Thin Solid Films 556 (2014) 9–12

Contents lists available at ScienceDirect

Thin Solid Films

j ourna l homepage: www.e lsev ie r .com/ locate / ts f

Epitaxial growth of kesterite Cu2ZnSnS4 on a Si(001) substrate bythermal co-evaporation

Byungha Shin, Yu Zhu, Talia Gershon, Nestor A. Bojarczuk, Supratik GuhaIBM T. J. Watson Research Center, Yorktown Heights, NY 10598, USA

0040-6090/$ – see front matter © 2014 Elsevier B.V. All rihttp://dx.doi.org/10.1016/j.tsf.2013.12.046

a b s t r a c t

a r t i c l e i n f o

Article history:Received 15 November 2013Received in revised form 17 December 2013Accepted 17 December 2013Available online 28 December 2013

Keywords:KesteriteCopper zinc tin sulfideEpitaxyEvaporation

Using thermal co-evaporation we have prepared epitaxial Cu2ZnSnS4 (CZTS) films on Si(001) substrates. A sub-strate temperature as high as 370 °C and proper substrate cleaning (HF-dip followed by thermal desorption ofsurface hydrogens) are found to be necessary for the epitaxial growth. Detailed transmission electronmicroscopymeasurements and X-ray diffraction studies are used to reveal the orientation relation of the CZTS films with theunderlying silicon substrate, and the formation of defects within the CZTS layer.

© 2014 Elsevier B.V. All rights reserved.

In recent years, there has been growing interests in Cu2ZnSnS4(CZTS) based polycrystalline thin film solar cell applications, as anearth abundant alternative to thin film absorbers such as Cu(In, Ga)Se2 (CIGS) and CdTe [1–3]. In the more extensively studied and relatedCIGS absorber, it is has been proposed that the presence of grainboundaries is beneficial in terms of solar cell performance, and itis often attributed to the reduced recombination at grain boundaries[4–6]. A similar claim has been made with regards to the role of grainboundaries in CZTS, however, there has not been enough study todraw solid conclusions [7,8]. The principle shortcoming of CZTS solarcells is an open circuit voltage (Voc) that is lower than expected: theVoc of the 8.4% record efficiency CZTS cell [1] is only ~60% of the thermo-dynamic limit. In general, very little is understood in terms of the ulti-mate performance capabilities of CZTS, given its complicatedpolycrystalline microstructure, composition, and defect structure. Inthe following, in order to isolate out the effects of grain boundaries,we show the initial results of a systematic study of epitaxial CZTSgrown on silicon, which is closely lattice matched to CZTS; the reporteda-axis lattice constant, a of CZTS ranges from 0.5426 to 0.5435 nm(c-axis lattice constant, c is twice of a in CZTS) while that of Si is0.5431 nm [9].

There appears to be only one report on the epitaxial growthof CZTS on Si, where Oishi and co-workers [10] deposited CZTSon Si(001) and demonstrated the dominance of c-axis and a-axisgrowth of CZTS, orientations expected for epitaxial overgrowth.In this work, we have carried out extensive transmission elec-tron microscopy (TEM) and x-ray diffraction (XRD) studies on ep-itaxial CZTS films grown on Si(001), and carry out a detailedstructural analysis of the epitaxial registry, defects and domainformation.

ghts reserved.

Epitaxial CZTS films were grown by molecular beam epitaxy. De-tails of the deposition system can be found in a previously publishedpaper [1]. Substrates used were Si(001) substrates (p++ withresistivity b 0.005 Ω-cm) and the deposition temperature was340 °C–450 °C. The substrate was cleaned by a standard RCA procedurefollowed by a 60 seconddip in 10 vol%HF in order to create a hydropho-bic surface. Prior to growth of the CZTS, the Si substratewas then heatedto ~650 °C in vacuum for 10 minutes in order to provide a clean singlecrystal surface for the epitaxial growth of CZTS. To examine the rele-vance of the final surface cleaning step, use of a HCl dip instead of HFwas also studied. Transmission Electron Microscopy (TEM) sampleswere prepared using a FEI Helios 400S dual-beam FIB, with ex-situ liftoff on a gold grid with carbon supporting film. TEM analysis wasperformed on a JEOL JEM-3000F microscope operated at 300 kV. XRDmeasurements were takenwith a CuKα X-ray source in θ–2θ geometry.

