The catalytic effect of iron(iii) on the etching of ZnO:Al front

Transcription

1 The catalytic effect of iron(iii) on the etching of ZnO:Al front contacts for thin film silicon solar cells Sascha E. Pust*, Janine Worbs, Gabrielle Jost, Jürgen Hüpkes Forschungszentrum Jülich GmbH, IEK5 Photovoltaik, Jülich, Germany * Corresponding author: Dr. Sascha E. Pust s.pust@fz-juelich.de Phone: Fax:

2 Abstract Sputter-deposited ZnO:Al thin films, used as front contact in thin-film Si solar cells, were etched in diluted HCl containing catalytic amounts of Fe(III) salts. The Fe(III) effectively catalyzes the HCl-based etch process, leading to a crater-like morphology that is qualitatively similar to the one generated by the uncatalyzed etching process at the same HCl concentrations. Utilizing this catalyzed process, an increase of the etching rate by a factor of approximately was observed. This allows for a well-controllable tuning of the etch duration without changing HCl concentration or temperature. The process has been evaluated with a selection of Fe(III) salts at different concentrations of the acid and of the catalyst. Optical, electrical and scanning force microscopic characterization of such catalytically etched ZnO:Al films has shown that the catalytical process leads to slightly smaller morphological features on the film compared to the uncatalyzed etching, accompanied by a shift of light scattering intensity to higher angles. This etching behavior may be used beneficially for light trapping in thin-film Si solar cells. In addition to these applicationoriented aspects, this approach provides a deeper insight into the mechanistical details of the ZnO thin-film dissolution. Keywords: Catalysis; Etching; Thin-film Si solar cells; Transparent conductive oxide (TCO); ZnO thin films 2

3 1. Introduction The front contact in thin-film Si solar cells has to fulfil three criteria: (i) it has to be highly transparent, (ii) a low series resistance is necessary, and (iii) it has to provide a certain roughness for light management in the device, thus helping to trap the light within the absorber layer stack by total internal reflection. The first two aspects are given for many doped, transparent conductive oxides (TCO) like SnO 2 :F, In 2 O 3 :SnO 2, ZnO:Al, ZnO:B, or ZnO:Ga [1-6]. The latter criterion, however, necessitates further texturing in the case of some TCO materials that are smooth in the as-deposited state. Sputter-deposited, polycrystalline ZnO:Al, for example, may be artificially roughened by an etching step in diluted HCl [7-10] (e.g., leading to the 'standard Jülich' type material [11, 12]), HF [13], NH 4 Cl [14-16], other acids in liquid [7-10, 17, 18], gaseous [19], or vaporized state [20], by electrochemical means [21], or by a combination of such processes [13, 22-26]. The variety of possible etching methods together with a model for the etching of polycrystalline ZnO:Al has been addressed in a recent review [27]. Besides the texture, one parameter of significant industrial relevance is the duration of the etching step. For inclusion of such a process into in-line production systems, adjustable etch times are desirable. This can be done to a certain extent by changing the concentration of the etchant or the temperature. However, the resulting morphology is changed significantly by such alterations as well [22, 28]. Readily applicable methods for the adjustment of the etch time are hence not at hand. One general option for accelerating chemical reactions is the utilization of catalytic processes. It is well known that catalysis is a concept of utmost relevance for chemical industry because of its ability to initiate, accelerate, or steer chemical reactions. Although the exact nature of the interfacial processes during acidic etching of ZnO:Al thin films is still not completely understood [27], it can be expected that this process is accelerated by suitable, catalytically active agents, similarly to what has been known for In 2 O 3 :SnO 2 thin films since many years [29]. In this study, we evaluated the potential of Fe(III) salts to catalyze and accelerate the etching of sputter-deposited ZnO:Al thin films in diluted HCl. A catalytic mechanism is proposed that is strongly supported by experimental evidence. Electrical measurements, angle-resolved scattering (ARS), and scanning force microscopy (SFM) were used to quantify 3

4 the surface morphologies resulting from the catalyzed etching. This new process renders very suitable morphologies for light scattering at increased etch rates, and their suitability for thinfilm Si photovoltaic devices is demonstrated. 2. Experimental 2.1 ZnO:Al thin-film preparation Approximately nm thick, polycrystalline ZnO:Al films were deposited on a (10 10) cm 2 glass substrate (Corning Eagle XG) using radio frequency magnetron sputtering in a vertical in-line system (cf. Fig. 1 in Ref. [30]) from a ceramic target consisting of ZnO with 1 w/w% Al 2 O 3 (Cerac Inc., Milwaukee, WI, USA). The deposition was carried out at a substrate temperature of 300 C, a discharge power density of 2 W cm -2, and an Ar pressure of 0.1 Pa. Details about the process and ZnO:Al film properties are reported elsewhere [12]. 2.2 Etching process and chemicals All solutions were prepared with deionised water ( >16 M cm) from an Elix 10 water purification system (Millipore Co., Schwalbach, Germany). HCl, FeCl 3, Fe(NO 3 ) 3, K 3 [Fe(CN) 6 ], FeCl 2 (all Merck KGaA, Darmstadt, Germany), and Fe 2 (SO 4 ) 3 (Sigma-Aldrich Chemie GmbH, Munich, Germany) were of analytical grade and were used without further purification. The temperature was kept constant at C during all etching experiments. After etching, substrates were rinsed with copious amounts of deionized water to remove salt residues originating from the etching solution. Special care has to be taken when an acid is added to a compound containing hexacyanidoferrate because very toxic HCN may be generated. Such experiments therefore have to be carried out with proper personal protective equipment in a closed fume hood and with a very small amount of CN-containing compounds only. 2.3 Film characterization Film thicknesses d have been measured with a Dektak 3030 surface profiler (Veeco Instruments Ltd., Santa Barbara, CA, USA), taking into account those points where the films have got the highest thickness. To correct for inhomogeneities in the initial measured film 4

