Hydrogenated amorphous silicon nitride (a-SiNx:H) layer

Get Complete Project Material File(s) Now! »

Passivating contacts: toward higher e ciency

Working principle

It has been mentioned in section 1.1.4 that the direct metal/absorber interface is a source of signi cant recombination losses due to a high defect density at that interface (Metal-Induced Gap States, dangling bonds, impurities, etc.) [24]. Passivating contact architec-tures allow to remove this interface by adding a thin insulating layer (most often intrinsic a-Si:H or SiOx) between the metallic contact and the c-Si absorber. This provides surface (chemical) passivation of the wafer, which limits recombination losses. It also acts as a potential barrier [25]. Then, an extraction layer is added in order to selectively transport one of the two types of charge carriers to the metallic contact while blocking the other one, which is necessary in order to achieve the carrier selectivity (see section 1.1.1) and keep a low contact resistivity. This structure prevents Fermi level pinning and the formation of a Schottky-barrier at the metal/semi-conductor interface and allows to signi cantly reduce the total J0 [26].

Heterojunction technology and limitations

The architecture that allowed to reach the highest power conversion e ciency is the het-erojunction with intrinsic thin layer (HIT) solar cell [27]. It relies on junctions between two di erent materials: c-Si and a-Si:H. The passivated contact is a stack of thin ( 5-10 nm)
(i) a-Si:H/doped a-Si:H ( 10-20 nm)/transparent conductive oxide (TCO) layer such as indium-doped tin oxide (ITO) and a low temperature screen printed silver grid (see Figure 1.10a). a high passivation is achieved thanks to the intrinsic a-Si:H layer which provides a lot of atomic hydrogen that binds to silicon dangling bonds. This hetero-structure also in-duces a strong inversion layer at the hetero-interface, which favors the tunneling of charge carriers through the potential barrier (see Figure 1.10b). Such a structure, along with an IBC architecture lead to the record high e ciency for a single junction solar cell: 26.7 % [20].
However, HIT solar cells have a poor blue response due to the strong parasitic absorp-tion in this spectral range of a-Si:H, and especially of (p) a-Si:H. This is why the emitter is placed at the rear side of the solar cell. However this inverted structure requires high-lifetime silicon wafers since most of the charge carriers are photogenerated at the front side of the solar cell (see equation 1.2) [28]: the minority carriers (holes) need to travel through the wafer all the way to the rear side of the solar cell to be collected. a-Si:H pas-sivation also degrades at temperatures higher than 300 °C and its conductivity is rather low due to a poor doping e ciency. This results in the necessity of using a TCO that is expensive (and often contains indium, a scarce material) and low temperature screen printing metallic pastes (not yet an industrial standard).

