Improved Silicon Surface Passivation by ALD Al2O3/SiO2 Multilayers with In‐Situ Plasma Treatments

Al2O3 is one of the most effective dielectric surface passivation layers for silicon solar cells, but recent studies indicate that there is still room for improvement. Instead of a single layer, multilayers of only a few nanometers thickness offer the possibility to tailor material properties on a nanometer scale. In this study, the effect of various plasma treatments performed at different stages during the ALD deposition of Al2O3/SiO2 multilayers on the silicon surface passivation quality is evaluated. Significant improvements in surface passivation quality for some plasma treatments are observed, particularly for single Al2O3/SiO2 bilayers treated with a H2 plasma after SiO2 deposition. This treatment resulted in a surface recombination parameter J0 as low as 0.35 fA cm−2 on (100) surfaces of 10 Ω cm n‐type silicon, more than a factor of 5 lower than that of Al2O3 single layers without plasma treatment. Capacitance‐voltage measurements indicate that the improved surface passivation of the plasma‐treated samples results from an enhanced chemical interface passivation rather than an improved field effect. In addition, a superior temperature stability of the surface passivation quality is found for various plasma‐treated multilayers.


Improved Silicon Surface Passivation by ALD Al 2 O 3 /SiO 2 Multilayers with In-Situ Plasma Treatments
Armin Richter,* Hemangi Patel, Christian Reichel, Jan Benick, and Stefan W. Glunz DOI: 10.1002/admi.202202469 total fixed negative charge density Q tot of ≈1 × 10 13 cm −2 in combination with a low interface defect density D it of ≈1 × 10 11 eV −1 cm −2 . [4][5][6][7][8][9] While the low D it represents a rather good chemical surface passivation, the high negative Q tot causes a reduction of the electron density at the surface, which results in an important field effect contribution to the c-Si surface passivation. Thus, this high negative Q tot induces an inversion layer on n-type Si surfaces, while an accumulation layer is formed on p-type surfaces. The inversion layer at n-type Si surfaces makes its application prone to parasitic shunting effects at n-type metal contacts. [10] Therefore, Al 2 O 3 is predominantly applied to p-type c-Si surfaces, such as the rear surface of passivated emitter and rear cell (PERC) passivated emitter and rear cell solar cells -the current mainstream cell design in high-volume production [11,12] -or the front-side boron-doped p + emitter of n-type c-Si tunneling oxide passivating contact (TOPCon) solar cells, which are becoming currently increasingly attractive due to their higher efficiency potential. [11,[13][14][15] Al 2 O 3 is also very interesting for advanced cell designs with efficiencies in the range of 26%, such as the rear emitter (TOPCon) cell [16] or the polycrystalline Si on oxide-interdigitated back contact (POLO-IBC) cell, [17] where a highly effective but transparent passivation of the bare (undiffused) p-type c-Si front surface is required.
Recently, a direct comparison of different surface passivation schemes indicated that there is still room for improvement for Al 2 O 3 [3] which becomes increasingly important as device performance improves. In contrast to single layers, multilayers with thicknesses of only a few nanometers of the individual layers open the opportunity to modify material properties on a nanometer scale. One interesting example are the so-called interface dipole layers, which are currently intensively investigated especially for the application in metal-oxide-semiconductor field-effect transistor (MOSFETs) to adjust the desired flat-band voltage. [18][19][20] They are multilayers consisting of two or three different dielectric layers and can provide the possibility of increasing the flat-band voltage simply by varying the number of the bi-or trilayers. The origin of this flat-band voltage shift are dipoles, which are formed only at specific interfaces of this multilayer with only one polarity. For instance, SiO 2 /Al 2 O 3 stacks have been reported, where dipoles are formed only at the SiO 2 /Al 2 O 3 interfaces with one polarity but not at the Al 2 O 3 / SiO 2 interfaces with the opposite polarity. [19] Al 2 O 3 is one of the most effective dielectric surface passivation layers for silicon solar cells, but recent studies indicate that there is still room for improvement. Instead of a single layer, multilayers of only a few nanometers thickness offer the possibility to tailor material properties on a nanometer scale. In this study, the effect of various plasma treatments performed at different stages during the ALD deposition of Al 2 O 3 /SiO 2 multilayers on the silicon surface passivation quality is evaluated. Significant improvements in surface passivation quality for some plasma treatments are observed, particularly for single Al 2 O 3 /SiO 2 bilayers treated with a H 2 plasma after SiO 2 deposition. This treatment resulted in a surface recombination parameter J 0 as low as 0.35 fA cm −2 on (100) surfaces of 10 Ω cm n-type silicon, more than a factor of 5 lower than that of Al 2 O 3 single layers without plasma treatment. Capacitance-voltage measurements indicate that the improved surface passivation of the plasma-treated samples results from an enhanced chemical interface passivation rather than an improved field effect. In addition, a superior temperature stability of the surface passivation quality is found for various plasma-treated multilayers.

