Printing technologies for silicon solar cell metallization: A comprehensive review

This paper presents a comprehensive overview on printing technologies for metallization of solar cells. Throughout the last 30 years, flatbed screen printing has established itself as the predominant metallization process for the mass production of silicon solar cells. For this reason, we will provide a detailed review on its history, its evolution over time, and how the continuous efforts of the scientific and industrial community for further improvements revolutionized the entire PV industry. Furthermore, we will guide the reader through the physics on silicon solar cell metallization, the fundamentals on contact formation, and what type of challenges and requirements these topics create for printing technologies. The main topic of this review addresses the flatbed screen‐printing process mechanics, its different process sequences, corresponding screen technology, and the very important impact of paste rheology on the printing result. Finally, we will compare alternative upcoming printing approaches to flatbed screen printing and discuss how they might solve upcoming challenges in metallization in terms of increasing the throughput rates at an ever‐decreasing grid line width and reduced silver consumption.


| INTRODUCTION
Throughout this review, we will attempt to present the reader a comprehensive overview on the unique road printing approaches for PV taken since the beginning of commercial solar cell production in the 1960s. As flatbed screen printing has evolved to become the predominant process when it comes to metallization, we will start by summarizing the history of screen and stencil printing. We will elaborate on the success story and show how this printing technology was able to overcome limitation and challenges over the years and then follow it up by presenting a comprehensive technological background on metallization of Si-solar cells. Further, we will guide the reader through the physical requirements for metallization from a solar cell physics point of view and then discuss those requirements in terms of the technical challenges they create for printing process. In Section 3, we are going to present a summary on the research around the physics and chemistry of the contact formation mechanism and how its understanding helped to push fine line metallization towards printing Ag-electrode widths below 50 μm. In Section 4, we will describe the screen-printing process mechanics and its variations. This is accompanied by an overview on the important aspects of the screen technology and the efforts of the scientific community to further improve its Sebastian Tepner and Andreas Lorenz contributed equally to this work.
design by simulations on screen architectures. In the following section, we present a comprehensive summary on metal pastes and its rheological behavior. The flow characteristics of the printing paste is one of the major aspects when it comes further advancing screenprinted metallization. Finally, we will present the reader an overview on promising alternative printing methods that show great potential to challenge flatbed screen printing in terms of throughput, achievable finger width, and Ag consumption.

| A short history of screen printing for solar cell metallization
The idea to use printing methods for the transfer of conductive circuits on electronic components dates back to the first half of the 20th century and to Paul Eisler, who is commonly-and sometimes controversially-known as the inventor of the printed circuit board (PCB). [1][2][3] In the early years of photovoltaics (PV) since the development of the first silicon solar cell at Bell Laboratories in 1954, 4 the metal contacts were realized in a cost-intensive and time-consuming process by electroplating, evaporation, or sputtering of various metal layers (i.e., Ti-Pd-Ag) through a photolithographic mask. [5][6][7] Until the mid-1970s, fabrication of silicon solar cells was a highly specialized, non-automated sequence of processes primarily for space applications with a very low annual production volume. The gradually increasing production volume in the 1970s and a growing terrestrial application of PV triggered a growing demand to automate and accelerate the production process. 8  contacts with a sufficiently low contact resistance and minority carrier recombination on high-resistivity p-type silicon. 9 This challenge was solved by forming an aluminum p + -doped BSF using high-temperature in-diffusion of aluminum (Al) from a screen-printed Al paste. On the front side, the contacts were screen printed using a metal paste consisting of silver powder, glass frit, and organic binders. 9 A major challenge of these early research activities was screen-printed metal contacts on the front side of high-ohmic n-type emitters with sufficiently low contact resistance and without shorting the p-n-junction. 8 During the 1980s, Al BSF solar cells rapidly evolved to the predominant industrially fabricated solar cell concept, which eventually lasted for more than 30 years. 10 In the late 1970s and early 1980s, a series of independent research activities quickly achieved progress by optimizing the paste formulation (glass frit, metal particles, and binders), 8,11,12 rapid-thermal processing (RTP) conditions, 8,11,12 and emitter profile. 11,12 Screen-printed front-side electrodes (also referred to as contact fingers in the following) typically had a width around 150 μm at that time. 11 To improve the lateral conductivity of the contacts and provide a well-solderable surface particularly on the rear side, it was common to apply a subsequent solder dip-in some cases on a previously electroplated Cu layer-after screen printing of the metal pastes. [11][12][13] This complicated process could be overcome by developing fire-through silver (Ag) pastes, which were able to obtain a sufficiently low contact and line resistance as well as a good adhesion strength after RTP without penetrating through shallow pn-junctions. 14 This could be realized by newly developed rapid thermal co-firing processes using high-energy infrared (IR) lamps to efficiently form the front side contact and the BSF on the rear side. 15 By the early 1980s, screen printing had already become a wellestablished method to apply the metal contacts on industrial scale. 16 Research activities at this time focused on replacement of costintensive silver by non-noble metals and the applicability of screen printing for other production steps like junction formation using dopant pastes 13,16 as well as anti-reflection and protective coating layers. 16 Furthermore, the impact of the metal-semiconductor contact on series resistance of the solar cell and the optimization of paste formulation and RTP process attracted increasing attention. [17][18][19] An early approach to overcome the limitations of the screen printing process with respect to the printed finger width (and thus shading losses) were so-called "buried contact solar cells." 20,21 This approach proposed mechanically or laser grooved trenches that were filled with Ag paste in a subsequent fine line screen printing process 22,23 and was investigated until the beginning of the early 2000s 24 until it has completely vanished. Further research activities in the 1990s focused on optimizing the screen printing process in order to achieve finer lines, that is, in combination with selective emitters. 25 Furthermore, first studies regarding the impact of paste rheology 26 and texture roughness 27 on the screen-printing results were carried out.
The development of the passivated emitter and rear contact (PERC) solar cell concept in the late 1980s based on the work of Blakers et al. 28 and previous conceptual considerations 29 was a further important milestone in the history of industrial PV mass production. The introduction of an additional dielectric passivation layer on the rear side between silicon and Al metallization effectively decreased recombination losses and improved the optical mirroring properties on the rear side. [30][31][32][33] A comprehensive overview of the working principle and current status of the PERC technology and related PERx cell concepts is provided by Preu et al., 31 Dullweber and Schmidt, 33 and Green. 34 In the early 2000s, the physical and chemical contact formation mechanisms of fire-through Ag pastes on n-type emitters as well as the microstructure of the metal-semiconductor contact interface attracted strong attention in the research community. 35, 36 Schubert et al. [37][38][39] proposed a comprehensive model to explain the chemical and physical formation as well as the current transport mechanisms between the screen printed silver contacts and the n-type emitter.
Multiple research studies in the following years formed the basis for today's understanding of the contact formation mechanisms and the development of high-performance fire-through pastes, which are today able to even contact extremely lowly doped emitters with low contact resistance. 38 The Al BSF solar cell was the unchallenged and predominant industrially fabricated cell concept until 2012/2013. Around this time, the technical possibilities for further optimization of this cell concept had been exhausted, and the cell efficiency stagnated at a level of $20 % on Cz-Si. 30 From this time on, the PERC solar cell gradually superseded the Al BSF solar cell as the predominant technology. The PERC cell concept enabled a further substantial increase of conversion efficiency beyond this level with an impressive average yearly improvement of approximately Δη ≈ 0.6 % abs throughout the last years. 30,33 Starting in 2002, new approaches to replace the originally applied evaporation process 40 for the rear side metallization of PERC cells with cost-effective and industrially feasible methods were investigated. A promising approach that finally did not reach the state of a broad industrial application was the so-called laser-fired contact (LFC) process. 41 Another approach focusing on the development of a firingstable SiN x passivation and diffusion barrier layer with local lasercontact opening (LCO), subsequent screen printing of Al paste, and high-temperature co-firing [42][43][44][45] has established itself as the standard technology for industrially fabricated PERC solar cells until today. 31 In the 2010 years, the alloying process of Al and Si as well as the formation of the local BSF has been intensely investigated [41][42][43] ; a comprehensive overview is provided by Dullweber and Schmidt, 33 Riegel et al., 46 and Rauer et al. 47 The ongoing efforts to further raise conversion efficiency of PERC solar cells demanded for fine line screen printing processes, which enable minimal shading and metal-related recombination losses as well as lower silver consumption by reducing the metallized fraction on the front side. However, such fine line metallization processes had to maintain a sufficient lateral conductivity of the grid as well as a sufficiently low contact resistance even on lowly doped emitters to keep series resistance losses at a minimum. Various approaches like screen printing of hotmelt Ag pastes, 48