Tunnel oxide passivating contacts

In order to maintain these good passivation properties while tackling the issues of high-temperature degradation and high parasitic absorption, a mainstream structure is cur-rently being extensively studied: TOPCon solar cells.
Main architecture tunnel oxide passivated contact (TOPCon) solar cells [29]{[31] use a stack of oxide and doped poly-Si on which the contact is taken (see Figure 1.11a). These structures are resistant to high temperature processes and the crystallization of the silicon layer allows to decrease the parasitic absorption losses due to a-Si:H [32]. The tunnel oxide layer reduces the interface defect density (chemical passivation) while allowing one type of carrier to reach the contact by quantum tunneling or di usion through pinholes in the oxide [33]{ [38]. The energy band diagram of such a contact is shown in Figure 1.11b.
The SiOx/poly-Si passivated contacts were rst implemented in bipolar junction transis-tors allowing a signi cant reduction of the recombination losses [40]. It was later adapted to solar cells for the rst time in the 1980’s under the semi-insulating poly-crystalline sili-con (SIPOS) denomination [41], [42]. An open circuit voltage of 720 mV was achieved (for a 2.5 nm-thick oxide layer and 100 nm of n+ poly-SiOx) but lead to low e ciencies due to a high series resistance. Other studies progressively improved the structure without achieving a major breakthrough in e ciency, until the tunnel oxide passivated contact (TOPCon) concept proposed by Feldmann et al. in 2014 [43]. Since then, numerous structures and denominations have been proposed by di erent research groups (POLO by ISFH [44], PERPoly by ECN Solar energy research group [45], monoPoly—and polyPoly— by SERIS [31], FPC by EPFL [46], etc.) which are all gathered under the denomination TOPCon in this thesis.
Record e ciencies of 25.7% for TOPCon [47] and 26.1 % for POLO with IBC [48] were achieved on lab scale solar cells ( 4 4 cm2), while on industrial wafers (244 cm2) SERIS achieved 22.8 % of PCE [49] and Trina Solar claimed a record e ciency of 24.58 % in mass production [50]. A thorough literature revue was carried out, and state of the art e ciencies along with process, main features and solar cell area are summarized in Table 1.1. The evolution of record e ciencies with time can also be found in Figure 1.12. The table was made in order to give some insight on the best e ciencies for TOPCon solar cells with di erent speci c features: higher constraints from the wafer bulk quality (p-type vs. n-type and Cz vs FZ silicon wafers, see Section 1.4 for the di erence in wafer types), from the size of the cell (lab scale vs. industrial size wafers), from the metallization process (high temperature re-through pastes), or from mass production industrial pilot lines.
A similar work was carried out for the passivation properties of these passivating stacks. Table 1.2 displays some of the highest passivation properties for n-type and p-type TOPCon on di erent kinds of substrate (n-type or p-type, Cz or FZ) and surface preparation (textured, after saw-damage removal (SDR), saw-damage etched (SDE), or double-side polished (DSP)). The results are mainly categorized by the doping of the passivating contact (n-type for the top part of the Table, p-type for the bottom part), and subdivided by bulk wafer type. It can be noticed that boron-doped TOPCon generally provide a lower passivation quality. A few propositions have been made to explain this phenomenon: a larger di usion of boron in silicon, boron-induced defects in the gate oxide [51], [52], as well as SiOx being a larger potential barrier for holes than for electrons [35]. Another limitation to TOPCon integration on c-Si solar cells is the lower passivation it provides on textured surfaces. It was shown to be due to a lower passivation of surfaces with a <111> crystal orientation compared with <100> [53], [54]. In the context of this thesis, the most relevant results are the last 3 lines corresponding to the maximum passivation properties for p-type TOPCon on p-type silicon wafers. These reach up to 735 mV and 5 fA/cm2 after a hydogenation step at the end of the process to boost the passivation properties [55].
Process for passivating contact integration
The large variety of solar cell architectures mentioned above also comes along with very di erent process ows. The doped semi-conductor layer can be made of poly-Si (most of the studies), silicon carbide (SiCx) [46], [55], [56], [68] or SiOx [49], with various deposition techniques such as plasma-enhanced chemical vapor deposition (PECVD) [29], [60], [69], [70], low pressure chemical vapor deposition (LPCVD) [45], [50], [57] or hot-wire chemical vapor deposition (HWCVD) [71], [72]. The doping of this layer can be achieved in situ in PECVD [60] and LPCVD deposition chambers, as well as ex situ by ion implantation [65], [73], [74], or di usion [75]{[77]. The fabrication of the thin oxide layer may be done by nitric acid oxidation of silicon (NAOS) [43], [65], [74], [78], [79], chemical oxidation by Radio Corporation of America (RCA) Cleaning [33], [80]{[82], thermal oxidation in quartz furnace [37], [83] or LPCVD chamber [31], [67], plasma-assisted nitrous oxide (PANO) oxidation [64], [84], [85], PECVD deposition [49], [69], [86], UV-O3 [87]{[89], H2O-O3 [66], [89], [90], atomic layer deposition (ALD) [91], [92]… or even replacing it by AlOx [80] or a-SiNx:H [93]. Many studies have been made on benchmarking the SiOx deposition process [54], [87], [94]{[97]. Thermal oxides usually lead to better passivation properties, but require higher process temperatures.
In the most standard processes, the thermal budget of such structures is high, often requiring a 5-to-15-min long annealing step between 800 and 1000 °C for crystallization and dopant activation (sometimes much more). This is not desirable in industrial processing, and especially for the integration of such structures on multi-crystalline or cast-mono silicon wafers [23]. One way to tackle this issue was presented by Ingenito et al. from EPFL through the development of red passivated contact (FPC) [46]. The aim of this approach is to remove the long-and-high-temperature annealing step, and to limit the high temperature processes to a single ring step (> 700 °C for a few seconds) necessary for the front side metallization. However, one problem that emerges from annealing the silicon layer is the formation of blisters, that are highly detrimental for the electrical properties of the device [62], [98]{[100]. To prevent this from happening, the silicon layer in the FPC must be mechanically resistant to the ring step. Their research group uses SiCx (with a low carbon content of approximately 2.5 at%), which has a larger mechanical resistance than a-Si:H, and a larger optical band gap, thus leading to less parasitic absorption. More details on this process will be given in section 1.5.