Introduction
The electronic passivation of crystalline silicon (c-Si) surfaces with dielectric layers is highly important for various semiconductor devices, in particular for c-Si solar cells. One of the most prominent and effective dielectric passivation layers is Al 2 O 3 , which has been intensively investigated and highly optimized. [1][2][3] The effectiveness of Al 2 O 3 is attributed to its high www.advmatinterfaces.de In a previous study, we investigated the c-Si surface passivation quality of such Al 2 O 3 /SiO 2 multilayers synthesized via atomic layer deposition (ALD) [21] -a deposition technique with a high relevance for high-volume production. [22] In that study, we did not find enhanced surface passivation quality of these layers when compared to Al 2 O 3 single layers, nor evidence of dipole effects in these layers, which would enhance the field-effect passivation. Instead, we found that the negative Q tot of the ALD Al 2 O 3 /SiO 2 multilayers could be quite substantially increased upon voltage stress. However, the increased negative Q tot did not result in an improved surface passivation quality because the chemical passivation degraded upon voltage stress. [21] In this work, we study systematically the effect of plasma treatments during the ALD multilayer deposition on the surface passivation quality of these Al 2 O 3 /SiO 2 multilayers. In general, the effect of plasma treatments on the surface passivation quality of dielectric layers has hardly been studied so far, in particular for layers prepared by ALD. Therefore, we studied the effect of different plasma treatments (H 2 , N 2 O 2 , N 2 /H 2, or Ar), with specific focus on whether they can induce a modification of the Al 2 O 3 /SiO 2 or SiO 2 /Al 2 O 3 interfaces so that interface dipoles are formed analogously to Ref., [19] and which might potentially enhance the field-effect passivation. Instead of starting these multilayers with SiO 2 at the c-Si interface, as usually done for MOSFET applications, [18][19][20] we studied in this work multilayers starting with Al 2 O 3 , which we found to be beneficial for c-Si surface passivation with our ALD layers. [23]

Results and Discussion
In the first experiment, we studied the influence of plasma treatments on the surface passivation quality for single and triple Al 2 O 3 /SiO 2 multilayers, i.e., c-Si/(Al 2 O 3 /SiO 2 ) n with n = 1 and n = 3, which were deposited with plasma-assisted ALD (PA-ALD). For the activation mainly of the chemical interface passivation, a post-deposition forming gas anneal was performed at temperatures between 350 and 550 °C. During this activation, the c-Si/Al 2 O 3 interface is hydrogenated with hydrogen mainly released from the Al 2 O 3 . [1,25] Our approach to provide similar hydrogen sources for all different multilayers was to deposit a 10 nm thick Al 2 O 3 layer on top of the multilayer stacks. This 10 nm thick Al 2 O 3 layer provides a quite good passivation and therefore also served as a reference passivation scheme in this study. The plasma treatments consisting either of pure H 2 , N 2 O 2 , or Ar or a mixture of N 2 /H 2 were conducted either i) after every Al 2 O 3 deposition, or ii) after every SiO 2 deposition, or iii) after both, as schematically illustrated in Figure 1 exemplarily for the single Al 2 O 3 /SiO 2 multilayers. The resulting effective minority charge carrier lifetime τ eff as a measure for the surface passivation quality is shown in Figure 2 as a function of the post-deposition annealing temperature T an ; on the left for the single and on the right for the triple Al 2 O 3 /SiO 2 multilayers, where the top row shows the results for plasma treatments after every Al 2 O 3 deposition, the middle row for plasma treatments after every SiO 2 deposition and the bottom row for plasma treatments after both, Al 2 O 3 and SiO 2 depositions. In addition, the results for multilayers and Al 2 O 3 single layers without plasma treatment are shown as a reference (gray data points). The τ eff data is evaluated at an excess charge carrier density of Δn = 1 × 10 15 cm −3 , which represents charge carrier injection conditions close to the maximum power point condition of highefficiency silicon solar cells, [13] , i.e., this τ eff assesses the surface passivation quality close to operation conditions of real devices.