5 thickness, this initial thickness has been determined by taking the maximum measured d value for each etch series as initial value instead of taking the measured values at each individual etching spot. SFM images have been recorded utilizing a NANOStation 300 (S.I.S. GmbH, Herzogenrath, Germany) in non-contact mode with NCHR-POINTPROBE Si cantilevers (NanoWorld AG, Neuchâtel, Switzerland). Image post-treatment was performed with the software SPIP, version (Image Metrology A/S, Hørsholm, Denmark). A linear background was subtracted from the images for removal of the sample tilt. A four-point probe was utilized for determining the sheet resistance R sh of the films. ARS measurements in transmission were performed in a home-built system at an optical wavelength of =550 nm. 2.4 Solar cell preparation and characterization Roughly 1 µm thick µc-si:h films for single junction solar cells were deposited in a (40 40) cm 2 in-line plasma-enhanced chemical vapor deposition (PECVD) system [31]. Si films for 420 nm a-si:h / 1.25 µm µc-si:h tandem p-i-n solar cells were prepared by PECVD in a (30 30) cm 2 reactor. Details of the Si PECVD process can be found elsewhere [32, 33]. The back contact consisted of sputter-deposited ZnO:Al and Ag from the same deposition system as the front contact films. A lab-scale laser patterning system (ROFIN-SINAR Laser GmbH, Hamburg, Germany) was used to determine a cell area of (1 1) cm 2. Solar cells were characterized with a Wacom WXS 140 S solar simulator (Wacom Electric Co., Saitama, Japan) under standard test conditions (AM1.5 spectrum, 100 mw cm -2, 25 C). External quantum efficiencies EQE were measured in a home-built system by differential spectral response (DSR) at zero bias voltage. The crystallinity of the Si absorber was determined from the back side of the cells via Raman measurements with an excitation wavelength of 647 nm which leads to information depths roughly of the thickness of the µc-si:h absorber. In the case of the tandem devices, the measured crystallinity of the bottom component cell also influenced by the top component cell. This is due to the penetration depth of the Raman RS I C RS I C,bottom is measurement and leads to an underestimation in RS I C, bottom compared to the real crystallinity of the bottom component cell. To access the Si, the ZnO:Al/Ag back contact has been removed at selected spots close to the measured cells with a wet-etching process. Details about the Raman setup [34] and the determination of the crystallinity [35] can be found in the respective references. 5

6 3. Results and discussion 3.1 Mechanistical considerations Van den Meerakker et al. [29] used Fe(III) species to catalyze the etching of In 2 O 3 :SnO 2 thin films in halogen acids. Taking into account the nature of the wurtzite-type ZnO:Al material, their proposed mechanism for the uncatalyzed [36] as well as for the catalyzed etching of In 2 O 3 :SnO 2 [29] may be adapted (Fig. 1). When etching ZnO in an aqueous, diluted HCl solution (Fig. 1a), a Zn-O-Zn bond (1) is attacked by an HCl moiety. This leads to the formation of one Zn-OH and one Zn-Cl bond. The resulting surface intermediate (2) reacts with a second HCl moiety in a condensation reaction to form two Zn-Cl bonds (3) and H 2 O. This process is repeated, yielding ZnCl 2 and H 2 O as the reaction products. Fig. 1. Proposed mechanism for the acidic dissolution of ZnO (a) without catalyst and (b) with Fe 3+ as a catalytically active species. If Fe 3+ is present in addition to HCl (Fig. 1b), the Fe 3+ ion interacts with one electron pair of the oxygen atom in 2, leading to a weakening of the Zn-O bond (2a). A chloride ion from an HCl moiety is then able to attack the Zn atom. This leads to a heterolytic fission of the Zn-O bond, yielding 3 and H 2 O, releasing the catalytically active Fe 3+ ion and making it available for further catalysis. The final reaction products are hence the same as in the uncatalyzed reaction pathway (Fig. 1a). 6

7 Van den Meerakker et al. [29] have shown for In 2 O 3 :SnO 2 etching that in the catalyzed case (Fig. 1b), the reaction of 1 to 2 is the rate-determining step, while the conversion of 2 to 3 is limiting the reaction rate if the process is not catalyzed (Fig. 1a). This means that, in the case of acidic ZnO etching, an increase in the etch rate can be expected when adding Fe 3+ to the etching solution if the mechanism of In 2 O 3 :SnO 2 etching [29, 36] is directly transferable to the ZnO:Al thin film. 3.2 Initial tests Before applying Fe(III) salts as catalysts in the acidic etching process of ZnO:Al, the immediate effect of those salts on the thin film had to be evaluated because aqueous solutions of FeCl 3, Fe(NO 3 ) 3, Fe 2 (SO 4 ) 3, or similar are slightly acidic by themselves. For example, taking into account that Fe(III) salts will exist in solution mainly as hydroxo and aqua complexes with acid dissociation constants pk a in the range of [37], a 0.5 w/w% solution of FeCl 3 will have a ph of approximately 2. This is about one ph unit higher than the value of an equally concentrated HCl solution. This means that etching in a diluted FeCl 3 solution should proceed slower than in similarly concentrated HCl if a possible catalytic effect of the Fe 3+ ions is not too pronounced. To check this assumption, ZnO:Al substrates have been etched for 40 s in FeCl 3 solutions of 0.5, 0.05, and w/w%, respectively, without addition of HCl. While the standard Jülich method with 0.5 w/w% HCl proceeds at etch rates of a few nm s -1 (depending on the material properties), the etch rates in FeCl 3 -containing solutions are indeed much lower: the 0.5 w/w% solution of FeCl 3 (corresponding to mol L -1 ) resulted in an etch rate of roughly 0.9 nm s -1. This value can be ascribed partly to the ph of approximately 2, and partly to a catalytic effect of the Fe 3+ in this slightly acidic solution. Almost no removal (in the range of few nm min -1 ) was observed for solutions with an even lower concentration. On the one hand, this meets the expectations concerning the acidic strength of Fe 3+ salts. On the other hand, it is an important prerequisite for further experiments as this observation implies that the addition of catalytic amounts of Fe 3+ to an HCl solution will be insignificant in terms of a ph decrease. Hence, Fe 3+ would not increase the etch rate in an HCl-based etching significantly if no catalytic reaction takes place. 7

