2. Experimental details

polymers can also exhibit a tunable work function from 3.0 to 5.8 eV by chemical modification [8]. As an example, Zielke et al. [52] has demonstrated a cell efficiency of 18.3 and 20.6% for both n-type silicon and p-type silicon solar cells, respectively, which deploys a rear-side tunnel layer passivated hole-extracting metal contact using their specifically adapted organic PEDOT:PSS blend as capping layer. Such findings could open up new opportunities for potentially low-cost novel material

Regarding contact passivation, however, it is to be noted that in most of these reports, the carrier-selective passivated contacts were mostly deployed at the rear side of the solar cell, while the front side composes of a conventionally diffused silicon surface followed by the standard anti-reflection coatings and screen-printed fire-through metal contacts. Since the rear-side deployed passivated contacts can achieve an excellently low contact recombination loss, instinctively the next focus will be to reduce the contact recombination loss at the front side as well in order to improve device performance. With varying degrees of success using either electronselective or hole-selective passivated contacts in a standalone configuration, the question arises on the feasibility to integrate both electron-selective and holeselective passivated contacts together in a typical silicon solar cell architecture. Regardless of the technological advances, the fundamental driving factors toward industry adoption will still be the same as outlined earlier (i.e., cost-effectiveness of the material and processes, manufacturing scalability, device performance, and product stability). Hence, it is of keen interest in this paper to evaluate the feasibility of combining our optimized electron-selective and hole-selective passivated contacts obtained via industrial relevant processes onto an otherwise conventional front and rear screen-printed silicon solar cells and comparing that to solar cells

In this work, we will investigate "conventional" SiOx/poly-Si passivated contacts to be deployed on both sides of the solar cell, instead of only being deployed rear side. Using different lifetime test structures and solar cell structures, the following topics are investigated: (i) the influence of the tunnel oxide choice on the passivation quality, comparing wet-chemically formed oxides (wet-SiOx), UV photooxidation-formed ozone oxides (ozone-SiOx), and in situ thermal oxidation-formed oxides (thermal-SiOx); (ii) the impact on the contact passivation quality after doped silicon capping layers were applied upon the tunnel oxide layers on the same lifetime test structures (formed via tube diffusion doping of low-pressure chemical vapor deposition (LPCVD) of intrinsic polysilicon layers, to serve as either electronselective or hole-selective capping layers); (iii) the influence of the surface conditions on the passivation quality by both types of electron- or hole-selective passivated contacts; (iv) the integration of the optimal passivated contacts onto a practical double-sided passivated contact solar cell structure and studies on the resulting passivation quality, both prior to and after subsequent anti-reflection passivation layers (i.e., SiNx layers) were applied; (v) the influence of the capping layer thickness on the absorbable cell current and various parasitic absorption losses via numerical analysis (SunSolve™) and our experimental approaches to realize ultrathin poly-Si capping layers; and (vi) the ability to apply screen-printed metal contacts on the developed electron-selective and hole-selective passivated contacts. In addition, from the measured passivation quality results on lifetime test structures, a numerical calculation of the practical solar cell efficiency potential adopting both of our developed electron-selective and hole-selective passivated contacts was performed by utilizing the measured saturation current density J<sup>0</sup> and the measured contact resistance R<sup>c</sup> from our investigated tunnel layer passivated contact test structures. This work demonstrates the feasibility and attractiveness of using industrial relevant processes to develop device quality tunnel oxide/doped

integration for high-efficiency solar cell concepts in the future.

Silicon Materials

with only a rear-side passivated contact scheme.

92

Firstly, the in-house development of device quality passivated contacts based on wet-SiOx/poly-Si(doped), ozone-SiOx/poly-Si(doped), or in situ thermal-SiOx/ poly-Si(doped) stack was established using simple planar symmetrical lifetime test structures as sketched in Figure 1. Such structures are convenient for assessing (i) the resulting tunnel layer/doped capping layer stack thickness; (ii) the passivation quality, attributing from the passivated contacts alone (i.e., determining minority carrier lifetime τeff, reverse saturation current density J0, and implied open-circuit voltage iVOC); and (iii) the tunneling resistance (i.e., determining the contact resistance Rc). Starting from bare diamond-wire cut Cz silicon wafers (NorSun, 190 μm thick, and

#### Figure 1.

Schematic of the (a-e) symmetrical lifetime and (f) contact resistance test structures utilized for assessing/ optimizing the passivation quality and minimizing the contact resistance of both, electron-selective and holeselective passivated contacts, developed on either a wet-SiOx, ozone-SiOx or an in-situ thermal-SiOx tunnel layer. Using the sketched symmetrically passivated contact test structures, the developed SiOx/poly-Si passivated contacts were characterized in terms of their passivation quality, doping profiles, film uniformity and contact resistance.

wafer resistivity of 3.4 Ω cm), these wafers received a saw damage etch removal process, followed by a standard RCA and HF clean process. The next step is the deposition of the various tunnel oxide layers (see Figure 1(a)). For lifetime test samples that require the wet-SiOx tunnel layers, these samples were subjected to one more round of RCA2 process for 5 min (using deionized water, HCl, and H2O2 in the volume ratio of 0.84:0.08:0.08) in order to form the wet-SiOx tunnel layer. Other selected lifetime test samples were deposited with a symmetrical ozone-SiOx (UVO-Cleaner® 42, Jelight Company Inc.). The samples planned for an in situ thermal-SiOx tunnel layer were processed using the low-pressure chemical vapor deposition (LPCVD) tool (TS-Series, Tempress) by flowing the oxidative gases prior to the deposition of the intrinsic poly-Si capping layers. Second, intrinsic poly-Si capping layers were deposited on top of all tunnel layers investigated, using LPCVD (TS-Series, Tempress) (see Figure 1(b)). These intrinsic poly-Si capped lifetime test structures were subsequently subjected to detailed doping optimization studies, using an industrial relevant high-throughput diffusion tool (Quantum, Tempress) to obtain device quality electron-selective and hole-selective passivated contacts (see Figure 1(c)). The increase in passivation quality after the deposition of an additional SiNx passivation layer was studied using test samples as sketched in Figure 1(d). In order to assess the total contact resistance Rc, some selected samples as sketched in Figure 1(c) and (e) (now using a p-type wafer instead of an n-type wafer) were further symmetrically contacted by thermally evaporated silver (System Control Technologies), i.e., processing symmetric Ag/poly-Si(n+ )/tunnel-oxide/n-Siwafer/tunnel-oxide/poly-Si(n+ )/Ag samples to study electron extraction and Ag/poly-Si(p+ )/tunnel-oxide/p-Si-wafer/tunnel-oxide/poly-Si(p+ )/Ag samples to study hole extraction (see Figure 1(f)). In such samples, an ohmic straight-line dark I–V curve can be obtained, from which the contact resistance R<sup>c</sup> on each side can be determined (after subtracting the resistance contribution from the silicon bulk).

