Solar Energy Materials & Solar Cells 117 (2013) 537–543

Contents lists available at ScienceDirect

Solar Energy Materials & Solar Cells

0927-02http://d

n CorrE-m

journal homepage: www.elsevier.com/locate/solmat

Using hydrofluoric acid to reduce the contact resistanceof screen-printed silicon solar cells – Its recombination impactand a method to eliminate it

Kee Soon Wang n, Alison J. Lennon, Budi S. Tjahjono, Ashraf Uddin, Stuart R. WenhamSchool of Photovoltaic and Renewable Energy Engineering, University of New South Wales, Sydney, NSW 2052, Australia

a r t i c l e i n f o

Article history:Received 18 September 2012Received in revised form10 June 2013Accepted 11 June 2013Available online 15 August 2013

Keywords:Screen-printingSolar cellSilver pasteContact resistanceRecombinationIdeality factor

48/$ - see front matter & 2013 Elsevier B.V. Ax.doi.org/10.1016/j.solmat.2013.06.017

esponding author. Tel.: +61 2 93856782.ail address: wangkeesoon@gmail.com (K.S. W

a b s t r a c t

Using hydrofluoric acid (HF) to improve the contact resistance of a screen-printed silicon solar cell with ahigh sheet resistance emitter was found to concurrently reduce its pseudo-fill factor. By treating the cellin phosphoric acid, this impact was found to be significantly reduced or eliminated. In the solar cellpresented in this work, the pseudo-fill factor reduced from 82.1% to 79.9% after treating the cell in HF, butincreased to 81.8% after a subsequent phosphoric acid treatment. Similar effects were found in threeother screen-printed solar cells that were fired at different peak furnace temperatures and belt speeds.It was also shown that the phosphoric acid treatment alone does not affect the pseudo-fill factor of a cell.HF treatment can now be used to reduce the contact resistance of a screen-printed silicon solar cell witha high sheet resistance emitter without suffering from a lower pseudo-fill factor so that the full benefit ofa HF treatment on a screen-printed silicon solar cell can be realized. This combination of HF andphosphoric acid treatments can potentially be a useful failure analysis method for screen-printed siliconsolar cells in the production line and the laboratory.

& 2013 Elsevier B.V. All rights reserved.

1. Introduction

In commercial screen-printed silicon solar cells, low contactresistance (Rcontact) is essential for high fill factors (FF) andefficiencies [1]. The Rcontact of a screen-printed silver (Ag) con-ductor is determined by a number of factors, one of which is thethickness of the insulating oxide layer that forms between thescreen-printed Ag conductor and the emitter surface [2–7]. Whenthe oxide is too thick, it retards the conduction of current into theAg fingers [2,8]. This can be circumvented by using dilute hydro-fluoric acid (HF) to thin or possibly eliminate such oxides. It waspreviously demonstrated that this HF treatment can lower theRcontact of such screen-printed cells, thus raising their FF andefficiencies [9–11]. With increasing interest in high sheet resis-tance emitters, the positive impact of an HF treatment on the FF ofscreen-printed cells with 100 Ω/□ emitters was reported [10].However, no negative electronic impacts were mentioned. Thisresearch reports that HF treatments on screen-printed silicon solarcells with such high sheet resistance emitters can increaserecombination in the cell and reduce its pseudo-FF (pFF) [12].Furthermore, a method to reduce or eliminate this impact is alsopresented.

ll rights reserved.

ang).