8 3.3 Quantification of the catalytic effect To quantify the catalytic effect of Fe 3+, several etch experiments on ZnO:Al substrates were performed. While the HCl concentration was kept constant in all experiments at 0.5 w/w% (0.137 mol L -1 ), different catalytic amounts of iron salts were added to the HCl, namely 1, 5, and 10 mol% with respect to the HCl concentration. Expressed as absolute iron salt concentrations, these values correspond to 1.37, 6.85, and 13.7 mmol L -1, respectively. The iron salt concentrations are hence at least about one order of magnitude smaller than in the previously elucidated, initial experiments. FeCl 3 was utilized as source for Fe 3+. Additionally, a cross-check was performed with HCl solutions containing FeCl 2 as a salt that does not contain Fe 3+, and K 3 [Fe(CN) 6 ] where the Fe 3+ is tightly bound as the metal center of a stable complex. The films were etched for 15, 30, 40, and 50 s with each catalyst concentration and each iron salt. In addition to that, one reference substrate has been etched in 0.5 w/w% HCl without any iron compound to compare the catalyzed with the uncatalyzed process. The first visual impression was that all etch attempts lead to more or less pronounced, hazy ZnO:Al surfaces. All etching solutions rendered visually homogeneous etch results. To quantify the amount of material that is etched away in a given time, one may take into account the film thickness d before and after etching, or the change in R sh that will increase when the film becomes thinner and more rough. Both parameters are good indicators for the change that is introduced to the film via etching, but the change in film thickness d represents the amount of removed material more exactly. Mainly due to the fact that the film is rough after etching, the profilometric measurements will be afflicted with a certain overestimation of d the rougher the films get [28]. For these experiments, we took into account only those points where the films have got the highest thickness (namely crater rims instead of etch pits) in the profilometric measurements to avoid a significant error originating from this overestimation. Considering variations in the initial film thickness, we estimated the overall error for d to be roughly 25 nm. The change in sheet resistance compared to the value before etching ( R sh ) has additionally been used to quantify the etch rates as the sheet resistance measurement is straightforward and may advantageously be used in terms of process control. The according R sh values are presented in the supporting material (Fig. S1). 8

9 Fig. 2. Decrease in film thickness d as a function of etch time t for etching in 0.5 w/w% HCl containing different concentrations of (a) FeCl 3, (b) FeCl 2, (c) K 3 [Fe(CN) 6 ]. As a reference, d of a substrate etched in HCl without addition of a catalyst is plotted in gray. An estimated measurement error for d of ±25 nm is included. Lines are solely meant to guide the eye. 9

10 Indeed, an acceleration occurs if FeCl 3 is used as a source for Fe 3+. As Fig. 2a clearly shows, the increase in d with etch time is higher compared to the reference (gray) whenever FeCl 3 is present. In all cases, the acceleration by far exceeds the result of an added-up acidity of the Fe(III) salt and the HCl. A catalytic effect of the Fe 3+ on the acidic dissolution of ZnO (Fig. 1b) is thus clearly evident. In general, the extent of acceleration seems to be dependent on the concentration of the catalyst (higher concentration leads to slightly stronger acceleration) as well as on the anion in the respective Fe(III) salt. While addition of FeCl 3 leads to a moderate, but well controllable acceleration of the etch process (Fig. 2a), the addition of Fe(NO 3 ) 3 or Fe 2 (SO 4 ) 3 accelerates the etching to such an extent that the ZnO film suffers an electrical breakdown after approximately 40 s of etching (not shown). Obviously, the presence of sulfate and nitrate ions contributes more strongly to the acceleration than chloride. A similar effect, especially concerning the sulfate influence, has been observed earlier in the context of electrochemical etching of ZnO (cf. Fig. 3b in Ref. [25]), and that has been ascribed to the strong interference of sulfate ions with Zn precipitation [38]. It is, however, unclear why the same, pronounced acceleration is observed in the presence of nitrate ions, as this was not observed in the aforementioned studies [25]. The effects when using nitrate or sulfate salts may also be attributed to solubility issues concerning the reaction products. A cross-check experiment has been performed with FeCl 2 where no Fe 3+ is available in the etch solution. An acceleration should thus not occur in this case. The differences in d between etch attempts with and without the iron salt are indeed negligible and within the measurement error (Fig. 2b). This is a clear evidence that Fe 2+ cannot influence the ZnO dissolution mechanism as postulated for Fe 3+ (Fig. 1b), most probably due to its lower oxidizing potential compared to Fe 3+. In the case of K 3 [Fe(CN) 6 ] where Fe 3+ is bound inside an octaedric complex ion, an acceleration indeed seems to occur at the utilized catalyst concentrations (Fig. 2c). The degree of acceleration is similar to that observed under the influence of FeCl 3. Although no free Fe 3+ should be available, the etch mechanism seems to be influenced by the presence of the hexacyanidoferrate complex. We assume that, due to the presence of HCl in the same solution, significant amounts of the complex are chemically destroyed. This makes Fe 3+ available in its ionic form to catalyze the etching process, ultimately leading to an increase in etch rate that is similar to that under the influence of FeCl 3 due to the presence of chloride ions originating from the HCl addition. Further experiments have shown that the catalytic effect of Fe(III) is present at higher HCl concentrations than 0.5 w/w%, although the reaction becomes less controllable as etch 10

