JOURNAL OF APPLIED PHYSICS 101, 014910 共2007兲

Investigation on the electrical properties and inhomogeneous distribution of ZnO:Al thin films prepared by dc magnetron sputtering at low deposition temperature X. B. Zhang, Z. L. Pei, J. Gong, and C. Suna兲 Division of Surface Engineering of Materials, Institute of Metal Research, Chinese Academy of Sciences, Shenyang 110016, People’s Republic of China

共Received 24 June 2006; accepted 26 October 2006; published online 11 January 2007兲 A study of the electrical properties and spatial distribution of the ZnO:Al 共AZO兲 thin films prepared by dc magnetron sputtering at low deposition temperature was presented, with emphasis on the origin of the resistivity inhomogeneity across the substrate. Various growth conditions were obtained by manipulating the growth temperature TS, total pressure PT, and ion-to-neutral ratio Ji / Jn. The plasma characteristics such as radial ion density and floating/plasma potential distribution over the substrate were measured by Langmuir probe, while the flux and energy distribution of energetic species were estimated through Monte Carlo simulations. The crystalline, stress and electrical properties of the films were found to be strongly dependent on TS and Ji / Jn. Under the low Ji / Jn 共⬍0.3兲 conditions, the TS exerted a remarkable influence on film quality. The films prepared at 90 ° C were highly compressed, exhibiting poor electrical properties and significant spatial distribution. High quality films with low stress and resistivity were produced at higher TS 共200 ° C兲. Similarly, at lower TS 共90 ° C兲, higher Ji / Jn 共⬃2兲 dramatically improved the film resistivity as well as its lateral distribution. Moreover, it indicated that the role of ion bombardment is dependent on the mechanism of dissipation of incident species. Ion bombardment is beneficial to the film growth if the energy of incident species Ei is below the penetration threshold Epet 共⬃33 eV for ZnO兲; on the other hand, the energy subimplant mechanism would work, and the bombardment degrades the film quality when Ei is over the Epet. The energetic bombardment of negative oxygen ions rather than the positives dominated the resistivity distribution of AZO films, while the nonuniform distribution of active oxygen played a secondary role which was otherwise more notable under conditions of lower TS and Ji / Jn. © 2007 American Institute of Physics. 关DOI: 10.1063/1.2407265兴 I. INTRODUCTION

Zinc oxide 共ZnO兲 film, especially aluminum-doped ZnO 共AZO兲, has emerged as one of the most important transparent conducting oxide 共TCO兲 thin films for a variety of promising applications such as flat panel displays 共FPDs兲, solar cells, and electrochromic glazing in recent years.1,2 Nowadays, the demands for ever lower substrate temperature TS are rapidly growing in device and product manufacturing such as polymer based flexible FPD devices which generally require TS below 100 ° C.3,4 However, the low TS, which is often lower than 0.2–0.3 of the film melting point Tm, is not favorable for depositing high quality films.5 It is easy to fabricate AZO films with a resistivity below 5 ⫻ 10−4 ⍀ cm at TS above 150– 200 ° C whatever preparation methods are taken. However, it is difficult to obtain AZO films with comparable properties at TS below 100 ° C, especially for the large area deposition.6,7 Many researchers have tried to produce lowresistivity AZO films at low TS by using a variety of deposition methods, such as sputtering,8 evaporation,9 and pulsed laser deposition.10 Among these methods, sputtering has been a robust and prevailing preparation method due to its ability to produce superior film quality, high deposition rate, and good uniformity for a large substrate. a兲

Author to whom correspondence should be addressed; electronic mail: [email protected]