| Screen printing meets carrier-selective contacts
While the impact of the bulk and rear surface as recombination channels has been effectively decreased in modern PERC solar cells, recombination losses related to the front side emitter and the metal contacts remain as important limitation factors for the electric performance of modern high-efficiency PERC cells. 85 Exceeding the efficiency of PERC cells towards the theoretical limit of 29.4 % 86 for silicon-based solar cells thus requires an effective suppression of surface recombination related to the metal contacts (expressed as dark saturation current density j 0,met ). 87 A promising path towards this goal are solar cell concepts with carrier-selective contacts (often also denoted as "passivating contacts"), which provide a good transport of majority carriers to the metal contacts on the one hand and effectively block the minority carriers on the other hand. 85 A comprehensive overview regarding carrier-selective contacts is provided by Hermle et al. 30 and Schmidt et al. 85 Until today, two particularly promising approaches for carrierselective contacts have found their way to an industrial large-scale production. The silicon heterojunction (SHJ) solar cell concept 88 using intrinsic and p-/n-type hydrogenated amorphous silicon (a-Si:H) layers for surface passivation has been originally developed 89 91 Thus, modules based on SHJ solar cells have successfully entered the global PV production landscape with an expected strongly increasing market share throughout the coming years. 92 Screen printing will again be the method of choice for metallization of SHJ solar cells on industrial scale. However, the temperature sensitivity of the a-Si:H layers limits the maximum processing temperature to $200-230 C. 93 Applying screen printing for the front and rear side metallization thus requires an application of specific low-temperature (LT) curing pastes, which have been intensely investigated and optimized on academic and industrial level throughout the last decade. [94][95][96] The second industrially relevant approach for carrier-selective contacts is based on a passivation stack consisting of an ultra-thin F I G U R E 1 (A) Evolution of screen and stencil printed finger width from 2008 to 2020 based on published results 32,53-83 on silicon solar cells as well as actual results using a test layout with advanced screen technology. The graph is modified and updated based on previously published versions. 31,79,84 (B-D) Comparison of SEM cross-sectional images of typical screen printed fingers using fire-through Ag paste from 2013 to 2020 SiO x tunneling layer, a heavily doped, partially crystallized polysilicon layer (poly-Si) and a SiN x capping layer. 85,87 This approach-in the following denoted as SiO x /poly-Si contact-was originally developed in the 1970s as passivation layer for silicon devices 97 and first applied for the so-called semi-insulating polysilicon solar cell (SIPOS) by Yablonowitch et al. 98 Further solar cell concepts based on SiO x /poly-Si contacts were evaluated in the following. [99][100][101] A milestone towards an industrial implementation was the development of the so-called tunnel oxide passivated contact (TOPCon) solar cell by Fraunhofer ISE, which was first presented in 2013. 102,103 Further related cell concepts like Poly-Si on Oxide (POLO), 104 monoPoly TM , 105 and PERPoly 106 are based on the same idea. The outstanding surface passivation quality, the compatibility with established high-temperature processes, and the comparatively easy integration of this cell concept into existing PERC production lines are particularly attractive for the massproduction on industrial level. Industrially fabricated i-TOPCon solar cells and modules have demonstrated impressive efficiency results and currently gain a rapidly growing market share within the PV landscape. 107,108 Once more, screen printing has proven to be the method of choice as a highly productive, reliable, and easy-to-handle metallization method for this solar cell concept.

