Insight into band alignment of Zn(O,S)/CZTSe solar cell by simulation?

Zhen-Wu Jiang(姜振武),Shou-Shuai Gao(高守帥),Si-Yu Wang(王思宇),Dong-Xiao Wang(王東瀟),Peng Gao(高鵬),Qiang Sun(孫強),Zhi-Qiang Zhou(周志強),Wei Liu(劉瑋),Yun Sun(孫云),and Yi Zhang(張毅),?

1Institute of Photoelectronic Thin Film Devices and Technology,Tianjin Key Laboratory of Photoelectronic Thin Film Devices and Technology,Tianjin 300071,China

2Tianjin Institute of Power Source,Tianjin 300384,China

1.Introduction

Kesterite structured Cu2ZnSnS4(CZTS) and Cu2ZnSnSe4(CZTSe)materials have received extensive attention due to their non-toxicity,large absorption coefficient(>104cm2),and earth abundant component elements in earth.[1–7]Up to now,the champion efficiency of Cu2ZnSn(S,Se)4(CZTSSe)solar cell with CdS buffer has reached to 12.6%.[8]For pure selenide CZTSe device,the device performance has been improved to 11.6%with CdS serving as a buffer layer.[9]However,the CdS buffer layer is a potential risk for CIGS and CZTS(e)solar cells[10–12]since the cadmium in CdS is toxic and the band gap of CdS is low(2.4 eV),which will cause the absorption to lose in a short-wavelength region.

Owing to these disadvantages,alternative buffer materials are used to replace the CdS buffer layer. Among them,Zn(O,S)is a promising alternative material due to its non-toxicity and large band gap which can be modified from 2.6 eV to 3.8 eV,depending on the S/(S+O)ratio in Zn(O,S).[13,14]Many groups have reported the kesterite structure-based device with Zn(O,S)serving as a buffer layer.Among them,Grenet et al.have reported that the barrier at the CZTSSe/Zn(O,S)interface decreased after light soaking treatment,and the device efficiency was increased to 5.8%.[15]Neuschitzer et al.optimized the band alignment between CZTSe absorber layer and Zn(O,S)buffer layer by changing the thiourea concentration in the CBD process.The device efficiency was increased up to 6.5%which was close to that of CdS reference device.[16]Recently,our group eliminated the secondary phases existing in the as-grown Zn(O,S)by sequential concentrated ammonium solution etching and low temperature annealing treatment.Finally,the Zn(O,S)/CZTSe efficiency was increased to 7.2%by our group recently.[17]However,we found that the band offset between Zn(O,S)and CZTSe is higher than 1 eV,which is very difficult for the carrier transport.After the temperature-dependent I–V study,we deduce that the carrier transport in such a system should be realized by a defect energy level close to Fermi level which acts as a shortcut for the carrier transport.

The performance of the Zn(O,S)/CZTSe device also suffers the low conductivity.[18,19]The efficiency of the device is seriously deteriorated with low buffer carrier concentration.The most commonly used method to enhance the carrier concentration is element doping.Many groups have improved the carrier concentration of the TCO by doping the group III elements like indium into ZnO.[20–23]Indium acts as a dopant since it has a similar atomic radius to Zn2+.It can substitute Zn2+in Zn Oto form In+Znsubstitution,which is an n-type dope for ZnO.[24]Recently,Jani et al.have doped different concentrations of indium into Zn(O,S)layer by the spray deposition method.[25]The carrier concentration of Zn(O,S)was significantly enhanced by doping 1 wt%–3 wt%indium.The electrical conductivity was improved due to the formation of In+Znsubstitution and reduction of hydroxyl group in the Zn(O,S)by annealing treatment in argon atmosphere.[25]Later,Mitzi et al.used In2S3/CdS double buffer layer in the traditional CZTSSe solar cells.The indium could diffuse into both CdS buffer layer and CZTSSe absorber layer by rapid thermal annealing treatment.Since the atomic radius of indium was similar to that of cadmium in CdS and that of tin in CZTSSe,indium could substitute cadmium and tin and form InCdand InSnsubstitution.Consequently,the carrier concentration of buffer layer and absorber layer were enhanced due to the formation of n-type doping and p-type doping,respectively.[26]Ikeda et al.used the In2S3as a buffer layer and they also reported that the CZTS carrier concentration was increased after rapid post-heat treatment.[27]

Kesterite structured solar cells with Zn(O,S)have made considerable progress.However,the study of detailed understanding of the CZTSe/Zn(O,S)is still rare.In this study,we investigate the band alignment between Zn(O,S)layer and CZTSe layer.The effect of the carrier concentration variation in the buffer on the performance of CZTSe solar cells is also studied.We find that double buffer layer In2S3/Zn(O,S)mainly increases the carrier concentration.