0021-8979/2007/101共1兲/014910/7/$23.00

Previous studies have suggested that the electrical properties of AZO thin films prepared by magnetron sputtering at low TS are inferior to that grown at elevated TS and strongly affected by the deposition conditions.11 Further, it always exhibits a resistivity inhomogeneity across the substrate, especially for dc magnetron sputtering.12–15 However, there still exist considerable controversies and ambiguity in the existing literature regarding the origin of resistivity distributions, which were mainly classified into two distinct mechanisms: the bombardment with energetic species,12–14 and the nonuniformity of active oxygen reaching the substrate,11,15,16 respectively. Tominaga et al.12,13 observed maxima of the resistance opposite to the erosion track, which were attributed to the bombardment on the growing film by the energetic negative oxygen ions originated from the cathode sheath. Herrmann et al.14 also found the similar phenomena for the AZO films prepared by dc dual magnetron sputtering. They considered that the different distributions of positive ion bombardment such as Ar+, Zn+, and O+ were also responsible for the resistivity distribution by mass spectrometry measurement. However, on the other hand, Minami et al.,15,16 proposed an opposite viewpoint that the inhomogeneity resulted from the activity and the spatial distribution of oxygen reaching the substrate instead of the energies bombardment. They observed that the resistivity distribution of AZO films prepared at lower TS by dc sputtering was im-

101, 014910-1

© 2007 American Institute of Physics

Downloaded 13 Jan 2007 to 210.72.130.156. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

014910-2

J. Appl. Phys. 101, 014910 共2007兲

Zhang et al. TABLE I. Deposition parameters.

FIG. 1. 共Color online兲 Schematic diagram of planar magnetron sputtering system.

proved as increasing the cathode voltage Vdc. It is difficult to explain this finding using energetic oxygen bombardment which was enhanced with the increase of Vdc.15 Note that Cebulla et al.17 have reported the opposite trend using the rf sputtering or simultaneous rf+ dc excitation. The lower specific resistance of AZO films facing the erosion track was observed and roughly attributed to the important role of low energy ion bombardment 共mainly Ar+兲 and higher plasma density in front of the substrate opposite to the plasma torus. This study is aimed to clarify the origin of resistivity distribution across the substrate by investigating the influence of deposition parameters on the electrical property and its lateral distribution of AZO thin films prepared by dc magnetron sputtering at low TS. In this work, AZO thin films were grown by dc magnetron sputtering with an external coil, and analyzed with respect to their crystalline structure, chemical composition, and electrical properties. The plasma characteristics over the substrate were measured by Langmuir probe, while the flux and energy distribution of energetic species were estimated through Monte Carlo simulations. The effects of TS, ion-toneutral Ji / Jn, total pressure PT, and oxygen partial pressure PO2 on the electrical properties of AZO films were investigated as well as their spatial distribution. Finally, based on the experiment and simulation, the origin of lateral resistivity distribution of AZO films was discussed in detail. II. EXPERIMENTAL DETAILS A. Film deposition

AZO thin films were grown in a dc circular magnetron sputtering system, as illustrated in Fig. 1. A coaxial solenoid coil positioned around the magnetron source was utilized to create an external magnetic field Bext for varying the nearsubstrate plasma density. The glass substrate was placed on the substrate holder with a resistive heater. Details of the deposition conditions are summarized in Table I. During the deposition, the oxygen flow was regulated by a mass flow controller, while the sputtering argon gas was directly introduced into the vicinity of the magnetron source, which was monitored by a capacitance manometer. The substrate temperature was measured by a thermocouple pressed against the substrate and controlled with an accuracy of 10 ° C. The sputtering power was kept at about 50 W with a typical Vdc of 320– 330 V.

Target to substrate distance Target diameter/thickness Target purity Magnetic flux parallel to the target surface Base pressure Working pressure Oxygen partial pressure Substrate potential Substrate temperature Substrate Deposition rate

70 mm 76/ 6 mm 4 N 共Zn: Al2 wt % 兲 550 G ⬍3 ⫻ 10−3 Pa 0.4/ 0.8 Pa 30– 60 mPa 共typically兲 Floating 90/ 200 ° C Soda-lime glass ⬃30 nm/ min