| Screen printing for tomorrow's highefficiency solar cell concepts
Boosting the conversion efficiency beyond the theoretical limit of silicon-based solar cells requires more than one absorber material.  111 and are already fabricated on industrial scale. 112 While the metallization of tandem solar cells is currently mainly realized by evaporation, 113 the scale-up of tandem solar cells for industrial mass production requires efficient and productive methods for the metallization. Not surprisingly, the "old bull" screen printing once again recommends itself as the method of choice due to its longstanding and captivating benefits. However, new challenges arise due to the sensitivity of Perowskites to environmental factors like oxygen, moisture, UV light, chemicals (i.e., solvents), and temperature, which lead to substantial degradation. 114,115 Applying screen printing for the large-scale metallization of this type of cells thus requires sufficient metallization pastes and drying/curing methods to guarantee a sufficiently low lateral resistance at curing temperatures as low as 100-150 C. 113 Research and development for this field of application are just at the beginning, and very few studies addressing this topic have been published so far. 113 However, it is very likely that screen printing will prove its powerful ability to cope even with highly sophisticated challenges and still prevail its unrivaled benefits for industrial high-volume metallization of tandem cells once again. • Sufficiently low lateral resistance (grid resistance) of the front/rear grid to maintain a low series resistance contribution depending on the selected interconnection scheme. 118 • Sufficiently low contact resistance of the metal-semiconductor contacts. 118 • Low recombination velocity (metal-related saturation current density) underneath the metal contacts. 118,119 • Sufficient adhesion of the interconnection ribbons or wires using soldering or conductive adhesive. 120 The optimal design and electrical performance of the metallic electrodes are closely connected to the selected interconnection scheme, the solar cell concept, and the material parameters of the semiconductor. All these factors as well as the technical limitations of the metallization process have to be considered when optimizing the metallization layout for a certain solar cell design.

| Interconnection concepts
Within the last decades, a multitude of different cell and interconnection concepts has been developed. To date, solar cell interconnection using 6 ribbons or 9-12 wires with lead-containing solder is the predominant method for cell interconnection in the PV industry with a cumulated market share of more than 90% in 2019. 89 The traditional front side H-pattern grid consists of 3-6 busbars with up to 150 fine narrow contact fingers. Using modern tabbing/stringing machines, solar cells with up to 6 ribbons down to a width of 0.6 mm can be interconnected. Throughout the last years, interconnection of solar cells by soldering 9-12 solder-coated wires onto regular pads on the front/rear side has gained a strongly increasing market share particularly in the Asian PV production. Modern high-efficiency solar cells with a full size format of 156 mm Â 156 mm or more usually have a comparatively high current, which induces substantial resistive power losses on module level. 121 An effective way to prevent these power losses is the reduction of the cell current by separating the cells on half instead of full size. 122,123 Using this approach, the cells with a suitable metallization layout are separated to half cells by scribing and mechanical cleaving or thermal laser separation (TLS). 124 Modules based on half cells have gained a strongly increasing market share within the last years. 92  host cell with a special, screen-printed metallization layout on the front and rear side. Each cell stripe has a printed solder pad/busbar along the edge, a full-area aluminum (Al) rear side metallization, and a grid of contact fingers on the front side. The cutting process of the strips from the host cell can be realized, that is, by laser scribing mechanical cleaving (LSMC) or TLS. 134 The separation process is critical for the cell performance with respect to micro-cracks and recombination losses of the cutting edge to obtain a high efficiency of the shingle cells. An additional post-metallization passivation of the cutting edges is thus important. 134 For module interconnection, the cell stripes are interconnected to shingled strings by overlapping and connecting the rear busbar of one stripe with the front busbar of the next stripe. Due to the constrained movement possibility of the interconnected shingle cells, thermo-mechanical stress on the joint interconnections within the module is much more critical compared to standard cell interconnection via soldered ribbons. 135 Electrically conductive adhesives (ECAs)-possibly in hybrid combination with nonconductive adhesive-are assumed to fulfill the mechanical and electrical requirements best. 135

| Optimization of the front side metallization: A narrow ridge
With focus on the screen printing process, realizing fine line grids on the front-and in case of bifacial solar cells on the rear-side is particularly challenging. The front side metallization of a solar cell has to combine an optimal trade-off between shading of the metal grid (finger shape and width, and number of contact fingers), series resistance contribution (lateral grid resistance and contact resistance), and silver consumption (uniformity of fingers). It is thus essential to optimize the grid layout individually depending on the selected interconnection concept, the material parameters of the semiconductor, and the performance of the selected metallization technology. To minimize optical losses due to shading, the grid should reflect as less of the incoming light as possible from the active area of the (encapsulated) solar cell in the module.
The optical quality of the contact fingers (considering the actual reflection losses in the module) is referred to as effective or shading-relevant finger width w f,eff . Models for the calculation of w f,eff have been proposed, that is, by Blakers 136 and Woehl et al. 137 The ideal case would be a nearly or totally transparent grid, which could theoretically be realized by a triangular shape of the contact fingers. 138 Yet realizing such a finger geometry using state-of-the-art printing technologies seems to be extremely challenging or even impossible. In reality, the crosssection of typical screen-printed contact fingers is Gaussian-shaped or parabolic ( Figure 2A). The electric quality-namely, a sufficiently low series resistance contribution r s,grid of the grid-is of great importance to achieve a high fill factor (FF) of the cell. The optimal trade-off between shading and series resistance contribution is strongly affected by the number of busbars/wires and contact fingers 140 and can be optimized using appropriate simulation tools. 141 The maximal tolerable lateral finger resistance R L for a defined series resistance contribution limit r s,f of the fingers can be calculated for a given interconnection scheme and grid layout according to Mette, 139 Fellmeth, 142 and Lorenz. 143 Table 1 and Figure 2B provide an impression of the maximal acceptable mean lateral finger resistance R L,max for a given finger series resistance contribution limit of r s,f = 0.1 Ωcm 2 calculated for various common interconnection concepts. The results underline the strong relation of the tolerable lateral finger resistance and thus the required electrical quality of the contacts with the selected interconnection concept. Consequently, interconnection concepts with an increasing number of busbars/wires are particularly attractive for solar cells with a higher lateral resistance of the front side contacts, which is the case, that is, for screen printed ultra-fine line grids and metallization concepts using LT pastes like SHJ and tandem cells.

