You are on page 1of 10

INSTITUTE OF PHYSICS PUBLISHING Semicond. Sci. Technol.

20 (2005) S35S44

SEMICONDUCTOR SCIENCE AND TECHNOLOGY doi:10.1088/0268-1242/20/4/004

Transparent conducting oxide semiconductors for transparent electrodes


Tadatsugu Minami
Optoelectronic Device System R&D Center, Kanazawa Institute of Technology, 7-1 Ohgigaoka, Nonoichi, Ishikawa 921-8501, Japan

Received 6 October 2004 Published 15 March 2005 Online at stacks.iop.org/SST/20/S35 Abstract The present status and prospects for further development of polycrystalline or amorphous transparent conducting oxide (TCO) semiconductors used for practical thin-lm transparent electrode applications are presented in this paper. The important TCO semiconductors are impurity-doped ZnO, In2O3 and SnO2 as well as multicomponent oxides consisting of combinations of ZnO, In2O3 and SnO2, including some ternary compounds existing in their systems. Development of these and other TCO semiconductors is important because the expanding need for transparent electrodes for optoelectronic device applications is jeopardizing the availability of indium-tin-oxide (ITO), whose main constituent, indium, is a very expensive and scarce material. Al- and Ga-doped ZnO (AZO and GZO) semiconductors are promising as alternatives to ITO for thin-lm transparent electrode applications. In particular, AZO thin lms, with a low resistivity of the order of 105 cm and source materials that are inexpensive and non-toxic, are the best candidates. However, further development of the deposition techniques, such as magnetron sputtering or vacuum arc plasma evaporation, as well as of the targets is required to enable the preparation of AZO and GZO lms on large area substrates with a high deposition rate.

1. Introduction
For most optoelectronic devices such as at panel displays, it is essential to use a transparent electrode consisting of a thin lm of a transparent conducting oxide (TCO) semiconductor. Although tin-doped indium oxide (commonly called indiumtin-oxide, or ITO) thin lms deposited by magnetron sputtering (MSP) have been in practical use for most transparent electrode applications, there are many reports on other TCO semiconductors as well as deposition methods [17]. A stable supply of ITO may be difcult to achieve for the recently expanding market for optoelectronic devices because of the cost and scarcity of indium, the principal material of ITO. In addition, recent developments in optoelectronic devices have frequently required thin-lm transparent electrodes with specialized properties. Recent research concerning thin-lm transparent electrodes using TCO semiconductors has focused on resolving these problems. For example, we have proposed the use of impurity-doped zinc oxide (ZnO) as an alternative to ITO [4, 8, 9] as well as the use of multicomponent oxide thin lms with properties suitable for specialized applications [4, 10]. In addition, lm deposition techniques
0268-1242/05/040035+10$30.00

suitable for fabricating these TCO semiconductors are being developed. This paper introduces the present status and prospects for further development of polycrystalline or amorphous TCO thin lms for transparent electrode applications. In particular, approaches for resolving the above problems and the relationship between TCO semiconductors and lm deposition techniques are presented in detail.

2. TCO semiconductors for thin-lm transparent electrodes


In general, TCO thin lms that are in practical use as transparent electrodes are polycrystalline or amorphous, except for single crystals grown epitaxially, and exhibit a resistivity of the order of 103 cm or less and an average transmittance above 80% in the visible range. Thus, TCO semiconductors suitable for use as thin-lm transparent electrodes should have a carrier concentration of the order of 1020 cm3 or higher and a band-gap energy above approximately 3 eV: i.e., degenerated n-type or p-type S35

2005 IOP Publishing Ltd Printed in the UK

T Minami

Table 1. TCO semiconductors for thin-lm transparent electrodes. Material SnO2 In2O3 ZnO CdO ZnOSnO2 ZnOIn2O3 In2O3SnO2 CdOSnO2 CdOIn2O3 MgIn2O4 GaInO3, (Ga, In)2O3 CdSb2O6 ZnOIn2O3SnO2 CdOIn2O3SnO2 ZnOCdOIn2O3SnO2 Dopant or compound Sb, F, As, Nb, Ta Sn, Ge, Mo, F, Ti, Zr, Hf, Nb, Ta, W, Te Al, Ga, B, In, Y, Sc, F, V, Si, Ge, Ti, Zr, Hf In, Sn Zn2SnO4, ZnSnO3 Zn2In2O5, Zn3In2O6 In4Sn3O12 Cd2SnO4, CdSnO3 CdIn2O4 Sn, Ge Y Zn2In2O5In4Sn3O12 CdIn2O4Cd2SnO4 Figure 1. Practical TCO semiconductors for thin-lm transparent electrodes.