2. Experimental

A screen-printed silicon solar cell, Cell 1, was fabricated on a2 in. square, ∼200 μm thick mono-crystalline 1 Ωcm (100) planarsilicon wafer. After saw damage etch and alkaline texturing, thewafer was cleaned in RCA1 and RCA2 [13] before undergoing POCl3diffusion to produce an ∼100 Ω/□ emitter. A subsequent dilute HFtreatment was used to remove the phosphosilicate glass layer.Approximately 75 nm of silicon nitride with a refractive index of2.0 was deposited on the emitter surface using a remote plasma-enhanced chemical vapor deposition system. The wafer was thenscreen-printed with aluminum (Al) paste on the rear side, dried at150 1C for 15 min, screen-printed with a commercial Ag pasteDuPont PV145 [14] on the emitter side, dried at 150 1C for 15 min,before being fired at a peak furnace-set temperature of 810 1C anda belt speed of 6200 mm/s in a Centrotherm 6-zone belt furnace.The cell was then edge-isolated by using a laser with an emissionwavelength of 1064 nm to partially-cleave the wafers from therear side before the redundant side pieces were manually snappedoff. The final cell dimension was ∼2.9�2.8 cm2. Cell 1 was thencharacterized using its light current voltage (IV) curve (at standardtesting conditions of 100 mW/cm2 of an AM1.5G spectrum at25 1C), dark IV curve, Suns–Voc curve and its pFF [12], andphotoluminescence series resistance (PL Rs) spatial image [15]using a BTi imaging tool [16]. Cell 1 was then immersed in a 1% HFsolution (thereafter referred as “HF treatment”) for 5 s before being

Fig. 1. Photoluminescence series resistance (PL Rs) images of Cell 1 (a) as-fired, (b) after 5 s, (c) after 10 s, and (d) after 20 s of treatment in a 1% HF solution. The resistancecolor scale is 0–2 Ωcm2. In (d), the Ag finger at the lower left corner peeled off due to prolonged HF treatment, resulting in higher local resistance.

Fig. 2. Light IV curves of Cell 1 as-fired, after a 25 s 1% HF treatment (“25 s HF”) andafter a subsequent 30 s of 85% phosphoric acid treatment (“25 s HF+30 s P”). Inset:magnified view of the “knee” of the IV curves between 0.35–0.55 V.

Table 1The 1-Sun power conversion output data of Cell 1: as-fired, after a cumulative of25 s of HF treatment in a 1% (w/v) HF solution (“25 s HF”), and after differentcumulative durations of treatments in 85% (w/v) phosphoric acid (“+P”). The Suns–Voc [12] pseudo-FF (pFF) and local ideality factor [18] at the maximum power point(SunsVoc mMPP), and the series resistance (Rseries) are also shown.

Cell1 treatments

Voc(mV)

Jsc (mA/cm2)

FF(%)

Eff.(%)

pFF(%)

SunsVocmMPP

Rseries

(Ωcm2)

K.S. Wang et al. / Solar Energy Materials & Solar Cells 117 (2013) 537–543538

rinsed in de-ionized (DI) water (∼18 MΩ) for 10 min and dried innitrogen (N2) gas. It was again characterized as described above.This cycle continued until the cumulative duration of the HFtreatment was 25 s.

After undergoing 25 s of the HF treatment, Cell 1 was subse-quently immersed in an 85% (w/v) phosphoric acid (H3PO4)solution (thereafter referred as “H3PO4 treatment”) for 10 s. Thecell was then characterized with the same methods as describedbefore. This cycle continued until the cumulative H3PO4 treatmentduration was 30 s.

To check the cross-sectional profile of the Ag fingers after theHF treatments, another cell, Cell 1A, was fabricated with the sameprocess condition as that of Cell 1. Cell 1A was then cleaved fromthe rear Al side in a direction perpendicular to the orientation ofthe Ag fingers before the two half pieces were manually snappedoff in the same way that edge-isolation of Cell 1 was done. In thisway, the Ag finger cross-sections are exposed. A cross-sectionalSEM image for the as-fired sample was taken from one of thecleaved Ag fingers. Subsequently, one of the halved pieces under-went the same cumulative 25 s of the HF treatment that was doneto Cell 1. After cleaving another randomly-picked spot along thesame Ag finger, another cross-sectional SEM image was taken forsame Ag finger.

In this work, the front Ag grid pattern had a finger pitch of3.5 mm. Such a non-optimally wide pitch was used because theobjective of the research was to fabricate laser-doped (LD) semi-conductor fingers (SCF) silicon solar cells [17], although no LD SCFwere incorporated into any of the cells presented in this work.Therefore, high series resistance and low FF levels should beexpected, although they did not affect the analyses presented here.