| Rear side metallization
The rear side of a solar cell fulfills several functions, namely, minority carrier transport, passivation, photon mirroring, mechanical adhesion, and-in case of bifacial solar cells-photon collection. Design, properties, and resulting processing parameters of the rear side metallization strongly depend on the selected cell and interconnection concept.
With an estimated market share of over 98% in 2020, 92 flatbed screen printing is the predominant method to apply the rear side metallization for most industrially fabricated silicon solar cells. Depending on the cell concept, the rear side metallization can be generally distinguished in full-area and structured designs. Monofacial cell concepts that collect the incoming light only from the front side currently dominate the market with a production share of 80% in 2020. 92 The rear side metallization of a typcial monofacial p-type PERC solar cell consists of separately printed Ag pads and surrounding full-area Al metallization.
In the first step, a layout of regularly arranged solder pads is printed using non-fire-through Ag paste. The dimension and number of the Ag pads are optimized with respect to the soldering process and the longterm stability of the interconnection. The impact on the electric performance of the solar cells can be neglected. 144 In the second step, the surrounding area of the backside with exception of a narrow edge region is coated by screen printing of Al paste, which is particularly unsophisticated from a printer's point of view. However, both printing steps require a stable and well-controlled application of a defined wet paste layer thickness within a defined production tolerance, which is determined by the selection of the screen mesh (see Section 3.5), the paste rheology (see Section 4), and-to a small extend-by the selected printing parameters (see Section 3.2). 145 During production, the layer thickness is usually controlled indirectly by monitoring the applied mass of wet paste on each cell using precision inline scales.
While the rear side metallization of monofacial cell concepts is technically unsophisticated, this is not the case if fine grid-like structures have to be printed on the rear side, which is the case for bifacial solar cell concepts or IBC solar cells. Bifacial solar cells-a concept originally developed in the 1960 146 and later applied for bifacial concentrator cells by Luque et al. 147 -collect the incoming light not only on the front side but also from the albedo, which significantly increases the output power density compared to monofacial cells/ modules. 125 Bifaciality can be applied for various cell concepts like PERC, 148,149 Heterojunction, 150 and TOPCon. 151 With respect to the rear side metallization, a grid is applied on the rear side instead of a full-area metallization. In an ideal case, the rear side grid should cover a small fraction of the cell area to minimize shading losses and maintain a sufficiently low lateral grid resistance and metal-semiconductor contact resistance to reduce series resistance losses. Applying a rear side grid on solar cells with rear side passivation requires a precise local opening of the passivation layer(s) or a fire-through approach to contact the semiconductor. Using screen printing, a precise alignment of the printed metal grid on the previously opened structures in the passivation is required. An alternative approach that avoids the previous local opening of structures has been demonstrated by screen printing of Al fire-through pastes, which etches through the rear side passivation and forms a local BSF on the p + -doped silicon surface within the subsequent co-firing process. 152 Using this approach, promising results have been demonstrated recently by applying multilayer screen printing or multi-nozzle dispensing for the rear side metallization of bifacial PERC solar cells. 153,154 2.5 | Metal-semiconductor contact and metalinduced recombination Realizing a low-ohmic contact between the (screen-printed) metal electrodes and the silicon semiconductor has been a major challenge since the beginnings screen-printed solar cell metallization. 19,145,155 Starting in the early 2,000 years, the chemical-physical nature of firethrough Ag contacts on n-type silicon gained strong attention [35][36][37]156 with Schubert et al. [37][38][39] being the first to present a comprehensive model of the contact formation process. The gradual establishment of the PERC technology as the dominating cell concept additionally required a deep understanding of contact formation and alloying processes using Al paste on p-type silicon surfaces, which was intensely investigated in the 2010s. 46,47,[157][158][159][160] The introduction of new cell concepts based on n-type material shifted an additional focus on contact formation of Ag/Al and Ag pastes on p + -emitter surfaces. [161][162][163][164][165][166][167][168][169][170][171][172] Realizing screen-printed contacts on passivation layers with carrierselective properties, that is, for TOPCon solar cells, represents the latest challenge regarding contact formation. 30,85,103,104,106 Beside the long-lasting challenge of forming a low-ohmic metal- the metallized fraction. 119 New approaches determine j oe,met with a simulative approach using PL imaging data combined with numerical PL simulations, 179,182 which has proven to be significantly more precise as they consider the non-uniformity of the excess carrier density Δn. The negative effect of metal-related recombination losses on the V oc of the solar cell can be reduced by various approaches 119 : the consequent reduction of the (fire-through) metallized fraction on the surface of the solar cell, that is, using non-contacting paste for the busbars, 183 applying metallization methods that induce a lower j oe,met , effectively shielding the minority carriers from the surface by a fieldeffect passivation (i.e., selective emitter), or introducing an effective surface passivation with carrier-selective properties ("passivating contacts"). Due to the complexity of this topic with respect to various specific metal-silicon interfaces and contact scenarios, this work can only provide a brief overview regarding the current state of science and technology.
Based on manifold studies, the contact formation process and the properties of the metal-semiconductor interface of fire-through silver paste on n-type silicon emitters have been deeply investigated. Today, it is mostly accepted that several current paths contribute to the current flow between the silicon semiconductor and the printed and fired silver grid, 38,[184][185][186][187][188][189] namely, local direct contacts between silver bulk and the silicon surface, 16,145 silver crystallites formed on the silicon surface during the contact formation process, 36 and tunneling of charge carriers through glass layers enriched with dissolved silver and silver colloids [190][191][192][193][194][195] as shown in Figure 3. A further important finding discovered the crucial role of electrons during the firing process. 196,197 Depending on the emitter type and the polarity of current injection, electron injection during firing enhances or suppresses the formation of silver crystallites on the emitter and silver colloids/nanocrystallites in the glass and thus strongly affects the resulting macroscopic contact resistance. [196][197][198] The presence of busbars in the grid layout significantly increases the probability and propagation of such local "short-circuit spots" and thus negatively affects the macroscopic contact resistance. 199 Constant optimization of the silver paste composition for n-type emitters throughout the last two decades enabled an impressive decline of specific constant resistance ρ c from values in the range 39,200 of 10 3 to 10 4 mΩm 2 to values in the lower one-digit range. [201][202][203] Actual silver fire-through silver pastes reliably obtain very low specific contact resistances below 1 mΩcm 2 on industryrelevant emitters ( Table 2). Intense effort to optimize the emitter, passivation, and paste properties enables a dark saturation current density of the passivated, non-metallized region ( j 0e,pass ) in the range of 20 fA/cm 2 ≤ j 0e,pass ≤ 60 fA/cm 2 for typical industrially fabricated p-type PERC solar cells. The metal-induced dark saturation current density can be estimated in the range of $500 fA/cm 2 ≤ j0,met _front ≤ 800 fA/cm 2 (see Table 2). 31,82,202 Beside the commonly established n-type emitter, p-type emitters with boron doping (p + -B-emitter) have gained strongly increasing attention throughout the last decade. Such p + -B-emitter is applied for various high-efficiency cell concepts, that is, n-type bifacial PERC solar cells. 233  emitter. 162,165,234 Actual Ag/Al pastes obtain a specific contact resistance in the range of a few mΩcm on typcial p + -B-emitters (see Table   2). However, fire-through Ag paste without additional Al content has also been successfully applied to contact p + -B-emitters 235-238 even with low contact resistance around ρ c ≈ 1 mΩcm 2 , which seriously questions the compelling necessity of Ag/Al pastes to contact p + -Bemitters. 40 The contact formation process of Ag/Al pastes on p + -Bemitters 172,239 exhibits fundamental differences compared to Ag pastes on n-type emitters. A major challenge is the mitigation of deep Ag/Al spikes during the contact formation process, 165,170,172 which can lead to significant FF losses due to local shunting 165,170 as well as considerable losses due to metal-induced surface recombination. 171 Similar to Ag pastes on n-type emitters, a strong impact of electron availability during the contact formation on p + -B-emitters process has been found. 197 While the challenge to realize a low-ohmic contact on p + -B-emitters has been largely overcome with actual pastes, further reduction of j0,met _front losses is still an existing challenge. A promising approach is the application of boron emitter profiles with considerably deeper junction depth. 171,240 Recently published results confirm the benefit of such deep emitter profiles with respect to lower j0,met _front losses without deteriorating ρ c . 213,214 Focusing on the rear side of the solar cell, the (screen-printed) electrode can either be applied in form of a full-area pattern (monofacial cell concepts) or a grid-like pattern (bifacial cell concepts). In both cases, the screen printed rear side electrode has to provide a lowohmic metal-semiconductor contact with preferably minimal j0,met _rear losses after the co-firing process. On a typical industrially fabricated monofacial p-type PERC solar cell, the full area rear side passivation stack (most commonly consisting of an Al 2 O 3 passivation layer covered by a hydrogenated SiN X capping layer 31 ) is locally removed by local point-or dash-shaped laser ablation (LCO). 241 Subsequently, the rear-side metallization consisting of more or less regularly arranged silver solder pads and a surrounding full area aluminum metallization is applied by a two-step screen printing process. 241 Within the following co-firing step, a complex Al-Si alloying process takes place, which provides a sufficiently low-ohmic electric contact between the rear side electrode and the p-type silicon bulk and in parallel induces the formation of a local aluminum-doped back surface field (Al-BSF), which effectively prevents surface recombination underneath the metal contacts. 33,47,160 This type or rear-side contact usually exhibits a specific contact resistance in the range of 3 mΩcm 2 ≤ ρ c ≤ 5 mΩcm 2 .
The metal-induced dark saturation current density can be estimated in the range of 500 ≤ j0,met _rear ≤ 800 fA/cm 2 (see Table 2).
Solar cell concepts based on carrier-selective SiO 2 /poly-Si contacts (see chapter 1.2) have established themselves as industrially feasible next generation high efficiency solar cells. 30,85 On industrial scale, the metal grid is applied on the front side p + -B-emitter and the rear side SiO 2 /poly-Si/SiN x passivation layer using screen printing.
N-type SiO 2 /poly-Si contacts are metallized with Ag paste, 107,242 p-type SiO 2 /poly-Si contacts either with Ag/Al, 223 or Ag paste. 228 Contact formation of fire-through metal pastes on a carrier-selective SiO 2 /poly-Si passivation is rather demanding. During the high-temperature co-firing process, the metal paste has to locally penetrate through the SiN x capping layer to form a low-ohmic contact. However, the penetration and degradation of the SiO 2 / poly-Si passivation layers by the aggressive glass frit must be avoided in order to preserve the outstanding passivation properties of the carrier-selective contact. 30 This challenging task requires specifically optimized metal pastes with adequate glass frit composition 108,224 as well as a careful adaption of the contact firing conditions. 223,229 The majority of current research activities focus on n-type SiO 2 /poly-Si contacts. More questions still arise with respect to screen printed metal contacts on SiO x /p + poly-Si passivation layers, a field of research that has so far been considered to a much smaller extend. 30,85,228 Table 2 provides an actual overview of the typical range of contact-relevant parameters ( j 0,pass , j 0,met , ρ c ) for selected silicon surfaces based on published results and recent achievements on industrial fabrication.
T A B L E 2 Overview of the typical passivation/metallization architecture and the typical currently obtained parameter range for screen-printed metallization on selected surfaces based on recently published results and empirical data