Next, considering that typical silicon solar cells are either single-sided textured or symmetrically textured, it is relevant to explore the passivation quality when these developed passivated contacts are deployed on textured surfaces as well, while comparing that to planar references, as sketched in Figure 2. The objective is to identify the suitability of our developed electron-selective and hole-selective passivated contacts for textured surfaces and to determine the optimum configuration for a silicon solar cell considering contact passivation for both the front and rear surfaces.

a conventional full-area or bifacial screen printing process using commercially available fire-through paste to contact the electron-selective and hole-selective passivated contacts, through a high-temperature co-firing process at 740°C in a fastfiring furnace (BTU) for 1 min. It is to be noted that the time of 1 min accounts for the total time spent within the fast-firing furnace, moving the intended sample across five temperature zones with increasing temperatures, with an estimated time of 5 seconds within the final peak temperature zone. As a final step, an edge isolation is carried out on the finished solar cell via a nanosecond laser process

Potential process flow for a silicon solar cell adopting double-sided passivated contacts and bifacial metal

Double-Sided Passivated Contacts for Solar Cell Applications: An Industrially Viable Approach…

DOI: http://dx.doi.org/10.5772/intechopen.85039

One potential issue with replacing the conventional diffused regions with carrier-selective passivated contacts (such as the poly-Si(doped)/tunnel oxide stack in this work) is the presence of parasitic absorption, similar to the case of transparent conductive oxides or amorphous silicon layers in a heterojunction silicon wafer solar cell concept. Hence, there is an optimization potential toward simultaneously achieving excellent passivation quality of both textured and planar surfaces while minimizing the doped poly-Si capping layer thickness as much as possible in order

Thus, it has been tested experimentally how thin our developed contact passivation layers can become while maintaining their excellent passivation quality. This has been realized by two different experimental approaches: (1) applying etch-back technology, thereby thinning down the already optimized thick layers, and (2) diffusion re-optimization for ultrathin LPCVD of intrinsic poly-Si layers.

To provide more insights into the influence of the doped capping layer thickness on the maximum absorbable current density Jabsorbed, cell attainable in a silicon solar cell, the simulation program SunSolve™, available on PV Lighthouse [53], was utilized. The SunSolve™ calculator combines Monte Carlo ray tracing with thinfilm optics to calculate the maximum potential photogeneration current in the solar cell for the standard AM1.5G spectrum, as well as the corresponding optical losses occurring elsewhere (i.e., front-reflected, front-escaped, rear-escaped, parasitic

(ILS500LT, InnoLas), followed by electrical characterization.

to minimize the parasitic absorption issue.

Figure 3.

contacts.

95

It will be shown in later sections that the optimum double-sided passivated contact scheme can be realized by deploying the electron-selective passivated contacts (i.e., poly-Si(n<sup>+</sup> )/tunnel oxide stacks) on the front textured surface while deploying the hole-selective passivated contacts (i.e., poly-Si(p<sup>+</sup> )/tunnel oxide stacks) on the rear planar surface. Subsequently, the silicon solar cell precursors with the optimum double-sided passivated contact scheme were experimentally realized according to the process flow shown in Figure 3 and characterized in terms of the passivation quality and doping profile, both prior to and after the standard anti-reflection/passivation dielectric coatings were deposited (i.e., step. 11 and 12, respectively) via microwave PECVD (MAiA, Meyer Burger), while comparing that to the symmetrical lifetime test structures. Selected samples were then subjected to

Figure 2.

Comparison of the passivation quality by both (a, b) hole-selective and (c, d) electron-selective passivated contacts on both planar and textured lifetime test structures.

Double-Sided Passivated Contacts for Solar Cell Applications: An Industrially Viable Approach… DOI: http://dx.doi.org/10.5772/intechopen.85039

Figure 3.

wafer resistivity of 3.4 Ω cm), these wafers received a saw damage etch removal process, followed by a standard RCA and HF clean process. The next step is the deposition of the various tunnel oxide layers (see Figure 1(a)). For lifetime test samples that require the wet-SiOx tunnel layers, these samples were subjected to one more round of RCA2 process for 5 min (using deionized water, HCl, and H2O2 in the volume ratio of 0.84:0.08:0.08) in order to form the wet-SiOx tunnel layer. Other selected lifetime test samples were deposited with a symmetrical ozone-SiOx (UVO-Cleaner® 42, Jelight Company Inc.). The samples planned for an in situ thermal-SiOx tunnel layer were processed using the low-pressure chemical vapor deposition (LPCVD) tool (TS-Series, Tempress) by flowing the oxidative gases prior to the deposition of the intrinsic poly-Si capping layers. Second, intrinsic poly-Si capping layers were deposited on top of all tunnel layers investigated, using LPCVD (TS-Series, Tempress) (see Figure 1(b)). These intrinsic poly-Si capped lifetime test structures were subsequently subjected to detailed doping optimization studies, using an industrial relevant high-throughput diffusion tool (Quantum, Tempress) to obtain device quality electron-selective and hole-selective passivated contacts (see Figure 1(c)). The increase in passivation quality after the deposition of an additional SiNx passivation layer was studied using test samples as sketched in Figure 1(d). In order to assess the total contact resistance Rc, some selected samples as sketched in Figure 1(c) and (e) (now using a p-type wafer instead of an n-type wafer) were further symmetrically contacted by thermally evaporated silver (System Control

)/tunnel-oxide/n-Si-

)/Ag samples to

)/tunnel oxide

)/Ag samples to study electron extraction and

Technologies), i.e., processing symmetric Ag/poly-Si(n+

)/tunnel-oxide/p-Si-wafer/tunnel-oxide/poly-Si(p+

study hole extraction (see Figure 1(f)). In such samples, an ohmic straight-line dark I–V curve can be obtained, from which the contact resistance R<sup>c</sup> on each side can be determined (after subtracting the resistance contribution from the silicon bulk). Next, considering that typical silicon solar cells are either single-sided textured or symmetrically textured, it is relevant to explore the passivation quality when these developed passivated contacts are deployed on textured surfaces as well, while comparing that to planar references, as sketched in Figure 2. The objective is to identify the suitability of our developed electron-selective and hole-selective passivated contacts for textured surfaces and to determine the optimum configuration for a silicon solar cell considering contact passivation for both the front and

It will be shown in later sections that the optimum double-sided passivated contact scheme can be realized by deploying the electron-selective passivated con-

stacks) on the rear planar surface. Subsequently, the silicon solar cell precursors with the optimum double-sided passivated contact scheme were experimentally realized according to the process flow shown in Figure 3 and characterized in terms of the passivation quality and doping profile, both prior to and after the standard anti-reflection/passivation dielectric coatings were deposited (i.e., step. 11 and 12, respectively) via microwave PECVD (MAiA, Meyer Burger), while comparing that to the symmetrical lifetime test structures. Selected samples were then subjected to

Comparison of the passivation quality by both (a, b) hole-selective and (c, d) electron-selective passivated

deploying the hole-selective passivated contacts (i.e., poly-Si(p<sup>+</sup>

contacts on both planar and textured lifetime test structures.

)/tunnel oxide stacks) on the front textured surface while

wafer/tunnel-oxide/poly-Si(n+

Ag/poly-Si(p+

Silicon Materials

rear surfaces.

Figure 2.