11 times for generating a feasible etch result are in the range of only a few seconds then. Concentrations as low as approximately 0.1 w/w% are possible as well. For example, while the etching of a substrate in 0.1 w/w% HCl for 40 s resulted in R sh =2.1, the etch time to reach a similar value ( R sh =1.9 ) could be reduced to 10 s when adding 5 mol% FeCl 3 to the etchant. In both cases, approximately 85 nm of the film were removed. Below that concentration, the ph value of the solution is shifted into a range where many Fe(III) salts are not soluble to an appropriate extent, making a catalytic acceleration impossible. Over the whole utilizable concentration range of HCl, however, the extent of acceleration is similar to the previously elucidated extents. This results in an overall acceleration factor (concerning d) of approximately for the application of FeCl 3, mostly depending on the catalyst concentration. It has been furthermore observed that the catalytic effect is not limited to HCl. Similar results have been achieved with other acids (e.g. HNO 3 or H 2 SO 4 ) at ph values of about 1 (not shown), although the degree of acceleration is different for all utilized compounds. 3.4 Influence on surface morphology Besides the advantage of an accelerated etching, the resulting layers have to have suitable morphologies for light management in thin-film Si solar cells. Concerning sputter-deposited ZnO:Al films, it is well known that the standard Jülich reference material [11] is well suited for this purpose, and this has also been confirmed by optical simulations [39]. It is hence desirable to keep the morphology after etching similar, even if the process is accelerated catalytically. This is an important issue, as Owen et al. [28] have pointed out recently that the etching of sputter-deposited ZnO:Al thin films at different HCl concentrations and different etchant temperatures influences not only the etch rate, but also the crater size after etching. To check for this issue, SFM images have been recorded of substrates etched with and without catalytic influence of FeCl 3 to a similar d. Figure 3 shows representative micrographs of one reference substrate etched without catalytic influence (Fig. 3a) and of one substrate etched shorter down to a similar film thickness under catalytic influence of FeCl 3 (Fig. 3b). Both etched films show qualitatively similar interfacial morphologies with craters in the range of less than 1 µm in diameter. In general, the catalyst obviously does not seem to alter the fundamental etch mechanism that has been postulated for such polycrystalline ZnO thin films [27]. However, the morphology of both films does differ quantitatively: in the catalytically etched case (Fig. 3b), medium crater diameters are lower compared to the 11

12 reference substrate (Fig. 3a), while the medium crater depths and the root mean square (RMS) roughnesses are similar. The respective quantitative values are given in Table I. Fig. 3. (10 10) µm 2 scanning force micrographs of ZnO:Al thin films etched in (a) 0.5 w/w% HCl for 50 s and (b) 0.5 w/w% HCl with 5 mol% FeCl 3 for 30 s. Table I. Statistical evaluation of the SFM-derived topographies of etched ZnO:Al thin films (cf. Fig. 3). The errors given are the standard deviations derived from the statistical evaluation of the crater dimensions. medium crater diameter [nm] medium crater depth [nm] RMS roughness [nm] etching conditions 50 s in 0.5 w/w% HCl (reference) 887 ± ± s in 0.5 w/w% HCl + 5 mol% FeCl ± ± Interestingly, this trend to smaller craters is also observed when etching without catalytic influence at elevated temperatures, i.e., higher etch rates, or lower concentrations of the acid, i.e., lower etch rates [28]. This implies that the morphology is not directly related to the etch rate, but the catalyst seems to lower the etching threshold for grain boundaries with a certain etch potential [27], leading to an increased crater density and, as a consequence, a decreased medium crater diameter. Hence, this new method is another tool in addition to the established procedures basing on different acids and different etching conditions that allows us to tune the size of surface features on etched ZnO:Al. 12

13 While SFM is a local technique and can deliver representative topographic informations of a microscopic fraction of the ZnO:Al surface, optical methods are suited to derive more integral informations about the substrate on an area of about 1 mm 2. Hence, we performed ARS measurements in transmission at =550 nm (Fig. 4). While, in general, longer wavelengths (>600 nm) are more relevant for scattering in thin-film Si solar cells, a of 550 nm has proven to resolve even small surface features properly. This is not the case any more if red lasers are used as a light source. =550 nm is thus a compromise between an appropriate sensitivity for small features and a relevance for light scattering. Fig. 4. ARS data of ZnO:Al thin films etched for different durations in 0.5 w/w% HCl with 5 mol% FeCl 3. The angular intensity distribution of a substrate etched in HCl without addition of a catalyst is plotted in gray (dashed line). It had been reported earlier that ARS is a powerful tool for gaining a characteristic surface fingerprint of etched ZnO:Al thin films [40-42]. Furthermore, the scattering intensity to higher angles in the range of roughly 60 to the substrate normal correlates with the shortcircuit current density J sc in thin-film Si solar cells, especially µc-si:h single junction devices [42]. The data presented in Fig. 4 show that the scattering is shifted from lower to higher angles when utilizing an etching process based on the catalytical effect of FeCl 3. It is known that smaller features in general lead to a scattering into higher angles, while bigger features scatter predominantly into lower angles [42]. Thus, the ARS data are coherent with the SFM-based observation of smaller surface features compared to the standard Jülich reference (Fig. 3 and Table I). A positive influence on J sc could hence be expected for µc-si:h single junction solar cells on these catalytically etched films. 13

14 3.5 Evaluation in solar cells To check the catalytically etched films in terms of suitability for an application as front contact, we co-deposited µc-si:h p-i-n single junction solar cells onto films etched for different durations in 0.5 w/w% HCl containing 5 mol% FeCl 3. As a reference, a standard Jülich-type ZnO etched for 40 s in 0.5 w/w% HCl was processed in the same deposition run. Note that the utilized ZnO:Al thin films are the very same substrates that have been characterized with ARS (Fig. 4). Table II shows the photovoltaic parameters of the prepared solar cells, namely the initial conversion efficiency init, the fill factor FF, the open-circuit voltage V oc, and the short-circuit current density J sc. Table II. Photovoltaic parameters and Raman crystallinity of µc-si:h p-i-n single junction solar cells on etched ZnO:Al thin films. The errors given are the standard errors of the mean of all measured, intact cells on one substrate. RS I C etchant without catalyst (reference) with 5 mol% FeCl 3 etch duration [s] init [%] FF [%] V oc [mv] J sc [ma cm -2 ] RS I C [%] ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Looking just at the overall efficiencies, the reference cells perform best in this comparison, but the solar cells with a front contact etched catalytically deliver very similar parameters, especially for longer etch durations. Increased etch times go along with an increase in V oc and FF. We assume that the growth of the microcrystalline Si is responsible for the V oc increase. The degree of crystallinity that is achieved during growth of µc-si:h strongly depends on factors such as the material properties and the topography of the underlying front contact. Table II shows that of the Si absorber in the reference cell is at 59.5% and decreases from a similar value at 15 s of FeCl 3 -aided etching to 55.7 and 55.3% at 25 und 30 s, respectively. A decrease in crystallinity means an increase in the band gap of the Si absorber that defines the theoretical limit of V oc [43]. Hence, the increase in V oc with etch time is coherent with the measured RS I C RS I C values. The reasons for this may be (i) the smaller feature sizes on the catalytically etched substrates and (ii) presumably residues of the iron salt or its reaction products at the ZnO:Al surface that may influence the Si growth. In J sc, there is an optimum after an etch duration of 25 s, being within the standard error of the reference's 14