B. Plasma and film characterization

Prior to the deposition, plasma characteristics were measured by Langmuir flat probe under the same growth conditions, from which the saturation ion density Ji, plasma potential V p, and floating potential V f were determined.18,19 The circular flat probe had a radius of 4 mm and was mounted on the substrate holder. The flux and energy distributions of the energies were estimated by Monte Carlo simulations. The energy distribution and yield of the species, such as sputtered Zn and Al atoms and backreflected Ar and O, emitted from the target were simulated by the transport of ions in matter SRIM 共Ref. 20兲 code. The transport of the species from target to substrate was calculated by SPATS code.21 Positions of ejected particles on the target surface were generated according to the measured target erosion track profile. Lennard-Jones potential was used to calculate the energy-dependent scattering cross sections, and a sampling number of 106 atoms for each case. The crystalline and phase structures of the films were determined by x-ray diffractometer 共XRD兲 with Cu K␣ radiation 共␭ = 0.154 056 nm兲. The intrinsic stress in the films was derived from the shift of the 共002兲 peak relative to the peak position of an annealed zinc oxide powder.17 X-ray photoelectron spectroscopy 共XPS兲 was performed for the analysis of the film composition 共Escalab 250, Thermo VG兲. The film thickness d was determined by a profilometer 共Surfcom 480A, Tokyo seimitsu兲 and a fitting of optical transmittance 共FTG software兲.22 The sheet resistance Rsq at room temperature was measured with a four-point probe and the resistivity ␳ of the films was calculated by the formula ␳ = Rsqd. The optical transmittance of the films was measured with an ultraviolet-visible near-infrared spectrophotometer 共Hitachi U-2800兲. III. RESULTS A. Plasma and films thickness distribution

The ion-to-neutral ratio Ji / Jn and the energy difference e 共V p − V f 兲 on the substrate are the important parameters in the ion-assisted film growth. Two near-substrate plasma conditions denoted by low plasma density 共LPD兲 and high plasma density 共HPD兲 were studied in this experiment, which were obtained without applying a Bext or with a Bext = 100 G assisting the outer pole of the magnetron, respectively. The ratios Ji / Jn in dependence on the position over the substrate

Downloaded 13 Jan 2007 to 210.72.130.156. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

014910-3

Zhang et al.

J. Appl. Phys. 101, 014910 共2007兲

FIG. 3. 共Color online兲 Normalized thickness distributions across the substrate for the deposited AZO films at three different conditions 共TS = 90 ° C兲.

= 80 mm兲 was within 40% and the minimum thicknesses were over 300 nm. Thus, the resistivity could be independent of film thickness. Distances of 0, 2, and 4 cm indicate the substrate locations which correspond to the center, the erosion track, and the border of the target, respectively. B. Effects of deposition parameters on the resistivity distribution of AZO films

FIG. 2. 共Color online兲 Lateral distribution of 共a兲 the ratio Ji / Jn, 共b兲 the difference V p – V f for different PT, and plasma density. Sputtering parameters: PO2 = 40 mPa, TS = 90 ° C.

for the different PT and plasma conditions were calculated from the ion current density and the deposition rate, as displayed in Fig. 2共a兲. The ratio Ji / Jn was found to be significantly inhomogeneous 共a factor of 5兲 for the HPD conditions, for which it was much stronger in the center, but varied little at the edges. Nevertheless, the total pressure had only very little effect on Ji / Jn. In addition, the parameters of V f and V p were also measured, which were in the ranges, of −23 to − 11 V and −5 – 2 V, respectively. Both had a similar variation trend over the substrate regardless of the pressure and plasma conditions. The difference V p − V f was always maintained at less than 20 V regardless of the substrate position 关see Fig. 2共b兲兴. The lateral thickness distributions were also measured to study the influence of deposition parameters, as shown in Fig. 3. It can be seen that the thickness was maximal at the center of the substrate and gradually reduced towards its edges under all sputtering conditions. The thickness uniformity was slightly affected by PT and improved to some extent at a higher pressure of 0.8 Pa. Although the inhomogeneity for HPD conditions was significantly severer, the thickness variation of AZO films 共measured diameter

The resistivity of AZO thin films prepared by magnetron sputtering as well as spatial distribution was strongly affected by the deposition conditions such as TS, PT, and Ji / Jn. As demonstrated in Fig. 4, under LPD conditions and at 90 ° C, the resistivity of AZO films prepared at either 0.4 or 0.8 Pa exhibited an obvious nonuniform distribution and, particularly, there existed a big difference of an order of magnitude between that grown in the center 共or erosion兲 and in the border area of the substrate. Expectedly, the increase in either TS or Ji had a significant improvement on the optoelec-

FIG. 4. 共Color online兲 Lateral distribution of the resistivity of AZO films at different conditions. Sputtering parameters: PO2 were 44, 33, 40, and 50 mPa from below to top, respectively.