As-fired 603.2 36.5 68.4 15.0 82.1 1.2 2.725 s HF 599.6 36.0 71.7 15.5 79.9 1.5 1.625 s HF+10 s P

600.0 36.0 72.3 15.6 81.7 1.2 1.6

25 s HF+20 s P

600.0 36.1 72.6 15.7 81.7 1.2 1.6

25 s HF+30 s P

600.6 35.9 72.7 15.7 81.9 1.2 1.5

3. Results and discussions

3.1. Impact of hydrofluoric acid treatment

Fig. 1 shows the PL Rs images of Cell 1 for different cumulativedurations of the HF treatment. The vertical dark lines are the Agfingers and the busbar is the top horizontal line. Fig. 1(a) showsthat the local series resistance was not uniform across the cellas-fired. Fig. 1(b)–(d) shows increasingly lower and improveduniformity of local resistance from longer HF treatments. As aresult, the slope of the IV curve (see Fig. 2) at voltage 40.45 Vbecame steeper after 25 s of the HF treatment, which indicatesthat the series resistance in Cell 1 has reduced. This is supportedby the cell data shown in Table 1, where the series resistance

reduced from 2.7 Ωcm2 to 1.6 Ωcm2 after 25 s of the HF treatment,and as a result the FF increased from 68.4% to 71.7%.

It can also be observed in Fig. 1(d) that the Ag finger at thelower left corner of Cell 1 has a higher local resistance comparedto adjacent fingers. This is because the Ag finger in this section haspeeled off, thus causing electrons generated in its vicinity to travel

Fig. 3. Cross-sectional SEM images of a screen-printed Ag finger in Cell 1A (a) as-fired, and (b) after 25 s of 1% HF treatment. Both images were taken at two different partsalong the length of the same Ag finger, and at approximately halfway in between the side and middle of the finger cross-section. The green arrows in (a) point to thepresence of interfacial oxide layers, while the yellow dashed circles in (b) highlight the obvious absence of interfacial oxides previously shown in (a).

Fig. 4. (a) Dark IV curves and (b) the corresponding dark m–V curves of Cell 1 as fired (“0 s”), and after 5 s, 15 s, and 25 s of 1% HF treatment (“5 s”, “10 s” and “25 s”respectively). Inset of (a): magnified view of the dark IV curves between 0.2–0.3 V.

K.S. Wang et al. / Solar Energy Materials & Solar Cells 117 (2013) 537–543 539

a longer distance in the emitter to the next nearest Ag finger to beconducted to the busbar. This localized peel-off is likely due tomore rapid local loss of adhesion under the same HF treatmentcompared to the rest of the Ag fingers. It is known that theinterfacial oxide that is formed during the screen-printing contactformation firing process in the belt furnace is responsible forholding the Ag finger to the solar cell [9]. Therefore, the localizedfinger peel-off in Fig. 1(d) may be due to a local defect during thecontact formation process that causes a thinner or more porouslayer of oxide to be formed that is more easily-removed by HFtreatments compared to the rest of the cell. Such investigation isbeyond the scope of this work. However, it can be shown thatdespite prolonged etching on the interfacial oxide layers after 25 sof the HF treatment, the Ag fingers can still be expected to remainintact. Fig. 3(a) shows the cross-sectional SEM image of theas-fired Ag fingers of Cell 1A. This image was taken at approxi-mately halfway in between the side and middle of the finger cross-section, and interfacial oxide layers, as expected, can be seenbetween the silicon wafer and the Ag finger. Fig. 3(b) showsapproximately the same spot across the cross-section of the sameAg finger as Fig. 3(a), but Fig. 3(b) was taken at a different sectionalong the length of the finger and only after 25 s of the HFtreatment. In contrast to Fig. 3(a), Fig. 3(b) shows that the oxidelayers were removed after 25 s of the HF treatment. However, thefinger remained intact for the SEM image to be taken. This isbecause further imaging of the finger after the 25 s HF treatment

revealed that the interfacial oxide layers towards the middle of thefinger cross-section were either only partially removed or notaffected at all, with the interfacial oxides in the middle resemblingthose shown in Fig. 3(a).