At annealing temperatures below 450-500 °C, τ eff of all multilayers is strongly increasing with increasing annealing temperatures, indicating a strongly improving surface passivation. This trend is consistent with thermal activation results of PA-ALD Al 2 O 3 single layers [29] and can be mainly attributed to the thermal activation of the chemical surface passivation, i.e., the reduction of interface defect states. [25,30] At higher annealing temperatures, τ eff of most multilayers starts to saturate and the difference in passivation quality of the various plasma-treated samples becomes visible. It can be clearly observed that there are plasma treatments which result in a significantly higher τ eff compared to the reference multilayers without plasma treatment, indicating improved surface passivation and thus, a beneficial effect of the plasma treatments. These effects are most pronounced for the single Al 2 O 3 /SiO 2 multilayers, while significantly weaker effects are obtained for the triple Al 2 O 3 /SiO 2 multilayers. Almost all of these plasma-treated triple Al 2 O 3 /SiO 2 multilayers show a maximum τ eff level, which is not significantly higher than that of the reference Al 2 O 3 single layer without plasma treatment. There is one clear exception, which is the triple Al 2 O 3 /SiO 2 multilayer treated with H 2 plasma after both the Al 2 O 3 and SiO 2 depositions (Figure 2f), which exceeds τ eff of the Al 2 O 3 single layer reference by about 40%. In contrast, there are various plasma-treated single Al 2 O 3 /SiO 2 multilayers which show an even higher τ eff , which is up to 65% higher than that of the Al 2 O 3 single layer reference, as observed, for instance, for the O 2 plasma treatment after SiO 2 deposition (Figure 2c). The lower performance of the plasma-treated triple Al 2 O 3 /SiO 2 multilayers might to some extent be attributed to their overall lower surface passivation quality even without plasma treatments, which is almost a factor of 2 lower than that of the single Al 2 O 3 /SiO 2 multilayers without plasma treatments.

www.advmatinterfaces.de
It can be also observed from Figure 2 that most plasma-treated samples show superior high-temperature stability. While τ eff of the reference multilayers and the single Al 2 O 3 layer without plasma treatments saturates at T an = 450-500 °C and even tends to degrade slightly above 500 °C, τ eff of various plasma-treated multilayers are at 550 °C not even in the saturation regime.