15 J sc. The cells after 30 s of FeCl 3 -catalyzed etching deliver the same init as the reference due to a gain in V oc of roughly 10 mv. Fig. 5. (a) EQE of representative µc-si:h p-i-n single junction solar cells deposited on a ZnO:Al front contact textured in 0.5 w/w% HCl with 5 mol% FeCl 3 for different etch times. (b) Individual EQE, including the individual J sc values, of the two component cells in a representative a-si:h/µc-si:h tandem p-i-n solar cell deposited on a ZnO:Al front contact textured in 0.5 w/w% HCl with 5 mol% FeCl 3 for 30 s. The sums of both component cells, being the overall EQE, are plotted as dashed lines. The EQE of reference cells from the same absorber deposition processes on substrates etched in HCl without addition of a catalyst are plotted in gray. The EQE data of representative cells from this series (Fig. 5a) confirm the observation that the reference cell (gray curve) as well as the cell with a front contact etched with addition of FeCl 3 for 25 s (red curve) perform similarly, especially at >600 nm. Especially the 15

16 variation in J sc for different etch times are almost within an error range that can occur just due to inhomogeneities in the deposition. Nevertheless, there are small variations in the EQE for different wavelength regions as a function of the etch time. Etching for only 15 s (black curve) leads to a loss in the EQE over the whole spectral range compared to the reference, while longer etching for 30 s (blue curve) results in a higher EQE at small wavelengths, but a slight deterioration above 600 nm. On the one hand, this can be correlated to the tendency of the cells to become more amorphous with increasing etch time (cf. V oc and in Table II). On the other hand, the ZnO:Al thickness and the development of the morphology of course plays a role as well. 15 s FeCl 3 -aided etching (black curve) means thick ZnO:Al and only slight structure development, leading to losses over the whole spectral range. At 25 s (red curve), the film is thinner and thus lets more light pass into the cell (short wavelength gain), and the structures are further developed (long wavelength gain due to light trapping). 30 s of etching (blue curve) mean that even more light passes through the very thin ZnO:Al (stronger gain at <600 nm), but the craters reach the glass substrate and thus the structure depth is reduced again (loss at long wavelengths due to deterioration of light trapping). RS I C Table III. Photovoltaic parameters and Raman crystallinity RS I C,bottom of a-si:h/µc-si:h tandem p-i-n solar cells on etched ZnO:Al thin films (cf. Fig. 3). The errors given are the standard errors of the mean of all measured, intact cells on one substrate. Pretreatment conditions init [%] FF [%] V oc [mv] J sc [ma cm -2 ] RS I C,bottom [%] 40 s in 0.5 w/w% HCl (reference) 10.5 ± ± ± ± s in 0.5 w/w% HCl + 5 mol% FeCl ± ± ± ± As the FeCl 3 -based process has proven its suitability to etch ZnO:Al thin films for µc-si:h single junction devices, we also prepared a-si:h/µc-si:h tandem p-i-n solar cells on FeCl 3 -etched ZnO:Al (Table III). This cell concept bears the advantage that the a-si:h top component cell (which is deposited directly onto the ZnO:Al thin film) is less sensitive in its growth than µc-si:h, meaning that the aforementioned crystallinity changes should not occur in the tandem cell concept. In this specific cell configuration, a representative reference cell deposited on ZnO:Al etched without influence of a catalyst delivered short-circuit current densities for the top (J sc,top ) and bottom (J sc,bottom ) component cell of 10.9 and 11.7 ma cm -2, respectively (Fig. 5b, gray curve). By adding FeCl 3 to the etchant, we were able to improve the light-trapping efficiency in both the top and bottom component cell which is seen as an 16

17 increase in J sc,top of 0.2 ma cm -2 and in J sc,bottom of 0.3 ma cm -2 (Fig. 5b, black curve), improving init compared to the reference as well. This improvement is at the cost of a loss of roughly 30 mv in V oc and at similar crystallinities of the bottom component cell absorber (Table III). In addition, µc-si:h single junction as well as a-si:h/µc-si:h tandem p-i-n solar cells have been prepared on some subtrates etched under the contribution of other Fe(III) salts. Those cell results are collected in the supporting material (Tables SI and SII, Figs. S2 and S3) and are generally in agreement with the expectations from the film characterization experiments. 4. Conclusions and outlook We have shown that the acidic etching of polycrystalline, sputter-deposited ZnO:Al thin films may be accelerated efficiently and in a well-controllable manner by adding catalytic amounts of an Fe(III) salt to the etchant. The Fe(III) acts as a catalyst in the etching mechanism and alters it in favor of a faster overall reaction, leading to an approximate doubling of the etch rate. The best and most homogeneous results have been achieved when utilizing FeCl 3 as Fe(III) source. As a result of the catalytic influence of Fe(III), the resulting morphology of the etched ZnO:Al thin film is altered towards slightly smaller features compared to non-catalytically etched substrates. This leads to a shift of the light scattering at the film/air interface to larger angles which can be advantageous for the light trapping in a thin-film Si solar cell if such ZnO:Al thin films are used as the front contact. The degree of acceleration as well as the extent of the change in morphology may be controlled by changing the catalyst concentration or the concentration of the etchant. Thus, we have got a method at hand to decrease the etch duration of an industrially relevant etching process with no adverse or, in some cases, even beneficial effects on the light trapping in the device. Such an additional process parameter is of significant practical importance, e.g., for the adjustment of process step times in an in-line production system, and it upvalues the industrial applicability of the well-established ZnO etching process. 17