Besides the reduced series resistance and higher FF observed afterthe HF treatment from IV data and PL Rs images, Fig. 4(a) shows thedark IV curves of Cell 1 with different durations of the HF treatment,where it can also be observed that after each HF treatment, the curveat voltage 40.5 V in general shifted up to higher current levels. Thisindicates that Cell 1's series resistance had been reduced, as expected.Interestingly, Fig. 4(a) shows that the dark IV curve also shifted tohigher current level at the voltage range below 0.5 V (defined here aslow-to-mid V range, “LTMVR”) after 5 s of the HF treatment. Whenthe cumulative duration of this HF treatment increased to 15 s and25 s, the dark IV curve in the LTMVR shifted progressively to evenhigher current levels. This indicates an increase in recombinationwithin the solar cell, and that the dark saturation current densities ofthe diode of the solar cell increased with increasing durations of theHF treatment.

For further analysis of this phenomenon, the local idealityfactor (m) was plotted against voltage to produce a dark m–Vcurve [18] of the cell (see Fig. 4(b)). Originally, Cell 1 as-fired(labeled ‘0 s’ in the graph) has the hump located in voltage rangebetween 0.05–0.5 V, with peak around 0.23 V. McIntosh [18]proposed that this “hump” in the m–V curve is a result ofresistance-limited enhanced recombination. This recombination

K.S. Wang et al. / Solar Energy Materials & Solar Cells 117 (2013) 537–543540

mechanism tends to occur when metal is contacting the lightlydoped base in localized points and therefore forms localizedSchottky contact diodes [18,19]. Interestingly, after 5 s, 15 s and25 s of the HF treatment, the hump in cell's m–V curve shiftsprogressively to the right to higher voltages, thereby extending thehump upto ∼0.55 V with peak at 0.3V. These shifts that haveoccurred at the LTMVR of both the dark IV and m–V curves havenot been reported before in screen printed solar cells that wentthrough a HF treatment.

Such localized Schottky diodes are detrimental to the efficiencyof a solar cell because they have a low turn-on voltage comparedto the main junction [19]. The possible origins of these localizedSchottky contacts will be discussed later. As a result of thisSchottky contact formation, the “knee” in the IV curve can becomemore rounded [20] and therefore the FF of the solar cell will bereduced. This effect is demonstrated in the difference between theSuns–Voc curves of Cell 1 as-fired and after 25 s of the HFtreatment (see Fig. 5). It can be observed that the “knee” of theSuns–Voc curve indeed became more rounded after the HFtreatment. Since this curve represents cell's light IV curve withoutthe effect of series resistance [12], this is clear evidence that the HFtreatment had reduced the maximum FF that Cell 1 can achieve.This was also reflected in Fig. 4(b) in the higher m values between0.50 and 0.55 V for Cell 1 after the HF treatment. Since themaximum power point (MPP) for a practical screen-printed solarcell is usually between 500 and 550 mV, the right-shift of thehump in the dark m–V curve would raise the m values at the MPP

Fig. 5. Suns–Voc curves of Cell 1 as-fired, after 25 s of treatment in 1% HF (“25 s”)and after an additional 30 s (“25 s+30 s P”) of treatment in 85% (w/v) H3PO4. Inset:magnified view of the curve between 0.45–0.6 V.

Fig. 6. (a) Dark IV curves and (b) the corresponding darkm–V curves of Cell 1 as fired (“030 s (“+30 s P”) of treatment in 85% (w/v) H3PO4. Inset of (a): magnified view of the da

of a screen-printed solar cell, resulting in reduced FF and efficien-cies. This analysis is consistent with the pFF trends, which droppedfrom the as-fired value of 82.1–79.9% after 25 s of the HF treat-ment. The Suns–Voc m at the MPP (mMPP) is also shown in Table 1to have increased from 1.2 to 1.5 after 25 s of the HF treatment. Inall, even though the HF treatment was shown to be able to reduceseries resistance loss, it also caused additional resistance-limitedenhanced recombination, which resulted in lower pFF and higherm values around the MPP.