With respect to the different plasma treatments being studied, it is found that almost every plasma can result in a beneficial effect. It appears, however, to be important at which interface the respective plasma treatment is performed. For instance, a N 2 /H 2 plasma is most beneficial when applied after the Al 2 O 3 deposition (Figure 2a,b), while a H 2 plasma is very beneficial when applied after both, the Al 2 O 3 and the SiO 2 deposition (Figure 2e,f). For plasma treatments after SiO 2 deposition, the O 2 plasma, the H 2 plasma, and even the N 2 plasma show quite substantial positive effects in the case of single Al 2 O 3 /SiO 2 multilayers (Figure 2c). Interestingly, the smallest positive or even detrimental effects are in almost any case observed for Ar plasma treatments. This indicates that the beneficial effects of the other plasma treatments might be related to chemical modifications taking place at the plasma-treated interfaces, rather than ion bombardment or UV irradiation-induced effects. In our ALD reactor with its remote plasma source, the kinetic energy of the ions was found to be low enough to prevent substantial film damage. [31][32][33] In the case of the H 2 containing plasma treatments even an incorporation of hydrogen into the layer might take place. Interestingly, we found that the surface passivation quality of 15 nm thick PA-ALD Al 2 O 3 single layers is not significantly modified when exposed to a post-deposition plasma treatment (see Figure S5, Supporting Information), which indicates that the beneficial effects observed here are a special property of these PA-ALD Al 2 O 3 /SiO 2 multilayers.
In order to study the underlying mechanism, we measured in a second experiment the interface properties using the capacitance-voltage (CV) technique. In this experiment, we studied the same variations as in the previous experiment, except that the multilayers were activated with a single annealing process at 450 °C for 25 min (instead of the 10 min used in the previous experiment), which is more compatible with our TOPCon solar cell processing. [14,16] Figure 3 displays the total fixed charge  The measured τ eff clearly confirms the beneficial effect of various plasma treatments when compared to the reference multilayers without plasma treatment (black data points in Figure 3a,b), in particular for the single Al 2 O 3 /SiO 2 multilayers. With respect to the Al 2 O 3 single layer reference (gray data points), these beneficial effects are, however, less pronounced than in the previous experiment. However, in the previous experiment, the beneficial effect was most pronounced at the highest annealing temperatures, while they were also less pronounced at the lower annealing temperature of 450 °C used in this experiment. A detailed comparison of the surface passivation qualities observed in both experiments indicate for some variations slightly different trends. The fact that the trends are reasonably reproduced in the next experiment (see the comparison in Figure S1, Supporting Information) indicates that the annealing conditions in particular the annealing duration is causing these differences.
With respect to the fixed charge density, the single Al 2 O 3 / SiO 2 multilayers tend to have a higher negative Q tot than the triple Al 2 O 3 /SiO 2 multilayers, which contribute to their overall higher surface passivation quality as observed from their higher τ eff . Interestingly, the observed Q tot of all plasma-treated multilayers is significantly lower than that of the Al 2 O 3 and Al 2 O 3 / SiO 2 multilayer reference samples without plasma treatment, in particular with respect to the Al 2 O 3 single layer, which reveals that they have a lower field-effect component contributing to the surface passivation. The fact that various plasma-treated multilayers display at the same time a superior surface passivation, indicates that they have a chemical passivation component which is significantly improved. However, exact D it values could not be extracted from our CV data due to the low D it level in combination with a limited measurement accuracy. Thus, this finding supports the hypothesis that, especially in the case of the H 2 containing plasma treatments, hydrogen is incorporated into the layers and/or at the c-Si/Al 2 O 3 interface, which then contributes to the chemical surface passivation during the thermal activation anneal. [34] Such H 2 plasma-induced silicon interface hydrogenation effects have also been reported, for instance, for amorphous silicon [35] or SiO 2 [36,37] layers. The mechanism behind the beneficial O 2 or N 2 plasma treatments is yet to be determined and requires further