| Characterization of screen printed contacts
Applying the metal contacts using screen printing is a high-precision production step that requires adequate characterization methods to assess the quality and reliability on cell and module level. A comprehensive characterization of the metallization quality has to consider the following aspects: • Optical quality of the metal grid on the front and-if necessary-on the rear side including the total metallized fraction of the active cell area and the (effective) contact finger width.
• Electric quality of the front and rear side metallization comprising grid resistance, lateral finger resistance, contact resistance, and-in case of full-area rear-side metallization-sheet resistance of the Al layer and the Ag pads.
• Surface-related recombination activity in the non-metallized and metallized areas (dark saturation current density of passivated and metallized areas).
The relevant parameter to assess the optical quality of the screen printed front (and respectively rear) side metallization is the total met- should be determined with adequate image analysis algorithms 243,244 to avoid subjective influences and measurement errors of manual measurements and ensure a reproducibility. Recommended threedimensional geometry parameters are shown in Figure 4A-D. In addition, scanning electron microscopy (SEM) images, that is, of the finger cross-section, are recommended to analyze the contact microstructure of bulk and interface. To assess the electric quality of the contact fingers related to the finger geometry, Pospischil et al. 245 have introduced the electric aspect ratio AR el : In an optimal case, contact fingers should have a large AR el and minor variations of finger height h f , width w f , and cross-sectional area A f along the length of the finger in order to maximize the optical and electrical performance and efficient use of silver paste.
The front-and rear-side metallization contributes to the total series resistance R s and thus the FF of the solar cell ( Figure 4E).
With respect to electrical characterization, several parameters of the front and rear side metallization are of importance (Table 3)  • Yield loss and uptime.
• Transport system (linear belt transport and shuttle system).
• Handling of critical wafer material, particularly very thin wafers.
• Amount of screen printing units per lane (three units for single printing and four units for double/dual printing).
• Design of printing system (fixed table, moving table, and rotary   table).
• Variability regarding screen format.
• Positioning system and alignment accuracy.
• Inline control systems (i.e., inline inspection of printed front side grid) and possibility for closed loop process control (i.e., for alignment correction).
• Flexibility and configuration of the sorting unit. This motion pushes the paste into the screen opening until contact with the substrate is reached. The screen tension causes the screen to snap-off back to its initial position, leaving behind the printed structure on the wafer.