It was mentioned earlier that the Schottky contact-inducedresistance-limited enhanced recombination could result frommetal penetrating in from the surface to contact regions oflightly-doped p-type base [20]. To identify the sources of metal,it is useful to understand that a HF treatment removes theinterfacial oxide in Ag fingers [9–11] and, as a result, can exposethe underlying emitter. Since Ag pastes contain many metallicelements such as Ag, lead (Pb), cadmium (Cd), bismuth (Bi) andzinc (Zn) [6,14,21], it is hypothesized that some of these elementsmay have been released from the interfacial oxide and come incontact with the exposed Si surface thus contributing to increasedrecombination. Out of these elements, Pb and Ag are perhaps themore likely contaminants. This is because a high density of Pbprecipitates had been shown to be embedded in such interfacialoxides [5–8], and because Ag is the major metallic content.Concentrated H3PO4 can dissolve Pb [22] and so it was hypothe-sized that immersing the HF-treated cells in a solution of H3PO4

could dissolve exposed Pb and therefore reduce or eliminate theapparent increase in recombination.

3.2. Impact of phosphoric acid treatment after hydrofluoric acidtreatment

In this section, the impact of H3PO4 treatment on screen-printed silicon solar cell samples that had previously undergonea HF treatment is studied. After undergoing 25 s of the HFtreatment as previously described in Section 3.1, Cell 1 wassubsequently immersed in an 85% (w/v) H3PO4 solution (H3PO4

treatment) for 10 s. The cell was then characterized with the samemethods as described before. This cycle continued until thecumulative H3PO4 treatment duration was 30 s.

As shown in Fig. 6(a), the LTMVR of Cell 1's dark IV curve shifteddown after the first 10 s of H3PO4 treatment and was approximatelyback to its as-fired level after 30 s. This came with no significantchange to the dark IV curve at 40.5 V, showing that Cell 1's seriesresistance remained approximately unchanged. This is substantiatedin Table 1, which shows that the series resistance dropped from1.6 to only 1.5 Ωcm2 after the H3PO4 treatment. In Fig. 6(b), after 30 sof the H3PO4 treatment the hump in the dark m–V curve shifted to

s”), after 25 s of 1% HF treatment (“25 s”), and after a subsequent 10 s (“+10 s P”) andrk IV curves between 0.2–0.3 V.

Fig. 7. (a) Dark IV and (b) darkm–V curves of the Cell 2 fired at 835 1C and 6000 mm/s as-fired, after 15 s and 25 s of treatment in 1% HF solution (“15 s HF” and “25 s HF”) anda subsequent 150 s of 85% H3PO4 treatment (“25 s HF+P”).

Fig. 8. (a) Dark IV and (b) darkm–V curves of the Cell 3 fired at 835 1C and 6400 mm/s as-fired, after 15 s and 25 s of treatment in 1% HF solution (“15 s HF” and “25 s HF”) anda subsequent 150 s of 85% H3PO4 treatment (“25 s HF+P”).

Fig. 9. (a) Dark IV and (b) darkm–V curves of the Cell 4 fired at 835 1C and 6600 mm/s as-fired, after 15 s and 25 s of treatment in 1% HF solution (“15 s HF” and “25 s HF”) anda subsequent 150 s of 85% H3PO4 treatment (“25 s HF+P”).

K.S. Wang et al. / Solar Energy Materials & Solar Cells 117 (2013) 537–543 541

the left and as a result the m values between the MPP range of 0.50–0.55 V returned back to the as-fired levels. Moreover, in Fig. 5 the“knee” of Cell 1's Suns–Voc curve became as sharp as its as-fired curveand the pFF of Cell 1 increased to 81.9% (i.e., only slightly less thanthe as-fired pFF, as shown in Table 1). The Suns–Voc mMPP alsodropped back to the as-fired level as shown in the same table.Moreover, in Fig. 2, a seemingly sharper knee in the light IV curvealso raised Cell 1's FF from 71.7% to 72.7%. These results clearlydemonstrated that H3PO4 can eliminate the increased recombinationcaused by the HF treatment and therefore improve the pFF and FF ofa screen-printed solar cell.

