the N/Si ratio of the films. This dependence is important because it gives the ability of tuning the characteristics of the SiN_x layers for specific applications. It is just a matter of using the deposition parameters to change the composition of the layers to obtain either N-rich, Si-rich or stoichiometric layers.

The experiments conducted in this study show that the N/Si ratio can be tuned between 0.6 and 1.7 by modifying either the SiH₄ flow, N₂ flow, RF power or chamber pressure used for deposition as illustrated in figure 5. In this case, the N/Si ratio decreases when either the SiH₄ flow or the pressure are increased, while it increases when the N₂ flow or the RF power are augmented. In all cases, it can be observed that the depletion of reactants causes a saturation trend in the N/Si ratio of the films. Nevertheless, the RF power is the parameter that has the strongest effect on the N/Si ratio of the films.

Zoom In Zoom Out Reset image size

The N/Si ratio decreases by 30% when the SiH₄ flow is doubled. This tendency was previously observed by other authors [18, 31] and indicates a higher incorporation of Si into the films due to the increased availability of Si species when the SiH₄ flow is high. On the contrary, the N/Si ratio slightly increases when the N₂ flow increases due to the reduction of the Si species available for deposition. This effect is equivalent to a reduction of the SiH₄ flow similar to the one observed in the deposition rates of the films.

The N/Si ratio doubles when the RF is increased by 5 times. This behaviour can be associated to the higher energies needed to dissociate N₂ molecules compared to SiH₄ molecules. N-rich materials are produced at high RF powers because there is enough energy available to increase the concentration of N radicals that are incorporated in the form of N bonds. This result is in line with the observations of Karouta when depositing PECVD SiN_x layers [18], but it is contrary to the behaviour described by Han et al [31]. However, Han's observations can be explained by their use of a different reactor and considerably higher RF powers that result in a fundamentally different experiment. In general, the results suggest that a RF power around 60 W is appropriate to obtain N/Si ratios that can be tuned by changing the SiH₄ and N₂ flow rates, as the N/Si ratio tends to saturate when the RF power is higher.

Finally, the N/Si ratio decreases slightly when the chamber pressure is increased. This behaviour is related to the increased residence time of the precursor gases at high pressures, because more SiH₄ is dissociated and incorporated in the films resulting in Si-rich compositions. This result is similar to what was observed by Lee et al on SiN_x films deposited by ICP-CVD [32].

The refractive index of the SiN_x layers is an important parameter for integrated photonic applications because it affects how the light is confined and propagated in the material. The experiments performed in the study show that the refractive index of the NH₃-free SiN_x layers is strongly related to their N/Si ratio regardless of the deposition parameters. For instance, the refractive index of the films can be tuned between 1.8 and 2.6 by changing their N/Si ratio from 0.6 to 1.7 as illustrated in figure 6. SiN_x layers with N/Si  >  1.33 have a N-rich structure that yields a lower refractive index useful for sensing, while SiN_x layers with N/Si  <  1.33 have a Si-rich composition that results in a relatively higher refractive index ideal for applications that require a higher confinement or higher non-linearity.

Zoom In Zoom Out Reset image size

A typical SiN_x FTIR spectrum, as the one shown in figure 7, reveals the presence of four major absorption peaks that can be used to extract the bonding configuration of a SiN_x film. The peak around 450 cm⁻¹ corresponds to the absorption of Si–Si bonds, the peak at 860 cm⁻¹ to Si-N bonds, and, finally, the peaks at 2100 and 3300 cm⁻¹ represent the absorption of Si-H and N-H bonds respectively [18, 33]. Both Si–Si and N-H covalent bonds are relevant to understand the propagation losses of SiN_x films. Si–Si bonds are important defects in the SiN_x matrix that lead to poor optical properties [17], while N-H bonds act as absorption centres that lead to undesirable losses in the 1510–1650 nm region [11]. As a result, the effect of both structural faults and their combination can produce high propagation losses in SiN_x films.

Zoom In Zoom Out Reset image size

The experiments show that the different bond configurations of the NH₃-free SiN_x layers have a more intrinsic relation with the N/Si ratio than with the deposition conditions. Hence, the following results are expressed in terms of the N/Si ratio which can explain the observed behaviours in a more significant way than the deposition parameters.

Figure 8 shows the Si–Si and Si-N bond concentrations for different N/Si composition ratios. The concentration of the Si–Si bonds decreases with increasing N/Si ratios, which means they are dominant in Si-rich films and practically non-existent in N-rich films. This happens because the number of back-bonded Si atoms decreases in favour of N bonds that can be either present in the form of Si-N or N-H. These results are consistent with what has been reported in the literature for standard PECVD films [23, 34]. In contrast, the concentration of the Si-N bonds increases with the N/Si ratio until it reaches its maximum near N/Si $\approx$ 1.33 and then decreases when the N/Si ratio is increased further. The maximum value coincides with the stoichiometric value for the N/Si ratio which is where the Si and N atoms are balanced out.

Zoom In Zoom Out Reset image size

Figure 9 shows the Si-H and N-H bond concentrations as a function of the N/Si ratio of the deposited films. In this case, the Si-H concentration decreases with increasing N/Si ratio,while the N-H concentration increases. This happens due to the competing nature of the formation of both types of bonds [34] and the availability of more Si or N radicals with the different N/Si ratios. In consequence, Si-rich films tend to have more Si-H bonds and N-rich films more N-H bonds.