94

tacts (i.e., poly-Si(n<sup>+</sup>

Potential process flow for a silicon solar cell adopting double-sided passivated contacts and bifacial metal contacts.

a conventional full-area or bifacial screen printing process using commercially available fire-through paste to contact the electron-selective and hole-selective passivated contacts, through a high-temperature co-firing process at 740°C in a fastfiring furnace (BTU) for 1 min. It is to be noted that the time of 1 min accounts for the total time spent within the fast-firing furnace, moving the intended sample across five temperature zones with increasing temperatures, with an estimated time of 5 seconds within the final peak temperature zone. As a final step, an edge isolation is carried out on the finished solar cell via a nanosecond laser process (ILS500LT, InnoLas), followed by electrical characterization.

One potential issue with replacing the conventional diffused regions with carrier-selective passivated contacts (such as the poly-Si(doped)/tunnel oxide stack in this work) is the presence of parasitic absorption, similar to the case of transparent conductive oxides or amorphous silicon layers in a heterojunction silicon wafer solar cell concept. Hence, there is an optimization potential toward simultaneously achieving excellent passivation quality of both textured and planar surfaces while minimizing the doped poly-Si capping layer thickness as much as possible in order to minimize the parasitic absorption issue.

Thus, it has been tested experimentally how thin our developed contact passivation layers can become while maintaining their excellent passivation quality. This has been realized by two different experimental approaches: (1) applying etch-back technology, thereby thinning down the already optimized thick layers, and (2) diffusion re-optimization for ultrathin LPCVD of intrinsic poly-Si layers.

To provide more insights into the influence of the doped capping layer thickness on the maximum absorbable current density Jabsorbed, cell attainable in a silicon solar cell, the simulation program SunSolve™, available on PV Lighthouse [53], was utilized. The SunSolve™ calculator combines Monte Carlo ray tracing with thinfilm optics to calculate the maximum potential photogeneration current in the solar cell for the standard AM1.5G spectrum, as well as the corresponding optical losses occurring elsewhere (i.e., front-reflected, front-escaped, rear-escaped, parasitic

absorption, edge absorption). Using SunSolve™, we provide a quantitative discussion of the influence of our doped poly-Si capping layer thickness on Jabsorbed, cell and Jabsorbed, parasitic which will ultimately affect the solar cell performance.

their deposition techniques as well as the time required to get tunneling relevant thicknesses. Starting from our wet-chemically formed oxides (wet-SiOx) via the standard RCA2 solution, Figure 4(a) shows that the resulting wet-SiOx tunnel oxide thickness is independent of the oxidation time utilized (1–10 min) and is well within the tunneling relevant thickness regime (1.2–1.5 nm). These wet-SiOx tunnel oxide layers were also found to exhibit a highly leaky interface toward the silicon bulk, as attempts to determine the wet-SiOx/c-Si interface properties Q<sup>f</sup> and Dit(E) were unsuccessful due to its inability to retain the deposited corona charges. Nonetheless, this is likely to be beneficial for the subsequent charge collection process, since the charge carriers to be collected can easily tunnel through such an oxide layer. It is also worthy to note that these symmetrical planar lifetime structures with only a wet-SiOx tunnel oxide do not passivate well (τeff 4 μs) and are only effective when subsequently coupled with a highly doped capping layer to form a wet-SiOx/poly-Si(doped) passivated contact scheme, as will be shown in the

Double-Sided Passivated Contacts for Solar Cell Applications: An Industrially Viable Approach…

In contrast, for the investigated UV/ozone photo-oxidation-formed ozone-SiOx tunnel oxides, Figure 4(b) shows that the resulting ozone-SiOx layer thickness shows a time dependence of the photo-oxidation time, which increases from 1.3 nm for an exposure time of 3 min to 2.5 nm for 10 min. Beyond 10 min, the thickness of the ozone-SiOx layer saturates at 2.7 nm (i.e., surface reaction limited). Hence, considering the need for tunneling relevant applications (<1.5 nm), the UV/ozone exposure time should be limited to ≤3 min. Similar to the wet-SiOx case, the ozone-SiOx tunnel oxides were also found to be leaky in the as-deposited state, evident from its inability to measure Q<sup>f</sup> and Dit(E). In terms of passivation, symmetrically ozone-SiOx passivated planar lifetime samples also do not passivate well (τeff 2 μs) and should be coupled with a highly doped capping layer to form

Finally, our investigated in situ thermal oxides were also found to exhibit a deposition time dependence on the measured oxide thickness, in which an in situ oxidation time of 30 secs at 570°C was already sufficient to achieve a tunneling relevant thickness of 1.0 to 1.2 nm. At higher deposition timings (e.g., 5 min), the thickness increases to 13 nm which is not suitable for tunneling relevant applications.

(a) Comparison of the wet-chemical (RCA2) oxidation time on the measured wet-SiOx tunnel oxide thickness. The wet-SiOx thickness does not exhibit a time dependence (1–10 min) and has a thickness range of 1.2– 1.5 nm, relevant for device integration. (b) For ozone-SiOx, the UV exposure time directly affects the ozone-SiOx tunnel oxide thickness, with a recommended exposure time of 3 min to achieve tunneling relevant

coming sections.

Figure 4.

thickness.

97

an effective contact passivation scheme as well.

DOI: http://dx.doi.org/10.5772/intechopen.85039

Finally, it is of keen interest to predict the impact of combining both of our developed electron-selective and hole-selective passivated contacts on the rear (and front side) of a silicon solar cell. To do this, we utilized Brendel's model [54] to predict the efficiency potential of a passivated contact and further enhanced the model to explicitly consider front-side conventional screen-printed contacts. This is done by additionally considering the combined front-side saturation current density J0, front (contributed by both the front-side metal-contacted regions and metalpassivated regions) and the front-side contact resistance Rc, front of the screenprinted contacts. Thus, practical iso-efficiency contour plots (under a variation of the J0, rear and Rc, rear values of the rear-side passivated contact) can be obtained, allowing us to predict a practical solar cell efficiency potential, given known J0, front, Rc, front, J0, rear, Rc, rear values. Subsequently, our individually measured J0, rear and Rc, rear values for our investigated passivated contacts in this work were inserted into this iso-efficiency contour plot, and a realistic prediction of the solar cell efficiency potential can be realized for both single-sided passivated contact scheme and double-sided passivated contact schemes.

Last but not least, the feasibility to contact our developed ultrathin contact passivation layers by an industrially suited method (i.e., aiming at conventional screen printing) is investigated, and the remaining issues, still to be solved in order to reach this goal, are addressed.

Concerning characterization metrology, we used the following tools: The average thickness and uniformity of the tunnel layers/doped poly-Si capping layers were determined by ellipsometry (SE-2000, Semilab) over a 9-point mapping measurement. The passivation quality was determined from the injection-dependent effective carrier lifetime measurements using a contactless flash-based photoconductance decay tester (WCT-120, Sinton Consulting) operated in both transient and quasi steady-state modes (QSSPC), which adopts an intrinsic carrier concentration of 8.6 109 cm<sup>3</sup> in the calculation of the saturation current densities. To provide further insights at the tunnel layer/silicon interface, the fixed interface charge density Q<sup>f</sup> and the interface defect density distribution Dit(E) were determined using time-resolved contactless corona charge—Kelvin probe measurements (PV-2000, Semilab). Considering our ultrathin dielectrics, and its high potential for charge leakage, PV-2000 utilizes a "Self-Adjusting SteadyState Technique (SASS)" which takes into consideration the SASS voltage obtained using both positive and negative corona charges in order to calculate a leakage index (equivalently a correction factor) which accounts for the dielectric leakage when present and applicable to both ultrathin and thicker dielectrics, in order to produce a reliable representation of the Q<sup>f</sup> and Dit(E) values across different samples. The film structure of the doped silicon capping layer was determined via Raman spectroscopy (SE-2000, Semilab). The light and dark I–V measurements were performed on an LED-based AAA-calibrated I–V tester (Sinus220, Wavelabs).

