development of high efficiency back passivated silicon solar cells with screen printed contacts

An industrial feasible process for these solar cells based on screen-printed aluminium layer and thermal oxide/silicon nitride passivation is described.. Although some works have brought

Energy Procedia 27 ( 2012 ) 598 – 603

1876-6102 © 2012 Published by Elsevier Ltd Selection and peer-review under responsibility of the scientifi c committee of the

SiliconPV 2012 conference.

doi: 10.1016/j.egypro.2012.07.116

SiliconPV: April 03-05, 2012, Leuven, Belgium Development of high efficiency back passivated silicon solar

cells with screen printed contacts

L Bounaasa, N Auriaca, B Grangea, J Jourdanb, S Mialonb, R Monnaa,

S De Magnienvilleb, M Pasquinellic, D Barakelc

a CEA-LITEN, INES, Le Bourget du Lac 73377, France

b MPO Energy, Domaine de Lorgerie, Averton 53700, France

c IM2NP - Université Aix-Marseille III UMR 6242, Marseille 13397, France

Abstract

This paper presents a detailed analysis of point-contacted aluminium rear side for silicon solar cells An industrial feasible process for these solar cells based on screen-printed aluminium layer and thermal oxide/silicon nitride passivation is described The local removal of the passivation stack by the mean of laser ablation is studied Laser conditions are found to selectively and locally ablate the layers and reduce the laser-induced damages in the Si and the passivation layer Results of these studies are applied on large area back passivated solar cells and electrical parameters highlight the importance of parasitic shunting on this type of structure

© 2012 Published by Elsevier Ltd Selection and peer-review under responsibility of the scientific

committee of the SiliconPV 2012 conference

Keywords: backpassivated solar cells; point contact; PERC; laser ablation; passivation; prasitic shunting

1 Introduction

Silicon solar cells integrating a full area screen-printed Al-BSF (Aluminum Back Surface Field) have been widely used in the photovoltaics industry over the last decades However this back side structure displays several major issues preventing the solar cells from reaching high electrical performances and using thinner wafers Indeed a strong bowing of thin wafers is observed and is a restraint in an industrial production perspective But most of all it has been noticed a lack of passivation quality with SRVrear (rear Surface Recombination Velocity) exceeding 300 cm/s and a low back side reflectivity [1] Although some works have brought successful improvements on full area Al-BSF solar cells [2], the development of alternative solutions such as Passivated Emitter and Rear Cell (PERC) [3] design has been intensively explored Nevertheless the complex fabrication process of PERC solar cells is still a limit towards the

© 2012 Published by Elsevier Ltd Selection and peer-review under responsibility of the scientifi c committee of the SiliconPV 2012 conference

production of low cost solar panels and represents a great challenge for the PV industry This paper first addresses the key process steps for the development of LBSF (Localized Back Surface Field) structures, among which surface passivation and laser ablation of the dielectric layers play an important part A surface recombination velocity of nearly 100 cm/s is achieved with SiO2/SiNx and a laser working point is

found to locally remove these layers Electrical performances of the device are analyzed in a second part, showing the importance of finding a trade-off between contacted and passivated areas and highlighting the role of parasitic shunts in the electrical losses

2 Experimental

Passivation properties of thermal oxide capped by a silicon nitride layer (SiO2/SiNx) are first studied

on Cz 3.6 Ω.cm boron doped silicon wafers by measuring the minority carrier lifetime with the Sinton QSSPC method [4] Samples are symmetrically passivated with a PECVD (Plasma Enhanced Chemical Vapor Deposition) silicon nitride layer capping a thermal silicon oxide grown at high temperature Afterwards, in order to investigate the effect of laser ablation on lifetime, we prepare 12 Ω.cm

boron-doped FZ polished silicon wafers showing a high bulk lifetime to free from bulk recombination We use a

20 nanosecond pulsed green laser (515 nm) displaying a gaussian spatial distribution and a 40μm spot size to selectively ablate the dielectrics stack on the front side All contaminants are then removed by a short cleaning step using an alkaline solution The quality of the laser ablation step was assessed with MW-PCD Semilab tool (MicroWave Photoconductance Decay) lifetime measurements as well as confocal microscopy characterization Based on these findings, we fabricated PERC-type solar cells on 156x156 mm² boron doped Cz 2 Ω.cm silicon based on the process flow depicted in figure 1

Fig 1 Process flow description of PERC-type solar cells (left) and schematic drawing of the solar cell architecture (right)

3 Results and discussion

3.1 Passivation and ablation of SiO2/SiNx layers

Replacing the full area Al-BSF by a LBSF and back side passivated structure involves the challenge

of developing passivation layers achieving a good passivation quality SiO2/SiNx stack layers have been widely studied and were found to provide a very low surface recombination velocity (<50 cm/s) [5, 6] To measure the minority carrier lifetime (τeff), symmetrically coated lifetime samples have been processed on

the same substrate material From these lifetime measurements the maximum effective surface recombination velocity Seff was determined by [12]

