Thin Solid Films 333-34-l (1999) 130-133
Transparent conductive ZnO:Al films by reactive co-sputtering from separate metallic Zn and Al targets F. Fenske”. W. Fuhsa, E. Nebauerb. A. Schijpke”. B. Selle”**. I. Sieber”
Abstract ZnO:Al films were depositedby reactive d.c. magnetron co-spurterin g from two separate metallic rargets of Zn and Al in an Ar/O? atmosphere. A first studyof theAl incorporation.of the film morphology,of the electricalandopticalpropertie:,wascarriedout considering rhemostrelevantdepositionparameters. the dissipation poua P,, of the Al sourceand the oxygen flow ratrAOz). The Al concentration in the iilms was found to increase with increasing P,, and to decrease nith increasingf(02). The films e.xhibirrd a textured columnar structure \+irhina well defined range of P,\, and,/102) where the optimum for high optical tranbm&ionandlow resistivitywas also obberved. a 1999
ElsevierScienceS.A. All rightsreserved.
resistivity and transmission)on P?,, and f(O:) at a fixed dischargepower Pzn of the Zn source.
1. Introduction Highly conductive ZnO:Al films are widely used in photovoltaic devices as transparent conductors (TCO) or as wide gap emitters in a heterojunction [l]. Al atoms on regular Zn sitesand/or interstitial Al atoms form donors in ZnO and thus. besidesof native donors (oxygen vacancies and/or Zn interstitials). contribute to the conductivity [3_.3]. A standarddeposition technique is reactive sputtering from metallic Zn/Al composite targets in an Ar/O? atmosphere [ 1.3.11.A shortcoming of this method is that the Al concentration in the film is fixed by the target composition. An auxiliary Al sourcewas usedin the r.f. sputtering technique to vary the Al doping continuously [j]. In this paper. we report on the reactive d.c. co-sputtering of ZnO:Al films from two separatemetallic targets of Zn and Al. This approach allo~vsus to change the Al concentration in a continuous manlier by adjusting the dissipation power PAI of the Al source. Becausein reactively sputtered ZnO:Al films the donor efficiency of Al was found to be strongly correlated with the oxygen partial pressure[ 1.3.11 we have varied the oxygen flow ratefl0:) asan additional parameter.We present resultsof a first study on the dependence of some film properties (Al content. microstructure.
* CorresPondq
author. Tel.:
+ 19-30-67053-3
67053-333.
E-nicti/ nddrrss:
[email protected] tB. Selle)
13: fax:
- 19-30-
2. Experiment The two magnetron sputter sources (50 mm target diameter) were mounted at confocal positions with a substrate/target distance of 150 mm. The substrateswere prepared by a conventional wet cleaning procedure with an HF dip immediately before loading into the deposition chamber. During deposition they were not intentionally heated, and no bias was applied. A high pumping speed was used to prevent instabilities of the deposition regime. The Zn sourcedissipationpower. the total pressureand the deposition time were kept constant at 100W. 6 X IO-’ mbar and 500 s. respectively. The Al target power PiI and the oxygen flow rate f102) were varied from 0 to 300 W and from 0 to 36 hccm. respectively. The total Al concentration in the films was measuredand cross-checkedby AES. EDX and RBS analysis. Details are describedelsewhere[6]. In this study. the RBS data are used becausethih method is standard-free.The structural characterisation of the films was carried out by scanningelectron microscopy (SEM) and by glancing incidenceX-ray diffraction (XRD). The film thicknesses(of the order of 200 nm) were determined from the SEM fracture cross-section views. The film resistivity (four-point probe) andthe optical transmissionwere measuredat room temperature.
00-10-6090/99/S - see front matter cc 1999 El
principal component analysis \vas applied to the AES spectra [6]. This Trocedure disclosed contribctions of metallic Zn to be present in the films gro\vn at lo~vj?Oz) and containing hisher amonnts of AI. In addition. films of high A1 concentr.ation. bet deposited a’: high,,fl,Oll, eshibited typical features of Xi:O- type bonding.