## 3. Results and discussion

#### 3.1 Screening of tunnel oxide layers for contact passivation

As mentioned earlier, the development and optimization of contact passivation layer stacks were initiated on symmetrical planar lifetime test structures as sketched in Figure 1. Prior to deposition of the doped capping layer, various tunnel oxide candidates were screened (i.e., wet-SiOx, ozone-SiOx, and thermal-SiOx) in terms of

### Double-Sided Passivated Contacts for Solar Cell Applications: An Industrially Viable Approach… DOI: http://dx.doi.org/10.5772/intechopen.85039

their deposition techniques as well as the time required to get tunneling relevant thicknesses. Starting from our wet-chemically formed oxides (wet-SiOx) via the standard RCA2 solution, Figure 4(a) shows that the resulting wet-SiOx tunnel oxide thickness is independent of the oxidation time utilized (1–10 min) and is well within the tunneling relevant thickness regime (1.2–1.5 nm). These wet-SiOx tunnel oxide layers were also found to exhibit a highly leaky interface toward the silicon bulk, as attempts to determine the wet-SiOx/c-Si interface properties Q<sup>f</sup> and Dit(E) were unsuccessful due to its inability to retain the deposited corona charges. Nonetheless, this is likely to be beneficial for the subsequent charge collection process, since the charge carriers to be collected can easily tunnel through such an oxide layer. It is also worthy to note that these symmetrical planar lifetime structures with only a wet-SiOx tunnel oxide do not passivate well (τeff 4 μs) and are only effective when subsequently coupled with a highly doped capping layer to form a wet-SiOx/poly-Si(doped) passivated contact scheme, as will be shown in the coming sections.

In contrast, for the investigated UV/ozone photo-oxidation-formed ozone-SiOx tunnel oxides, Figure 4(b) shows that the resulting ozone-SiOx layer thickness shows a time dependence of the photo-oxidation time, which increases from 1.3 nm for an exposure time of 3 min to 2.5 nm for 10 min. Beyond 10 min, the thickness of the ozone-SiOx layer saturates at 2.7 nm (i.e., surface reaction limited). Hence, considering the need for tunneling relevant applications (<1.5 nm), the UV/ozone exposure time should be limited to ≤3 min. Similar to the wet-SiOx case, the ozone-SiOx tunnel oxides were also found to be leaky in the as-deposited state, evident from its inability to measure Q<sup>f</sup> and Dit(E). In terms of passivation, symmetrically ozone-SiOx passivated planar lifetime samples also do not passivate well (τeff 2 μs) and should be coupled with a highly doped capping layer to form an effective contact passivation scheme as well.

Finally, our investigated in situ thermal oxides were also found to exhibit a deposition time dependence on the measured oxide thickness, in which an in situ oxidation time of 30 secs at 570°C was already sufficient to achieve a tunneling relevant thickness of 1.0 to 1.2 nm. At higher deposition timings (e.g., 5 min), the thickness increases to 13 nm which is not suitable for tunneling relevant applications.

#### Figure 4.

absorption, edge absorption). Using SunSolve™, we provide a quantitative discussion of the influence of our doped poly-Si capping layer thickness on Jabsorbed, cell and Jabsorbed, parasitic which will ultimately affect the solar cell performance. Finally, it is of keen interest to predict the impact of combining both of our developed electron-selective and hole-selective passivated contacts on the rear (and front side) of a silicon solar cell. To do this, we utilized Brendel's model [54] to predict the efficiency potential of a passivated contact and further enhanced the model to explicitly consider front-side conventional screen-printed contacts. This is done by additionally considering the combined front-side saturation current density J0, front (contributed by both the front-side metal-contacted regions and metalpassivated regions) and the front-side contact resistance Rc, front of the screenprinted contacts. Thus, practical iso-efficiency contour plots (under a variation of the J0, rear and Rc, rear values of the rear-side passivated contact) can be obtained, allowing us to predict a practical solar cell efficiency potential, given known J0, front, Rc, front, J0, rear, Rc, rear values. Subsequently, our individually measured J0, rear and Rc, rear values for our investigated passivated contacts in this work were inserted into this iso-efficiency contour plot, and a realistic prediction of the solar cell efficiency potential can be realized for both single-sided passivated contact scheme

Last but not least, the feasibility to contact our developed ultrathin contact passivation layers by an industrially suited method (i.e., aiming at conventional screen printing) is investigated, and the remaining issues, still to be solved in order

Concerning characterization metrology, we used the following tools: The average thickness and uniformity of the tunnel layers/doped poly-Si capping layers were determined by ellipsometry (SE-2000, Semilab) over a 9-point mapping measurement. The passivation quality was determined from the injection-dependent effective carrier lifetime measurements using a contactless flash-based photoconductance decay tester (WCT-120, Sinton Consulting) operated in both transient and quasi steady-state modes (QSSPC), which adopts an intrinsic carrier concentration of 8.6 109 cm<sup>3</sup> in the calculation of the saturation current densities. To provide further insights at the tunnel layer/silicon interface, the fixed interface charge density Q<sup>f</sup> and the interface defect density distribution Dit(E) were determined using time-resolved contactless corona charge—Kelvin probe measurements (PV-2000, Semilab). Considering our ultrathin dielectrics, and its high potential for charge leakage, PV-2000 utilizes a "Self-Adjusting SteadyState Technique (SASS)" which takes into consideration the SASS voltage obtained using both positive and negative corona charges in order to calculate a leakage index (equivalently a correction factor) which accounts for the dielectric leakage when present and applicable to both ultrathin and thicker dielectrics, in order to produce a reliable representation of the Q<sup>f</sup> and Dit(E) values across different samples. The film structure of the doped silicon capping layer was determined via Raman spectroscopy (SE-2000, Semilab). The light and dark I–V measurements were performed on an LED-based

and double-sided passivated contact schemes.

AAA-calibrated I–V tester (Sinus220, Wavelabs).

3.1 Screening of tunnel oxide layers for contact passivation

As mentioned earlier, the development and optimization of contact passivation layer stacks were initiated on symmetrical planar lifetime test structures as sketched in Figure 1. Prior to deposition of the doped capping layer, various tunnel oxide candidates were screened (i.e., wet-SiOx, ozone-SiOx, and thermal-SiOx) in terms of

3. Results and discussion

96

to reach this goal, are addressed.

Silicon Materials

(a) Comparison of the wet-chemical (RCA2) oxidation time on the measured wet-SiOx tunnel oxide thickness. The wet-SiOx thickness does not exhibit a time dependence (1–10 min) and has a thickness range of 1.2– 1.5 nm, relevant for device integration. (b) For ozone-SiOx, the UV exposure time directly affects the ozone-SiOx tunnel oxide thickness, with a recommended exposure time of 3 min to achieve tunneling relevant thickness.