To confirm these findings, experiments were repeated onanother three screen-printed silicon solar cells that were fabri-cated in the same way as Cell 1 but with different firing conditions.Cells 2, 3 and 4 were fired at a peak furnace-set temperature of835 1C and belt speeds of 6000 mm/s, 6400 mm/s and 6600 mm/s,respectively. The dark IV and m–V curves of Cells 2, 3 and 4 areshown in Figs. 7, 8 and 9, respectively. The dark IV and m–V curvesof all three cells responded in approximately the same way asthose of Cell 1 after the same HF treatments, and the shifts in thedark IV and m–V curves after the HF treatment were significantlyreduced or eliminated after immersion in an 85% (w/v) H3PO4

K.S. Wang et al. / Solar Energy Materials & Solar Cells 117 (2013) 537–543542

solution for 150 s. The H3PO4 treatment was extended to 150 s tomaximize the impact of H3PO4 treatment.

3.3. Impact of phosphoric acid treatment alone

The experimental results shown so far have proven that aH3PO4 treatment can undo the detrimental recombination impactcaused by a prior HF treatment. This effect has been comprehen-sively demonstrated by dark IV curves and pFF data in Cells 1–4.To corroborate this observation, there is a need to prove that aH3PO4 treatment alone (in the absence of a prior HF treatment)does not affect the recombination within the solar cell, an effectthat can be characterized by a comparison using Suns–Voc curves,the pFF data and the dark IV curves.

In order to understand at the device level the impact that aH3PO4 treatment alone (with no prior HF treatment) has onas-fired screen-printed silicon solar cell samples, a screen-printed silicon solar cell Cell 5 was fabricated in the same way asCell 1 and was subsequently treated in an 85% (w/v) H3PO4

solution for 150 s. Table 2 shows that the pFF and Suns–Voc mMPP

of Cell 5 before and after the H3PO4 treatment remains approxi-mately the same. Fig. 10 also shows that the Suns–V and dark IVcurves of Cell 5 before and after the H3PO4 treatment are almostidentical. It is therefore proven that a H3PO4 treatment alone (inthe absence of a prior HF treatment) does not affect the recombi-nation within the solar cell.

Note from Table 2 that at 10.4 Ωcm2, Cell 5 as-fired has veryhigh Rseries, much higher than Cell 1's 2.7 Ωcm2 (see Table 1). Thisis attributed to the variability of the experimental outcome forcells fabricated in different batches. Since this work focuses oncell's recombination impact rather than its Rseries and that therecombination impact can be effectively isolated from Rserieseffects with the Suns–Voc technique, the analysis on the recombi-nation impact in this section is not affected by cell's high Rserieslevels. On the other hand, it can be observed from Table 2 that theRseries of Cell 5 improved from 10.4 Ωcm2 to 8.5 Ωcm2, resulting inan approximately 4% absolute increase in the FF. This suggests that

Table 2The 1-Sun power conversion output data of Cell 5: as-fired, and after 150 s oftreatment in 85% (w/v) phosphoric acid (“P”). The Suns–Voc [12] pseudo-FF (pFF)and local ideality factor [18] at the maximum power point (SunsVoc mMPP), and theseries resistance (Rseries) are also shown.

Cell5 treatments

Voc(mV)

Jsc (mA/cm2)

FF(%)

Eff.(%)

pFF(%)

SunsVocmMPP

Rseries

(Ωcm2)

As-fired 602.7 34.0 43.9 9.0 82.8 1.0 10.4150 s P 603.5 35.4 48.0 10.3 83.0 1.0 8.5

Fig. 10. (a) Suns–V and (b) dark IV curves of Cell 5 as-fire

the H3PO4 treatment can also potentially reduce the Rseries of ascreen-printed silicon solar cell, although the extent of theimprovement in Cell 5 appears to be very limited. More work willbe done to investigate this effect.