Fig. 1 presents θ–2θ XRD scans of CZTS films grown on Si using theabove technique between 340 °C–450 °C, compared to that of a conven-tional polycrystalline film grown on molybdenum coated soda limeglass (Mo-SLG) at 150 °C, followed by a recrystallization and graingrowth inducing anneal at 570 °C. We have previously used this tech-nique to prepare CZTS solar cells with efficiencies of up to 8.4% [1].The polycrystalline film exhibits peaks corresponding to many differentorientations with the (112) peak being dominant [Fig. 1(a)]. When theSi wafer is HCl-treated prior to growth at 370 °C, the X-ray pattern re-mains that of a polycrystalline material (Fig. 1b): the strong peaksmarked * are due to the (002) and (004) reflections from the Sisubstrate. On the other hand, when the Si wafer is HF-treated, the addi-tional peaks such as (112), (220), and (224) due to the polycrystallinenature of the film disappear as the growth temperature is increasedfrom 340 °C to 450 °C [Fig. 1(c), (d) and (e)], with the emergence of a

Mo

(110

)

25 30 35 40 45 50 55 60 65 70 75

(a)

2 Theta (o)

(b)

(d)

(e)

(004

)/(2

00)

(220

)

(312

)(2

24)

(112

)

(008

)/(4

00)

**

*

Mo

(220

)

(220

)

(312

)(2

24)

(006

)(0

06)

(c)

(224

)

(220

)*(1

12)

(112

)(1

12)

(112

)

Fig. 1. θ–2θ XRDmeasurements (in log scale) of (a) poly-crystalline CZTS grown onMo at150 °C followed by post-deposition annealing at 570 °C, (b) CZTS grown on HCl-treatedSi(001) at 370 °C, (c) CZTS grown on HF-treated Si(001) at 340 °C, (d) CZTS grown onHF-treated Si(001) at 370 °C, and (e) CZTS grown on HF-treated Si(001) at 340 °C,(f) CZTS grown on HF-treated Si(001) at 450 °C. Indexes, (hkl) are from CZTS except Mo(111) and Mo (220) in Fig. (a). Both CZTS and Si contribute to those reflections with anasteroid mark (*); c-axis oriented CZTS (004), a-axis oriented CZTS (200) and Si (002) at~33°, and c-axis oriented CZTS (008), a-axis oriented CZTS (400), and Si (004) at ~69.2°.

Si (001) substrate

Epitaxial CZTS

[110]

[112]

2 nm

Fig. 2. High resolution TEM image taken from 370 °C epi-CZTS illustrating epitaxialrelationship between CZTS and Si(001).

10 B. Shin et al. / Thin Solid Films 556 (2014) 9–12

CZTS (006) peak in addition to the peaks marked * which are contribu-tions from the c-axis oriented (004), (008) reflections and a-axis orient-ed (200), (400) reflections of CZTS, as expected in the case of epitaxy, aswell as from the (002) and (004) reflections of Si. At 450 °C, the non-epitaxially related CZTS grains such as those aligned along [112] or[220] direction is negligible. As has been pointed out by Oishi, thereare 3 epitaxial variants for CZTS on Si, one c-axis and two a-axis orienta-tions—these give rise to the (004) and (200) peaks, respectively. Sinceboth peaks appear at the same 2θ position at 33.1°, their relative contri-butions cannot be distinguished. The peak at 2θ = 50.6° corresponds tothe lattice spacing of either (300) or (006) planes but (300) reflection iskinematically prohibited from the kesterite CZTS, and, therefore, onlythe c-axis oriented CZTS contributes to the 50.6° peak. The (006) peakbecomes stronger with the increasing substrate temperature from370 °C to 450 °C, suggesting that c-axis orientation is more preferredover a-axis at a higher growth temperature or CZTS crystallinity in gen-eral improves with the temperature. The results indicate that the CZTSdeposited at 340 °C still contains observable polycrystalline grains,and a growth temperature of 370 °C or higher is required for epitaxialgrowth.We further observe that when the starting Si surface was treat-edwith anHCl dip insteadof aHF dip, the Si surfacewas hydrophilic andcoveredwith a thin oxide layer [11]. The presence of this oxide layer en-sures the growth of a polycrystallinefilm, and theXRD reveals similarityto the polycrystalline pattern; compare Fig. 1(b) to (a). It is important tonote that the XRD θ–2θ scans of Fig. 1(d) and (e), while consistent with

that expected from an epitaxial CZTS film on Si(100), is not a sufficientindicator of an epitaxial relationship, since the scans cannot reveal anyin-plane rotational mismatch that might exist. We therefore proceededwith the examination of the films by TEM.

A high-resolution TEM image taken at the CZTS/Si(001) interfacefrom the 370 °C epi-CZTS is shown in Fig. 2. An epitaxial registrybetween the CZTS and the Si(001) is obvious. Selected area diffractionpatterns (SADPs) were also collected to gather structural informationover an area (the aperture size used is ~10 μm) much larger thanavailable from the high-resolution imaging.

There are three possible epitaxial configurations of CZTS on a Si(001)substrate—growth along c-axis where the CZTS c-axis is oriented along[001] of the Si substrate and growth along a-axis where the CZTS c-axisis oriented either along [100] or [010]. The expected diffraction patternfor epitaxial Si(001)/CZTS, with the incident electron beam parallel tothe Si [110] zone axis (ZA), are shown in Fig. 3. Two different sets ofindexing are given for two possible epitaxial orientations of a-axisgrowth in Fig. 3(c), however, the resulting patterns are identical inboth cases. The SADPs of Si and a-axis oriented epitaxial CZTS maponto one other while additional spots such as (002) and (220) arepresent in the case of c-axis oriented epitaxial CZTS. Fig. 4 presents ex-perimental SADPs taken from the 370 °C epi-CZTS sample. ComparingFig. 4(a) from the Si substrate and Fig. 4(b) from the CZTS layer, wenote the alignment of the CZTS diffraction pattern with that of the Sisubstrate, confirming that the epitaxial registry seen from the high-resolution TEM image (Fig. 2) extends over a larger area. The presenceof (002) and (220) diffraction spots in Fig. 4(b) indicates the pattern isfrom c-axis oriented epitaxial CZTS. The CZTS diffraction spots are la-beled in Fig. 4(c), which presents the same SADP as in Fig. 4(b) but ona magnified scale. Fig. 4(b) and (c) show additional spots other thanthose expected from a defect-free epitaxial CZTS film: these are labeledwith blue and orange circles in Fig. 4(c). The spots placed in red circlesare those expected from a kesterite CZTS with no structural disorder;the spots placed in blue circles arise from twinning on (112) planes[12]. The remaining spots encircled in orange can be accounted for ifdouble diffraction is considered. The presence of a high density ofmicrotwins on (112) inferred from the streaking in the SADP and alsoconfirmed by bright field images (not shown here).

We now turn to the results from the 450 °C epi-CZTS sample. At thegrowth temperature of 450 °C, strong three dimensional growth(Volmer–Weber) sets in so that the Si surface is not completely coveredby the CZTS even for micron thick films; the areal coverage of CZTS isapproximately 90%. Fig. 5(a) consists of three bright field images fromadjacent areas stitched together to show the epitaxial CZTS layer alongan extended length of ~5 microns. The epitaxial CZTS layer consists ofdomains across which a subtle change in contrast is noticeable; thered arrow in the figure marks a boundary between two domains, for

(002)

(111)(111)

(222)

(222)

(a) Si substrate

(220)

(222)

(111)(111)

(002)

(220)

(b) Epitaxial CZTS along c-axis (c) Epitaxial CZTS along a-axis

(004)

(112)(112)

(224)

(224)(220)

(112)(112)

(004)

(220)

(114)

(114)

(222)(222) (222)(224)

(110)(114)

(002)(002)

(114)

(222) (222) (224)

(224)/(224)(024)/(024)

(112)/(112)

(224)/(224)

(112)/(112)

(000) (000) (200)/(200)(000)/(000)(200)/(200)

(112)/(112) (112)/(112)

(224)/(224)(024)/(024)(224)/(224)

(110)

Fig. 3. Expected diffraction pattern of (a) Si substrate and epitaxially grown CZTS films (b) along c-axis, and (c) along a-axis, with the incident electron beamparallel to the Si [110] ZA. Twosets of index are given in 3(c) since two different epitaxial configurations are possible for a-axis growth.

(a) from Si substrate (b) from CZTS (c) from CZTS

(112)

(112)

(004)

(004)

(112)

(112)

( )

(002)(002)

(110)

(110)

Fig. 4. SADPs from 370 °C epi-CZTS sample: (a) from Si substrate, (b) from CZTS, (c) magnified pattern shown in (b). Circles drawn in (a) and (b) are on the exact same spots illustratingepitaxial relationship between Si and CZTS. In (c), red, blue, and orange circles represent regular CZTS reflections, twinning on (112) planes, and double diffractions, respectively.

11B. Shin et al. / Thin Solid Films 556 (2014) 9–12

instance. Selected area diffraction patterns from two different domainswithin the CZTS as well as from the Si substrate were taken as indicatedby the dotted circles shown in Fig. 5(a). As in the case of the 370 °C epi-

1 mµ

SADSAD 2

SAD 1

(002)

(111)

(220)(220)

(111)

(111) (111)

(002)

(004

(112)

(220)

(112)

(00

(002

(00

(110)

(c) SAD 2 from C(b) SAD 1 from Si

Fig. 5. (a) Bright-field TEM image of 450 °C epi-CZTS sample showing a typical grain that is largDotted circles indicates areas where SADPs shown in (b)–(d) are taken. (b) SAD1P from Si sub

CZTS sample, the periodicity of the Si substrate is maintained in theCZTS; compare the position of red spots in Fig. 5(b)–(d). Other thanthose in red circles, spots (in blue circles) arising from c-axis oriented

Si substrate

CZTS

3

(004)

(112)

(220) (220)

(112)

(112) (112)

(004)

(002)

(002)

(110)(110)

)

(220)

(112)

(112)

4)

)

2)

(110)

(d) SAD 3from CZTS ZTS

(a)

er than 5 μm (three images taken from areas next to each others were stitched together).strate, (b) and (c) from two different areas of CZTS.

12 B. Shin et al. / Thin Solid Films 556 (2014) 9–12

CZTS are also evident in Fig. 5(c) and (d), indicating that both domainsare c-axis oriented. It is likely that these domains are anti-phasedomains where the vertical ordering of metal elements in CZTS alongc-axis is reversed, giving rise to the domain contrast. An interesting ob-servation can bemade by comparing the CZTS SADPs of 370 °C epi-CZTSand 450 °C epi-CZTS—Fig. 4(c) vs. Fig. 5(c) or (d). Spots arising fromtwinning on the (112) planes that are seen from 370 °C epi-CZTS areabsent in 450 °C epi-CZTS, suggesting that the structural disorder man-ifested by a high density of twins can be significantly reduced by using ahigher growth temperature.

In conclusion,we have demonstrated the growth of epitaxialfilms ofCZTS on Si(001) substrates. Our study reveals the following: (i) a properSi surface treatment and a growth temperature of 370 °C or higher isnecessary for epitaxial growth of CZTS on Si; (ii) growth is three dimen-sional inmorphologywith poorwetting of the silicon substrate; (iii) theCZTS film consist of domains and these domains have epitaxial relation-ship to the Si substrate–predominantly, c-axis oriented epitaxy out ofthree possible orientations (c-axis and two a-axis orientations)–as con-firmed by SADPs in comparison to that from the Si substrate; (iv) a highdensity of twins on (112) planes exist in some grains of 370 °C epi-CZTSwhile this issue can be greatly alleviated by employing a higher temper-ature (450 °C) although this presents an increased wetting problem.Two future research directions ought to be pursued. First, improving

the quality of epitaxial films including a better wetting is planned.Second, building a solar device from an area entirely comprised of epi-taxial CZTS (i.e., free of grain boundaries) will be undertaken, whichshould unambiguously answer the role of grain boundaries in CZTSsolar cells.

References

[1] B. Shin, O. Gunawan, Y. Zhu, N.A. Bojarczuk, S.J. Chey, S. Guha, Prog. Photovolt. Res.Appl. 21 (2013) 72.

[2] T.K. Todorov, J. Tang, S. Bag, O. Gunawan, T. Gokmen, Y. Zhu, D.B. Mitzi, Adv. EnergyMater. 3 (2013) 34.

[3] B. Shin, Y. Zhu, N.A. Bojarczuk, S.J. Chey, S. Guha, Appl. Phys. Lett. 101 (2012)053903.

[4] M. Hafemeister, S. Siebentritt, J. Albert, M.C. Lux-Steiner, S. Sadewasser, Phys. Rev.Lett. 104 (2010) 196602.

[5] D. Abou-Ras, S.S. Schmidt, O. Cojocaru-Miredin, Adv. Energy Mater. 2 (2012) 992.[6] U. Rau, K. Taretto, S. Siebentritt, Appl. Phys. A 96 (2009) 221.[7] J.B. Li, V. Chawla, B.M. Clemens, Adv. Mater. 24 (2012) 720.[8] A.R. Jeong, W. Jo, S. Jung, J. Gwak, J.H. Yun, Appl. Phys. Lett. 99 (2011) 082103.[9] D.B. Mitzi, O. Gunawan, T.K. Todorov, K. Wang, S. Guha, Sol. Energy Mater. Sol. Cells

95 (2011) 1421.[10] K. Oishi, G. Saito, K. Ebina, M. Nagahashi, K. Jimbo, W.S. Maw, H. Katagiri, M.

Yamazaki, H. Araki, A. Takeuchi, Thin Solid Films 517 (2008) 1449.[11] K. Prabhakaran, Y. Kobayashi, T. Ogino, Surf. Sci. 290 (1993) 239.[12] X.J. Wu, F.H. Li, H. Hashimoto, Philos. Mag. B 63 (1991) 931.