Correspondingly, an in situ thermal oxide growth rate of 2.4–2.6 nm/min can be expected. Interestingly, in contrast to the wet-SiOx and ozone-SiOx tunnel layers, our as-deposited thermal-SiOx tunnel oxides were able to retain the deposited charges from the contactless corona charge—Kelvin probe measurements, allowing the fixed interface charge density and the interface defect density distribution to be determined (see Figure 5). At the first glance, this already suggests that the in situ thermal-SiOx exhibits a higher film quality (i.e., non-leaky) than both wet-SiOx and ozone-SiOx. It is also likely that the thermal-SiOx film structure is more dense, which can be beneficial when coupled with a highly doped silicon capping layer, which could reduce the out-diffusion of dopants into the c-Si bulk. Table 1 summarizes the measured Q<sup>f</sup> and Dit for our investigated tunnel oxides (wet-SiOx, ozone-SiOx, thermal-SiOx) and compares that to literature-reported values. As plotted in Figure 5, and summarized in Table 1, the as-deposited in situ thermal-SiOx (1.2 nm) in this work exhibited a <sup>Q</sup><sup>f</sup> of 4.3 1011 cm<sup>2</sup> and a minimum <sup>D</sup>it of 2.5 1012 cm<sup>2</sup> eV<sup>1</sup> . The energy distribution of the interface defect density Dit(E) as a function of the silicon band-gap energy for our in situ thermal-SiOx layers is plotted in Figure 5(b), showing a minimum Dit closer to the valence band, instead of the midgap position. This could be due to the increase of surface micro-roughness from the processing conditions, leading to a higher density of dangling bond defects in the higher part of the silicon energy gap, similar to the observation by Angermann et al. [55] who observed a skewing

As compared to other thermally grown silicon oxides [56–59] which exhibited significantly lower Dit values (1–2 orders), their film thickness was however also significantly higher at 50–240 nm, making it inappropriate for tunnel layer applications. On the other hand, the wet-SiOx and ozone-SiOx tunnel oxides reported in Refs. [55, 57, 60–64] do exhibit measurable Q<sup>f</sup> and Dit values, unlike our investigated samples, which can be attributed to the deposition method and the postdeposition annealing conditions. Our wet-SiOx and ozone-SiOx tunnel oxides were unable to retain the deposited corona charges due to its leaky interface, which nonetheless could be beneficial for the purpose of tunneling carrier transport. Another noteworthy tunnel oxide candidate is atomic layer deposition (ALD) of aluminum oxide (AlOx), whereby in one of our earlier publications [28], we experimentally realized ultrathin ALD-AlOx films in the tunneling regime (1.5 nm) which is capable of exhibiting a significantly higher negative <sup>Q</sup><sup>f</sup> of <sup>6</sup> <sup>10</sup><sup>12</sup> cm<sup>2</sup>

Double-Sided Passivated Contacts for Solar Cell Applications: An Industrially Viable Approach…

initial passivation quality prior to the doped capping layers as compared to a conventional wet-SiOx tunnel oxide layer using our test structures (i.e., from 2 to 218 μs) [28]. This finding positions ultrathin ALD-AlOx as a highly attractive tunnel oxide candidate for hole-extracting selective contacts. In contrast, although the Dit values of our in situ thermal-SiOx and ALD-AlOx films are comparable, the thermal-SiOx in this work exhibited one order lower negative fixed interface charge density which do suggest a reduced field-effect passivation and overall surface passivation prior to the doped capping layers, which is observed experimentally as well

(τeff 4.7 μs). Nevertheless, this positions thermal-SiOx as a tunnel oxide candidate

As compared to the TOPCon approach by the Fraunhofer ISE's team, which deposited doped amorphous silicon films followed by a suitable annealing condition and hydrogenation process to convert the highly doped amorphous silicon to highly doped polysilicon capping layers, we implement an alternative approach by first depositing intrinsic silicon films via the LPCVD approach, followed by either a phosphorus or boron diffusion process to convert it to a highly doped poly-Si(n<sup>+</sup>

) capping layer, respectively. The optimization goal is to incorporate as

As a start, Raman spectroscopy was utilized to monitor the structural evolution of our in-house deposited silicon capping layers, both in the as-deposited intrinsic case and after the optimal diffusion process (boron or phosphorus doped). Figure 6 shows that our LPCVD as-deposited intrinsic silicon films were amorphous in film structure, evident by a single Raman peak centered at a Raman shift of 480cm<sup>1</sup> [66]. Nonetheless, upon either a boron diffusion process or a phosphorus diffusion process, which takes place at temperatures between 850 and 950°C, these doped silicon films fully crystallize as evident by a single Raman peak centered at a Raman shift of 520.5 cm<sup>1</sup> with a full width at half maximum (FWHM) of 5.3 and

, respectively. These findings were comparable to our crystalline silicon

wafer reference (Raman shift centered at 520.6 cm<sup>1</sup> and a FWHM of 3.5 cm<sup>1</sup>

The slightly higher FWHM measured for our doped silicon films indicated a marginally higher structural disorder than a perfect crystalline silicon wafer bulk which

suitable for both electron-extracting and hole-extracting selective contacts.

much active dopants within the poly-Si layers as possible while reducing or avoiding the out-diffusion of dopants into the c-Si wafer bulk, which will increase the surface recombination rates and reduce the device performance, as also

3.2 Screening LPCVD poly-Si capping layers for contact passivation

. This resulted in a 110-fold increase in the

) or

).

toward the conduction band when p-type Si substrates are utilized.

and a <sup>D</sup>it of 2.7 <sup>10</sup><sup>12</sup> cm<sup>2</sup> eV<sup>1</sup>

DOI: http://dx.doi.org/10.5772/intechopen.85039

poly-Si(p<sup>+</sup>

4.0 cm<sup>1</sup>

99

reported in Ref. [65].

#### Figure 5.

The in-situ thermal silicon oxides deposited within the LPCVD process prior to the intrinsic poly-Si layers exhibited (a) a negative fixed interface charge density Q f of 4.3 <sup>10</sup><sup>11</sup> cm<sup>2</sup> and (b) a minimum interface defect density Dit(min) of 2.5 <sup>10</sup><sup>12</sup> cm<sup>2</sup> eV<sup>1</sup> .


#### Table 1.

Comparison of the fixed interface charge density Qf and the interface defect density distribution Dit(E) for different investigated tunnel oxides (thermal-SiOx, wet-SiOx, ozone-SiOx) in this work as compared to literature.

Double-Sided Passivated Contacts for Solar Cell Applications: An Industrially Viable Approach… DOI: http://dx.doi.org/10.5772/intechopen.85039

gap, similar to the observation by Angermann et al. [55] who observed a skewing toward the conduction band when p-type Si substrates are utilized.