Seff bulk

eff

2 1

1



|

W

with W the wafer thickness and τbulk the bulk lifetime assumed to be 5ms in the case of Cz material

In this experiment, an average SRV of 120cm/s was measured on five different samples even after high temperature firing (figure 2), which makes this layer very competitive with the Al-BSF passivation (>300 cm/s)

Fig 2 Lifetime measurements showing the surface recombination velocities (SRVrear) of thermal grown oxide SiO2, SiNx capped grown oxide before and after high temperature firing (>800°C)

However integrating a selective ablation step of the previous passivation stack layers within the process fabrication of PERC-type solar cells, as earlier studied [7], requires defining suitable laser parameters A characterization routine is carried out to tune the laser parameters We used spatially-resolved lifetime measurements to evaluate the quality of the openings Different parameters (fluence; overlap) are tested in order to find the best laser conditions enabling the ablation of the dielectrics stack SiO2/SiNx without damaging the Si surface or the surrounding passivation layer To that end, we performed square shaped openings on the samples using parameters summarized in Table 1 The overlap was defined as the percentage of irradiated areas between two successive laser pulses compared to the laser spot area A negative value means that the distance between two successive pulses is larger than the laser pulse diameter

Fig 3 MW-PCD lifetime measurements of samples according to the positions and parameters referenced on Table 1: a) asdep after SiO2/SiNx double side passivation; b) after laser opening; c) after wet cleaning and SiNx repassivation

Figure 3 shows MW-PCD lifetime measurements carried out on these samples after passivation with the SiO2/SiNx stack and after laser opening Although lifetime is not homogeneous over the whole wafer, one can clearly notice that the laser beam has interacted with the layers at all studied conditions [7]

0 40 80 120 160 200 240

Nevertheless after cleaning and subsequent SiNx deposition, lifetime is recovered for several conditions in which case we can assume the dielectric stack has been properly ablated and defects have been removed From condition 1 to 3, we conclude that the laser ablation is better for an overlap as high

as 50% Indeed, a proper overlap is needed to completely remove the dielectric layers and for instance in

the case of a negative value the surrounding passivation might have been damaged by the laser It might account for the presence of low lifetime regions around each squared shape openings In addition, we notice that lifetime depends on the chosen fluence, indeed a value lower than 0.90J/cm² is not enough to properly remove the dielectric layers because they might have been only partially ablated On the contrary

a fluence as high as 1.75 J/cm² would induce too many surface defects largely degrading the minority carrier lifetime Therefore we can conclude that a laser fluence around 1 J/cm² is appropriate to perform the ablation step which such a laser

In addition, figure 4 shows the influence of the cleaning step on the surface rugosity and highlights the

necessity to remove most of the residual damage Indeed, these defects are most likely highly recombination active defects and should involve a strong increase of the surface recombination velocity

Fig 4 Confocal microscope images of the ablated area (right side) before (left) and after cleaning step (right)

Table 1 List of the laser parameters used for the square-shaped openings of SiO2/SiNx layers

Condition Fluence Overlap Condition Fluence Overlap

3.2 Solar cells results

Based the process flow depicted on figure 1 and using the selected passivation layers and laser parameters defined in the previous section, 156x156 mm² boron doped Cz 2 Ω.cm silicon were fabricated

Figure 5 displays the electrical I-V characteristics open circuit voltage (Voc) and fill factor (FF) of the fabricated PERC-type solar cells with different metalized fraction on the rear side The laser contact opening width was set to a fixed value and the contact spacing was changed It was then observed a clear

improvement of the FF when increasing this parameter, which can be easily explained by the reduction of

the series resistance Resistance losses are lowered (FF≈78%) when the contacted area is higher and reaches a maximum (FF≈79%) in the case of a full area Al-BSF On the other hand, the increase of the metallization fraction induces a Voc decrease because the SiO2/SiNx passivated area gets smaller It has been observed elsewhere [8] that the optimal spacing is the result of the competition between the resistive and the contact recombination losses One can clearly notice that Voc values measured on fabricated solar cells are at least 20mV lower than expected Since these devices integrate a SiO2/SiNx layer achieving very low surface recombination velocity, that may not be caused by passivation issues The phenomenon has already been noticed in dielectric back side passivated solar cells but is only relevant for dielectric displaying high amount of positive charge [9,11] Indeed in that case, the underlying p-type silicon base is under inversion and induces floating junction between the contacts If the floating junction is not properly isolated from the contacts, it gives rise to a parasitic shunting effect between the inversion layer and the metal contact as described on figure 6