3.4. Discussion

The ability of a H3PO4 treatment to undo the enhancedrecombination impact that a HF treatment can have on a screen-printed silicon solar cell fabricated in this work was comprehen-sively shown in Section 3.2. This was based on a hypothesis thatthe recombination source is Pb as described in Section 3.1. Onepossible mechanism that may explain the observed effects is asfollows: a HF treatment etches the interfacial oxide under thescreen-printed Ag grid lines and both releases Pb from the etchedoxide and exposes the silicon surface under the oxide; subse-quently, at least some of the released Pb precipitates in such a wayto form Schottky diodes which cause resistance-limited enhancedrecombination; a subsequent H3PO4 treatment then eliminates theimpact of this recombination by effectively dissolving or “leaching”the Pb from sites of recombination. Further work in ongoing toinvestigate the actual mechanism.

The main significance of this work is that HF treatments cannow be used with subsequent H3PO4 treatments to lower theRseries of a screen-printed silicon solar cell with a high sheetresistance emitter without concurrently reducing its pFF so thatthe full benefit of the HF treatment can be realized. While HFtreatments on screen-printed silicon solar cells can cause the gridlines to peel off [9,10], in the course of this work it was observedthat with a combination of sufficiently low HF concentration andshort immersion time, the gridlines can remain intact after the HFtreatments and its subsequent rinsing process. While the tendencyfor screen-printed grid lines to peel off means this HF treatment isnot suitable as a means of improving the Rseries in the productionline, the combination of HF treatment together with a subsequentH3PO4 treatment can be useful as a failure analysis method to, forexample, identify howmuch higher the FF can efficiency of screen-printed silicon solar cells in a production line can attain so as tohelp to improve the commercial production process.

4. Conclusion

This work reports that using HF to reduce the Rcontact of ascreen-printed silicon solar cell with a high sheet resistanceemitter can concurrently induce resistance-limited enhancedrecombination in the cell and reduce its pFF. It was also demon-strated that a subsequent H3PO4 treatment at room temperature

d and after 150 s of 85% H3PO4 treatment (“150 s P”).

K.S. Wang et al. / Solar Energy Materials & Solar Cells 117 (2013) 537–543 543

can significantly reduce or eliminate this recombination impact. Itwas also shown that the H3PO4 treatment alone does not affect thepFF of a cell. The main significance of this work is that HFtreatments can be used with subsequent H3PO4 treatments tolower the Rseries of a screen-printed silicon solar cell with a highsheet resistance emitter without concurrently reducing its pFF sothat the full benefit of the HF treatment can be realized. Thiscombination of HF and H3PO4 treatments can potentially be auseful failure analysis method for screen-printed silicon solar cellsin a production line and the laboratory.

Acknowledgments

The authors would like to extend their appreciation to DuPontfor providing paste samples used in this work. The AustralianResearch Council, the NSW State Government through the DSRD,Silex Solar Pty Ltd and Sunrise Global Solar Energy are acknowl-edged for their various forms of support and assistance. The firstauthor would also like to express his gratitude to the NationalResearch Foundation of Singapore for funding his Ph.D. studies.

References

[1] M.A. Green, Solar cells: operating principles, technology and system applica-tions, University of New South Wales, 1998, Kensington, NSW 2033, Australia,pp. 96–98.

[2] M.M. Hilali, B. To, A. Rohatgi, A review and understanding of screen-printedcontacts and selective-emitter formation, in: Proceedings of the 14th Work-shop on Crystalline Silicon Solar Cells and Modules NREL, Winter Park,Colorado, USA, 2004, pp. 80401.

[3] M.M. Hilali, K. Nakayashiki, C. Khadilkar, R.C. Reedy, A. Rohatgi, A. Shaikh,S. Kim, S. Sridharan, Effect of Ag particle size in thick-film Ag paste on theelectrical and physical properties of screen-printed contacts and silicon solarcells, Journal of the Electrochemical Society 153 (2006) A5–A11.

[4] M.M. Hilali, M.M. Al-Jassim, B. To, H. Moutinho, A. Rohatgi, S. Asher, Under-standing and formation of high-quality thick-film Ag contacts on high sheet-resistance Si emitters for solar cells, Journal of the Electrochemical Society 152(2005) G742–G749.