As compared to other thermally grown silicon oxides [56–59] which exhibited significantly lower Dit values (1–2 orders), their film thickness was however also significantly higher at 50–240 nm, making it inappropriate for tunnel layer applications. On the other hand, the wet-SiOx and ozone-SiOx tunnel oxides reported in Refs. [55, 57, 60–64] do exhibit measurable Q<sup>f</sup> and Dit values, unlike our investigated samples, which can be attributed to the deposition method and the postdeposition annealing conditions. Our wet-SiOx and ozone-SiOx tunnel oxides were unable to retain the deposited corona charges due to its leaky interface, which nonetheless could be beneficial for the purpose of tunneling carrier transport. Another noteworthy tunnel oxide candidate is atomic layer deposition (ALD) of aluminum oxide (AlOx), whereby in one of our earlier publications [28], we experimentally realized ultrathin ALD-AlOx films in the tunneling regime (1.5 nm) which is capable of exhibiting a significantly higher negative <sup>Q</sup><sup>f</sup> of <sup>6</sup> <sup>10</sup><sup>12</sup> cm<sup>2</sup> and a <sup>D</sup>it of 2.7 <sup>10</sup><sup>12</sup> cm<sup>2</sup> eV<sup>1</sup> . This resulted in a 110-fold increase in the initial passivation quality prior to the doped capping layers as compared to a conventional wet-SiOx tunnel oxide layer using our test structures (i.e., from 2 to 218 μs) [28]. This finding positions ultrathin ALD-AlOx as a highly attractive tunnel oxide candidate for hole-extracting selective contacts. In contrast, although the Dit values of our in situ thermal-SiOx and ALD-AlOx films are comparable, the thermal-SiOx in this work exhibited one order lower negative fixed interface charge density which do suggest a reduced field-effect passivation and overall surface passivation prior to the doped capping layers, which is observed experimentally as well (τeff 4.7 μs). Nevertheless, this positions thermal-SiOx as a tunnel oxide candidate suitable for both electron-extracting and hole-extracting selective contacts.

## 3.2 Screening LPCVD poly-Si capping layers for contact passivation

As compared to the TOPCon approach by the Fraunhofer ISE's team, which deposited doped amorphous silicon films followed by a suitable annealing condition and hydrogenation process to convert the highly doped amorphous silicon to highly doped polysilicon capping layers, we implement an alternative approach by first depositing intrinsic silicon films via the LPCVD approach, followed by either a phosphorus or boron diffusion process to convert it to a highly doped poly-Si(n<sup>+</sup> ) or poly-Si(p<sup>+</sup> ) capping layer, respectively. The optimization goal is to incorporate as much active dopants within the poly-Si layers as possible while reducing or avoiding the out-diffusion of dopants into the c-Si wafer bulk, which will increase the surface recombination rates and reduce the device performance, as also reported in Ref. [65].

As a start, Raman spectroscopy was utilized to monitor the structural evolution of our in-house deposited silicon capping layers, both in the as-deposited intrinsic case and after the optimal diffusion process (boron or phosphorus doped). Figure 6 shows that our LPCVD as-deposited intrinsic silicon films were amorphous in film structure, evident by a single Raman peak centered at a Raman shift of 480cm<sup>1</sup> [66]. Nonetheless, upon either a boron diffusion process or a phosphorus diffusion process, which takes place at temperatures between 850 and 950°C, these doped silicon films fully crystallize as evident by a single Raman peak centered at a Raman shift of 520.5 cm<sup>1</sup> with a full width at half maximum (FWHM) of 5.3 and 4.0 cm<sup>1</sup> , respectively. These findings were comparable to our crystalline silicon wafer reference (Raman shift centered at 520.6 cm<sup>1</sup> and a FWHM of 3.5 cm<sup>1</sup> ). The slightly higher FWHM measured for our doped silicon films indicated a marginally higher structural disorder than a perfect crystalline silicon wafer bulk which

Correspondingly, an in situ thermal oxide growth rate of 2.4–2.6 nm/min can be expected. Interestingly, in contrast to the wet-SiOx and ozone-SiOx tunnel layers, our as-deposited thermal-SiOx tunnel oxides were able to retain the deposited charges from the contactless corona charge—Kelvin probe measurements, allowing the fixed interface charge density and the interface defect density distribution to be determined (see Figure 5). At the first glance, this already suggests that the in situ thermal-SiOx exhibits a higher film quality (i.e., non-leaky) than both wet-SiOx and ozone-SiOx. It is also likely that the thermal-SiOx film structure is more dense, which can be beneficial when coupled with a highly doped silicon capping layer, which could reduce the out-diffusion of dopants into the c-Si bulk. Table 1 summarizes the measured Q<sup>f</sup> and Dit for our investigated tunnel oxides (wet-SiOx, ozone-SiOx, thermal-SiOx) and compares that to literature-reported values. As plotted in Figure 5, and summarized in Table 1, the as-deposited in situ thermal-SiOx (1.2 nm) in this work exhibited a

<sup>Q</sup><sup>f</sup> of 4.3 1011 cm<sup>2</sup> and a minimum <sup>D</sup>it of 2.5 1012 cm<sup>2</sup> eV<sup>1</sup>

Figure 5.

Silicon Materials

Table 1.

literature.

98

defect density Dit(min) of 2.5 <sup>10</sup><sup>12</sup> cm<sup>2</sup> eV<sup>1</sup>

Tunnel oxide Thickness (nm) Qf (cm<sup>2</sup>

distribution of the interface defect density Dit(E) as a function of the silicon band-gap energy for our in situ thermal-SiOx layers is plotted in Figure 5(b), showing a minimum Dit closer to the valence band, instead of the midgap position. This could be due to the increase of surface micro-roughness from the processing conditions, leading to a higher density of dangling bond defects in the higher part of the silicon energy

The in-situ thermal silicon oxides deposited within the LPCVD process prior to the intrinsic poly-Si layers exhibited (a) a negative fixed interface charge density Q f of 4.3 <sup>10</sup><sup>11</sup> cm<sup>2</sup> and (b) a minimum interface

Thermal-SiOx 1.2 4.30 <sup>10</sup><sup>11</sup> 2.50 <sup>10</sup><sup>12</sup> This work Wet-SiOx 1.5 Not measurable Not measurable This work Ozone-SiOx 1.3 Not measurable Not measurable This work Thermal-SiOx <sup>50</sup>–240 +3.00 1011 1010 <sup>7</sup> <sup>10</sup><sup>11</sup> [56–59] Wet-SiOx <sup>1</sup>–<sup>2</sup> +1.28 <sup>10</sup><sup>12</sup> 5.17 <sup>10</sup><sup>12</sup> [57, 60–64] Ozone-SiOx <sup>1</sup>–<sup>2</sup> No data 1.00 <sup>10</sup><sup>13</sup> [55] ALD-AlOx 1.5 6.10 1012 2.70 <sup>10</sup><sup>12</sup> [28]

Comparison of the fixed interface charge density Qf and the interface defect density distribution Dit(E) for different investigated tunnel oxides (thermal-SiOx, wet-SiOx, ozone-SiOx) in this work as compared to

) Dit (cm<sup>2</sup> eV<sup>1</sup>

) References

.

. The energy

#### Figure 6.

Raman spectra for our in-house LPCVD of silicon films (250 nm), comparing the film crystallinity in the asdeposited state, post-POCl3 diffusion process, and post-BBR3 diffusion process.

out-diffusion of dopants is expected to lead to increased surface recombination rates and a corresponding drop in the overall passivation quality as well.

) layers, which was found to limit the potentially achievable implied-VOC values (see Table 3).