Downloaded 13 Jan 2007 to 210.72.130.156. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

014910-4

Zhang et al.

FIG. 5. 共Color online兲 X-ray diffraction patterns of AZO films of different sputtering parameters at 0.8 Pa: 共a兲 LPD, and 共b兲 90 ° C, respectively.

tronic quality and its uniformity of the films. The resistivity of AZO films grown on the center and erosion area was almost reduced by an order of magnitude. XRD scans were performed on those samples, as given in Fig. 5. All films exhibited the 共002兲 preferred crystallographic orientation. In addition, the 共002兲 peaks of the films prepared at either 200 ° C or HPD conditions were much stronger and narrower than that prepared at 90 ° C and under LPD conditions, which indicated a notable improvement of crystallinity of the films with the increase of TS and/or Ji. This maybe contributed to the decrease of the resistivity. Moreover, Fig. 6 depicts that the films, especially, that faced the center and erosion of the target, exhibited a strong compressive stress ␴. It is well known that this behavior can be interpreted in terms of the atomic peening mechanism,23 that is, lattice distortions produced from bombardment on the growing film by energetic species. Further, for the films prepared at 90 ° C, whether under the LPD or HPD conditions, the values of compressive stress and full with at half maxi-

FIG. 6. 共Color online兲 Lateral distribution of the stress and FWHM of AZO films shown in Fig. 4.

J. Appl. Phys. 101, 014910 共2007兲

mum 共FWHM兲 were maximal at the center of the substrate and gradually reduced towards its edges. This suggested that the AZO films deposited at the central substrate suffered from a stronger energies bombardment. It should be noted that the stress variations of AZO films were in good agreement with their resistivity distribution 共Fig. 4兲, especially for that with larger stress. This supported the validity of the energy subimplant mechanism for the resistivity inhomogeneity. From Figs. 5 and 6, it can also be observed that increasing either TS or Ji can improve the crystallinity and relaxing the stress of the film prepared at low TS. Consequently, the resistivity was pronouncedly decreased in the same trend, as shown in Fig. 4. This implied that although the high energy bombardment degraded the properties of the films; simultaneously there was a remarkable recovering course which occurred under these conditions during the film growth. The reasonable explanation could be that, at 90 ° C 共TS / Tm ⬃ 0.16兲 and low LPD condition 共Ji / Jn ⬍ 0.3兲, mass transport and defect mobility were low enough to freeze the lattice distortion induced by the energy bombardment in place. The lattice distortion resulted in a considerable stress and severely degraded electrical properties.12 For the deposition either at higher TS = 200 ° C or HPD condition 共Ji / Jn ⬃ 2兲, additional energy caused an increased mobility of the adatoms trapped in subsurface positions of the growing layer, improved the lattice perfection, then considerably released the internal stress and reduced the resistivity. Effect of PO2 on the resistivity distribution was also investigated for determining whether the active oxygen nonuniformity dominated the electrical distribution. Since the distribution of active oxygen across the substrate is inhomogeneous as a result of the confined magnetron plasma, a location dependence of the minimum resistivity ␳min on PO2 would be observed, supposing a ␳min corresponds to a certain PO2 and the active oxygen mechanism is predominant. Figure 7 exhibits the dependence ␳ of different substrate positions on PO2 at 0.8 Pa under three typical conditions: 共i兲 LPD, 200 ° C; 共ii兲 HPD, 90 ° C; and 共iii兲 LPD, 90 ° C. It is clear that under either high TS or HPD conditions, there was no distinct location dependence of ␳min on PO2 关Figs. 7共a兲 and 7共b兲兴. It suggested that the resistivity was not significantly sensitive to the active oxygen distribution, thus the active oxygen mechanism was minor under these conditions. However, in the case 共iii兲, the resistivity of AZO films at the border of the substrate always kept minimum over entire substrate independent of PO2, and their values were almost lower by an order of magnitude compared with that of the center and erosion region, as shown in Fig. 7共c兲. It is noted that the locations of local minima which were separately named after center small 共CS兲 and center big 共CB兲 varied with PO2, if only considering the films of the center and erosion area of the substrate. Their XRD patterns showed no notable difference in two cases, as given in Fig. 8. This can be understood from the manner in which oxygen is incorporated into the growing film. As TS increased, the adatoms mobility and the activity of oxygen reacting to the zinc atom increased, thus the sensitivity to the active oxygen inhomo-