2.Numerical modeling

The devices are simulated by wxAMPS software,which is updated from the original AMPS program.[28]Based on the main physical principles of AMPS,the wxAMPS software induces two new tunneling models and improves the algorithm by combining the Newton method and the Gummel method.Compared with other simulation software,the wxAMPS provides a good stability and running speed,and is very suitable for simulating materials with defect densities,band tails.[29]The device parameters used in the simulation process are listed in Table 1.The temperature and the illumination spectrum are set to be 300 K and AM1.5,respectively.

Table 1.Device parameters used in simulation.

3.Results and discussion

3.1.S/(S+O)ratio in Zn(O,S)

The band gap of Zn(O,S)can be adjusted from 2.6 eV to 3.8 eV by modifying the S/(S+O)ratio.[13,14]Sharbati S et al.reported that the band gap of the Zn(O,S) first decreases and then increases,and the value of the conduction band and the valence band both change with S fraction variety in the Zn(O,S).[30]The variation of electron affinity(χe)and band gap energy(Eg)with S ratio x=S/(S+O)in Zn(O1?xSx)can be approximately calculated from the following equations:[13,31]

where x is the S fraction,Egis the band gap,and the band gap value of ZnS and ZnO are 3.6 eV and 3.2 eV,respectively,and b is the bowing factor,which is about 3.Figure 1 shows the corresponding values and variation tendencies of Egand χewith different S/(S+O)ratios.Obviously,the minimum band gap appears in the region where the S and O fraction are almost equal.The value of Egincreases no matter whether the S fraction is higher or lower than 0.5.

Fig.1. Electron affinity and band gap energy of Zn(O,S)versus S/(S+O)ratio.

As the band gap and the electron affinity of Zn(O,S)vary with the S fraction,[32]the S/(S+O)ratio becomes a key factor to affect the conduction band offset(CBO)at the interface between CZTSe absorber layer and Zn(O,S)buffer layer.[31]The Zn(O,S)electron affinity tends to slide downward as the S fraction increases(Fig.1).So the CBO between CZTSe and Zn(O,S)is negative when the buffer layer has a lower S content(x≤0.5)which is named the cliff-like band alignment.In this case,the CZTSe conduction band edge is higher than that of Zn(O,S),the recombination in the CZTSe/Zn(O,S)interface increases.[33,34]As a result,the FF andVocbecome undesirable,and the performance of the device will deteriorate.In contrast,the CBO in the CZTSe/Zn(O,S)interface is spikelike alignment when the S fraction is higher than 0.5(x>0.5).Too large a CBO will induce a barrier that hinders the photogenerated electrons from transferring across the interface,resulting in Jscdecreasing.However,when the value of CBO is set to be in a range from 0 eV to 0.4 eV,the negative effect of the spike-like CBO is negligible.[33,34]

Fig.2.Band alignments of Zn(O,S)/CZTSe devices with different S fractions.

Fig.3.J–V curves of Zn(O,S)/CZT Sedevices with different S fractions.

Figure 2 shows the band structures of the CZTSe/Zn(O,S)device with different S/(S+O)ratios(x=0.2,0.7,0.9).The thickness of the Zn(O,S)used in simulation is 20 nm.The value of CBO at CZTSe/Zn(O,S)interface changes from negative to positive with the increase of S content.As the S fraction increases in Zn(O,S),the barrier induced by large CBO becomes higher.[31]From Fig.3,when x is 0.9,the CBO is far beyond 0.4 eV,which is considered to be difficult for electron to flow across the interface.We can see that when the S content is higher than 0.7,the J–V characteristic curve of device is damaged due to the large barrier for photogenerated electrons.In our simulation process,the optimal S content appears at 0.7,and in this case,the CBO between CZTSe and Zn(O,S)is about+0.2 eV,which is in a reasonable range(0 eV–0.4 eV).[33,34]The performance of the device is improved due to the small CBO value at the interface and high light absorption.

3.2.Thickness of Zn(O,S)

The optimal band gap of Zn(O,S)simulation(2.8 eV)is larger than that of CdS(2.4 eV).Consequently,the absorption loss in the short-wavelength region of the visible spectrum is reduced by using the Zn(O,S) film to replace the traditional CdS buffer material,thus improving the short circuit current of the device.The EQE responses of the devices with different buffer layers(CdS layer and Zn(O,S)layer)are showed in Fig.4.The short wavelength response of device is obviously improved by replacing the CdS layer with Zn(O,S)layer.Figure 5 shows the band structures of solar cells with different Zn(O,S)thickness values for the S/(S+O)=0.7.Neuschitzer et al.have fabricated a 6.5%efficient CBD-Zn(O,S)/CZTSe device after light soaking treatment,[16]while their CBDZn(O,S)was less than 10 nm in thickness.According to the results reported previously,too thin a buffer layer can induce pinholes and shunt channels which deteriorate the device parameters.Meanwhile,as the carrier concentration of the untreated CBD-Zn(O,S)is usually low,[18,19]the barrier of the carrier transport will become higher with the thickness of the Zn(O,S)increasing.

Fig.4.EQE responses to wave of devices with different buffer layers.

Fig.5.Energy band structure diagram of Zn(O,S)/CZTSe devices for various Zn(O,S)layer thickness.

Figure 6 shows the variation of the short circuit current density(Jsc)and fill factor(FF)with the Zn(O,S)thickness increase.This demonstrates that the Jscand FF dramatically collapse as the Zn(O,S)thickness increases when the carrier concentration of Zn(O,S)is low(5×1017cm?3).However,the performance of the device is almost constant when the carrier concentration is adequately high(5×1018cm?3).This phenomenon is due to the low buffer carrier concentration,leading to the undesirable conductivity in Zn(O,S)layer.[19]However,it will become serious when Zn(O,S) film is thicker.This indicates that the effect of thickness is dependent on the Zn(O,S)carrier concentration.

Fig.6.Curves of(a)Jscand(b)FF of Zn(O,S)/CZTSe solar cells versus Zn(O,S)layer thickness with different Zn(O,S)carrier concentrations.

Similarly,we can see from Fig.7(a)that the poor conductivity makes the photo-generated electrons hard to collect by the buffer layer in the case of constant Zn(O,S)layer thickness when the carrier concentration of Zn(O,S)layer is low.The recombination rate increases in the absorber region as thickness increases.Thus the Jscand FF are deteriorated seriously respectively as the thickness reaches to 100 nm.From Fig.7(b),when the carrier concentration is as high as 5×1018cm?3,the recombination rate in absorber layer is low and unchanged with the thickness of Zn(O,S)layer.As the photo-generated carrier is mainly generated at CZTSe absorber layer,the recombination rate at the buffer layer has almost no effect on the device performance,so the variation of Jscand FF caused by the variable thickness are negligible.To reduce the light absorption in the buffer layer,the thickness of Zn(O,S)should not be too thick.To ensure that the absorber layer is completely covered,the optimum Zn(O,S)thickness is 20 nm in our simulation,which accords with our previous experiment result.[17]Besides,we find that the effect of buffer thickness depends on carrier concentration of the buffer.The high carrier concentration is propitious to the device performance.

Fig.7.Recombination rates of Zn(O,S)/CZTSe devices with different Zn(O,S)thickness and carrier concentrations of Zn(O,S)of(a)5×1017cm?3,(b)5×1018cm?3.

3.3.Double buffer layer

TheIn2S3/Zn(O,S)double buffer layer is selected for simulation.Obviously,In3+ions will diffuse into Zn(O,S)buffer layer.As a consequence,the carrier concentration of Zn(O,S)is improved due to the dopant of the n-type In+Zn.Thus,we first try to disclose the effect of In2S3/Zn(O,S)double buffer layer on carrier concentration On the assumption that the In2S3layer is ultrathin,it diffuses into Zn(O,S)layer completely.Figure 8 shows the conduction band alignment of Zn(O,S)/CZTSe heterojunction with different Zn(O,S)carrier concentrations.According to the simulation results and our previous experimental result,[17]the thickness of Zn(O,S)is set to be 20 nm.,The distance from Fermi level to the conduction band minimum when carrier concentration is high(5×1018cm?3)is greater than that when the carrier concentration of Zn(O,S)is as low as 1×1016cm?3.Relatively,the barrier at the interface of Zn(O,S)/CZTSe is high without doping,which impedes the photo-generated electrons flowing across the interface.With the increase of the carrier concentration,the barrier height be-comes lower.The parameters of Zn(O,S)/CZTSe solar cells with different Zn(O,S)carrier concentrations are listed in Table 2,demonstrating that the performance of the device is observed to improve as the carrier concentration increases due to the forming of the In+Znsubstitutions in Zn(O,S)layer.

Fig.8.Conduction band alignments of Zn(O,S)/CZTSe devices with different Zn(O,S)carrier concentrations.

Table 2.Performances of Zn(O,S)/CZTSe cells with different Zn(O,S)carrier concentrations.

Some In2S3layers ineluctably remain on the surface of Zn(O,S)/CZTSe solar cell in experiment.The effect of residual In2S3layers should be further considered.The band gap of In2S3is tunable from 2.1 eV to 2.9 eV,which should be attributed to the variation of the temperature and the oxygen content doped into the In2S3layer.[35–37]Furthermore,the electron affinity of In2S3will change with the variation of band gap.[38]The relationship between the band gap and electron affinity is shown below.[38]

The optimal band gap(2.83 eV)and electron affinity(4.16 eV)of Zn(O,S)are fixed.The properties of Zn(O,S)layer and CZTSe absorber layer keep constant in this simulation.The χerange of In2S3varies from 3.85 eV to 4.65 eV.The band gap of In2S3layer is larger(2.8 eV–2.9 eV)or smaller(2.1 eV–2.8 eV)than that of Zn(O,S)layer.To make clear the effect of In2S3/Zn(O,S)double buffer layer and ensure that the incident light is fully absorbed by CZTSe solar cell,the In2S3layer with a band gap of 2.9 eV should be deposited on the absorber layer and the In2S3layer with a band gap of 2.1 eV–2.8 eV should be deposited between the absorber layer and Zn(O,S)layer. And the thickness of the remaining In2S3layer is set to be 5 nm.Figure 9 shows the J–V curve of Zn(O,S)/CZTSe device with and without 2.9-eV In2S3layer,to compare with those of the device without remaining In2S3layer,and it can be seen that the Vocof the device with 5-nm-thick remaining In2S3layer decreases.The efficiency decreases from 16.18%to 15.48%.Figure 10 shows the J–V curves of Zn(O,S)/In2S3/CZTSe devices with different band gaps of In2S3layer band gap.To make the value of CBO between CZTSe layer and In2S3layer varied in a range from 0 eV to+0.4 eV,the value of In2S3layer band gap is changed from 2.4 eV to 2.8 eV.The performance parameters of the solar cells are almost unchanged at this time.However,the performance of the device deteriorates when the In2S3layer band gap is below 2.4 eV due to the fact that the band alignment between CZTSe layer and In2S3layer is cliff-like.So the optimal efficiency is 15.71%when the In2S3band gap is 2.7 eV.And the parameters are listed in Table 3.

Fig.9.The J–V curve of devices with and without 5-nm In2S3layer.

Fig.10.The J–V curves of devices combined with In2S3layer with different band gaps.

Table 3.Performances of Zn(O,S)/In2S3/CZTSe devices with 2.7-eV band gap In2S3layer.

Figure 11 shows J–V curves of CZTSe devices in different buffer conditions,which are,respectively,for the following cases:ND=5×1017cm?3without In2S3layer,ND=5×1018cm?3with In2S3diffusing into the Zn(O,S)layer completely by annealing treatment,and ND=5×1018cm?3with 5-nm-thick In2S3layer remaining on the top of Zn(O,S)layer or CZTSe layer.The detailed parameters are listed in Table 4,indicating that the device efficiency increases to 16.18%when the NDis improved by In2S3doping.The device performance becomes slightly worse when ultrathin In2S3layer remains on the top of Zn(O,S)layer or CZTSe layer,while the parameters are still better than those without In2S3layer due to the lower NDin Zn(O,S)layer.Therefore,to achieve optimal device performance,the thickness of In2S3layer should be very thin.

In summary,we have analyzed the performance of Zn(O,S)/CZTSe device from the perspective of band structure and carrier concentration by numerical simulation.The band alignment between Zn(O,S)layer and CZTSe layer is+0.2 eV and the carrier concentration of Zn(O,S)layer is improved by In2S3doping.The efficiency of Zn(O,S)/CZTSe device is improved to 16.18%,when no In2S3layer remains on CZTSe layer.

Fig.11.J–V curves of devices for ND=5×1017cm?3,ND=5×1018cm?3without In2S3layer,ND=5×1018cm?3with 5-nm-thick In2S3layer remaining on Zn(O,S)layer,ND=5×1018cm?3with 5-nm-thick In2S3layer remaining on CZTSe layer.

Table 4.Performances of devices for ND=5×1017cm?3,ND=5×1018cm?3without remaining In2S3layer,ND=5×1018cm?3 with 5-nm-thick In2S3layer remaining on Zn(O,S)layer,ND=5×1018cm?3with 5-nm-thick In2S3layer remaining on CZTSe layer.

4.Conclusions

In this study,the Zn(O,S)/CZTSe solar cell performance is simulated and studied by using the wxAMPS software.To achieve the optimal performance,the S/(S+O)ratio in Zn(O,S),Zn(O,S)thickness,and carrier concentration are discussed.By numerical simulation,the optimum S fraction is about 0.7,here the value of CBO in the Zn(O,S)/CZTSe interface is about+0.2 eV.The CBO is in a range of 0 eV–0.4 eV as S/(S+O)=0.6–0.8 in Zn(O,S).Meanwhile,the In2S3/Zn(O,S)double buffer layer is innovatively induced into the CZTSe device.The carrier concentration of the Zn(O,S)layer is improved by employing the In2S3/Zn(O,S)double buffer layer on CZTSe absorber layer upon annealing treatment.The device efficiency increases to 16.18%when the carrier concentration of Zn(O,S) layer is improved without any remaining In2S3.The efficiency of Zn(O,S)/5 nm-In2S3/CZTSe and 5 nm-In2S3/Zn(O,S)/CZTSe devices decrease to 15.71%and 15.48%,respectively.However,they are still higher than that of device without In2S3layer(14.93%).These promising results indicate that the performance of Zn(O,S)/CZTSe device has great potential improvement and they provide guidance for further study of CZTSe solar cells.

[1]Li J,Kim S,Nam D,Liu X,Kim J,Cheong H,Liu W,Li H,Sun Y and Zhang Y 2017 Sol.Energy Mater.Sol.Cells 159 447

[2]Polizzotti A,Repins I L,Noufi R,Wei S H and Mitzi D B 2013 Energy Environ.Sci.6 3171

[3]Mitzi D B,Gunawan O,Todorov T K,Wang K and Guha S 2011 Sol.Energy Mater.Sol.Cells 95 1421

[4]Walsh A,Chen S,Wei S H and Gong X G 2012 Adv.Energy Mater.2 400

[5]Barkhouse D,Aaron R,Gunawan O,Gokmen T,Todorov T K and Mitzi D B 2012 Prog.Photovoltaics Res.Appl.20 6

[6]Bag S,Gunawan O,Gokmen T,Zhu Y,Todorov T K and Mitzi D B 2012 Energy Environ.Sci.5 7060

[7]Gao S,Jiang Z,Wu L,Ao J,Zeng Y,Sun Y and Zhang Y 2018 Chin.Phys.B 27 018803

[8]Wang W,Winkler M T,Gunawan O,Gokmen T,Todorov T K,Zhu Y and Mitzi D B 2014 Adv.Energy Mater.4 1301465

[9]Lee Y S,Gershon T,Gunawan O,Todorov T K,Gokmen T,Virgus Y and Guha S 2015 Adv.Energy Mater.5 1401372

[10]Klenk R,Steigert A,Rissom T,Greiner D,Kaufmann C A,Unold T and Lux-Steiner M C 2014 Prog.Photovoltaics Res.Appl.22 161

[11]Gautron E,Buf fi`ere M,Harel S,Assmann L,Arzel L,Brohan L,Kessler J and Barreau N 2013 Thin Solid Films 535 175

[12]Yagioka T and Nakada T 2009 Appl.Phys.Express 2 072201

[13]Meyer B K,Polity A,Farangis B,He Y,Hasselkamp D and Wang C 2004 Appl.Phys.Lett.85 4929

[14]Chua R H,Li X,Walter T,The L K,Hahn T,Hergert F and Wong L H 2016 Appl.Phys.Lett.108 043505

[15]Grenet L,Grondin P,Coumert K,Karst N,Emieux F,Roux F,Fillon R,Altamura G,Fournier H,Faucher,P and Perraud S 2014 Thin Solid Films 564 375

[16]Neuschitzer M,Lienau K,Guc M,Barrio L C,Haass S,Prieto J M and Izquierdo-Roca 2016 J.Phys.D:Appl.Phys.49 125602

[17]Li J,Liu X,Liu W,Wu L,Ge B,Lin S,Gao S,Zhou Z,Liu F,Sun Y,Ao J,Zhu H,Mai Y and Zhang Y 2017 Sol.RRL.1 1700075

[18]Hsieh T M,Lue S J,Ao J,Sun Y,Feng W S and Chang L B 2014 J.Power Sources 246 443

[19]Steirer K X,Garris R L,Li J V,Dzara M J,Ndione P F,Ramanathan K,Repins I,Teeter G and Perkins C L 2015 Phys.Chem.Chem.Phys.17 15355

[20]Palimar S,Bangera K V and Shivakumar G K 2013 Appl.Nanosci.3 549

[21]Llican S,Caglar Y,Caglar M and Demirci B 2008 J.Optoelectron.Adv.Mater.10 2592

[22]Pham A T T,Ta H K T,Liu Y R,Aminzare M,Wong D P,Nguyen T H,Pham N K,Le T B N,Seetawan T,Ju H,Cho S,Chen K H,Tran V C and Phan T B 2018 J.Alloys Compd.747 156

[23]Benramache S,Benhaoua B and Bentrah H 2013 Nanostrut.Chem.3 54

[24]Gonc?alves G,Elangovan E,Barquinha P,Pereira L,Martins R and Fortunato E 2007 Thin Solid Films 515 8562

[25]Jani M,Raval D,Pati R K,Mukhopadhyay I and Ray A 2018 Bull.Mater.Sci.41 22

[26]Kim J,Hiroi H,Todorov T K,Gunawan O,Kuwahara M,Gokmen T,Nair D,Hopstaken M,Shin B,Lee Y S,Wang W,Sugimoto H and Mitzi D B 2014 Adv.Mater.26 7427

[27]Jiang F,Ozaki C,Guna wan,Harada T,Tang Z,Minemoto T,Nose Y and Ikeda S 2016 Chem.Mater.28 3283

[28]Liu Y,Sun Y and Rockett A 2012 Sol.Energy Mater.Sol.Cells 98 124

[29]Omer and Mohamed B 2015 Chin.Phys.Lett.32 088801

[30]Sharbati S,Keshmiri S H,McGoffin J T and Geisthardt R 2015 Appl.Phys.A 118 1259

[31]Persson C,Platzer-Bjorkman C,Malmstrom J,Torndahl T and Edoff M 2006 Phys.Rev.Lett.97 46403

[32]Grimm A,Kieven D,Klenk R,Lauermann I,Neisser A,Niesen T and Palm J 2011 Thin Solid Films 520 1330

[33]Huang T J,Yin X,Qi G and Gong H 2014 Phys.Status Solidi RRL.08 735

[34]Minemoto T,Matsui T,Takakura H,Hamakawa Y,Negami T,Hashimoto Y and Kitagawa M 2001 Sol.Energy Mater.Sol.Cells 67 83

[35]Barreau N,Bernede J C,Marsillac S and Mokrani A 2002 J.Crystal Growth 235 439

[36]Revathi N,Prathap P,Subbaiah Y P V and Ramakrishna Reddy K T 2008 J.Phys.D:Appl.Phys.41 155404

[37]Saadallah F,Jebbari N,Kammoun N and Yacoubi N 2011 Int.J.Photoenergy 1–4

[38]Khoshsirat N and Md Yunus N A 2016 J.Electron.Mater.45 5721