analysis. One hypothesis might be that the plasma treatments result in more mobile hydrogen during the post-deposition thermal annealing, where the interface hydrogenation takes place. [25,30] The fact that the plasma-treated multilayers show a lower Q tot means also that the plasma treatments did not result in a formation of interface dipoles which enhance the field-effect passivation. Instead, this might indicate that the plasma treatment induces interface dipoles of the opposite polarity reducing thereby the apparent Q tot , as speculated by Irikawa et al. for their Q tot reduction observed for c-Si/Al 2 O 3 interfaces after H 2 plasma exposure. [38] Alternative explanations for the Q tot reduction might be that the reduction of interface defects inherently results also in a reduction of trapped charges at the interface, or that oxide charging effects take place upon UV irradiation during plasma treatments, [39] although for UV radiation of the c-Si/Al 2 O 3 interface an increase of Q tot is expected. [40,41] So far, the plasma exposure time was constant at 10 s. In a third experiment, we studied the influence of the plasma exposure time for H 2 , O 2, and N 2 plasma treatments. Figure 4 displays the result of this experiment where the 10 s exposure is compared to a 60 s exposure for single Al 2 O 3 /SiO 2 multilayers annealed at either at 450 or 500 °C. The experiment confirms again that a H 2 plasma treatment shows the most significant improvement, in particular when applied after the SiO 2 deposition. It can also be clearly observed that there are plasma treatments for which the longer plasma exposure time results in increased τ eff , i.e., further improved surface passivation. This is particularly the case for the H 2 plasma after SiO 2 deposition, where the 60 s plasma results in ≈40% increased τ eff . With respect to the annealing temperature, it is interesting that for 10 s plasma treatments annealing at 500 °C results in superior passivation quality for all variations consistently with the results of the first experiment, while for the 60 s plasma treatments annealing at 500 °C is only beneficial if the plasma treatment is performed after SiO 2 deposition. Thus, the highest τ eff of 23.8 ms is observed for 60 s H 2 plasma treatment after SiO 2 deposition annealed at 500 °C. This translates to a surface recombination parameter J 0 of 0.35 ± 0.34 fA/cm 2 for this sample while the Al 2 O 3 reference with a max. τ eff of 13.2 ms has a corresponding J 0 of 2.09 ± 0.90 fA/cm 2 (see also Figure S2, Supporting Information and Table S1), which highlights the quite substantially improved surface passivation.
However, for other plasma treatments such as the O 2 or the N 2 plasma after Al 2 O 3 deposition, τ eff of the extended www.advmatinterfaces.de 60 s exposure is actually lower than after 10 s exposure. In the case of the 60 s N 2 plasma treatment, τ eff is even significantly lower than that of the Al 2 O 3 /SiO 2 reference multilayer without plasma treatment, which indicates that this long plasma treatment can also degrade the surface passivation quality. Such plasma-induced degradation effects can be caused by the high energy of the vacuum UV photons, as reported, for instance, for O 2 plasma treatments. [4,32] The fact that these detrimental effects are most pronounced when the plasma treatment was performed after Al 2 O 3 deposition (see, e.g., Figure 4: 60 s N 2 plasma), i.e., when it took place very close to the c-Si interface (see Figure 1), might, however, also indicate that some kind of plasma-induced damage occurs even if a remote plasma source with low ion energies is used.
Since the H 2 plasma treatment showed the strongest surface passivation improvement after extended plasma exposure, we studied the effect of plasma duration for the H 2 plasma treatment in more detail. Figure 5 shows the resulting τ eff and Q tot of a systematic plasma duration variation between 10 s and 120 s again for single Al 2 O 3 /SiO 2 multilayers annealed at 450 °C. It can be clearly observed that for the plasma treatment after SiO 2 deposition, as well as after Al 2 O 3 and SiO 2 deposition, there is a pronounced maximum τ eff for a plasma exposure time ≈60 s to 80 s. The positive effect of plasma treatments after Al 2 O 3 deposition seems to saturate after 60 s at a slightly lower τ eff . The Q tot determined for these samples is not significantly affected by the plasma exposure time, indicating again that the improved surface passivation being observed by the maximum τ eff is caused by an improved chemical passivation, i.e., interface hydrogenation, rather than by improved field-effect passivation.