Downloaded 13 Jan 2007 to 210.72.130.156. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

014910-5

J. Appl. Phys. 101, 014910 共2007兲

Zhang et al.

FIG. 8. 共Color online兲 X-ray diffraction patterns of AZO films at different PO2 shown in Fig. 7共c兲.

parency for the HPD conditions at fixed TS. Then, the source of reactive atoms also results in more efficient oxygen incorporation and lower sensitivity to the nonuniform active oxygen. Therefore, the spatial phenomena in dependence on PO2 should be more obvious under low TS and LPD conditions. This conclusion is also consistent with the experimental results of indium tin oxide 共ITO兲 films by Ichihara et al.24 Therefore, it revealed that although nonuniform active oxygen could not be the main reason, it play a secondary role under the conditions of low TS and low plasma density. In addition, the irregular variation of composition distribution on PO2, displayed in Fig. 9, also confirmed it. IV. DISCUSSION

As suggested above, the energy subimplant mechanism was mainly responsible for the resistivity distribution of AZO films, while the nonuniform distribution of active oxygen played a minor role, more obvious under conditions of lower TS and plasma density. As follows, we will explore the role of ion bombardment, determine the energies responsible for the subimplant mechanism, and further discuss the origin of the resistivity distribution.

FIG. 7. 共Color online兲 Relations between PO2 and ␳ of different substrate positions at 0.8 Pa under three typical conditions: 共a兲 LPD, 200 ° C; 共b兲 HPD, 90 ° C; and 共c兲 LPD, 90 ° C.

geneity was reduced. Similarly, HPD conditions were likely to result in an increased flux of active oxygen to the growing surface.7 This can be deduced from the fact 关see Figs. 7共b兲 and 7共c兲兴 that less PO2 was required to obtain a given trans-

FIG. 9. 共Color online兲 Lateral distribution of the Zn, O, and Al content of AZO films shown in Fig. 7共c兲.

Downloaded 13 Jan 2007 to 210.72.130.156. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

014910-6

J. Appl. Phys. 101, 014910 共2007兲

Zhang et al. TABLE II. Energetic particles in AZO dc-magnetron sputtering process. E f 共eV兲 共0.4 Pa兲 Energetic particles

Eini 共eV兲

Relative fluxa

Sputtered Zn Sputtered Al Reflected Ar Reflected O Ar+ / O+ / Zn+ O−

9 12 7–25 30–100 11–18 ⬃300

1 1.4⫻ 10−2 1 ⫻ 10−2 3 ⫻ 10−3 0.2–2 ⬃0.01– 0.1

0 cm

2 cm

222 共1兲b

5.5 5 5–16 22–78 11–18 216 共0.90兲

E f 共eV兲 共0.8 Pa兲 4 cm

212 共0.67兲

0 cm

2 cm

4 cm

165 共1兲

3.5 3 3–10 17–60 11–18 163 共0.88兲

148 共0.61兲

Relative to the flux 共=1兲 of sputtered Zn. The numbers in brackets represent the relative flux of O−.