semiconductors. Historically, most research to develop TCO thin lms as transparent electrodes has been conducted using n-type semiconductors; in practice, TCO thin lms used as transparent electrodes are n-type semiconductors consisting of metal oxides. On the other hand, the preparation of TCO thin lms consisting of p-type semiconductors was rst reported in 1993 [11]. Since this report of p-type NiO thin lms deposited by rf magnetron sputtering (rfMSP), there have been many reports on the preparation of p-type semiconducting thin lms using new TCO semiconductors [1215]. Nevertheless, there has been no report on the preparation of a p-type TCO thin lm suitable for use as a practical transparent electrode. For the purpose of obtaining lower resistivities, various TCO semiconductor materials have been developed; n-type TCO semiconductors now available for thin-lm transparent electrodes are listed in table 1, grouped by compound type [4, 10]. One advantage of using binary compounds as TCO materials is the relative ease of controlling the chemical composition in lm depositions compared to using ternary compounds and multicomponent oxides. Up to now, various TCO thin lms consisting of binary compounds such as SnO2, In2O3, ZnO and CdO have been developed, with impuritydoped SnO2 (SnO2:Sb and SnO2:F), impurity-doped In2O3 (In2O3:Sn, or ITO) and impurity-doped ZnO (ZnO:Al and ZnO:Ga) lms in practical use. In addition, it is well known that highly transparent and conducting thin lms can also be prepared using metal oxides without intentional impurity doping [13]. The resulting lms are n-type degenerated semiconductors with free electron concentrations of the order of 1020 cm3 provided by native donors such as oxygen vacancies and/or interstitial metal atoms. However, since undoped oxide lms were found to be unstable when used at a high temperature, binary compounds without impurity doping have proved unusable as practical transparent electrodes [9, 16]. The reported effective dopants are also listed in table 1 along with their associated binary compounds. In addition to binary compounds, ternary compounds such as Cd2SnO4, CdSnO3, CdIn2O4, Zn2SnO4, MgIn2O4, CdSb2O6 and In4Sn3O12 have been developed [13, 1720]. However, there are few reports on the effect of impurity doping due to the lack of an effective dopant. As a result, TCO lms fabricated from these ternary compounds have yet to S36

be used widely. In order to obtain TCO lms suitable for specialized applications, new TCO semiconductor materials have been actively studied in recent years. In the 1990s, new TCO semiconductor materials consisting of multicomponent oxides, such as combinations of binary compound TCOs, were developed [21, 22]. In addition, new TCO semiconductor materials consisting of multicomponent oxides composed of combinations of ternary compound TCOs have been developed [23]. The use of multicomponent oxide materials makes possible the design of TCO lms suitable for specialized applications because the electrical, optical, chemical and physical properties can be controlled by altering the chemical composition [4, 10]. Table 1 shows that TCO semiconductors for thin-lm transparent electrodes have been developed only with metal oxides containing at least one of the following metal elements: Zn, Cd, In and Sn. Although producing a low resistivity thin lm, it should be noted that Cd-containing TCO semiconductors such as In-doped CdO (CdO:In), Cd2SnO4, CdSnO3 and CdIn2O4 [13, 5, 7, 19] are of no practical use because of the toxicity of Cd. For thin-lm transparent electrodes, TCO semiconductors such as impurity-doped ZnO, In2O3 and SnO2 and multicomponent oxides composed of combinations of these binary compounds are the best candidates for practical use, as shown in gure 1.

3. Challenge to obtain lower resistivity


ITO thin lms prepared by MSP with a resistivity of the order of 1 104 cm are commercially available at present. Although the obtainable electrical properties are strongly dependent on the deposition method as well as the deposition conditions, TCO lms with a low resistivity of the order of 105 cm can only be prepared with impurity-doped binary compounds. Figure 2 shows the change in minimum resistivity of impurity-doped binary compound TCO lms reported in recent years. The impurity-doped ZnO, In2O3 and SnO2 lms were prepared on glass substrates under various deposition conditions using various deposition methods. It should be noted that the obtained minimum resistivities of impurity-doped ZnO lms is still decreasing, whereas those of impurity-doped SnO2 and In2O3 lms have essentially remained unchanged for more than the past twenty years. As also seen in gure 2, a resistivity of the order of 105 cm

Transparent conducting oxide semiconductors for transparent electrodes

103 HALL MOBILITY (cm2/Vs)

Ionized impurity scattering (BHD theory)

102

101

Figure 2. Reported resistivity of impurity-doped binary compound TCO lms, 1972present: impurity-doped SnO2 ( ), In2O3 ( ) and ZnO (). Table 2. Electrical properties of ITO lms with a resistivity of the order of 105 cm. Resistivity ( cm) 7.4 105 8.9 105 7.2 105 9.5 105 8.45 105 Carrier concentration (cm3) 1.38 1021 1.3 1021 2.5 1021 1.8 1021 1.38 1021 Hall mobility (cm2 (V s)1) 103 54.1 33.2 40 53.5 Deposition method (year) [Reference] Zone conning (1996) [24] Pulsed laser deposition (2000) [25] Pulsed laser deposition (2001) [26] Spray pyrolysis (2002) [27] Pulsed laser deposition (2002) [28]

100 19 20 21 22 10 10 10 10 3) CARRIER CONCENTRATION n (cm

Figure 3. The n relationship of reported AZO lms: broken line shows the n relationship calculated using the BHD theory. Table 3. Electrical properties in low-resistivity AZO lms prepared by PLD or MSP. Resistivity ( cm) 1.3 104 8.5 105 Carrier concentration (cm3) 1.31 1021 1.5 1021 Hall mobility (cm2 (V s)1) 36.7 47.6 Deposition method (year) [Reference] Pulsed laser deposition (1999) [29] Pulsed laser deposition (2003) [30] rf magnetron sputter (1984) [9] dc magnetron sputter (1990) [31]

1.9 104 2.7 104

1.5 1021 9.0 1020

22.0 25.0

has been recently reported in ITO and ZnO:Al (AZO) thin lms. The obtained electrical properties (resistivity, carrier concentration and Hall mobility) and the deposition method used for the past ten years to produce ITO thin lms with a resistivity of the order of 105 cm are summarized in table 2; the best results seem to result from using depositions at a high temperature and/or a pulsed laser deposition (PLD) method. In addition, Suzuki et al have recently reported AZO thin lms prepared by PLD with a resistivity of the order of 105 cm [29]: the rst report of impurity-doped ZnO thin lms with this level of resistivity, the same as that found for ITO. It should be noted that the obtained minimum resistivities in ITO and AZO lms prepared by PLD were lower than those resulting from MSP, the conventional deposition method in practical use. As an example, table 3 shows a comparison of the obtained electrical properties between low-resistivity AZO thin lms prepared by PLD and MSP. The difference of obtained resistivity was ascribed to that of Hall mobility rather than carrier concentration: PLD Hall mobility, about 40 cm2 (V s)1 [29, 30] and MSP Hall mobility, 2030 cm2 (V s)1 [9, 31]. It is well known that the Hall mobility of AZO lms with a carrier concentration of the order

of 10201021 cm3 is dependent on the carrier concentration because Hall mobility is mainly dominated by ionized impurity scattering [32]. Figure 3 shows the relationship between the obtained Hall mobility () and carrier concentration (n) for AZO lms with a carrier concentration of the order of 10201021 cm3 and a resistivity () of 5 104 to 8.5 105 cm prepared using various deposition methods such as PLD and MSP; also shown is the n relationship calculated using the BrooksHerringDingle (BHD) theory by taking into account degeneracy and assuming that ionized impurity scattering dominates the mobility [32]. As can be seen in gure 3, the measured mobility in AZO lms with a carrier concentration of the order of 10201021 cm3 was lower than that calculated by the BHD theory. It has been reported that the results of the experimental examination of the n relationship can be explained by modied ionized impurity scattering using the BHD theory by taking into account degeneracy and the non-parabolicity of the conduction band [32]. However, as can be seen in gure 3 at a carrier concentration around 1 1021 cm3, the data measured for Hall mobility were scattered in the range from approximately 10 to 50 cm2 (V s)1. With the carrier concentration S37

T Minami

CARRIER CONCENTRATION n (cm-3)

RESISTIVITY ( cm)

40

102

10

21

101

30 40 20

103

1020

100

10

10

10 200 600 1000 1400 FILM THICKNESS (nm)

19

30 200

600

1000

0 1400

FILM THICKNESS (nm)


Figure 5. Hall mobility ( ) and crystallite size ( ) as functions of lm thickness for AZO lms prepared at 200 C by PLD.

Figure 4. Resistivity ( ), carrier concentration ( ) and Hall mobility ( ) as functions of lm thickness for AZO lms prepared at 200 C by PLD.

approximately constant, the variation of Hall mobility cannot be solely explained by ionized impurity scattering. This suggests that the mobility of AZO lms is dominated by not only ionized impurity scattering but also another scattering mechanism. In addition, we have found the Hall mobility of AZO lms prepared with various lm thicknesses by PLD exhibiting the same n relationship as shown in gure 3. As an example, the thickness dependences of electrical properties (, n and ) are shown in gure 4 for AZO lms prepared with an Al content (Al/(Al+Zn) atomic ratio) of 1.5 at% on glass substrates at 200 C [33]. The Hall mobility increased from approximately 20 to 50 cm2 (V s)1 as the thickness was increased, whereas the carrier concentration of approximately 1 1021 cm3 was relatively independent of the thickness. The prepared AZO lms were polycrystalline and predominantly c-axis oriented perpendicular to the substrate surface, as evidenced by x-ray diffraction analyses. It was also found that the thickness dependence of Hall mobility is closely related to that of crystallinity estimated from the full width at half maximum (FWHM) of the (0002) diffraction peak. Figure 5 shows Hall mobility and crystallite size as functions of lm thickness for the AZO lms shown in gure 4. It should be noted that Hall mobility was estimated by the van der Pauw method using the carrier transport in the direction parallel to the lm surface whereas the crystallite size was estimated in the direction perpendicular to the lm surface. However, it is usually believed that the crystallite size estimated in the direction perpendicular to the lm surface may be correlated with the crystallite size in the direction parallel to the lm surface, as evidenced from the observation of surface morphology by an atomic force microscope (AFM). The variation in Hall mobility shown in gures 3 and 5 suggests that the mobility of polycrystalline AZO thin lms with a carrier concentration of the order of 10201021 cm3 is not solely dominated by ionized impurity scattering but is also affected by crystallinity. S38

In addition to the above, it has also been reported that Hall mobility in polycrystalline AZO and undoped ZnO lms are affected by grain boundary scattering [32]; this scattering is characterized by an increase in Hall mobility caused by increases in carrier concentration up to approximately 1021 cm3, resulting from increases of the potential barrier height as well as the trapping of free electrons due to oxygen adsorbed on grain boundaries and the lm surface. In contrast, grain boundary scattering may be unimportant for AZO lms with a carrier concentration of the order of 1021 cm3 because of the lowering of potential barrier height resulting from the carrier screening effect. Nevertheless, the increase of mean free path from approximately 2 to 10 nm, corresponding to the increase of mobility from 10 to 50 cm2 (V s)1 in AZO lms with a carrier concentration of 1 1021 cm3, may be assigned to the improvement of crystallinity, which is related to carrier scattering mechanisms such as grain boundary scattering and dislocation scattering [33]. However, details concerning crystal imperfections such as point defects and dislocations that dominate carrier scattering as well as the relationship between imperfections and scattering mechanisms have yet to be investigated sufciently. From the results described above, the higher obtained mobility in AZO lms prepared by PLD than in lms prepared by MSP may be explained by differences in thinlm crystallinity. In addition, it is clear that obtaining a resistivity of the order of 105 cm in AZO lms with a carrier concentration of the order of 1 1021 cm3 requires an improvement of crystallinity. On the other hand, this way of obtaining AZO thin lms with a resistivity of the order of 105 cm may also be effective for the preparation of ITO thin lms. In addition to the investigation of Hall mobility described above, it should be noted that the preparation of thin lms with a carrier concentration of the order of 1021 cm3 is essential to obtain a resistivity of the order of 105 cm. It should be noted that the preparation of impurity-doped ZnO thin lms with a carrier concentration of the order of 1021 cm3 is much more difcult than with ITO lms. Since the oxygen adsorption on

HALL MOBILITY (cm2/Vs)

HALL MOBILITY (cm2/Vs)

101

1022

102

50

50

CRYSTALLITE SIZE D (nm)

Transparent conducting oxide semiconductors for transparent electrodes

the surface and grain boundaries of the lms is affected by the oxidizing atmosphere during deposition, controlling the oxidizing atmosphere during the deposition is very important for the preparation of highly transparent and conducting AZO thin lms with a carrier concentration of the order of 1021 cm3 [34]; a weaker oxidizing atmosphere is required than with the deposition of other TCO materials such as ITO [4]. It should also be noted that an oxygen vacancy in the ZnO lattice usually acts as a donor that generates a single free electron, whereas the one in the In2O3 lattice acts as a donor that generates doubly free electrons [3]. Because of the differences in the chemical and physical properties between ITO and AZO described above, ITO thin lms can even be prepared by various deposition methods on high temperature substrates with a resistivity of the order of 105 cm. However, in order to prepare AZO lms with a resistivity of the order of 105 cm using a deposition method other than PLD, it is necessary to develop deposition techniques that provide a controlled weak oxidizing atmosphere during the deposition.

4. TCO lms as alternatives to ITO


Recently, the demand for ITO thin-lm transparent electrodes has dramatically increased in the eld of optoelectronic devices. If the increase in usage of ITO lms for at panel displays and solar cells continues, not only will the price of ITO continue to rise but also the availability of In may be jeopardized in the near future. The development of alternative TCO materials is necessary to resolve this serious problem. For obtaining a resistivity of the order of 105 cm, impurity-doped binary compounds such as Al-doped ZnO (AZO) and In-doped CdO have been proposed as alternative materials. With consideration for the environment, it appears that impurity-doped ZnO, particularly AZO, is a promising alternative because of its use of ZnO or Zn, both being inexpensive, abundant and non-toxic materials, and its ability to exhibit resistivity comparable to ITO, as described above. In the following, the present status and prospects for further development of impurity-doped ZnO lms are surveyed. In addition, various problems and solutions associated with using impurity-doped ZnO as a practical alternative to ITO are described. In the 1980s, transparent conducting ZnO lms with a resistivity of the order of 104 cm were prepared by impurity doping [9, 35, 36]. However, a low resistivity below approximately 2 104 cm has only been obtained in AZO and GZO lms, as shown in gure 6; table 4 summarizes the minimum resistivity and the maximum carrier concentration obtained for typical impurity-doped ZnO lms prepared with optimal doping content for various dopants and deposition methods. In general, the obtainable electrical properties of impuritydoped ZnO lms are strongly dependent on the deposition methods and conditions. Impurity-doped ZnO lms with a resistivity of the order of 104 cm have been prepared by vacuum arc plasma evaporation (VAPE), metal organic molecular beam deposition (MOMBD) [37] and metal organic chemical vapour deposition (MOCVD) [6] as well as by MSP [9, 10] and PLD [29]. Recently, AZO and GZO lms with a

Figure 6. Reported resistivity of impurity-doped ZnO lms, 1983present: AZO ( ), GZO ( ) and other impurity-doped ZnO ().

Table 4. Resistivitiy, carrier concentration and dopant content for typical ZnO lms doped with various impurities. Dopant content (wt%) 12 27 2 2 6 0.53 0.5 (at%) 0 Resistivity 104 ( cm) 0.85 1.2 2.0 3.1 4.8 5.0 4.0 4.5 Carrier concentration 1020 (cm3) 15.0 14.5 5.4 6.7 8.8 4.9 5.0 2.0

Dopant Al2O3 Ga2O3 B2O3 Sc2O3 SiO2 V2O5 F None

resistivity of the order of 1 104 cm were prepared using PLD and VAPE [38]. Although VAPE using an Uramoto gun as the arc plasma source [39], similar to activated reactive evaporation (ARE), has attracted much attention as a new deposition technique with a high deposition rate on large area substrates, vacuum evaporation of AZO is difcult to achieve because the vapour pressure of Al2O3 is too low in comparison with that of ZnO. AZO lms with a resistivity of the order of 105 cm have been prepared by PLD, but preparing lms on large substrates with a high deposition rate is also very difcult to achieve. When preparing highly conductive and transparent impurity-doped ZnO lms, controlling the oxidation of Zn is much more difcult than with other binary compounds such as SnO2 and In2O3, because Zn is more chemically active in an oxidizing atmosphere than either Sn or In. Because of this binding energy of Zn and O, the activity and amount of oxygen must be precisely controlled during the deposition. As a result, ZnO lms with low resistivity are attainable only by depositions in atmospheres that are less oxidizing than in depositions of In2O3 and SnO2 lms. In general, the preparation of ITO and impurity-doped SnO2 lms by MSP using oxide targets usually requires an introduction of O2 gas during the sputter deposition in order to obtain a low resistivity S39

T Minami

[3]. In contrast, the introduction of O2 gas is unnecessary for the sputter deposition of AZO and GZO lms. As a result, the obtainable resistivities in AZO and GZO lms are considerably affected by the oxygen content in the sintered oxide pellets in PLD and VAPE and in the targets in MSP lm preparations. For example, the oxygen content in the AZO target is very important in the preparation of lower resistivity AZO lms by MSP, because the oxidization occurring on the surface of substrates is mainly dominated by the substrate temperature and the activity and quantity of the oxygen generated from the oxide target reaching the substrate surface [37]. At the same substrate temperatures, controlling oxidization during VAPE impurity-doped ZnO lm depositions is easier than with MSP, because there is less oxidation occurring on the substrate surface. The preparation of AZO lms on large area substrates by dc MSP (dcMSP) using an oxide target is particularly difcult because a spatial distribution of resistivity corresponding to the target erosion area pattern is created on the substrate surface [40]. This problem is mainly attributed to the distribution pattern of the activity and quantity of oxygen reaching the substrate surface, which depends on the sputter deposition conditions as well as the target preparation conditions. For the purpose of improving this distribution pattern, various sputtering techniques have been developed. For example, lower resistivities in AZO lms prepared by MSP have been achieved by placing substrates perpendicularly to the target [8, 9] and by conducting depositions under an applied magnetic eld [8, 31], but the problem has not yet been resolved completely. Therefore, further development of deposition techniques that can produce more spatially uniform low resistivity lms on large area substrates is required to overcome the problems associated with MSP. At present, resistivities of 23 104 cm, a refractive index of approximately 2.0 and an average transmittance above 85% in the visible range can be obtained in AZO and GZO lms prepared on large area substrates with a high deposition rate at a temperature above approximately 200 C by MSP and VAPE. The transparent conducting AZO and GZO lms have characteristics that can be either an advantage or a disadvantage depending on the application. As an example, there are several problems at present with the use of these lms as the transparent electrode of at panel displays, replacing ITO. For example, good patterns can be etched in amorphous ITO lms deposited on low temperature substrates by doping with Zn [41] or using a H2O [42] atmosphere. In contrast, it is difcult to apply the presently used photolithography process with wet treatments, because doped ZnO lms are more easily etched by both acid and alkaline solutions than ITO lms. Of course, in other applications their high etching rates might be advantageous. In addition, the etching rate of doped ZnO lms is considerably affected by not only the crystallinity and the kind of dopant used but also the deposition conditions. As an example, gure 7 shows etching rates in HCl (0.2 mol l1 at 20 C) and KOH (3 mol l1 at 20 C) solutions as functions of Co content for Co-codoped AZO (AZO:Co) lms prepared at 200 C by MSP [43]; the etching rate of the undoped ZnO (ZO) lm in the HCl solution decreased markedly with the addition of Al doping. The etching rate for AZO lms in both HCl and KOH solutions was further decreased with Co S40

10

20

AZO:Co
(Al2O3:2wt.%) Etchant Temp. 20C RE(H) 0.2 mol./1 HCL 3 mol./1 KOH RE(OH)

10

10

10

RE(H)
5

RE(OH)
10
4

0 ZO AZO 1 2

CoO CONTENT (wt.%)


Figure 7. Resistivity () and etching rates in HCl ( ) and HOH ( ) solutions as functions of Co content for AZO:Co lms.

Figure 8. Resistivity () and etching rate ( ) in HCl solution for undoped and impurity-doped ZnO lms: FGO, GZO and GZO:F.

codoping. As another example, the dopant dependence of etching rate in a HCl solution (0.2 mol l1 at 20 C) is shown in gure 8 for various impurity-doped ZnO lms prepared at 100 C by VAPE [44]; the etching rate of both F- and Gacodoped ZnO (FZO and GZO) lms was also lower than that of the undoped ZnO (ZO) lm. The decreased etching rate of F-codoped GZO (GZO:F) lm was also accompanied by a decreased obtained resistivity. In addition, we have proposed that using low temperature etchants is very effective for reducing the etching rate. On the other hand, an all dry

ETCHING RATE RE (nm/s)

15

RESISTIVITY ( cm)

Transparent conducting oxide semiconductors for transparent electrodes

Table 5. Comparison of thin-lm transparent electrode properties between AZO and ITO. Doped ZnO Low resistivity ( cm) Practical resistivity ( cm) Eg (eV) Index of refraction Work function Cost 105 23 104 3.3 2 4.6 Inexpensive ITO 105 1 104 3.7 2 4.85.0 Very expensive

Stability Acid solution <Good (stable) Alkali solution <Good (stable) Oxidizing atmosphere at high <Good (stable) temperature (or oxygen plasma) Reducing atmosphere at high Good (stable)> temperature (or hydrogen plasma)

process using an oxygen ashing process can be applied to AZO lms [45]. It should be noted that for applications involving reducing atmospheres at high temperatures and hydrogen plasma, doped ZnO lms are more stable than ITO lms [46]. In contrast, for use in oxidizing atmospheres at high temperatures, ITO lms are more stable [37]. As another specialized application example, AZO lms can be advantageously used with thin-lm solar cells, because milky ZnO lms with a textured surface structure are more easily produced on relatively low temperature substrates than milky SnO2 lms [47, 48]. The use of milky AZO transparent electrodes for solar cells can produce an improvement in efciency resulting from a light connement effect. Several researchers have reported the preparation of milky AZO, GZO and B-doped ZnO (BZO) lms using various deposition methods such as MSP and atmospheric pressure or low pressure MOCVD and CVD [4, 49]. Using the comparison of AZO and ITO thin-lm transparent electrode characteristics summarized in table 5, AZO and GZO lms are shown to have thin-lm transparent electrode properties comparable to those of ITO lms. GZO lms prepared by VAPE and AZO lms by MSP will become widely used in optoelectronic device applications as alternatives to ITO lms. However, further development of MSP deposition techniques is necessary in order to obtain a more uniform spatial distribution, a less oxidizing atmosphere and AZO targets with a lower oxygen content that would enable the preparation of lower resistivity AZO thin lms.

Figure 9. Resistivity as a function of In content for ZnOIn2O3 lms prepared at RT by dcMSP () and at 200 C by VAPE ( ).

5. TCO lms suitable for specialized applications


Recent developments in optoelectronic device applications have frequently required improvements in the physical and chemical properties of TCO lms used as thin-lm transparent electrodes. In order to develop TCO lms suitable for specialized applications, previously proposed material development using multicomponent oxides has recently been attracting much attention as a source of new TCO semiconductors [10]. Minami et al reported new TCO semiconductors in 1994, ZnOSnO2 multicomponent oxides, that not only had the advantages of ZnO but also those of

SnO2 [10, 49]. In addition, TCO lms using multicomponent oxides composed of combinations of binary compound TCO materials and/or ternary compound TCO materials were developed [10]. Several examples of multicomponent oxides composed of combinations of binary compound TCO materials, ZnOIn2O3, In2O3SnO2 and SnO2ZnO systems, are shown in gure 1. Figure 9 shows resistivity () and carrier concentration (n) as functions of chemical composition (In content, or In/(In+Zn) atomic ratio) for multicomponent oxide ZnO In2O3 thin lms prepared on glass substrates at room temperature (RT) by dcMSP using an oxide powder target [50] and at 200 C by VAPE using a sintered oxide target [44]. With dcMSP, the surface temperature of substrates at RT rose up to about 180 C during the sputter deposition. The In content in the deposited lms was approximately equal to that existing in the targets used, as measured by energy dispersive x-ray (EDX) spectroscopy analyses. It should be noted that transparent multicomponent oxide conducting lms could be prepared in all possible composition ratios of In2O3 and ZnO. The minimum resistivity, corresponding to the maximum carrier concentration, was obtained by altering the In content. The carriers were generated from native donors such as oxygen vacancies. ZnOIn2O3 lms prepared at RT with an In content of about 75.5 and 90 at% by dcMSP and VAPE, respectively, exhibited a resistivity as low as 3 104 cm, comparable to ITO lms. In addition, all deposited ZnOIn2O3 lms with thicknesses below 400 nm exhibited an average transmittance above 80% in the visible range. Figure 10 shows etching rate (RE) and band-gap energy (Eg) as functions of In content for the lms prepared by dcMSP [50]: RE measured using 0.2 M HCl solution at 25 C. The RE of ZnOIn2O3 lms increased as the In content was decreased from about 80 at%; lms prepared with In contents in the range from about 80 to 100 at% were not etched. The minimum band-gap energy of dcMSP deposited lms, 2.9 eV, occurred at an In content around 80 at%. As evidenced from x-ray diffraction analyses, ZnOIn2O3 lms with In contents in the S41

T Minami

Figure 10. Etching rate ( ) and band-gap energy () as functions of In content for ZnOIn2O3 lms prepared at RT by dcMSP.

Figure 11. Resistivity () and etching rate ( ) as functions of Zn2In2O5 content for In4Sn3O12Zn2In2O5 lms prepared at RT by rfMSP.

range from about 40 to 80 at% were amorphous. Regarding the composition dependence of the electrical, optical and chemical properties of other multicomponent oxides, results similar to those described above for ZnOIn2O3 lms were also obtained in In2O3SnO2 and SnO2ZnO systems; transparent conducting thin lms were prepared in multicomponent oxides composed of all composition ratios of ZnO and SnO2 or of SnO2 and In2O3 [4, 10]. In addition, the chemical stability and etching rate changed as the composition was altered in In2O3SnO2 and ZnOSnO2 lms prepared by MSP and VAPE. As mentioned above, dcMSP deposited ZnOIn2O3 lms with In contents around 80 at% exhibited minimums in band-gap energy as well as in resistivity. It is well known that many compounds such as ZnmIn2O3+m (m = 27) exist in the ZnOIn2O3 system. In order to investigate their crystallographical properties, ZnOIn2O3 lms were prepared on high temperature substrates with various In contents. For example, dcMSP deposited ZnOIn2O3 lms prepared at 350 C with In contents in the range from about 40 to 90 at% were identied as a ternary compound, Zn2In2O5, a new TCO semiconductor, as evidenced from x-ray diffraction analyses [50]. The Zn2In2O5 thin lms have a band-gap energy of 2.9 eV and a refractive index of approximately 2.1 to 2.4 in the visible range, higher than the refractive index of about 2.0 seen in conventional TCO thin lms such as ZnO, ITO and SnO2. For the purpose of obtaining lower resistivity, an impurity such as Sn was doped into the Zn2In2O5 lms; however, a signicant decrease of resistivity could not be attained. Minami et al have reported that SnO2ZnO lms prepared by MSP and VAPE with Zn contents (Zn/(Zn+Sn) atomic ratio) of approximately 50 at% could be identied as a ternary compound, ZnSnO3, another new TCO semiconductor [49]. In addition, Minami et al have also reported that In2O3 SnO2 lms with Sn contents (Sn/(In+Sn) atomic ratio) in the range from 40 to 60 at% could be identied as a ternary compound, In4Sn3O12, another new TCO semiconductor [20]. cm, comparable to ITO lms, A resistivity of 2 104 was obtained in In4Sn3O12 lms prepared with a Sn content of S42

50 at% on substrates at 350 C. The In4Sn3O12 lms were very stable in acid solutions and in an oxidizing atmosphere at high temperatures. Thus, In4Sn3O12 lms, which are lower in cost than ITO lms with their higher In content, are very promising as TCO lms. In addition, it has been reported that TCO semiconductors could also be obtained using a ternary compound such as (Ga, In)2O3 in the In2O3Ga2O3 system or MgIn2O4 in the MgOIn2O3 system, composed of a TCO semiconductor and an insulator. Unno et al have reported that transparent conducting MgIn2O4 lms were prepared by rf sputtering [18]; Phillips et al have reported that transparent conducting undoped and impurity-doped GaInO3 thin lms prepared by PLD and reactive sputtering exhibited resistivities of the order of 103 cm, very low optical absorption coefcients in the visible range, and a refractive index of 1.65 [51]. In In2O3 Ga2O3 and MgOIn2O3 systems, however, the resistivity of the resulting lms increased considerably as the insulating material (Ga2O3 or MgO) content was increased [10]. TCO semiconductors could be prepared in all compositions of the In2O3(Ga, In)2O3 and the In2O3MgIn2O4 systems when the components were TCO semiconductors. In developing TCO lms suitable for specialized applications, Minami et al prepared Zn2In2O5MgIn2O4 (or ZnMgInO) multicomponent oxide, a new TCO semiconductor material, in 1995 [52]. Various TCO lms were prepared by MSP using combinations of ternary compound semiconductors such as MgIn2O4, ZnSnO3, GaInO3, Zn2In2O5 and In4Sn3O12. As an example, gure 11 shows and RE as functions of Zn2In2O5 content for (In4Sn3O12)1X(Zn2In2O5)X (or Zn-In-Sn-O) thin lms prepared on RT substrates by rfMSP [10]; the electrical and chemical properties changed monotonically as the composition was altered. The etching rate was measured using 0.2 M HCl solution at 25 C. It should be noted that all transparent conducting thin lms prepared on low temperature substrates using multicomponent oxides composed of combinations of ternary compound semiconductors were amorphous, regardless of the composition. It was also found that the work functions of

Transparent conducting oxide semiconductors for transparent electrodes

multicomponent oxide thin lms usually decreased as their carrier concentration was increased [4, 10]. As described above, TCO lms could always be obtained in multicomponent oxides composed of combinations, at all composition ratios, of ternary compound TCO semiconductors such as Zn2In2O5MgIn2O4, GaInO3Zn2In2O5, Zn2In2O5 In4Sn3O12, ZnSnO3In4Sn3O12, ZnSnO3Zn2In2O5 and GaInO3In4Sn3O12 [10]. It should be noted that there are many recent reports on TCO thin lms using ternary compounds as well as multicomponent oxides composed of CdO [5, 7] that are usable in rather specialized applications but not introduced here. By using TCO lms made of multicomponent oxides, the electrical, optical and chemical properties as well as the physical properties such as band-gap energy and work function can be controlled by altering the chemical composition. All the TCO developed lms discussed in this paper were prepared with metal oxides containing at least one of the following metal elements: Zn, In and Sn. In fact, it is difcult to name an important TCO semiconductor that can exhibit improved TCO lm properties other than those contained in the triangle shown in gure 1. Thus, only the materials contained in the triangle shown in gure 1 are usable for developing TCO semiconductors for practical thin-lm transparent electrodes: impurity-doped ZnO, In2O3 and SnO2 as well as multicomponent oxides consisting of combinations of ZnO, In2O3 and SnO2, including some ternary compounds existing in their systems. From the results described above, the following points should be taken into account when designing TCO semiconductors for practical use [4, 10]: The composition dependence of resistivity in multicomponent oxide lms is similar to that in metal alloys. The composition dependence of optical properties in multicomponent oxide lms is similar to that in mixed lms of dielectric materials. The chemical properties of multicomponent oxide lms are basically determined by the kind and amount of metal elements they contain. For example, the etching rate in an acid solution increased as the Zn content was increased, but it decreased as the Sn content was increased. Importantly, the obtainable properties of TCO lms using multicomponent oxides can be estimated from those of their binary and/or ternary compound TCO semiconductor constituents. However, the development of deposition techniques that can enable reliable high rate large area depositions of multicomponent oxide thin lms has yet to be achieved.

important TCO semiconductors for use as constituents of practical thin-lm transparent electrodes: impurity-doped ZnO, In2O3 and SnO2 and multicomponent oxides consisting of combinations of ZnO, In2O3 and SnO2, including some ternary compounds existing in their systems. Al- and Gadoped ZnO (AZO and GZO) semiconductors are promising as alternatives to ITO for thin-lm transparent electrode applications. In particular, AZO thin lms, composed of inexpensive and non-toxic source materials and comparable to ITO lm with its low resistivity of the order of 105 cm, can be considered the best candidate. However, further development of deposition techniques, such as magnetron sputtering or vacuum arc plasma evaporation, as well as of targets is required to enable the preparation of AZO and GZO lms on large area substrates with a high deposition rate. On the other hand, multicomponent oxides are suitable for specialized applications because their electrical, optical, chemical and physical properties such as band-gap energy and work function can be controlled by altering the chemical composition. However, both deposition technique and sintered targets have to be developed further in order to produce transparent conducting multicomponent oxide lms for specialized applications.

Acknowledgment
The author would like to thank Professor T Miyata for fruitful discussions.

References
[1] Dawar A L and Joshi J C 1984 J. Mater. Sci. 19 1 [2] Chopra K L, Major S and Pandya D K 1983 Thin Solid Films 102 1 [3] Hartnagel H L, Dawar A L, Jain A K and Jagadish C 1995 Semiconducting Transparent Thin Films (Philadelphia, PA: Institute of Physics Publishing) [4] Minami T 2000 MRS Bull. 25 38 [5] Freeman A J, Poeppelmeier K R, Mason T O, Channg R P H and Marks T J 2000 MRS Bull. 25 45 [6] Gordon R G 2000 MRS Bull. 25 52 [7] Coutts T J, Young D L and Li X 2000 MRS Bull. 25 58 [8] Minami T, Nanto H and Takata S 1982 Appl. Phys. Lett. 41 958 [9] Minami T, Nanto H and Takata S 1984 Japan. J. Appl. Phys. 23 L280 [10] Minami T 1999 J. Vac. Sci. Technol. A 17 1765 [11] Sato H, Minami T, Takata S and Yamada T 1993 Thin Solid Films 236 27 [12] Kawazoe H, Yasukawa M, Hyodo H, Kurita M, Yanagi H and Hosono H 1997 Nature 389 939 [13] Tate J, Jayaraj M K, Draeseke A D, Ulbrich T, Sleight A W, Vanaja K A, Nagarajan R, Wager J F and Hoffman R L 2002 Thin Solid Films 411 119 [14] Windisch C F, Exarhos G J, Ferris K F, Engelhard M H and Stewart D C 2001 Thin Solid Films 398399 45 [15] Suzuki S, Miyata T and Minami T 2003 J. Vac. Sci. Technol. A 21 1336 [16] Minami T, Nanto H, Shooji S and Takata S 1984 Thin Solid Films 111 167 [17] Enoki H, Nakayama T and Echigoya J 1992 Phys. Status Solidi a 129 181 [18] Unno H, Hikuma N, Omata T, Ueda N, Hashimoto T and Kawazoe H 1993 Japan. J. Appl. Phys. 32 L1260 [19] Yanagawa K, Ohki Y, Omata T, Hosono H, Ueda N and Kawazoe H 1994 Appl. Phys. Lett. 65 406 [20] Minami T, Takeda Y, Takata S and Kakumu T 1997 Thin Solid Films 308309 13

6. Conclusion
Further development of practical thin-lm transparent electrodes using polycrystalline or amorphous transparent conducting oxide (TCO) semiconductors is necessary since a stable supply of indium-tin-oxide (ITO) cannot be assured because indium is a very expensive and scarce material. In addition, recent developments in optoelectronic device applications frequently require thin-lm transparent electrodes with specialized properties. The following are the most

S43

T Minami

[21] Minami T, Sonohara H, Takata S and Sato H 1994 Japan. J. Appl. Phys. 33 L1963 [22] Minami T, Sonohara H, Kakumu T and Takata S 1995 Japan. J. Appl. Phys. 34 L971 [23] Minami T, Takata S, Kakumu T and Sonohara H 1995 Thin Solid Films 270 22 [24] Rauf I A 1996 J. Appl. Phys. 79 4057 [25] Durodija F O, Izumi H, Ishihara T and Yoshioka H 2000 Japan. J. Appl. Phys. 39 L277 [26] Suzuki A, Matsushita T, Aoki T and Yoneyama Y 2001 Japan. J. Appl. Phys. 40 L401 [27] Sawada Y, Kobayashi C, Seki S and Funakubo H 2002 Thin Solid Films 409 46 [28] Suzuki A, Matsushita T, Aoki T, Mori A and Okuda M 2002 Thin Solid Films 411 23 [29] Suzuki A, Matsushita T, Wada N, Sakamoto Y and Okuda M 1996 Japan. J. Appl. Phys. 35 L56 [30] Agura H, Suzuki A, Matsushita T, Aoki T and Okuda M 2003 Thin Solid Films 445 263 [31] Minami T, Oohashi K, Takata S, Mouri T and Ogawa N 1990 Thin Solid Films 193194 721 [32] Minami T, Suzuki S and Miyata T 2001 Mat. Res. Soc. Symp. Proc. 666 F1.3.1 [33] Tanaka H, Ihara K, Miyata T, Sato H and Minami T 2004 J. Vac. Sci. Technol. A 22 1757 [34] Minami T, Sato H, Nanto H and Takata S 1985 Japan. J. Appl. Phys. 24 L781 [35] Minami T, Miyata T and Yamamoto T 1999 J. Vac. Sci. Technol. A 17 1822 [36] Minami T, Sato H, Nanto H and Takata S 1986 Japan. J. Appl. Phys. 25 L776

[37] Sato H, Minami T, Takata S, Miyata T and Ishii M 1993 Thin Solid Films 236 14 [38] Minami T, Ida S and Miyata T 2002 Thin Solid Films 416 92 [39] Uramoto J 1979 Res. Rep. Inst. Plasma Phys. IPP 406 [40] Minami T, Sato H, Imamoto H and Takata S 1992 Japan. J. Appl. Phys. 31 L257 [41] Minami T, Yamamoto T, Toda Y and Miyata T 2000 Thin Solid Films 373 189 [42] Nishimura E, Ohkawa H, Song P K and Shigesato Y 2003 Thin Solid Films 445 235 [43] Minami T, Suzuki S and Miyata T 2001 Thin Solid Films 398399 53 [44] Minami T, Ida S, Miyata T and Minamino Y 2003 Thin Solid Films 445 268 [45] Miyata T, Minami T, Sato H and Takata S 1992 Japan. J. Appl. Phys. 31 932 [46] Minami T, Sato H, Nanto H and Takata S 1989 Thin Solid Films 176 277 [47] Nakada T, Ohkubo Y and Kunioka A 1991 Japan. J. Appl. Phys. 30 L3344 [48] Minami T, Sato H, Takata S, Ogawa N and Mouri T 1992 Japan. J. Appl. Phys. 31 L1106 [49] Hu J and Gordon R 1992 J. Appl. Phys. 71 880 [50] Minami T, Kakumu T, Takeda Y and Takata S 1996 Thin Solid Films 290291 1 [51] Phillips J M, Kwo J, Thomas G A, Carter S A, Cava R J, Hou S Y, Krajewski J J, Marshall J H, Peck W F, Rapkine D H and van Dover R B 1994 Appl. Phys. Lett. 65 115 [52] Minami T, Takata S, Kakumu T and Sonohara H 1995 Thin Solid Films 270 22

S44

You might also like