So far, the surface passivation was studied on planar surfaces. However, the surfaces of c-Si solar cells are often textured with random pyramids in particular on the front surface for improved light absorption. Therefore, we studied in a further experiment the effect of H 2 plasma treatments for single Al 2 O 3 / SiO 2 multilayers on textured surfaces. Figure 6 summarizes the resulting τ eff of this experiment, where in addition to the plasma duration (10 s/60 s), the ALD deposition temperature (180 °C/250 °C) was also varied. It can be observed that for the multilayers deposited at 250 °C, the 60 s plasma treatment after SiO 2 resulted again in the most significantly improved surface passivation, as indicated by a τ eff which is almost a factor of 4 higher than that of the Al 2 O 3 /SiO 2 multilayer reference without plasma treatment, and almost a factor of 2 higher than that of the Al 2 O 3 reference. The samples deposited at 180 °C are much less affected by the plasma treatments. Interestingly, the 10 s plasma treatments did not show any effect on the surface passivation of these textured samples, neither after SiO 2 nor after Al 2 O 3 deposition, indicating that on textured surfaces with larger surface area longer plasma exposure times are required. The same variations have also been studied on p-type FZ Si wafers (1 Ω cm, 250 µm thick) with textured surfaces, where very similar trends have been obtained (see Supplementary Figure S4). This shows that the enhanced surface passivation is not only obtained for n-type Si surfaces but also for p-type Si surfaces.

Conclusion
We have studied the effect of interface plasma treatments during the ALD deposition of Al 2 O 3 /SiO 2 multilayers on their  www.advmatinterfaces.de c-Si surface passivation properties. By varying, for instance, the plasma gas (H 2 , O 2 , N 2, or Ar), the interfaces where the plasma treatment took place, or the number of Al 2 O 3 /SiO 2 bilayers, we identified plasma treatments resulting in a significant improvement in the surface passivation quality in different independent experiments. These positive effects were found to be most pronounced for a single Al 2 O 3 /SiO 2 bilayer, where a 60 s H 2 plasma treatment was applied after SiO 2 deposition. This treatment resulted in J 0 values as low as 0.35 fA cm −2 which is more than a factor of 5 lower than that of the Al 2 O 3 single layer reference without plasma treatment. This is among the lowest J 0 values reported in the literature. [1] Interestingly, such improved surface passivation upon plasma treatments was not observed for the Al 2 O 3 references, indicating that the positive effect strongly depends on the type of layer system and the interface, at which the treatment is applied during film deposition.
CV measurements indicate that the positive effect of the plasma-treated multilayers is not caused by an increased field-effect passivation. In fact, their fixed charge density was found to be significantly lower than that of the Al 2 O 3 reference without plasma treatment, indicating an even lower field-effect contribution. Instead, the CV results indicate an improved chemical surface passivation for multilayers with plasma treatments, which supports the hypothesis that in the case of the H 2 plasma-treated samples hydrogen is incorporated into the film and/or at the c-Si/Al 2 O 3 interface, which can then contribute to the interface hydrogenation during the post-deposition thermal anneal. In addition, we found indications that the plasma-treated multilayers show a superior high-temperature stability. Although the multilayers were deposited in a lab-type PA-ALD reactor, they are of high relevance for mass production as large-batch PA-ALD reactors are available. [22] As such, these plasma-treated dielectric multilayers with superior c-Si surface passivation properties identified in this study might pave the path to new surface passivation schemes for high-efficiency c-Si solar cells with a high transparency as usually required for the front side.