a

b

The first goal is to understand the role of the energetic species bombardment and determine the energy threshold of subimplant mechanism. The damage caused to films induced by ion bombardment should be considered when the energy of incident particles, Ei, was over the penetration threshold Epet of the deposited films. The Epet is the sum of the displacement threshold Edisp and the surface binding energy Esb25 The Edisp could be approximately equal to the sputtering threshold energy Esp of the films.23 Yamamura and Itoh26 gave a semiempirical formula for the Esp, Esp = 8Esb共M 1 / M 2兲2/5, where M 1,2 is atom weight of incidence ion and target atom, respectively. Generally, for the covalent compounds ZnO the formation energy of ⬃3.6 eV is a good estimate of Esb, and the value of Esp is about 29 eV for Ar+ bombardment, which is in good agreement with the experimental threshold of ⬃30 eV in the studies of Cebulla et al.17 Thus, the net penetration threshold for ZnO is Epet ⬃ Edisp + Esb ⬃ 33 eV. In addition, a series of SRIM computations of the energetic species 共Ar+, O+, and Zn+兲 impinging the growing film surface was performed. The simulations indicated that all the ion penetration depth was limited in the top ZnO monolayer, equal to c / 2 ⬃ 0.26 nm for the preferred 共002兲 orientation, if Ei was less than ⬃30 eV 共⬃Epet兲. For higher energy, the energetic species would be implanted below the ZnO surface, which was likely to induce high intrinsic stresses.27 Therefore, it can be concluded that the role of ion bombardment is dependent on the mechanism of dissipation of incident species. Ion bombardment is beneficial to the film growth if Ei is below Epet of the deposited films, while the energy subimplant mechanism should be considered and the bombardment is harmful to the film quality in case that Ei is over the Epet. The following section is to investigate the lateral distribution of energetic species and determine the energies responsible for the energy subimplant mechanism. During the AZO deposition, the energetic particles toward the substrate are mainly comprised of sputtered Zn and Al atoms, reflected neutral gas 共Ar, O兲, positive ions 共Zn+, Ar+, O+2 , and O+兲, and negative O− ions, etc, originating from three mechanisms.13 In this work, for sputtered atoms and reflected neutral gas, their initial energies were calculated from SRIM simulation of the sputtering process considering a Zn:Al target model in the atomic ratio of 95–5 and the energy of 0.733Vdc

⬃ 235 eV for incident Ar ions, while the final energy at the film surface was determined from the SPATS code.21 The relative flux of sputtered Zn and Al atoms was taken equal to their sputtering yields. The flux of backscattered Ar and O neutrals was also evaluated from SRIM simulation, scaling back the flux O neutrals by the ratio, PO2 / PAr. Secondly, for the positive ions which were from the plasma region and accelerated in the plasma sheath in front of the substrate, their energy can be considered as e共V p − V f 兲 共eV兲, assuming no collisions in the sheath,19 while their flux were estimated from the ratio Ji / Jn. The third mechanism is that negative O− ions are emitted from the target surface and accelerated along the potential gradient in the cathode sheath.13,28 The transport from the target to the substrate was determined by SPATS simulation, regarding the O− ions as O neutrals. Then by subtracting e共V p − V f 兲 from the simulation results, the final energy was obtained. Further, the flux of O− ions was estimated at a reasonable range of 5–50% of the sputtered O atoms which simulated from SRIM,3,13 and assuming that 50% of metallic target was oxidized for our case that the reactive sputtering processes were carried out in nearcompound mode. According to the above analysis, the relative flux and initial/final energies for each species was tabulated in Table II. As expected, the energy of sputtered Zn, Al atoms, and backscattered Ar neutrals were far below Epet, which was ruled out as a possible reason for the film damage. Although the energy of reflected O neutrals was likely to exceed Epet, their flux was much lower by several orders of magnitude and thus also less consequential. Secondly, for positive ions Zn+, Ar+, and O+, etc., their mean energies were approximately 11– 18 eV, also less than Epet, which were similar to the results of Herrmann et al.14 Thus, the lattice distortion induced by positive ions could be neglected. However, it should be questioned that Herrmann et al. contributed the spatial distribution of positive ions bombardment to the reason of the lateral distribution of AZO film properties because their energies were not enough to implant the film surface. Experimental evidence also confirmed this finding. The resistivity distribution of AZO films prepared at HPD conditions was significantly improved, as shown in Fig. 4, contradictory to the change predicted by the subimplant mechanism, because the ion density, i.e., the bombardment,