(4..
0
20:
!-
10
50 100 150 Al target power (W) 1
1
I
1
I
200
I
Al target power (W)
12
14
16
18
I
-, 0)
20
Oxygen flow (seem)
The growth morphology of the ZnO:Ai f.lms lvac inspected by SEhl cross-section images of cieaved samples. As sho\sn schematically in Fig. 2 three basic types of rilm mor?hoIogy v:ere obser\red: grar;ular i‘or lo~.Jf‘iO~;>and high PA,:(the left and upper part oi the parameter field,, coiumnar for mediumJ\O:i and not too high Pii (tile central part of the saramet.:r field) and flaky for th? highestj,Ol, : the right part of the parameter kid). According IO the results of the precedin_r section (see insc1.t in Fig. ia‘i the existence tie% for the columnar structure is divided into an AI doped re,uion and a region \<,here the -Al conte:lt is beiou, the detection Ikit of RBS i-O.15 at.5 I, Different kinds of transitionai morphologies appear in the \kirAt!, of the upper boundar! of the columnar range field. .As r.n example :he crosssectional vie\vs of five tiIms are displayed in Fig. 3. These films refer to the l&t qpzr bo:mdar): 02 the columnar range and are marked in Fig. 2 by
3. Results 3.1. Fi!nr composiiiorz The total Al concentration C,+: in the films is shown in Fig. 1 as a function of PA, and of f(O& respccti\.el!,. X wilting fact is that a detectable a.motm: of AI is on!’ observed if PA,, has passed a threshold \,alue. Above this threshold a !inear increase of Cg, with increasing P,l;l is found. The threshold moves to higher vaIues with increasing oxygen fiow rates (see insert in Fig. 1a). As shoivn in Fig. 1b the Al content decreases lviih increasing JO:). Since the discharge voltage of the Xl :arger decreases simultaneously, this effect is attributed to an increasing osidation of the XI target surface (target poisoning) ; 117]. AES depth proiiles have shown that the X1 zooms were dikbuted homogeneously in ali of the in\:estigated films. They did not rel.eal any significant co~rrIation to chrqes of the gro\vth mol~holog~. Since the Auger line shapes in\.oI\.e information on the bonding state of the detected elements a
s, I
Granular
I
ii
.....
132
F. Fetuhe
Fig. 3. SEM cross-sectional view Iilk sre Indicated in Fi:. 2.
of ZnO:.LU
et ni. / Thin So!itl Filters 343-344
films \\ith granular
(3). columnx
indicating the well established c-axis orientation of columnar textured ZnO:.Al films [-l,8]. In the lower curve. referring to flm (d) in Fig. 3 (C,tI = 7.5 at.%). the (002) peak disappeared. and the (100) reflex became most intensive. The peak positions of all the remaining reflexes were shifted to higher angle values. This was also observed by Park et al. [8] \vho esplained the effect by a lattice constant decrease due to the Al incorporation. The XRD spectra of the granular films showed patterns of weak intensities only and. in accordance with the AES results. indicated contributions of metallic Zn for films grown at lowerf(O?).
Fig. 5 shows the film resistivity
and the optical transmis-
f 1999)
me) and iniermediate
130-133
(b-d)
morphologirs.
The deposition
parameters
of these
sion as a function ofJO2) for a sample series deposited at a fixed P.,, = 100 W. The resistivity curve exhibits a minimum of about lo-’ f2 cm atf(O?) = 12 seem and increases with higher.flO;). This behaviour is well known from the reactive sputtering of Zn0:.41 using compound targets Zn/ Al [3.1] and was attributed to an inadequate incorporation of Al at higher f(O1). With increasing fro,) the optical transmission increased steeply and forf(O?) L 12 seem attained values in excess of 90R. The optimum for use as a TCO is in the vicinity of f(O$ = 13 seem. The resistivity can be lowered by a factor of three by a post-deposition annealing. A further improvement offilm properties by optimising their deposition \vith respect to the total pressure. substrate temperature and Zn source dissipation power will be addressed in the future.