Fig 5 Electrical parameters (Voc and FF) measured on 156x156 mm² solar cells as a function of the metallisation fraction area (fmet) compared to a full area Al-BSF

The detrimental effect of parasitic shunts on electrical is highlighted when plotting the local ideality factor according to the voltage across the device [10] extracted from dark I-V curves A significant shoulder appears on the curve but only for the dielectric back passivated solar cell A consequence of the shunt is a surface recombination velocity that is injection level dependant It has been explained by the presence of two different regimes [10]: a shunt dominated regime at low voltage when the electrons in the inversion layer flow through the shunt into the contact which influences the overall SRV of the cell; a diode dominated regime when the shunt is saturated by the electron flow, and then the SRV of the cell is only controlled by the SRV achieved by the passivation layer

Fig 6 Schematic drawing of the possible mechanism of parasitic shunting observed on dielectric back passivated solar cells (left) Local ideality factor plotted as a function of the voltage across the cell for a PERC-type (blue) and a full area BSF (pink) solar cell

75 76 77 78 79

fmet (%)

604 608 612 616 620 624

FF (%) Voc (mV)

4 Conclusion

In conclusion, it was demonstrated the relevance of using a SiO2/SiNx dielectric stack layers passivation because of a SRV achieved lower than 110 cm/s It was also described the possibility to integrate a laser ablation step using a nanosecond green laser in order to fabricate PERC-type silicon solar cells Laser parameters were selected to properly remove the SiO2/SiNx passivation stack Nevertheless it has the disadvantage to include a post-laser cleaning step which needs to be modified for industrial requirements Additionally, the fabricated dielectric back passivated solar cells showed the relevance of taking into account the parasitic shunting detrimental effects on the electrical performances Therefore, future work will focus on the implementation of alternative passivation layers to reduce these effects on back passivated solar cells

Acknowledgements

The authors would like to thank IM2NP, CEA and OSEO funding group for financial support

References

[1] Narasimha S, Rohatgi A An optimized rapid aluminum back surface field technique for silicon solar cells, Photovoltaic

Specialists Conference, 1997, Conference Record of the Twenty-Sixth IEEE , vol., no., pp.63-66, 29 Sep-3 Oct 1997

[2] Ristow A Screen-printed back surface reflector for light trapping in crystalline silicon solar cells, 2001

[3] Green MA, Blakers AW, Zhao J Wang A Characterization of 23-percent efficient silicon solar cells, Electron Devices,

IEEE Transactions on , vol.37, no.2, pp.331-336, Feb 1990

[4] Sinton R., Cuevas A, and Stuckings M Quasi-Steady-State Photoconductance, A New Method for Solar Cell Material

and Device Characterization, Proc of the 25th IEEE Photovoltaic Specialists Conference, pp 457-460, 1996

[5] Mack S, Jäger U, Kästner G Towards 19% efficient industrial PERC devices using simultaneous front emitter and rear

surface passivation by thermal oxidation, Photovoltaic Specialists Conference (PVSC), 2010 35th IEEE , vol., no., pp.34-38, 20-25

June 2010

[6] Rohatgi A, Narasimha S, and Ruby DS Effective passivation of the low resistivity silicon surface by a rapid thermal

Oxide/PECVD silicon nitride stack and its application to passivated rear and bifacial si solar cells In 2nd World Conference and

Exhibition on Photovoltaic Solar Energy Conversion, 1999

[7] Jaffrennou P Laser Ablation of SiO 2 /SiN x and AlO x /SiN x Back Side Passivation Stacks for Advanced Cell Architectures,

In 26 th World Conference and Exhibition on Photovoltaic Solar Energy Conversion, 2011

[8] Lai J, Upadhyaya A, Ramanathan S, Das A, Rohatgi A High-Efficiency Large-Area Rear Passivated Silicon Solar Cells

With Local Al-BSF and Screen-Printed Contacts, Photovoltaics, IEEE Journal of , vol.1, no.1, pp.16-21, July 2011

[9] Prajapati V, Cornagliotti E, Russell R High efficiency industrial silicon solar cells on Si mono cast material using dielectric

passivation and local BSF In 24th European Photovoltaic Solar Energy Conference, 2009

[10] Dauwe S Ph.D Thesis, Low-Temperature Surface Passivation of Crystalline and its Application to the Rear Side of Solar

Cells, ISFH, 2004

[11] Cesar I, Mewe AA, Manshanden P, Romijn IG Effect of Al contact pitch on rear passivated solar cells, Energy Procedia,

8, pp 672-680, 2011

[12] Sproul AB Dimensionless solution of the equation describing the effect of surface recombination on carrier decay in

semiconductors, Journal of Applied Physics , vol.76, no.5, pp.2851-2854, Sep 1994