[5] G. Schubert, F. Huster, P. Fath, Physical understanding of printed thick-filmfront contacts of crystalline Si solar cells—review of existing models andrecent developments, Solar Energy Materials and Solar Cells 90 (2006)3399–3406.

[6] M. Hörteis, T. Gutberlet, A. Reller, S.W. Glunz, High-temperature contactformation on n-type silicon: basic reactions and contact model for seed-layer contacts, Advanced Functional Materials 20 (2010) 476–484.

[7] S. Kontermann, G. Willeke, J. Bauer, Electronic properties of nanoscale silvercrystals at the interface of silver thick film contacts on n-type silicon, AppliedPhysics Letters 97 (2010) 191910.

[8] D. Pysch, A. Mette, A. Filipovic, S.W. Glunz, Comprehensive analysis ofadvanced solar cell contacts consisting of printed fine-line seed layersthickened by silver plating, Progress in Photovoltaics: Research and Applica-tions 17 (2010) 101–114.

[9] K. Firor, S. Hogan, Effects of processing parameters on thick film inks used forsolar cell front metallization, Solar Cells 5 (1981) 87–100.

[10] A. Ebong, D.S. Kim, A. Rohatgi, W. Zhang, Understanding the mechanism oflight induced plating of silver on screen-printed contacts for high sheetresistance emitters with low surface phosphorus concentration, in: Proceed-ings of the 33rd IEEE Photovoltaic Specialists Conference, San Diego, 2008,pp. 1–5.

[11] ⟨http://www.secondmetal.eu/fileadmin/secondmetal/docs/Session1/4_recaman_epi_emitters.pdf⟩, (accessed 23 02 2013).

[12] R.A. Sinton and A. Cuevas, A quasi-steady-state open-circuit voltage methodfor solar cell characterization, in: Proceedings of the 16th European Photo-voltaic Solar Energy Conference, Glasgow, 2000, pp. 1152–1155.

[13] W. Kern, Handbook of Semiconductor Wafer Cleaning Technology, NoyesPublications, Park Ridge, New Jersey, 1993, p. 49.

[14] DuPont PV145 silver paste Material Safety Data Sheet.[15] H. Kampwerth, T. Trupke, J.W. Weber, Y. Augarten, Advanced luminescence

based effective series resistance imaging of silicon solar cells, Applied PhysicsLetters 93 (2008) 202102.

[16] ⟨http://www.btimaging.com/?page_id=81⟩, January 2012.[17] D. Wang, B.S. Tjahjono, A. Uddin, S.R. Wenham, Modelling the cell output of

the laser-doped semiconductor fingers solar cell design, in: Proceedings of the5th World Conference on Photovoltaic Energy Conversion, Valencia, 2010,pp. 1401–1404.

[18] K.R. McIntosh, Humps, lumps and bumps: three detrimental effects in thecurrent-voltage curve of silicon solar cells, Ph.D. thesis, Department ofPhotovoltaic Engineering, University of New South Wales, Sydney, Australia,2001.

[19] A. Sugianto, B.S. Tjahjono, J.H. Guo, S.R. Wenham, Impact of laser-induceddefects on the performance of solar cells using localised laser doped regionsbeneath the metal contacts, in: Proceedings of the 22nd European Photo-voltaic Solar Energy Conference, Milan, 2007, pp. 1759–1762.

[20] B.S. Tjahjono, Laser doped selective emitter silicon solar cells, Ph.D. thesis,School of Photovoltaic and Renewable Energy Engineering, University of NewSouth Wales, Sydney, Australia, 2010.

[21] D. Kim, S. Shim, S. Hwang, H. Kim, Electrical properties of screen printedsilicon solar cell dependent upon thermophysical behaviour of glass frits in Agpastes, Japanese Journal of Applied Physics 48 (2009) 05EC06.

[22] J. Yang, D.E. Mosby, S.W. Casteel, R.W. Blanchar, Lead immobilization usingphosphoric acid in a smelter-contaminated urban soil, Environmental Scienceand Technology 35 (2001) 3553–3559.