ECV profiles for both (a) electron-selective passivated contacts comprising thermal-SiOx/poly-Si(n<sup>+</sup>

and (b) hole-selective passivated contacts comprising thermal-SiOx/poly-Si(p<sup>+</sup>

metrical lifetime test structures with the optimized doped poly-Si(n<sup>+</sup>

were exhibiting implied-VOC values of 719 mV and <sup>J</sup><sup>0</sup> of <sup>6</sup>–9 fA cm<sup>2</sup>

implied-VOC values of 729 mV, despite a similar <sup>J</sup><sup>0</sup> of 9 fA cm<sup>2</sup>

implied-VOC values for thermal-SiOx/poly-Si(n<sup>+</sup>

sivation quality (implied-VOC increases by 10 mV).

with single-sided <sup>J</sup><sup>0</sup> values down to 2.5 fA cm<sup>2</sup>

tures, our wet-SiOx/poly-Si(n<sup>+</sup>

thermal-SiOx/poly-Si(n<sup>+</sup>

Figure 7.

Si(n+

of 2 times. Poly-Si(p<sup>+</sup>

from the poly-Si(n<sup>+</sup>

101

Table 2 summarizes our measured passivation quality results on planar sym-

passivated contacts exhibited a higher peak doping concentration than the hole-selective counterpart by a factor

Double-Sided Passivated Contacts for Solar Cell Applications: An Industrially Viable Approach…

DOI: http://dx.doi.org/10.5772/intechopen.85039

layers on various investigated tunnel oxide candidates (i.e., wet-SiOx, ozone-SiOx, thermal-SiOx). Table 2 shows that using planar symmetrical lifetime test struc-

with the earlier discussion on the tunnel oxides, in which a thermal-SiOx tunnel oxide is likely more effective in reducing the out-diffusion of phosphorus dopants

Since a typical silicon solar cell would be further coated with suitable antireflection layers (such as SiNx or AlOx/SiNx stacks) prior to metallization, the influence of these layers on our symmetrical lifetime samples were evaluated as well, by capping the passivated contacts with an additional 70-nm-thick SiNx films symmetrically and its resulting passivation quality evaluated. As summarized in Table 2, the measured passivation quality further improves with the additional SiNx capping layers upon all three investigated lifetime test structures with electron-selective passivated contacts. In particular, the thermal-SiOx/poly-Si(n<sup>+</sup>

SiNx-capped lifetime structure exhibits high implied-VOC approaching 740 mV,

best results from the Fraunhofer ISE team [69]. Concurrently, similar studies were conducted on lifetime test structures with hole-selective passivated contacts, and

) and ozone-SiOx/poly-Si(n<sup>+</sup>

) passivated contact stack exhibited an even higher

) layers also exhibited a higher out-diffusion of dopants into the c-Si bulk than the poly-

) into the c-Si wafer bulk, hence providing better overall pas-

) capping

. The enhanced

, while the

) stacks

)/

) passivated samples

) stacks. The electron-selective

) passivation stack were consistent

, which is already on par with the

is not too surprising, given the high quantities of dopants (1019–10<sup>20</sup> cm<sup>3</sup> ) incorporated in the former.

The corresponding dopant profile within these highly doped silicon capping layers can be extracted from ECV measurements as shown in Figure 7. After an optimized diffusion doping process to convert the thermal-SiOx/a-Si(intrinsic) capping layer stack toward either an electron-selective passivated contact (i.e., thermal-SiOx/poly-Si(n<sup>+</sup> ) stack) or a hole-selective passivated contact (i.e., thermal-SiOx/poly-Si(p<sup>+</sup> ) stack), the ECV measurements revealed a peak doping concentration within the poly-Si(n<sup>+</sup> ) and poly-Si(p<sup>+</sup> ) capping layers as 1.5 <sup>10</sup><sup>20</sup> and <sup>5</sup> <sup>10</sup><sup>19</sup> cm<sup>3</sup> , respectively.

The tunnel oxide in the passivated contact stack not only serves as passivation/ tunneling purposes, but it also likely serves as a blocking layer to reduce the outdiffusion of dopants from the highly doped silicon capping layer into the crystalline silicon wafer bulk. The lower active dopant concentration within the poly-Si(p<sup>+</sup> ) layer can be partially attributed to the lower doping efficiency of boron atoms than phosphorus atoms [67] based on the theoretical prediction of impurity formation energies and partially attributed to the higher diffusivity of the boron dopants [68] into the silicon bulk which resulted in a deeper boron-diffused junction (see Figure 7). Similar to other reports [65], we also observed experimentally that it is preferable to concentrate all the dopants within the poly-Si layers, as the

Double-Sided Passivated Contacts for Solar Cell Applications: An Industrially Viable Approach… DOI: http://dx.doi.org/10.5772/intechopen.85039

Figure 7.

is not too surprising, given the high quantities of dopants (1019–10<sup>20</sup> cm<sup>3</sup>

The corresponding dopant profile within these highly doped silicon capping layers can be extracted from ECV measurements as shown in Figure 7. After an optimized diffusion doping process to convert the thermal-SiOx/a-Si(intrinsic) capping layer stack toward either an electron-selective passivated contact (i.e.,

Raman spectra for our in-house LPCVD of silicon films (250 nm), comparing the film crystallinity in the as-

) and poly-Si(p<sup>+</sup>

The tunnel oxide in the passivated contact stack not only serves as passivation/ tunneling purposes, but it also likely serves as a blocking layer to reduce the outdiffusion of dopants from the highly doped silicon capping layer into the crystalline silicon wafer bulk. The lower active dopant concentration within the poly-Si(p<sup>+</sup>

layer can be partially attributed to the lower doping efficiency of boron atoms than phosphorus atoms [67] based on the theoretical prediction of impurity formation energies and partially attributed to the higher diffusivity of the boron dopants [68]

(see Figure 7). Similar to other reports [65], we also observed experimentally that it is preferable to concentrate all the dopants within the poly-Si layers, as the

into the silicon bulk which resulted in a deeper boron-diffused junction

) stack) or a hole-selective passivated contact (i.e.,

) stack), the ECV measurements revealed a peak doping

porated in the former.

Figure 6.

Silicon Materials

thermal-SiOx/poly-Si(n<sup>+</sup>

thermal-SiOx/poly-Si(p<sup>+</sup>

and <sup>5</sup> <sup>10</sup><sup>19</sup> cm<sup>3</sup>

100

concentration within the poly-Si(n<sup>+</sup>

, respectively.

deposited state, post-POCl3 diffusion process, and post-BBR3 diffusion process.

) incor-

)

) capping layers as 1.5 <sup>10</sup><sup>20</sup>

ECV profiles for both (a) electron-selective passivated contacts comprising thermal-SiOx/poly-Si(n<sup>+</sup> ) stacks and (b) hole-selective passivated contacts comprising thermal-SiOx/poly-Si(p<sup>+</sup> ) stacks. The electron-selective passivated contacts exhibited a higher peak doping concentration than the hole-selective counterpart by a factor of 2 times. Poly-Si(p<sup>+</sup> ) layers also exhibited a higher out-diffusion of dopants into the c-Si bulk than the poly-Si(n+ ) layers, which was found to limit the potentially achievable implied-VOC values (see Table 3).

out-diffusion of dopants is expected to lead to increased surface recombination rates and a corresponding drop in the overall passivation quality as well.