18 Acknowledgments The authors thank Mengfei Wu for ARS measurements, Simone Bugdol, Joachim Kirchhoff, Daniel Weigand, and Thomas Zimmermann for assistance in solar cell deposition and characterization, Markus Hülsbeck for Raman measurements as well as Nicole Lühmann, Matthias Meier, and Eerke Bunte (all Forschungszentrum Jülich GmbH) for fruitful discussions. Financial support by the German Federal Environment Ministry (BMU, grants A and A) is gratefully acknowledged. 18

19 References [1] K. Sato, Y. Gotoh, Y. Wakayama, Y. Hayashi, K. Adachi, H. Nishimura, Highly Textured SnO 2 :F TCO Films for a-si Solar Cells, Rep. Res. Lab. Asahi Glass Co., Ltd., 42 (1992) [2] J.K. Rath, Y. Liu, M.M. de Jong, J. de Wild, J.A. Schuttauf, M. Brinza, R.E.I. Schropp, Transparent conducting oxide layers for thin film silicon solar cells, Thin Solid Films, 518 (2010) e129-e135. [3] H.N. Wanka, M.B. Schubert, E. Lotter, Growth of a-si:h on transparent conductive oxides for solar cell applications, Sol. Energy Mater. Sol. Cells, (1996) [4] J. Hüpkes, J. Müller, B. Rech, Texture Etched ZnO:Al for Silicon Thin Film Solar Cells, in: K. Ellmer, A. Klein, B. Rech (Eds.) Transparent Conductive Zinc Oxide: Basics and Applications in Thin Film Solar Cells, Springer, Berlin, 2008, pp [5] J. Meier, J. Spitznagel, U. Kroll, C. Bucher, S. Faÿ, T. Moriarty, A. Shah, Potential of amorphous and microcystalline silicon solar cells, Thin Solid Films, (2004) [6] V. Bhosle, A. Tiwari, J. Narayan, Electrical properties of transparent and conducting Ga doped ZnO, J. Appl. Phys., 100 (2006) [7] B.-H. Seo, S.-H. Lee, J.H. Seo, J.-H. Jeon, H. Choe, Study on the wet etch behavior of a zinc-oxide semiconductor in acid solutions, J. Korean Phys. Soc., 53 (2008) [8] D.-G. Yoo, S.-H. Nam, M.H. Kim, S.H. Jeong, H.-G. Jee, H.J. Lee, N.-E. Lee, B.Y. Hong, Y.J. Kim, D. Jung, J.-H. Boo, Fabrication of the ZnO thin films using wetchemical etching processes on application for organic light emitting diode (OLED) devices, Surf. Coat. Technol., 202 (2008) [9] W.T. Yen, Y.C. Lin, J.H. Ke, Surface textured ZnO:Al thin films by pulsed DC magnetron sputtering for thin film solar cells applications, Appl. Surf. Sci., 257 (2010) [10] W.-L. Lu, K.-C. Huang, C.-H. Yeh, C.-I. Hung, M.-P. Houng, Investigation of textured Al-doped ZnO thin films using chemical wet-etching methods, Mater. Chem. Phys., 127 (2011) [11] M. Kubon, E. Boehmer, F. Siebke, B. Rech, C. Beneking, H. Wagner, Solution of the ZnO/p contact problem in a-si:h solar cells, Sol. Energy Mater. Sol. Cells, (1996)

20 [12] M. Berginski, J. Hüpkes, M. Schulte, G. Schöpe, H. Stiebig, B. Rech, M. Wuttig, The effect of front ZnO:Al surface texture and optical transparency on efficient light trapping in silicon thin-film solar cells, J. Appl. Phys., 101 (2007) [13] J.I. Owen, J. Hüpkes, H. Zhu, E. Bunte, S.E. Pust, Novel etch process to tune crater size on magnetron sputtered ZnO:Al, Phys. Status Solidi A, 208 (2011) [14] J. Sun, J. Bian, H. Liang, J. Zhao, L. Hu, Z. Zhao, W. Liu, G. Du, Realization of controllable etching for ZnO film by NH 4 Cl aqueous solution and its influence on optical and electrical properties, Appl. Surf. Sci., 253 (2007) [15] S. Fernández, O. de Abril, F.B. Naranjo, J.J. Gandía, Etching process optimization using NH 4 Cl aqueous solution to texture ZnO:Al films for efficient light trapping in flexible thin film solar cells, Thin Solid Films, 520 (2012) [16] S. Fernández, S.E. Pust, J. Hüpkes, F.B. Naranjo, Development of two-step etching approach for aluminum doped zinc oxide using a combination of standard HCl and NH 4 Cl etch steps, Thin Solid Films, 520 (2012) [17] Y.C. Lin, Y.C. Jian, J.H. Jiang, A study on the wet etching behavior of AZO (ZnO:Al) transparent conducting film, Appl. Surf. Sci., 254 (2008) [18] N. Kumar, G. Zhang, N. Kalyankar, M.A. Nguyen, J. Li, H. Yang, Z.-W. Sun, High productivity combinatorial study of wet chemical texture etch of sputter deposited Aldoped ZnO thin films for thin film Si solar cells, Photovoltaics International, 9 (2010) [19] S.R. Droes, T.T. Kodas, M.J. Hampden-Smith, Etching of ZnO films with Hexafluoroacetylacetone, Adv. Mater. (Weinheim, Ger.), 10 (1998) [20] T. Kawaharamura, T. Hirao, Development and Research on the Mechanism of Novel Mist Etching Method for Oxide Thin Films, Jpn. J. Appl. Phys., 51 (2012) [21] S.O. Klemm, S.E. Pust, A.W. Hassel, J. Hüpkes, K.J.J. Mayrhofer, Electrochemical texturing of Al-doped ZnO thin films for photovoltaic applications, J. Solid State Electrochem., 16 (2012) [22] J. Hüpkes, J.I. Owen, E. Bunte, H. Zhu, S.E. Pust, J. Worbs, G. Jost, New Texture Etching of Zinc Oxide: Tunable Light Trapping for Si Thin Film Solar Cells, in: Proceedings of the 25th European Photovoltaic Solar Energy Conference, Valencia, Spain, WIP, Munich, 2010, pp [23] J.I. Owen, J. Hüpkes, E. Bunte, S.E. Pust, A. Gordijn, Etching Modulated Surface Textures into Sputtered ZnO:Al Films, in: Proceedings of the 25th European 20