| Printing sequence
Reducing the finger width w f and in parallel maintaining a low lateral resistance of the grid requires an increasing electric aspect ratio AR el and in consequence a sufficient finger height h f and cross-section area A f . An approach to fulfill both goals-reducing w f and increasing AR elis the application of two subsequent screen printing steps for the front side metallization instead of one. This approach, which is commonly known as double or dual printing, requires modern screen printing machines with a very small alignment tolerance between the two printing steps. A further advantage of two subsequent printing steps is the possibility to use different Ag pastes for the first and second print. The paste used for the first printing process can be optimized with respect to contact formation, while the paste for the

| Double printing
Printing the same grid layout in two consecutive printing steps onto the front side of the solar cell is commonly known as double printing or print-on-print (PoP) process. Usually, the full H-pattern grid (fingers and busbars) is printed in the first step and a finger grid without busbars in the second step to increase the aspect ratio of the contact fingers but reduce the overall silver consumption. 256 Applying double printing enables narrow contact fingers with high aspect ratio due by increasing the finger height. 50 Schematic illustration of the screen printing process. In Step A, the squeegee is moving over the screen and fills the mesh with paste. In Step B, a printing squeegee pushes the screen down to the substrate until contact is reached. The mechanics of this movement induce a snap-off behind the squeegee, where the screen moves back to its initial position, while paste remains on the substrate at the defined location. 81 substantial efforts. During the production process, the alignment quality between the first and second print deteriorates with increasing number of process cycles. The failure of one screen thus requires an exchange of both screens, which is economically unfavorable.

| Dual printing
Printing the contact fingers and the busbars in two separate, consecutive steps is commonly known as dual printing. Dual printing has been originally applied for stencil printing processes as intersecting elements like busbars and fingers cannot be combined in one stencil layout. 256 However, dual printing is also applied for screen printing processes using two different screens or hybrid processes combining screen and stencil printing. 67 Printing the front side metallization of solar cells with a dual printing sequence means that the busbars are printed in the first step, that is, with non-contacting silver paste. The contact fingers are printed in a second step using, that is, conventional fire-through Ag paste. Dual printing offers several benefits compared to double printing. The possibility to use separate pastes for busbars and fingers has several positive effects on the performance of the solar cell and the Ag consumption, particularly for cell concepts with fire-through contacts on the front side: Using non-contacting Ag paste for the busbars, the total contact area between front side metallization and the emitter surface is substantially reduced. Thus, losses related to V oc due to metal-induced recombination 119,173 and to FF due to the short-circuit effect are significantly lower. 199 Finally, the requirements regarding alignment precision are less critical compared to double printing. Specific paste compositions for dual printing are commercially available today. Furthermore, it offers the opportunity to use separate, optimal screen configurations for finger and busbar printing, which allows for an individual optimization of the printing result and silver consumption. Fine mesh screens can be used to print the fingers with narrow width, high aspect ratio, and good uniformity.
The printing of busbars can be realized using suitable screens to optimize layer thickness, lateral conductivity, adhesion and soldering properties, and low silver consumption. Dual printing processes currently have a market share of around 20%, which is expected to increase throughout the next years. 92

| Screen technology
The screen is one of the most important components when it comes to fine line screen printing because it serves as the guiding geometry for the entire paste flow. In today's state of the art metallization of Si-F I G U R E 7 Gasket seal effect of the emulsion on the textured surface of the silicon wafer: On a rough wafer surface/texture, the sealing is less effective, which leads to substantial spreading along the channel edge (A). A smoother surface/texture of the wafer leads to a better sealing of the channel edges and thus decreased paste spreading (B). 260 Solar cells, the industry uses mainly screens consisting out of a structured emulsion mask layer on top of a quadratic woven metal mesh. Table 4 presents an overview on common screen and mesh parameters which have been investigated in various scientific studies.
This parameter is crucial in screen printing because it dictates the screen life time, which is one of the most important parameters in an industrial production environment. 266 The wire-to-wire distance d 0 can be calculated with the known mesh count MC (in 1/inch) by using Equation (4).
After the mesh is defined by its mesh count MC, the corresponding wire diameter d, the wire material with its maximal tensile strength σ uts_wire_mat , and the corresponding rate of calendaring, the emulsion layer with a certain thickness EOM (emulsion over mesh), are applied and structured on top. 264 The combination will result in a screen opening with a width w n and a certain length l (usually defined by the dimensions of the substrate or desired pattern). Furthermore, Figure 8A shows that the opening channel is aligned at a certain angle φ in respect to the quadratic mesh (or frame, respectively).
The opening rate OA is the most important parameter in order to describe this geometric apparatus and is defined by Equation (5). 267 OA State of the art screens are usually made with a screen angle