Table 2 summarizes our measured passivation quality results on planar symmetrical lifetime test structures with the optimized doped poly-Si(n<sup>+</sup> ) capping layers on various investigated tunnel oxide candidates (i.e., wet-SiOx, ozone-SiOx, thermal-SiOx). Table 2 shows that using planar symmetrical lifetime test structures, our wet-SiOx/poly-Si(n<sup>+</sup> ) and ozone-SiOx/poly-Si(n<sup>+</sup> ) passivated samples were exhibiting implied-VOC values of 719 mV and <sup>J</sup><sup>0</sup> of <sup>6</sup>–9 fA cm<sup>2</sup> , while the thermal-SiOx/poly-Si(n<sup>+</sup> ) passivated contact stack exhibited an even higher implied-VOC values of 729 mV, despite a similar <sup>J</sup><sup>0</sup> of 9 fA cm<sup>2</sup> . The enhanced implied-VOC values for thermal-SiOx/poly-Si(n<sup>+</sup> ) passivation stack were consistent with the earlier discussion on the tunnel oxides, in which a thermal-SiOx tunnel oxide is likely more effective in reducing the out-diffusion of phosphorus dopants from the poly-Si(n<sup>+</sup> ) into the c-Si wafer bulk, hence providing better overall passivation quality (implied-VOC increases by 10 mV).

Since a typical silicon solar cell would be further coated with suitable antireflection layers (such as SiNx or AlOx/SiNx stacks) prior to metallization, the influence of these layers on our symmetrical lifetime samples were evaluated as well, by capping the passivated contacts with an additional 70-nm-thick SiNx films symmetrically and its resulting passivation quality evaluated. As summarized in Table 2, the measured passivation quality further improves with the additional SiNx capping layers upon all three investigated lifetime test structures with electron-selective passivated contacts. In particular, the thermal-SiOx/poly-Si(n<sup>+</sup> )/ SiNx-capped lifetime structure exhibits high implied-VOC approaching 740 mV, with single-sided <sup>J</sup><sup>0</sup> values down to 2.5 fA cm<sup>2</sup> , which is already on par with the best results from the Fraunhofer ISE team [69]. Concurrently, similar studies were conducted on lifetime test structures with hole-selective passivated contacts, and

selected results are highlighted in Table 3, which demonstrates the potential of our developed hole-selective contact passivation layers as well (i.e., thermal-SiOx/poly-Si(p<sup>+</sup> ) or ALD-AlOx/poly-Si(p<sup>+</sup> ) stacks) with implied-VOC approaching 700 mV in the as-deposited state and a further enhancement to 713 mV with single-sided J<sup>0</sup> values down to 4 fA /cm<sup>2</sup> after applying symmetrical SiNx capping layers. This can be attributed to the hydrogenation process which occurs spontaneously during the deposition of the SiNx capping layer, which helps to reduce the interface defect densities and directly improves the passivation quality [70]. Comparing our results to the excellent results from the Fraunhofer ISE team [69], which adopts PECVD of p-doped a-Si:H layers followed by sintering and SiNx capping (with a high implied-VOC values up to 732 mV and single-sided J<sup>0</sup> values <1 fA cm<sup>2</sup> ), we do identify optimization potential for our LPCVD of intrinsic silicon capping layer and the associated boron diffusion optimization thereafter.

3.3 Evaluation of developed contact passivation stacks on textured surfaces

electron-selective (thermal-SiOx/poly-Si(n<sup>+</sup>

DOI: http://dx.doi.org/10.5772/intechopen.85039

ering contact passivation for both the front and rear surfaces.

Si(p<sup>+</sup>

27.5 fA cm<sup>2</sup>

Figure 8.

103

(thermal-SiOx/poly-Si(p<sup>+</sup>

approaching 700 mV, compare Table 3).

sided J0, rear of 5.4 fA cm<sup>2</sup>

Given the excellent passivation quality from our developed electron-selective and hole-selective passivated contacts on planar Cz silicon wafers, it is then of research and commercial interest to evaluate the performance of these layers on textured surfaces as well, to determine its viability for deployment on a conventional silicon solar cell structure which adopts a front-side textured surface and either a rear-side planar or textured surface. To evaluate that, the lifetime test structures as shown in Figure 2 are utilized, featuring either symmetrical planar surfaces or symmetrical textured surfaces and symmetrically capped by either the

Double-Sided Passivated Contacts for Solar Cell Applications: An Industrially Viable Approach…

)) passivated contacts. The objective is to identify the suitability of our developed electron-selective and hole-selective passivated contacts for textured surfaces as well and to determine the optimum configuration for a silicon solar cell consid-

The highlight of this evaluation is plotted in Figure 8. Firstly, considering the influence of surface conditions on the passivation quality, it can be observed consistently from Figure 8 and summarized in Table 4 that both the electron-selective and hole-selective passivated contact stacks exhibited significantly better passivation quality on planar surfaces than on textured surfaces and which were consistent with the best results shown in Tables 2 and 3. Based on a batch average of 18 samples for each investigated lifetime test structure shown in Figure 8, the holeselective passivated contacts on symmetrical planar lifetime test structures demonstrated an effective minority carrier lifetime τeff of 1650 μs, a single-sided J0, rear of

over the textured case (τeff of <sup>170</sup> <sup>μ</sup>s, single-sided <sup>J</sup>0, rear of 265 fA cm<sup>2</sup>

implied-VOC of 628 mV). Effectively, upon deploying the hole-selective passivated contact on a textured surface, the τeff and implied-VOC reduce by 90 and 8.9%, respectively. Similarly, while the electron-selective passivated contacts continued to exhibit excellent passivation quality on planar surfaces (τeff of 6030 μs, single-

Comparison of the passivation quality (i.e. (a) effective carrier lifetime at 10<sup>15</sup> cm<sup>3</sup> injection level, (b) rear

test structures. It can be observed that electron-selective passivated contacts are suitable for applications on both planar and textured surfaces (with implied-VOC > 720 mV and > 695 mV respectively), while the holeselective passivated contacts are only suitable for planar surfaces at the moment (with implied-VOC

side J0 values and (c) implied-VOC values) by electron-selective (thermal-SiOx/poly-Si(n<sup>+</sup>

, and an implied-VOC of 689 mV, which is a significant improvement

, implied-VOC of 723 mV), the passivation quality

)) passivated contacts on both symmetrical planar and symmetrical textured lifetime

)) or hole-selective (thermal-SiOx/poly-

, and

)) and hole-selective


## Table 2.

Comparison of the passivation quality of electron-selective passivated contacts on planar Cz n-Si symmetrical lifetime samples, both prior to and after the additional hydrogenation process step via the symmetrical addition of the SiNx capping layers.


#### Table 3.

Comparison of the passivation quality of hole-selective passivated contacts on planar Cz n-Si symmetrical lifetime samples, both prior to and after the additional hydrogenation process step via the symmetrical addition of the SiNx capping layers.

Double-Sided Passivated Contacts for Solar Cell Applications: An Industrially Viable Approach… DOI: http://dx.doi.org/10.5772/intechopen.85039