21 Photovoltaic Solar Energy Conference, Valencia, Spain, WIP, Munich, 2010, pp [24] E. Bunte, H. Zhu, J. Hüpkes, J. Owen, Novel texturing method for sputtered zinc oxide films prepared at high deposition rate from ceramic tube targets, EPJ Photovolt., 2 (2011) [25] S.E. Pust, J.-P. Becker, J. Worbs, S.O. Klemm, K.J.J. Mayrhofer, J. Hüpkes, Electrochemical Etching of Zinc Oxide for Silicon Thin Film Solar Cell Applications, J. Electrochem. Soc., 158 (2011) D413-D419. [26] H. Zhu, J. Hüpkes, E. Bunte, J. Owen, S.M. Huang, Novel etching method on high rate ZnO:Al thin films reactively sputtered from dual tube metallic targets for silicon-based solar cells, Sol. Energy Mater. Sol. Cells, 95 (2011) [27] J. Hüpkes, J.I. Owen, S.E. Pust, E. Bunte, Chemical Etching of Zinc Oxide for Thin- Film Silicon Solar Cells, ChemPhysChem, 13 (2012) [28] J.I. Owen, S.E. Pust, E. Bunte, J. Hüpkes, ZnO Etch-Feature Control via Concentration and Temperature of Various Acids, ECS J. Solid State Sci. Technol., 1 (2012) P11-P17. [29] J.E.A.M. van den Meerakker, P.C. Baarslag, M. Scholten, On the Mechanism of ITO Etching in Halogen Acids: The Influence of Oxidizing Agents, J. Electrochem. Soc., 142 (1995) [30] J. Hüpkes, H. Zhu, J.I. Owen, G. Jost, E. Bunte, Instabilities in reactive sputtering of ZnO:Al and reliable texture-etching solution for light trapping in silicon thin film solar cells, Thin Solid Films, 520 (2012) [31] T. Zimmermann, C. Strobel, M. Albert, W. Beyer, A. Gordijn, A.J. Flikweert, J. Kuske, J.W. Bartha, Inline deposition of microcrystalline silicon solar cells using a linear plasma source, Phys. Status Solidi C, 7 (2010) [32] B. Rech, T. Roschek, T. Repmann, J. Müller, R. Schmitz, W. Appenzeller, Microcrystalline silicon for large area thin film solar cells, Thin Solid Films, 427 (2003) [33] A.J. Flikweert, T. Zimmermann, T. Merzhanova, D. Weigand, W. Appenzeller, A. Gordjin, Microcrystalline thin-film solar cell deposition on moving substrates using a linear VHF-PECVD reactor and a cross-flow geometry, J. Phys. D: Appl. Phys, 45 (2012) [34] F. Köhler, S. Schicho, B. Wolfrum, A. Gordijn, S.E. Pust, R. Carius, Gradient etching of silicon-based thin films for depth-resolved measurements: The example of Raman crystallinity, Thin Solid Films, 520 (2012)

22 [35] C. Smit, R.A.C.M.M. van Swaaij, H. Donker, A.M.H.N. Petit, W.M.M. Kessels, M.C.M. van de Sanden, Determining the material structure of microcrystalline silicon from Raman spectra, J. Appl. Phys., 94 (2003) [36] M. Scholten, J.E.A.M. van den Meerakker, On the Mechanism of ITO Etching: The Specificity of Halogen Acids, J. Electrochem. Soc., 140 (1993) [37] A.F. Holleman, E. Wiberg, N. Wiberg, G. Fischer, Lehrbuch der Anorganischen Chemie, 102 nd ed., Walter de Gruyter, Berlin, New York, [38] S.O. Klemm, J.-C. Schauer, B. Schuhmacher, A.W. Hassel, A microelectrochemical scanning flow cell with downstream analytics, Electrochim. Acta, 56 (2011) [39] C. Rockstuhl, S. Fahr, F. Lederer, K. Bittkau, T. Beckers, R. Carius, Local versus global absorption in thin-film solar cells with randomly textured surfaces, Appl. Phys. Lett., 93 (2008) [40] G. Jost, E. Bunte, J. Worbs, H. Siekmann, J. Hüpkes, Surface Characterization of Sputtered ZnO:Al for Silicon Thin-Film Solar Cells, in: Proceedings of the 25th European Photovoltaic Solar Energy Conference, Valencia, Spain, WIP, Munich, 2010, pp [41] W. Dewald, V. Sittinger, B. Szyszka, D. Wippler, J. Hüpkes, P. Obermeyer, F. Hamelmann, H. Stiebig, F. Säuberlich, D. Severin, S. Klein, M. Rhode, U. Schmidt, Evaluation of Textured TCOs for a-si:h/µc-si:h Thin Film Solar Cells by Angular Resolved Light Scattering Measurements, in: Proceedings of the 26th European Photovoltaic Solar Energy Conference, Hamburg, Germany, WIP, Munich, 2011, pp [42] G. Jost, T. Merdzhanova, T. Zimmermann, J. Hüpkes, Process control of texture-etched high-rate ZnO:Al front contact layers for application in silicon thin-film solar cells, in: Proceedings of the ICCG 9, Breda, The Netherlands, [43] A.V. Shah, M. Vaněček, J. Meier, F. Meillaud, J. Guillet, D. Fischer, C. Droz, X. Niquille, S. Faÿ, E. Vallat-Sauvain, V. Terrazzoni-Daudrix, J. Bailat, Basic efficiency limits, recent experimental results and novel light-trapping schemes in a-si:h, μc-si:h and 'micromorph tandem' solar cells, J. Non-Cryst. Solids, (2004)