| Screen simulation approach
As discussed above, the opening rate of the screen opening w n is one of the most important parameters when it comes to evaluating the expected printing quality of a screen. However, it only describes the average opening rate. As seen in Figure 9, the regularity of the opening rate OA actually depends on the screen parameters (e.g., screen angle φ). This was studied by Ney et al. 263 and Tepner et al., 83,270,276 who both developed a mathematical model which is able to calculate the average deviation of the opening rate OA across the length l of the screen opening. They found that increasing the screen angle will reduce the average deviation, suggesting that screen architectures with screen angles above the overall industry standard of φ = 22.5 should be explored in more detail. On the other hand, it should be noted that an increase of the screen angle will also increase the overall production cost of the screen because more mesh will be wasted due to the angled cutting into respective sheets. Tepner et al. presented a full analytical description of the cutting losses based on the screen angle and the dimensions of the weaving machine.

| Process modeling and simulation approaches for screen and stencil printing
In the 1980s, the first significant attempts to provide a theoretical model of the screen printing process were done by Riemer, [253][254][255]278 who mathematically described how the squeegee orientation creates a pressure distribution within the paste volume in front of the squeegee. Figure 10 306 Hsu et al., 307 Thibert et al., 308,309 Jewell et al., 273,310 Potts et al., 311  • Which values on the right y axes for the shear rate are relevant in today's screen printing process and how are they linked to the process and screen parameters?
• Which values then result on the left y axes for the viscosity of the investigated paste and how to measure it in the relevant shear rate regime?
• How exactly unfolds the presented curve over time during the printing process?
• How to directly test this model by experiment?
• What are the limits of this model and what other phenomena may have to be included in order to predict the flow behavior of the investigated paste during the process steps (e.g., slip effects)?
These five questions summarize the current and recent research activities in this field and the above mentioned researches published over the last ten years significant contributions in answering them.
They examine very similar types of rheological measurements to determine the shear rate depending viscosity in low and high shear regimes, the corresponding yield stress, the thixotropic and viscoelastic behavior, and surface interaction related measurements such as determination of slip phenomena or contact angles. Thibert et al. 308 was focusing on the first two questions by studying how various screen-printing parameters (e.g., typical screen parameters) influences the paste rheology and therefore the shape of printed structures.
They found that in order to optimize the interaction between paste and screen during printing, the screen parameters can be chosen in such a way that an optimal compromise between reducing the paste consumption and increasing the homogeneity of the printed structure will emerge. However, in today's mass production environment, the screen parameters are mainly determined by economic choices rather than meeting the given paste rheology for an optimal printing. Fur- When in the future, this model is utilized to calculate the exact shear rate across the screen opening, a sufficient answer to question 1 will be found.
Hsu et al. 307 and Hoornstra et al. 306  is desired. In Figure 11B, the flow curve for two commercial PERC Agpastes for almost seven orders of magnitude is presented.
Furthermore, he worked on generalizing the correlation between finger geometry and paste rheology by deriving a very interesting relationship of the Bingham Capillary number Ca y and the electrical aspect ratio AR el of the contact finger by the dispensing approach as presented in Figure 12. 315 The Bingham capillary number describes the F I G U R E 1 1 On the left, a schematic diagram for the viscosity and apparent shear rate during the different stages of the screen printing process is presented. The paste experiences a reduction of its viscosity by a few orders of magnitude when being pushes through the screen. Afterwards, depending on the thixotropic behavior of a given paste, the internal structure of the printed paste volume recovers over time. This figure is slightly adapted from the original by Trease and Dietz. 321 Furthermore, it was discussed by Reichl 322 and more recently by Reinhardt. 323 On the right, a flow curve for commercial Ag-paste obtained from rotational and capillary data is shown. Further, the corresponding Bingham number is highlighted, marking the point at which viscous stress starts to dominate and a static flow regime is established. Additionally, Herschel and Bulkley modeling of the experimental data for both pastes is shown. 315 F I G U R E 1 2 Dependency of the electrical aspect ratio AR el on the Bingham Capillary number Ca y for various combinations of paste and nozzle diameter on three different types of wafer textures 315 ratio of yield stress to capillary pressure of a yield stress fluid as described in Equation (6).
Even though this relationship applies for the dispensing process, it reveals a more general guideline for the design of paste rheology.
Surface effects cannot be neglected in the micrometer range of today's printing anymore. The presented model that links rheology to printing parameters may need to be expanded by surface effects on the substrate and that of the interacting surfaces of the screen and squeegee.