Experimental Section
The surface passivation of Al 2 O 3 single layers and Al 2 O 3 /SiO 2 multilayer stacks was mainly studied on shiny-etched (100)-oriented n-type FZ Si wafers with a resistivity of 10 Ω cm and a thickness of 200 µm. Some of the wafers were textured with random pyramids in an alkaline solution to study the surface passivation quality on surfaces similar to those usually utilized at the front surface of silicon solar cells. These textured samples were made of (100)-oriented n-type FZ Si wafers with a resistivity of 1 Ω cm and a thickness of 200 µm. First, the wafers were subjected to a standard RCA cleaning procedure, followed by a thermal oxidation at 1050 °C. This high-temperature process was performed to remove the recombination-activity of the common FZ defects. [24] Prior to the multilayer deposition, the thermally-grown SiO 2 was etched in buffered oxide etching solution. The Al 2 O 3 /SiO 2 bilayers were then deposited with plasma-assisted ALD (PA-ALD) using trimethylaluminum (TMAl) and bis-diethyl aminosilane (BDEASi) as aluminum and silicon precursors, respectively. The deposition was performed in a single wafer reactor (FlexAL, Oxford Instruments) with a remotely placed inductively coupled plasma source operated at a frequency of 13.56 MHz. Two different multilayer variations were prepared: single and triple sequences of Al 2 O 3 /SiO 2 , i.e., c-Si/(Al 2 O 3 /SiO 2 ) n with n = 1 and n = 3. If not stated otherwise, the samples were deposited at 250 °C and with fixed Al 2 O 3 and SiO 2 layer thicknesses of 3 and 2 nm respectively, which were found to result in a promising surface passivation. [21] Some of the samples were treated with a plasma during the deposition, either i) after every Al 2 O 3 deposition, or ii) after every SiO 2 deposition, or iii) after both, as schematically illustrated in Figure 1 exemplarily for the single Al 2 O 3 /SiO 2 multilayers. The plasma consists either of pure H 2 , N 2 O 2 , or Ar or a mixture of N 2 /H 2 and was performed at 150 W for 10 s, if not stated otherwise. The gas flow rates during plasma treatments were either 30 sccm H 2 , 50 sccm N 2 , 60 sccm O 2 , 60 sccm Ar or 15 sccm/15 sccm N 2 /H 2 . The pressure during these treatments was around 50 mTorr. The deposition of the complete multilayer stack, including the plasma treatments, took place without breaking the vacuum. After deposition, the samples were annealed in forming gas (N 2 /H 2 mixture) in a tube furnace (from ASM).
Photoconductance decay (PCD) measurements were performed with the lifetime tester WCT-120 (Sinton Instruments) to determine the effective minority charge carrier lifetime (τ eff ), using either the transient mode or the quasi-steady-state mode (with generalized analysis). [26,27] For PCD measurements, symmetrically coated (passivated) samples were prepared. All τ eff values were evaluated at an excess charge carrier density of Δn = 1 × 10 15 cm −3 . The surface recombination parameter J 0 was determined via modeling the full injection-dependent effective lifetime data τ eff as described in Ref., [42] considering also the intrinsic bulk recombination model from Niewelt et al.. [28] This J 0 fitting was performed in a Δn range from 6 × 10 14 cm −3 to 8 × 10 15 cm −3 . The J 0 values were evaluated for a temperature of 25 °C taking the effective intrinsic charge carrier concentration n i,eff as a function of Δn according to Altermatt et al. [44] (with the equilibrium intrinsic charge carrier concentration n i,0 = 8.29 × 10 9 cm −3 at 25 °C [45] and band gap narrowing according to Schenk [46] ) into account. In addition to the PCD measurements, we also performed τ eff calibrated photo-luminescence imaging (PLI) using the Fraunhofer ISE modulum setup. [43] Reasonable homogeneity was observed in the PLI data, thus a dominating random effect on the observed trends in surface passivation quality due to inhomogeneity issues can be excluded. Exemplary PLI data is shown in Figures S6 and S7, Supporting Information. Metal-insulator-semiconductor (MIS) structures were prepared for capacitance-voltage (CV) measurements by thermally evaporating Al metal dots (≈1 mm diameter) on the multilayer surface and full area Al contact (≈300 nm thick) on the rear side. The CV measurements were performed with the semiconductor analyzer B1500A (Agilent Technologies) to determine Q tot , which was evaluated from the flat band voltage and the maximum capacitance of the Al 2 O 3 layer or Al 2 O 3 /SiO 2 multilayer stacks was attained by the low-frequency curve of the samples. The low-frequency curve was measured at 1 kHz and the high-frequency curve at 1 MHz.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.