23 Supporting material The catalytic effect of iron(iii) on the etching of ZnO:Al front contacts for thin-film silicon solar cells Sascha E. Pust *, Janine Worbs, Gabrielle Jost, Jürgen Hüpkes Determination of sheet resistances a increase in sheet resistance R sh [ ] mol% FeCl 3 05 mol% FeCl 3 01 mol% FeCl 3 00 no addition etch time t [s] b b increase in sheet resistance R sh [ ] mol% FeCl 2 05 mol% FeCl 2 01 mol% FeCl 2 00 no addition increase in sheet resistance R sh [ ] mol% K 3 [Fe(CN) 6 ] 05 mol% K 3 [Fe(CN) 6 ] 01 mol% K 3 [Fe(CN) 6 ] 00 no addition etch time t [s] etch time t [s] Fig. S1. Increase in sheet resistance R sh as a function of etch time t for etching in 0.5 w/w% HCl containing different concentrations of (a) FeCl 3, (b) FeCl 2, (c) K 3 [Fe(CN) 6 ]. As a reference, R sh of a substrate etched in HCl without addition of a catalyst is plotted in gray (dashed line). Lines are solely meant to guide the eye. * Corresponding author: Dr. Sascha E. Pust, s.pust@fz-juelich.de 1

24 In addition to the measured film thicknesses, the sheet resistances R sh have also been used to control and quantify the ZnO:Al thin film etching with and without catalytic influence. The increases in sheet resistance ( R sh ) as a function of the etch time t are reproduced above for the experiments with FeCl 3 and for the reference experiments with FeCl 2 and K 3 [Fe(CN) 6 ] (Fig. S1). In general, the results of the determined trends in R sh and the thickness changes d (cf. Fig. 2 of the paper) are qualitatively coherent, although the trends in the dependence on catalyst concentrations are not as clear as they have been in the case of d. 2

25 Solar cells on ZnO:Al etched catalytically with Fe(NO 3 ) 3 and Fe 2 (SO 4 ) 3 Table SI. Photovoltaic parameters of µc-si:h p-i-n single junction solar cells on etched ZnO:Al thin films. The errors given are the standard errors of the mean of all measured, intact cells on one substrate. etch etchant duration [s] init [%] FF [%] V oc [mv] J sc [ma cm -2 RS ] I C [%] without catalyst (reference) ± ± ± ± with 5 mol% ± ± ± ± Fe(NO 3 ) ± ± ± ± with 5 mol% ± ± ± ± Fe 2 (SO 4 ) ± ± ± ± Table SII. Photovoltaic parameters of a-si:h/µc-si:h tandem p-i-n solar cells on etched ZnO:Al thin films (cf. Fig. 3 in the paper). The errors given are the standard errors of the mean of all measured, intact cells on one substrate. Pretreatment conditions init [%] FF [%] V oc [mv] J sc [ma cm -2 ] RS I C,bottom [%] 40 s in 0.5 w/w% HCl (reference) 10.5 ± ± ± ± s in 0.5 w/w% HCl + 5 mol% Fe(NO 3 ) ± ± ± ± s in 0.5 w/w% HCl + 5 mol% Fe 2 (SO 4 ) ± ± ± ± a 1.0 b 1.0 external quantum efficiency EQE ma cm ma cm ma cm -2 no addition (40 s) 5 mol% Fe(NO 3 ) 3 (15 s) 5 mol% Fe(NO 3 ) 3 (20 s) external quantum efficiency EQE ma cm ma cm ma cm -2 no addition (40 s) 5 mol% Fe 2 (SO 4 ) 3 (15 s) 5 mol% Fe 2 (SO 4 ) 3 (20 s) wavelength [nm] wavelength [nm] Fig. S2. EQE of a representative µc-si:h p-i-n single junction solar cell deposited on a ZnO:Al front contact textured in 0.5 w/w% HCl (a) with 5 mol% Fe(NO 3 ) 3, (b) with 5 mol% Fe 2 (SO 4 ) 3. The EQE of a reference cell from the same absorber deposition process on a substrate etched in HCl without addition of a catalyst is plotted in gray (dashed line). 3

26 a external quantum efficiency EQE ma cm ma cm ma cm -2 no addition (40 s) 5 mol% Fe(NO 3 ) 3 (15 s) 9.9 ma cm -2 b external quantum efficiency EQE ma cm ma cm -2 no addition (40 s) 5 mol% Fe 2 (SO 4 ) 3 (15 s) 11.7 ma cm ma cm wavelength [nm] wavelength [nm] Fig. S3. Individual EQE, including the individual J sc values, of the two component cells in a representative a-si:h/µc-si:h tandem p-i-n solar cell deposited on a ZnO:Al front contact textured in 0.5 w/w% HCl (a) with 5 mol% Fe(NO 3 ) 3 for 15 s, (b) with 5 mol% Fe 2 (SO 4 ) 3 for 15 s. The EQE of a reference cell from the same absorber deposition process on a substrate etched in HCl without addition of a catalyst is plotted in gray. The sums of both component cells, being the overall EQE, are plotted as dashed lines. The data in Tables SI, SII, Figs. S2a and S3a clearly show that the cell parameters and EQE deteriorate slightly over the whole spectral range when using Fe(NO 3 ) 3. In the case of tandem cells, this is especially true for the bottom component cell. For the Fe 2 (SO 4 ) 3 -treated substrates (Figs. S2b and S3b), this effect is even stronger with losses of up 2.7 ma cm -2 compared to the reference cells. However, this should not surprise on account of the inhomogeneities and salt residues that come with the Fe 2 (SO 4 ) 3 treatment. The Fe(NO 3 ) 3 and Fe 2 (SO 4 ) 3 -treated tandem cells are clearly bottom-limited and show increased bottom component cell crystallinities RS I C,bottom, explaining elevated FF and decreased V oc values (Table SII). 4