| The importance of wall slip phenomena
The into the emulsion channel in Figure 13B. During that motion, the entire screen is pushed onto the substrate until the so called nip contact has been reached in Figure 13C. At this position, the paste wets the surface and creates the substrate-paste interface. As soon as the squeegee moves beyond that position, the screen snap-off is initiated because the screen tension forces the screen back to an equilibrium position. During that motion, wall slip at the emulsion interface ( Figure 13D) will be helpful in order avoid additional spreading of the paste after screen snap-off because any additional shearing force at the emulsion-paste interface will cause a reduction of the paste vis- which is partly opened according to the printing layout ( Figure 14A).
The openings can be realized by electroforming, laser-cutting, or wetchemical etching. Single-layer stencils do not allow a combined printing of intersecting elements like fingers and busbars. Thus, a dual printing process using two stencils or a screen and a stencil has to be applied. An alternative is the so-called double-layer or hybrid stencil. 343 This type of stencil has been first developed in the mid 1970s 346 but originally not applied for solar cell metallization due to challenges of realizing fine contact openings. Double-layer stencils consist of a partly opened metal foil with regular bridges to stabilize the open areas and an emulsion layer, which defines the printing layout ( Figure 14B).
Double-layer stencils allow a combined printing of fingers and busbars using one printing form. Compared to single-layer stencils, a better edge homogeneity of the printed fingers can be achieved due to an optimal sealing between the soft emulsion along the finger channel edges and the textured silicon surface (see Section 3.2). In contrast to screens, the metal foil of a stencil is hardly deformed during the printing process, which leads to a better alignment precision and a significantly longer lifetime when handled adequately. Stencilprinted fingers usually show a significantly better uniformity (no "mesh mark" effect) and a high aspect ratio. Furthermore, the cleaning of stencils is easier compared to screens. 344  They have been constantly improving the finger geometry while at the same time scaling up the process from a single nozzle continuous basic dispensing approach towards a multi nozzle dispensing print head for an intermittent process at very competitive industrial process speeds ( Figure 15B). In Figure 15C-E, the evolution of printed finger width using different PERC front side high temperature Ag pastes is presented, revealing an impressive reduction from 65 μm reported in 2012 to only 17μm in 2019. They showed that this approach led to cell efficiency increases of around 1%rel. compared to screen printing and further reduction of Ag-laydown by over 20% at the same time.

| Dispensing technology
This advantage relies on the fact that the approach intrinsically solves a major problem in screen printing: due to mesh marks (further discussed in Section 2.6), it becomes impossible to reach optimal Ag-consumption. The dispensing process provides intrinsically a perfectly homogenous finger geometry across its length removing an entire degree of freedom (deviation of the finger height across its length) from the optimization problem. 315,351 Besides the front side metallization for PERC, the dispensing approach was further successfully utilized by Erath et al. 359

| Rotary printing
Rotary printing methods have been used since decades for the largevolume production in many fields of industrial applications. Due to their high precision and throughput capability, these proven technologies can also be applied for a wide range of technical applications or production steps, which require precise and fast coating processes.  Figure 16A shows on textured silicon solar cells. Due to a limited layer thickness, flexography is particularly suited for a seed and plate metallization with subsequent reinforcement of a seed layer grid with electroplated Ag or other metal stacks. 365 However, the successful application of flexography for front side metallization without subsequent plating achieving contact finger widths down to around 30 μm has also been demonstrated. 366 Gravure or rotogravure printing is usually applied for large-scale production of packaging and graphic arts products.
Gravure printing cylinders consist of a steel core, which is coated with a several millimeter thick, electroplated copper layer. The printing layout is usually engraved into the comparatively soft copper layer by mechanical or laser engraving methods and subsequently covered by a protective, extremely resistible electroplated chromium layer. 367 The fabrication of the cylinders is cost-intensive and challenging with respect to environmental and safety aspects and is thus only reasonable for large-volume printing products. Gravure printing is able to Several studies investigated the application of the front side grid using inkjet technology. Noticeable results were achieved with inkjetprinted seed layer metallization and subsequent reinforcement using electroplating. 365,378 Promising results were also demonstrated using inkjet for the front side metallization of Al BSF 379 A further approach is the so-called Pattern Transfer Printing (PTP TM ) technology, which has been developed and commercialized by the Israeli company Utilight. PTP TM is a two-step contactless laser deposition process to transfer a fine line grid on the wafer surface 384 : In the first step, a transparent web-based polymer substrate with preembossed trenches is filled with high-viscous printing paste using doctor blading. The trenches in the foil substrate can be embossed with a depth and width of 20 μm. In a second step, the foil is placed in a distance of $100-300 μm to the wafer surface, and the transparent backside of the foil is irradiated with laser light. The laser irradiation instantly evaporates the solvent at the walls of the trenches leading to an overpressure, which accelerates the paste towards the wafer surface. 384 Using the PTP TM approach, finger widths down to $20 μm with a high aspect ratio of more than 0.5 have been demonstrated. 385,386 Furthermore, Cu plating is an interesting alternative to printing Increasing the number of parallel production lines is not the most appropriate option as it significantly increases the footprint and investment costs of the required equipment. Following this, the current overall assumption relies on the fact that the screen-printing machine manufacture needs to provide the tools necessary for a high throughput potential. As this is obviously true, the R&D does not end with a fast screen printer, rather than starts with it. The paste-screen interaction is the crucial physical mechanism that dictates how fast printing at ever decreasing structure sizes becomes possible. Further, the solution needs to be reliable, long living, cheap, and scalable for mass production. This economic pressure dictates paste and screen development in an iterative fashion. For example, the emergence of a robust, cheap, and long living fine-line screen will indirectly force paste manufacture to adjust their paste rheology to that screen emulsion to ensure sufficient printability of their product. This optimization cycle proved to be very successful over the last 20 years as it creates many opportunities for disruptive events in the market.
When looking into further paste development, the challenge lies in adjusting the paste rheology in such a way that paste penetration of the screen becomes viable and reproducible at a minimal shearing of the paste volume. Presented research studies around wall slip phenomena are a promising candidate to achieve that. Further, the clogging of individual openings within the screen needs to be prevented to guarantee a stable printing process across thousands of wafers.
When approaching the 20 μm level for finger widths, this seems to become one of the major problems. Adjusting the particle size distribution of such paste is very sensitive to overall production cost and electrical performance. Therefore, this tool of solving the clogging problem does not seem to be a promising option. Latest research has shown that the screen development has on the other hand the great

AUTHOR CONTRIBUTIONS
S. Tepner and A. Lorenz both contributed equally to all aspects of this work. S. Tepner primarily authored chapter 4 (the role of screen printing in the PV industry) and chapter 5 (paste rheology) and also contributed to the other chapters within this work. A. Lorenz primarily authored chapter 1 (history) and chapter 2 (fundamental aspects) and also contributed to the other chapters within this work.