READ  PHOTO-REALISM IN VIRTUAL ENVIRONMENTS

Transparent Metal Oxides

Another technology that is being investigated by the PV research community is based on the dopant-free passivating contacts, such as Transparent Metal Oxides (TMOs). The « dopant-free » denomination implies that highly-doped silicon layers are not used in such devices. The optical and electrical properties of TMOs can be tuned in order to meet speci c needs in the silicon solar cells. Noticeably, these materials are wide band-gap semiconductors allowing to reduce parasitic absorption and are dopant-free therefore sup-pressing free-carrier absorption thus tacking the two main limiting factors in poly-Si and a-Si:H-based passivated contacts. TMOs were rst explored in the 1970’s as an alterna-tive to homojunction solar cells [101]. There has recently been a regain of interest in these technologies due to improvements of the passivating layers, the current limitations of solar cell e ciencies by parasitic absorption, and improvements in the manufacturing of high/low work function materials [102] from organic electronics [103]. A thin passivat-ing layer between the TMO and the c-Si bulk is usually required: an (i) a-Si:H layer is deposited or a thin SiOx layer grows at the Si/TMO interface (either naturally [104] or voluntarily), thus providing a good passivation of the silicon wafer surface.
MoOx [30] [102], [105]{[107], WOx [107] and VOx [104] hole selective layers have high work functions that induce a bending of the silicon energy bands, which allows carrier selectivity (see Figure 1.13a). E ciencies as high as 23.5 % were reached for MoOx carrier selective contacts [108].

Table of contents :

Introduction 
Global energy context
Purpose and structure of this work
1 Context, background and objectives 
1.1 Basic concepts of solar cells
1.1.1 Working principle of a solar cell
1.1.2 Characteristics of a solar cell
1.1.3 Limits of conversion eciency
1.1.4 Homojunction solar cells
1.2 Passivating contacts: toward higher eciency
1.2.1 Working principle
1.2.2 Heterojunction technology and limitations
1.2.3 Tunnel oxide passivating contacts
1.2.4 Transparent Metal Oxides
1.3 Metallization of TOPCon solar cells
1.3.1 Screen printing
1.3.2 Low-temperature metallization
1.3.3 Fire-through metallic paste
1.3.4 n-type and p-type TOPCon FT metallisation
1.4 Silicon Wafer manufacturing
1.4.1 From quartzite to polycrystalline silicon rods
1.4.2 Multi-crystalline silicon
1.4.3 Mono-crystalline silicon
1.4.4 Cast-mono wafers
1.5 Objectives of the study
2 Materials development and optimization 
2.1 Deposition tools and methods
2.1.1 Plasma-enhanced chemical vapor deposition
2.1.2 Radio Corporation of America (RCA) cleaning
2.1.3 Substrates and characterizations
2.2 Tunnel oxide layer
2.2.1 PECVD oxide
2.2.2 Wet-chemical oxide
2.3 Boron-doped silicon layer resistant to ring
2.3.1 Avoiding blistering
2.3.2 Impact of µc-Si:H deposition on the tunnel oxide layer
2.3.3 Optimization of the passivation properties of (p) µc-Si:H
2.3.4 µc-SiOx:H
2.4 Phosphorous-doped silicon layer
2.5 Capping layer
2.5.1 Hydrogenated amorphous silicon nitride (a-SiNx:H) layer
2.5.2 Comparison of dierent capping layers
2.6 Summary
3 In situ modulated photoluminescence for AlOx 
3.1 In situ MPL coupled to a PECVD reactor
3.1.1 Introduction to modulated photoluminescence
3.1.2 In situ modulated photoluminescence experimental setup
3.1.3 Diusion phenomena
3.1.4 Temperature correction
3.2 AlOx passivated samples: a case study
3.2.1 Context
3.2.2 Experimental methods
3.2.3 In situ MPL measurements during annealing
3.2.4 Chemical passivation
3.2.5 Field-eect passivation
3.2.6 Investigation of the discrepancy between QSSPC and MPL
3.3 Summary
4 In situ MPL for red passivating contacts 
4.1 Hydrogenation after ring
4.1.1 Context and experimental methods
4.1.2 Annealing temperature and hydrogen diusion
4.1.3 Identication of the dierent contributions
4.2 Hydrogenation before ring
4.2.1 In situ MPL setup for acquisition during a PECVD process
4.2.2 (p) µc-Si:H and (p) µc-SiOx:H under NH3 + H2 plasma exposure
4.2.3 Evolution of the passivation during a-SiNx:H deposition
4.2.4 Cell precursors with p-type and n-type FPC
4.3 Summary
5 Towards large area FPC cells 
5.1 Process transfer on Cz and cast-mono wafers
5.1.1 Context and experimental methods
5.1.2 Textured wafers
5.1.3 Polished surfaces
5.2 Solar cell fabrication
5.3 Summary
Conclusion 119
Optimization of the FPC stack
In situ Modulated Photoluminescence for AlOx
In situ Modulated Photoluminescence for FPC
Towards large area FPC cells
Bibliography 

GET THE COMPLETE PROJECT

Related Posts