Downloaded 13 Jan 2007 to 210.72.130.156. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

014910-7

J. Appl. Phys. 101, 014910 共2007兲

Zhang et al.

was pronouncedly enhanced on the center and more inhomogeneous under HPD conditions 关Fig. 2共a兲兴. Therefore, it is more reasonable that the increased Ji 共mainly positive ions兲 could be favorable for adatoms diffusion and distortion recovering rather than the film damage due to the penetration of these ions 共Ei ⬍ Epet兲 was limited in the top monolayer of AZO films. Their unconvincing explanation probably originated from not considering the bombardment by energetic negative ions such as O− ions in their studies and the beneficial role of low energy ions. Table II shows that the negative oxygen ions are most likely to be responsible for high stress and degradation of the film, since their energy far exceeded the Epet. Furthermore, according to forward sputtering model proposed by Windischmann,23 the compressive intrinsic stress ␴ is in proportion to the flux ␾ p and square root kinetic energy 冑E p of energetic projectiles. The distribution of energetic O− ions also exhibited a maximum at the center of the substrate and reduced towards its edges, which was in agreement with the stress distribution from the qualitative perspective as, demonstrated in Fig. 6共a兲. Thus, it further confirmed the subimplant mechanism of O− ions as a major origin of the spatial distribution for our case, which also supported the viewpoints of Tominaga et al.12,13 Although Minami et al.15 had questioned the validity of the subimplant mechanism for the inhomogeneity because it did not completely explain their experiment, the subimplant mechanism did not contradict essentially with their results. According to the previous discussion, if we take into account that in their case the sputtering voltage was still high enough to cause damage to the film even though it decreased from 288 to 250 V, and the lower TS and LPD conditions made the atomic oxygen effect more notable. V. CONCLUSIONS

This paper presented a study of the electrical property and its spatial distribution of AZO thin films prepared by dc reactive magnetron sputtering under different conditions. The crystalline, stress, and resistivity of the films were found to be strongly dependent on TS and Ji / Jn. AZO films prepared at lower TS 共90 ° C兲 and Ji / Jn 共⬍0.3兲 exhibited a significant distribution of stress and resistivity, which could be remarkably improved by increasing either TS or Ji / Jn. The experiment suggested that the energy subimplant mechanism dominated the resistivity distribution of AZO films, while the nonuniform distribution of active oxygen played a secondary role, whose influence became more significant under conditions of lower TS and plasma density. Furthermore, the simulation indicated that the role of ion bombardment is associated with the penetration threshold Epet ⬃ 33 eV for ZnO. It may improve or degrade the film quality depending on whether Ei is over the Epet. This study also revealed that the energetic O− rather than positive ions made main contribution to the subimplant mechanism during the sputtering. Finally, this study also proposed some feasible methods to improve electrical property and distribution of AZO films prepared at low TS. The essential way is to avoid the high energy ion bombardment, which could be achieved by using

a low sputtering voltage, for example, by applying an rf or rf+ dc excitation.17,29 This has already been testified by the fact that AZO films produced by rf sputtering has somewhat higher quality than that by dc sputtering, especially at low TS. But this method might be weakened by the lower deposition rate, higher costs, and difficult scalability. Alternatively, other effective methods such as increasing nearsubstrate plasma density, adding a low energy 共below Epet兲 ion bombardment, and applying an oxygen ion source are believed to be able to further improve the property and uniformity. ACKNOWLEDGMENTS