Zoom In Zoom Out Reset image size

The relative H concentration in atomic % (H%) can be determined from the bond concentrations calculated from the FTIR spectrum. This value is of the uttermost importance because it has been demonstrated that films with lower H concentrations tend to have lower propagation losses [11]. The results in figure 10 suggest that the H% decreases with increasing N/Si ratio until it reaches its minimum value near N/Si $\approx$ 1.33 and then increases back again. As a result, both Si-rich and N-rich films have a higher H% than stoichiometric films, which have the minimum H concentration and the maximum Si-N incorporation as demonstrated previously. Moreover, the results indicate that the lowest H% concentration obtained with the NH₃-free recipe is 14.8%. This value is almost 10% lower than the H% typically observed in SiN_x films deposited with the standard recipe [18]. This reduction in H% can only be attributed to the decreased Si-H and N-H absorbance peaks obtained with the NH₃-free recipe as illustrated in figure 7. Hence, it is expected to achieve lower propagation losses with the NH₃-free approach.

Zoom In Zoom Out Reset image size

The roughness is another important material parameter because it is one of the main sources of scattering losses that impact negatively the propagation losses of the deposited layers [21]. Figure 11 shows the RMS roughness measurements obtained from PECVD SiN_x layers deposited on Si substrates using the standard and the NH₃-free recipe with RF  =  60 W and $\text{P}=650$ mTorr. The RMS roughness of NH₃-free layers ranges between 1.5 and 2.5 nm whereas that of standard PECVD layers tends to be below 1.5 nm. In general, the study suggests that the RMS roughness of the NH₃-free films tends to be higher compared to that of the standard SiN_x layers.

Zoom In Zoom Out Reset image size

Materials with low propagation losses are needed to improve the performance of passive devices in integrated photonic circuits. As a result, it is desirable to understand how the deposition parameters and the structure of the deposited films influence the propagation losses of the NH₃-free SiN_x layers. The results obtained in this study suggest that no direct relation can be established between the varied deposition parameters and the measured propagation losses. However, in all cases the losses at 1310 and 1550 nm are intrinsically correlated with the N/Si ratio of the films as illustrated in figure 12. This means that the propagation losses of the deposited layers are primarily a consequence of the structural properties of the films.

Zoom In Zoom Out Reset image size

The propagation losses at 1310 nm decrease with increasing N/Si ratio in such a way that N-rich films have lower propagation losses at this wavelength. On the other hand, the propagation losses at 1550 nm follow a trend similar to that of the H concentration of the films. They decrease with increasing N/Si ratio until they reach a minimum value and then increase back again as the films become more N-rich. We believe that these behaviours suggest that the propagation losses at 1310 nm are the product of the Rayleigh scattering inherent to the microstructure of the deposited materials, whereas the propagation losses at 1550 nm are mostly the product of the absorption losses produced by the N-H bonds incorporated into the films. In this case, the propagation losses of Si-rich films are higher at 1310 nm because they have a microstructure with more defects due to the incorporation of both Si-H and Si–Si dangling bonds, while their losses at 1550 nm are lower because of the absence of N-H bonds in their structure. Alternatively, the propagation losses of N-rich films are higher at 1550 nm because of the presence of N-rich bonds in their structure that act as absorption centres at that particular wavelength. As a result, the lowest propagation losses are obtained simultaneously at both wavelengths with films that have a N/Si ratio close to stoichiometry because at this point the presence of H bonds is minimised and the microstructure defects produced by the Si dangling bonds are low.

The lowest propagation losses obtained are of 2 dB cm⁻¹@1310 nm with N-rich layers and 5 dB cm⁻¹@1550 nm with nearly stoichiometric films. Under this circumstances, the propagation losses at 1550 nm are still high compared to the losses of almost 4 dB cm⁻¹@1550 nm measured on PECVD SiN_x layers by Huang et al [12], even though the H content of NH₃-free PECVD layers is almost 10% lower. This result suggests that the propagation losses of the NH₃-free SiN_x layers are not only affected by the microstructure and the H bonds of the material, but also by other external factors.

One of the factors that can potentially increase the propagation losses of the SiN_x strip waveguides is the roughness of the SiO₂ used to isolate them from the Si substrate. In fact, any surface roughness at the bottom or top of the films can lead to scattering that will increase considerably the propagation losses of the deposited materials. The PECVD SiO₂ layers used in the previous experiments had a RMS roughness of almost 4 nm which is considerably higher than the RMS roughness of the Si substrate (0.2 nm) that was used on the roughness experiments. To rule out the effect of the SiO₂ roughness, we changed the underlying layers for thermal SiO₂ with a roughness below 2 nm. Figure 13 shows the propagation losses obtained for NH₃-free PECVD SiN_x layers with different N/Si ratios grown on the commercial SiO₂. In this case, the propagation losses at both wavelengths have a behaviour respect to their N/Si ratio similar to the one observed in the films grown on the PECVD SiO₂. However, the propagation losses at both wavelengths are reduced to values below 2 dB cm⁻¹. In fact, we achieved propagation losses around 1.5 dB cm⁻¹@1550 nm and below 1 dB cm⁻¹@1310 nm with NH₃-free PECVD SiN_x layers just by improving the roughness and quality of the underlying SiO₂.

Zoom In Zoom Out Reset image size

Figure 14 shows the propagation losses obtained in this study compared to losses observed in the literature. The propagation losses of the NH₃-free films are lower than those obtained with similar PECVD layers that have not been subjected to high temperatures. In this case, the propagation losses at 1310 nm are almost the same as for the standard layers, while the losses at 1550 nm are more than 2 dB cm⁻¹ lower for the NH₃-free layers. This result is significant because it shows that lower losses at 1550 nm are possible with the NH₃-free films because of their lower H concentration. Hence, lower losses can potentially be achieved with this approach if the deposition conditions are tuned to obtain more stoichiometric films with even lower H concentration and by improving further the quality of the SiO₂ used.

Zoom In Zoom Out Reset image size