F. Fcnske
30
50
40
'0
60 idegj
70
Fig. 4. XRD spectra of a ZnO:Al fi!m with srructure and CAI = 1.3 ar.% (upper curve) and ate structure and Car = 7.5 zt.%, film (d) of ordinate is a square-roof scale in order to show more clearly.
et al. /Thin
80
Solid
90
v&I deveioped columnar of a film with an intermediFig. 3 (iouer cun!e). The the peaks of Iower intensity
4. Discussion Although reactive co-sputtering from metallic targets is a complex and non-linear process [7] simple and monotonous reIationships between the total amount of Al, CA,, in our films and the two investigated deposition parameters were found. The pumping speed applied in our sputtering system seemed to be sufficiently high that the hysteresis phenomena [7] did not occur. The transition from metallic to compound mode sputtering at rhe Al target is we11 reflected in the shift of the power threshold (see Fig. la) and in the decrease of
Fib
343-344
(1999)
130-133
133
C,+, with increasing fi0,). The growth morphology of the films as well as the doping efficiency of Al depend on a complex interplay of more processing parameters than those considered in the present study. In particular. the effect of the different reactivity of Zn and Al to the reaction gas oxygen needs further investigation. According to our results from AES, SEM and XRD analysis, we conclude that the columnar texture, which is a typical feature of the unperturbed ZnO thin film growth, begins to break down if other phases than ZnO (such as metallic Zn or oxidized Alj are present in the films. This occurs iffl0:) is too low or too high and if P,: and, consequently, Cd exceed critical values (see Fig. 2). Here, the breakdown of the columnar structure appears to be related with a lattice distortion induced by the Al incorporation (see Fig. 4).
5. Conclusions ZnO:Al films with a well controlled total ti concentration were deposited successfully by reactive d.c. magnetron co-sputtering from tlvo separate metallic targets of Zn and Al. The influence of the most important deposition parameters, the power PAi of the Al source and the oxygen flow rate JO:), on some fundamental film properties was studied. The textured columnar film growth appears in a distinct range of both these quantities. Optimum conditions for films having low resistivity and high oprical transmissions are found within this range field.
Acknowledgements This work was supported in part by the Bundesministeium fiir Forschung und Bildung (contracr 0329773).
References [II
0
2
4
6
8
10 12 14 16 18 Oxy&cn now [Scm)
20
22
24
26
28
Fig. 5. Film resistiviiy and optical transmission (measured at h = 800 nm) as a function of the oxygen fiowA0~ for depositions SI fixed target powers P, = 100 W and PA, = IO0 W.
A.E. Delahoy, 51. Chrmy, Mater. Res. Sot. Symp, Proc. 426 (1996) 467. PI T. ?&nami,H. Sato, H. Nanto. S. T&da, Jpn. J. Appl. Phys. 24 i 1985) LX1. [31 G.L, Harding, B. Window, E.C. Horrigan. Solar Energy Mater. 22 (1991) 69. PI K. Ellmer, F. KudelIa, R. Itientus, R. Schieck, S. Fiechrer. Thin Solid FiIrns 247 (1994) 15. PI K. Tominaga. hf. Kataoka, H. Manabe, T. Ueda, i. Mori, Thin Solid Films 290 (1996) 84. ra A. Schb;pke. F. Fen&e. B. SeUe. I. Sieber. in: I. Olefjord: L. Nyburg, D. Briggs, (E&s.), ECASIA ‘97, Wiiey. Chichester. 1997, p, 852. [:I S. Berg, T. Nyberg, H.O. Blom, C. Nender, J. I’ac. Sci. Technol. A 16 (199Sj 1277. @I KC. Park. D.Y. Ma, K.H. Kim. Thin Solid Fiis 305 (1997) 201.