This work was supported by the National Natural Science Foundation of China 共Grant No. 50172051兲. The authors are grateful to Dr. K. Macak of Sheffield Hallam University, UK, for his kind advice about the SPATS program. D. S. Ginley and C. Bright, MRS Bull. 25, 15 共2000兲. B. G. Lewis and D. C. Paine, MRS Bull. 25, 22 共2000兲. 3 P. F. Carcia, R. S. Mclean, M. H. Reilly, Z. G. Li, L. J. Pillione, and R. F. Messier, Appl. Phys. Lett. 81, 1800 共2002兲. 4 T. W. Kelley, P. F. Baude, C. Gerlach, D. E. Ender, D. Muyres, M. A. Haase, D. E. Vogel, and S. D. Theiss, Chem. Mater. 16, 4413 共2004兲. 5 I. Petrov, P. B. Batma, L. Hultman, and J. E. Greene, J. Vac. Sci. Technol. A 21, S117 共2003兲. 6 R. J. Hong, X. Jiang, G. Heide, B. Szyszka, V. Sittinger, and W. Werner, J. Cryst. Growth 249, 461 共2003兲. 7 N. W. Schmidt, T. S. Totushek, W. A. Kimes, D. R. Callender, and J. R. Doyle, J. Appl. Phys. 94, 5514 共2003兲. 8 K. Ellmer, J. Phys. D 33, R17 共2000兲. 9 K. L. Chopra, S. Major, and D. K. Pandya, Thin Solid Films 102, 1 共1983兲. 10 A. V. Singh and R. M. Mehra, J. Appl. Phys. 90, 5661 共2001兲. 11 D. Song, P. Widenborg, W. Chin, and A. G. Aberle, Sol. Energy Mater. Sol. Cells 73, 1 共2002兲. 12 K. Tominaga, K. Kuroda, and O. Tada, Jpn. J. Appl. Phys., Part 1 27, 1176 共1988兲. 13 K. Tominaga, M. Chong, and Y. Shintani, J. Vac. Sci. Technol. A 12, 1435 共1994兲. 14 D. Herrmann, M. Oertel, R. Menner, and M. Powalla, Surf. Coat. Technol. 174–175, 229 共2003兲. 15 T. Minami, T. Miyata, T. Yamamoto, and H. Toda, J. Vac. Sci. Technol. A 18, 1584 共2000兲. 16 T. Minami, Y. Takeda, S. Takata, and T. Kakumu, Thin Solid Films 308– 309, 13 共1997兲. 17 R. Cebulla, R. Wendt, and K. Ellmer, J. Appl. Phys. 83, 1087 共1998兲. 18 I. Petrov, I. Ivanov, V. Orlinov, and J. Kourtev, Contrib. Plasma Phys. 30, 223 共1990兲. 19 N. Hellgren, K. Macak, E. Broitman, M. P. Johansson, L. Hultman, and J. E. Sundgren, J. Appl. Phys. 88, 524 共2000兲. 20 J. F. Ziegler, J. P. Biersack, and U. Littmark, The Stopping Range of Ions in Matter (SRIM) 共Pergamon, New York, 1985兲. 21 K. Macak, P. Macak, and U. Helmersson, Comput. Phys. Commun. 120, 238 共1999兲. 22 See http://www.ftgsoftware.com 23 H. Windischmann, Crit. Rev. Solid State Mater. Sci. 17, 547 共1992兲. 24 K. Ichihara, N. Inoue, M. Okubo, and N. Yasuda, Thin Solid Films 245, 152 共1994兲. 25 J. Robertson, Mater. Sci. Eng., R. 37, 129 共2002兲. 26 Y. Yamamura and N. Itoh, Ion Beam Assisted Film Growth 共Elsevier, Amsterdam, 1989兲. 27 P. Patsalas, C. Gravalidis, and S. Logothetidis, J. Appl. Phys. 96, 6234 共2004兲. 28 J. P. Krumme, R. A. A. Hack, and I. J. M. M. Raaijmakers, J. Appl. Phys. 70, 6743 共1991兲. 29 T. Minami, T. Miyata, Y. Ohtani, and Y. Mochizuki, Jpn. J. Appl. Phys., Part 2 45, L409 共2006兲. 1 2

Downloaded 13 Jan 2007 to 210.72.130.156. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp