ARTICLE IN PRESS

Journal of Crystal Growth 311 (2009) 2621–2625

Contents lists available at ScienceDirect

Journal of Crystal Growth

0022-02

doi:10.1

�Corr

E-m

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

Synthesis of CIGS absorber layers via a paste coating

Jong Won Park a, Young Woo Choi a, Eunjoo Lee a, Oh Shim Joo a, Sungho Yoon b, Byoung Koun Min a,�

a Clean Energy Research Center, Korea Institute of Science and Technology, 39-1 Hawolgok-dong, Seongbuk-gu, Seoul 136-791, South Koreab Department of Bio & Nano Chemistry, Kookmin University, 861-1 Jeongneung-dong, Seongbuk-gu, Seoul 136-702, South Korea

a r t i c l e i n f o

Article history:

Received 7 January 2009

Received in revised form

20 February 2009

Accepted 23 February 2009

Communicated by R. Kerncomposition was observed during thermal annealing at 450 1C in the absence of a gas-phase selenium

Available online 4 March 2009

PACS:

71.20.Nr

81.05.Hd

82.80.Ej

61.05.F�

61.66.Dk

68.37.Hk

68.55.ag

Keywords:

A1. Thin films

A1. X-ray diffraction

A3. Solar cells

B2. Semiconducting quarternary alloys

B3. Solar cells

48/$ - see front matter & 2009 Elsevier B.V. A

016/j.jcrysgro.2009.02.038

esponding author. Fax: +82 2 958 5859.

ail address: bkmin@kist.re.kr (B.K. Min).

a b s t r a c t

CuInxGa1�xSe2 (CIGS) thin films were prepared by a paste coating with the aim of developing a simpler

and the lower cost method for fabricating the absorber layers of thin film solar cells. In particular, a

paste of a Cu, In, Ga, and Se precursor mixture was first prepared, followed by a reaction between them

at elevated temperatures after depositing the paste onto a glass substrate. No apparent change in

source. A pre-annealing process at 250 1C under ambient conditions performed before annealing

(450 1C) under reduction conditions reduced the level of carbon deposition in/on the films without

perturbing the stoichiometry of the CIGS thin films.

& 2009 Elsevier B.V. All rights reserved.

1. Introduction

Among the various thin film solar cells in the market,CuInxGa1�xSe2 (CIGS) thin film solar cells have been consideredthe most promising alternatives to crystalline silicon solar cells onaccount of their potential high solar to electricity-conversionefficiency, reliability, and stability [1–4]. The CIGS thin film solarcells fabricated by conventional vacuum-based routes, such as co-evaporation and sputtering techniques, have already achievedcomparable efficiency (�19%) to polycrystalline silicon solar cells,which is much higher than that of other thin film solar cells [2–4].However, as noted by many other researchers, the currentmethods for fabricating CIGS thin film solar cells need consider-able improvements considering the manufacturing cost and scaleas well as environmental issues [1,5–8]. Since the currentmethods for manufacturing CIGS thin film solar cells are basedon vacuum processes, they require initial high capital investment

ll rights reserved.

as well as maintenance of capital expense arising from thenecessity of vacuum equipment. Moreover, there is significant lossof resource material (20–50%), which again increases themanufacturing cost [5]. In addition, the large-scale productionof photovoltaic (PV) panels would be restricted due to the limitingsize of vacuum equipment.

Non-vacuum processes have been suggested to solve theproblems of the current CIGS solar cell fabrication methods, suchas printing, electroplating, spraying, etc. [5–13]. In particular,precursor-based printing methods, e.g. screen printing and doctor-blade, would be competitive in terms of processing capital costs,efficiency of resource material usage, and processing speed [5–8].In addition, these methods can be applied to continuous roll-to-roll deposition processes as well as to large-scale panel fabrica-tion. In precursor-based printing methods, various types ofelemental sources have been applied. For example, metal nitrates,metal chlorides, metal iodides, or elemental metals have beenused as Cu, In, and Ga precursors. In most cases, the cationic formof the metals is generated from these precursors [5–11]. Incontrast, gas-phase Se precursors, such as H2Se or Se vapor, aregenerally used as Se sources instead of the salt-type of precursors.

ARTICLE IN PRESS

J.W. Park et al. / Journal of Crystal Growth 311 (2009) 2621–26252622

Although the salt-type of precursors (e.g. Na2Se) have been usedto synthesize of CIGS alloys, a selenization process using H2Se orSe vapor is subsequently applied [9,10].

Selenization may be an essential process in the synthesis of aCIGS absorber layer. It was reported that selenization helps toincrease the grain size of CIGS crystal and compensates for theloss of Se during high-temperature annealing process [1].However, despite these beneficial effects of selenization for CIGSthin film synthesis, there are several drawbacks [1]. First, therepresentative Se precursor, H2Se, which is a very toxic chemical,is unavoidably released into the atmosphere during the seleniza-tion process. Second, it may be a barrier for the overall cost-effective fabrication process of CIGS solar cells because it requiresseparate facilities from the synthesis of Cu-In-Ga alloy layers.

To produce CIGS thin film solar cells in a cost-effective manner,we have been carrying out a long-term research plan in that allthe fabrication processes of CIGS are accomplished usingprecursor-based non-vacuum processes. Furthermore, we areaiming to develop processes that can replace some environmentalunfriendly ones, such as selenization with hazardous gases, in thesynthesis of the CIGS absorber layer. In this study, as a first step ofthis project, we suggest a new method for fabricating the CIGSabsorber layer using a paste coating. With this method, a cationicSe precursor rather than elemental or anionic Se was used toprepare the paste, and no further selenization using gas-phaseprecursors was employed to synthesize the CIGS absorber layer.

Inte

nsity

(a.u

.)

: CuIn0.5Ga0.5Se2: CuGa3Se5: In2Se3: CuInGaO4

400°C

300°C

2. Experimental

A precursor mixture was prepared by dissolving the appro-priate amounts of Cu(NO3)2 �3H2O (99.5%, Kanto, 0.50 g),Ga(NO3)3 � xH2O (99.9%, Alfa Aesar, 0.27 g), In(NO3)3 � xH2O(99.9%, Alfa Aesar, 0.31 g), and SeCl4 (99.5%, Alfa Aesar, 0.92 g) inanhydrous ethanol (60 mL), followed by the addition into asolution of terpineol (Fruka, 10.0 g) and ethyl cellulose (Alfa aesar,0.75 g) in ethanol (20 mL). After condensing the solution at 40 1Cunder reduced pressure, a viscose paste with rheological proper-ties suitable for a doctor-blade coating was prepared.

After printing the paste on the glass substrates, the filmthickness was adjusted to �2mm using two stripes of a 3 M scotchtape, applied to both sides of the glass substrates. The sampleswere then placed into a wind blower (Leister) for pre-annealingand/or a furnace for high-temperature annealing. Pre-annealingwas performed below 250 1C for 1 h under ambient conditions,and high-temperature annealing was carried out at 300–500 1Cfor 40 min in a H2(5%)/Ar gas mixture at a flow rate of 80 mL/min.

Structural characterization of the films was performed byscanning electron microscopy (SEM, Hitachi, S-4200) and X-raydiffraction (XRD, Shimadzu, XRD-6000). Composition analysis wascarried out using energy dispersive X-ray spectroscopy (EDX,Horiba) and ICP–AES (Perkin-elmer). Thermogravimetric analysis(TGA, Ta Instruments Inc.) was also performed to determineappropriate annealing conditions as well as to determine thethermal stability of the films.

202 Theta (degree)

40 60 80

200°C

100°C

Fig. 1. X-ray diffraction (XRD) data of the various films prepared from the

Cu–In–Ga–Se precursor paste with respect to the annealing temperatures under

ambient conditions.

3. Results and discussion

The use of a cationic Se precursor, SeCl4, is unusual since thefinal product, CuInxGa1�xSe2, is a chalcogenide compound whereSe has an oxidation state of �2. Na2Se is most frequently used as aSe source for precursor-based CIGS thin film synthesis [6]. Areduction process is necessary to change the oxidation state of Sefrom +4 to �2. Alcohols, e.g. ethanol and/or terpineol in thissynthetic method may play a role as reducing agents at elevated

temperatures similar to the polyol process, which is well known inmetal nanoparticle synthesis [14,15].

Cu, In, and Ga nitrates and SeCl4 were dissolved completely inethanol at room temperature to generate the precursor mixture.An ethanol solution of terpineol and ethyl cellulose was alsoprepared and combined with the precursor mixture. It should benoted that the pH of the mixture was approximately 1, indicatingproton generation from alcohols Eq. (2). This suggests that Se4+

ions coordinate with alkoxy species to form Se(OR)4 Eq. (2).

SeCl4 ! Se4þ þ 4Cl� (1)

Se4þ þ 4ROH! SeðORÞ4 þ 4Hþ (2)

4Hþ þ 4Cl� ! 4HClm (3)

Cu2þ þ xIn3þ þ ð1� xÞGa3þ þ SeðORÞ4

þ ROH! CuInxGað1�xÞSe2 þ residual organicsm (4)

Although a viscous paste with suitable rheological propertieswas prepared by condensing the mixture solution under reducedpressure, the reaction between the precursors to the CIGS alloyparticles does not occur in this stage. However, some HCl gas wasgenerated, resulting in an increase in pH to �3 Eq. (3).

A reaction Eq. (4) between the metal ions in the precursormixture occurred during heat treatment of the printed pasteabove 100 1C. The temperature effects on the compositionalchange were examined by carrying out the heat treatments underambient conditions using a hot wind blower for 1 h. As shown inFig. 1, heat treatment at 100 1C resulted in the appearance of anXRD peak at 28.11 2y with a weak peak at 46.51 2y . At highertemperatures (200 1C), the intensity of the XRD peak at 28.11 2yincreased with the appearance of new peaks at 29.21, 47.81, and55.81 2y. Heat treatment at 300 1C resulted in another new intensepeak at 31.11 2y with a weak peak at 49.91 2y. Simultaneously,weak peaks at 26.51 and 45.51 2y were observed. However, asignificantly different XRD pattern was observed when the printedfilms were annealed at 400 1C, showing only weak peaks at 30.61and 35.51 2y. The color of the films also changed from black togrey.

ARTICLE IN PRESS

J.W. Park et al. / Journal of Crystal Growth 311 (2009) 2621–2625 2623

Based on the XRD database (PDF-2), the XRD peaks wereassigned tentatively as follows. The XRD peaks at 28.11 and 46.512y were assigned to CuGaxSey (most likely x=3 and y=5) accordingto the reference (PDF # 5112230). In addition, the peaks at 29.21and 47.81 2y, which evolved during annealing at 200 1C, may resultfrom the formation of InxSey (most likely x=2 and y=3) accordingto the reference (PDF # 721469), implying the existence of twodifferent crystalline structures. The XRD pattern of the filmprepared at 300 1C showed more complicated features. The peaksat 26.51 and 45.51 2y were assigned to stoichiometric CIGS, andthe intense peak at 31.11 2y may have originated from CuInGaoxide (most likely CuInGaO4, PDF # 731866). Above 300 1C, Sewould begin to evaporate as SeO2 because the sublimationtemperature of SeO2 is relatively low (315 1C), resulting in aCuInGaO4 film. The formation of CuInGa oxide was also confirmedby EDX where oxygen was observed to be the major species withthe substantial loss (490%) of Se. Overall, with respect to theheating temperature, the films from the CIGS precursor pasteinitially formed CuGa3Se5 at 100 1C followed by the formation ofIn2Se3 at 200 1C. At higher temperatures (300 1C), a variety ofcrystalline structures co-existed in the film including CuInGaO4

and a stoichiometric CIGS alloy. At above 400 1C, most of the Sehad evaporated resulting in only CuInGaO4 remaining in the film.

To prevent oxidation of the CIGS alloy films, the annealingprocess was carried out under reduction conditions. Afterdepositing the Cu–In–Ga–Se mixture paste onto the glasssubstrate using the doctor-blade method, the films were annealedat three different temperatures (300, 400, and 500 1C) for 40 min

20

Inte

nsity

(a.u

.)

2Theta (degree)

300 °C

400 °C

(112)

(220)/(204)

(116)/(312)

(400)/(008) (424)/(228)

40 60 80

500 °C

Fig. 2. X-ray diffraction (XRD) data of the various films prepared from the

Cu–In–Ga–Se precursor paste with respect to the annealing temperatures under

reduction conditions (H2/Ar).

in H2(5%)/Ar gas environment. The thin films prepared byannealing at 300 1C showed three major peaks at 26.71, 44.51,and 52.81 2y, which is different from the XRD pattern observed inthe film prepared under ambient conditions (see Fig. 1). In theXRD patterns, the most intense peak at 26.71 2y indicates thestoichiometric CIGS alloy has a (112) orientation. The otherprominent peaks correspond to the (2 0 4)/(2 2 0) and (116)/(31 2)phase. In addition to these peaks commonly observed in CIGS,several weak peaks, such as (4 0 0)/(0 0 8) and (4 2 4)/(2 2 8), werealso present in the XRD patterns (PDF # 401488). The presence ofthese peaks clearly indicates the perfect chalcopyrite structure ofCIGS, which is in good agreement with other reported values [3,5].As the annealing temperature was increased to 500 1C, theintensity of the XRD peaks also increased, indicating enhancedcrystallinity of the film, as shown in Fig. 2.

The crystalline grain size of the CIGS thin films prepared atdifferent annealing temperatures was also estimated using theDebye–Scherrer formula

D � l=ðb cos yÞ (5)

where l is the X-ray wavelength (1.54 A for CuKa radiation), b thefull peak-width at half-maximum (FWHM) in radians, and y thepeak position. The average grain size in the CIGS thin films wascalculated to be ca. 21, 28, and 36 nm at 300, 400, and 500 1C,respectively, which indicates that a higher annealing temperature

00

20

40

60

80

100

Wei

ght (

%)

Temperature (°C)

A B

60

80

100

Wei

ght (

%)

200 400 600 800 1000

0Temperature (°C)

200 400 600 800 1000

Fig. 3. The thermogravimetric analysis (TGA) using (a) a paste of Cu–In–Ga–Se

precursor mixture, (b) CIGS powder prepared by the same way as CIGS thin films.

Regions A and B in (a) indicate the temperature ranges of evaporation of organic

addictives and decomposition of CIGS thin film, respectively.

ARTICLE IN PRESS

J.W. Park et al. / Journal of Crystal Growth 311 (2009) 2621–26252624

induces particle growth. Since the paste preparation requiredorganic additives, such as ethyl cellulose and terpineol, thedeposition of residual carbon impurities would be expectedduring the annealing processes. Indeed, annealing under reduc-tion conditions resulted in significant carbon deposition on/in theCIGS thin films. In order to reduce the presence of residual carbon,a pre-annealing process for 1 h under ambient conditions wasperformed before high-temperature annealing at 450 1C inreduction conditions. According to thermogravimetric analysisusing a paste of a Cu–In–Ga–Se precursor mixture, most of theorganic compounds would be eliminated during the heat treat-ment up to 300 1C, as shown in Fig. 3a (region A).

Fig. 4 shows the XRD patterns of the films prepared by pre-annealing at different temperatures under ambient conditionsfollowed by subsequent annealing at 450 1C under reductionconditions. As reported earlier, the films prepared by heatingbelow 200 1C under ambient conditions did not show thestoichiometric structure of the CIGS alloy, rather they revealedthe formation of a CuGa3Se5 and In2Se3 mixture crystal structure.However, subsequent high-temperature annealing underreduction conditions generated the stoichiometric CIGS alloy, asshown in Fig. 4. In contrast, the films pre-annealed above 300 1Cresulted in the formation of CuInGaO4 due to annealing at 450 1Cunder reduction conditions.

0.0

0.5

1.0

1.5

2.0 Cu In Ga Se

Elem

ents

/Cu

Rat

io

Precursor mixture Thin film

Fig. 5. The comparison of the compositions between the precursor pastes and the fi

(b) Cu:In:Ga:Se=1:0.3:0.7:2.

20

Inte

nsity

(a.u

.)

2 Theta (degree)

(a)

(b)

(c)

(d)

(112)

(220) /(204)(116) /(312)

(101) (104) (015)

100°C

200°C

300°C

400°C

40 60 80

Fig. 4. X-ray diffraction (XRD) data of the various films prepared by the high-

temperature annealing at 450 1C under reduction conditions after the pre-

annealing at (a) 100 1C, (b) 200 1C, (c) 300 1C, and (d) 400 1C.

The most important aim of this study was to develop asynthetic method for CIGS thin films from an all salt-typeprecursor in the absence of an additional selenization process.To validate the proposed method, there should be no substantialloss of selenium from the films during high-temperature anneal-ing. TGA (Fig. 3a) in a nitrogen atmosphere shows little weightloss between 300 and 500 1C indicating insignificant componentlosses due to the heat treatments. The thermal stability of the

0.0

0.5

1.0

1.5

2.0 Cu In Ga Se

Elem

ents

/Cu

Rat

io

Precursor mixture Thin film

lms prepared by two-step annealing processes. (a) Cu:In:Ga:Se=1:0.5:0.5:2 and

15.0kV x 10.0K 3.0 µm

15.0kV x 100K 300 nm

Fig. 6. (a) Planar and (b) cross-sectional SEM micrographs of CIGS thin film

prepared by two-step annealing processes. The sample was coated by Pt prior to

the SEM measurement.

ARTICLE IN PRESS

J.W. Park et al. / Journal of Crystal Growth 311 (2009) 2621–2625 2625

CIGS alloy was also tested by TGA using the CIGS powder that wassynthesized by annealing the paste at 450 1C under reductionconditions. As shown in Fig. 3b, there was negligible weight loss ofthe CIGS powder up to 500 1C, indicating no decomposition of theCIGS alloy up to 500 1C. The composition of the pastes and thefilms prepared using the two-step annealing processes (pre-annealing at 250 1C and high-temperature annealing at 450 1C)was also compared. The composition of the CIGS film wasestimated by EDX. As shown in Fig. 5, there were only smallvariations in composition indicating no significant loss ofelements including Se.

Fig. 6 shows the representative morphology of the CIGS filmprepared using the two-step annealing processes. The grain sizeobserved in the SEM image was estimated to be 20–30 nm, whichis in agreement with that calculated from the Debye–Scherrerformula from the XRD data. However, as shown in the cross-section micrograph, there were virtually no crystalline grains,suggesting that the CIGS particles were still embedded in residualcarbon. It would be necessary to efficiently remove the carbonresidues in order to apply the CIGS thin films prepared from thepaste coating to the fabrication of CIGS solar cell devices. This maybe achieved using organic binders that can be burned out at lowertemperatures.

4. Conclusion

Stoichiometric CIGS thin films were synthesized using a pastecoating on glass substrates. A paste of a Cu–In–Ga–Se mixture wasprepared from an all salt-type of precursor and the actual reactionto form CIGS alloy occurred at elevated temperatures where thealcohol used may act as a reducing agent. Pre-annealing processesat 250 1C under ambient conditions followed by annealing at

450 1C under reduction conditions was found to be optimum forpreserving the stoichiometry of the films with reduced residualcarbon deposition. Importantly, the composition was relativelyunaffected by the thermal-annealing process even in an absenceof a gas-phase Se environment.

Acknowledgement

This work is supported by the Hydrogen Energy R&D Center,one of the 21st century frontier R&D programs, founded by theMinistry of Science and Technology of Korea.

References

[1] M. Kemellm, M. Ritala, M. Keskela, Crit. Rev. Solid State Mater. Sci. 30 (2005) 1.[2] P. Jackson, R. Wurz, U. Rau, J. Mattheis, M. Kurth, T. Schlotzer, G. Bilger, J.H.

Werner, Prog. Photovolt. Res. Appl. 15 (2007) 507.[3] M.A. Contreras, M.J. Romero, R. Noufi, Thin Solid Films 511–512 (2006) 51.[4] I. Repins, M.A. Contreras, B. Egaas, C. DeHart, J. Schart, C.L. Rerkins, B. To, R.

Noufi, Prog. Photovolt. Res. Appl. 16 (2008) 235.[5] M. Kaelin, D. Rudmann, F. Kurdesau, H. Zogg, T. Meyer, A.N. Tiwari, Thin Solid

Films 480–481 (2005) 486.[6] C. Eberspacher, C. Fredric, K. Pauls, J. Serra, Thin Solid Films 387 (2001) 18.[7] V.K. Kapur, A. Bansal, P. Le, O.I. Asensio, Thin Solid Films 431–432 (2003) 53.[8] M. Kaelin, D. Rudmann, A.N. Tiwari, Sol. Energy 77 (2004) 749.[9] S.J. Ahn, C.W. Kim, J.H. Yun, J.C. Lee, K.H. Yoon, Sol. Energy Mater. Sol. Cells 91

(2007) 1836.[10] S.J. Ahn, K.H. Kim, Y.G. Chun, K.H. Yoon, Thin Solid Films 515 (2007) 4036.[11] M. Ganchev, J. Kois, M. Kaelin, S. Bereznev, E. Tzvetkova, O. Volobujeva, N.

Stratieva, A. Tiwari, Thin Solid Films 511–512 (2006) 325.[12] S.L. Castro, S.G. Bailey, R.P. Raffaelle, K.K. Banger, A.F. Hepp, Chem. Mater. 15

(2003) 3142.[13] R.N. Bhattacharya, J.F. Hiltner, W. Batchelor, M.A. Contreras, R.N. Noufi, J.R.

Sites, Thin Solid Films 361–362 (2000) 396.[14] B. Blin, F. Fievet, D. Beaupere, M. Figlarz, New J. Chem. 13 (1989) 67.[15] J. Sun, Y. Jing, Y. Jia, M. Tillard, C. Belin, Mater. Lett. 59 (2005) 3933.