* The preview only display some random pages of manuals. You can download
full content via the form below.
Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335 www.elsevier.com/locate/pcrysgrow
Solvothermal growth of ZnO Dirk Ehrentraut a,*, Hideto Sato b, Yuji Kagamitani a, Hiroki Sato a, Akira Yoshikawa a, Tsuguo Fukuda a a
Institute of Multidisciplinary Research for Advanced Materials, Tohoku University, 2-1-1 Katahira, Aoba-ku, Sendai 980-8577, Japan b Murata Mfg. Co., Ltd., 1-10-1 Higashikotari, Nagaokakyo, Kyoto 617-8555, Japan
Abstract The growth of ZnO single crystals and crystalline films by solvothermal techniques is reviewed. Largest ZnO crystals of 3 inch in diameter are grown by a high-pressure medium-temperature hydrothermal process employing alkaline-metal mineralizer for solubility enhancement. Structural, thermal, optical and electrical properties, impurities and annealing effects as well as machining are discussed. Polyand single-crystalline ZnO films are fabricated from aqueous and non-aqueous solutions on a variety of substrates like glass, (100) silicon, a-Al2O3, Mg2AlO4, ScAlMgO4, ZnO and even some plastics at temperatures as low as 50 C and ambient air conditions. Film thickness from a few nanometers up to some tens of micrometers is achieved. Lateral epitaxial overgrowth of thick ZnO films on Mg2AlO4 from aqueous solution at 90 C was recently developed. The best crystallinity with a full-width halfmaximum from the (0002) reflection of 26 arcsec has been obtained by liquid phase epitaxy employing alkaline-metal chlorides as solvent. Doping behavior (Cu, Ga, In, Ge) and the formation of solid solutions with MgO and CdO are reported. Photoluminescence and radioluminescence are discussed. Ó 2006 Elsevier Ltd. All rights reserved. PACS: 68.35.Dv; 68.55.Jk; 68.55.Nq; 71.55.Gs; 78.55.Et; 81.05.Dz; 81.15.Lm; 81.10.Dn; 81.20.Wk Keywords: A1. Solvents; A1. Substrates; A2. Hydrothermal crystal growth; A3. Liquid phase epitaxy; B1. Zinc compounds; B2. Semiconducting IIeVI materials
* Correponding author. Fax: þ81 22 217 5102. E-mail address:
[email protected] (D. Ehrentraut). 0960-8974/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.pcrysgrow.2006.09.002
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
281
1. Introduction Over the last few years an enormous effort has been dedicated towards the film growth of ZnO in order to meet the needs of a large variety of applications [1,2]. The quality of ZnO films has significantly improved and recently a room temperature (RT) ZnO-based light-emitting diode (LED) has been demonstrated by Tsukazaki et al. [3]. Often due to the lack of high-quality ZnO substrates, most of the device structures are based on heteroepitaxial ZnO films, i.e. ZnO deposited on foreign substrates with different crystallographic and thermal properties. The resulting disadvantages in ZnO films on a-Al2O3 or GaN involve, i.e. a temperature-dependent lattice misfit that often makes buffer layers indispensable, and the out-diffusion of ions like Al and Ga from the substrate into the ZnO film [4,5]. This has accelerated the quest for large-size ZnO substrates of excellent crystallinity and low defect concentration. The growth has been carried out by chemical vapor transport (CVT) [6], pressure-melt [7], flux (Table 1) and the hydrothermal technique (Table 2). The largest high-quality crystals are currently produced by the hydrothermal technique, which now is capable of producing specimens 3 inch in diameter (i.e. perpendicular to the h0001i direction). Nearly lattice matched solid-solution films can be grown within a large area of the wurtzitetype MgOeZnOeCdO system [8], e.g. the in-plane lattice mismatch between (0001) Mg0.2Zn0.8O and (0001) ZnO is as small as 0.13% [9]. The growth of ZnO layers is dominated by technologies employing the vapor phase, i.e. pulsed laser deposition (PLD), molecular beam epitaxy (MBE), Metal-organic chemical vapor deposition (MOCVD), and physical vapor deposition (PVD). A very comprehensive overview is given by Triboulet and Perrie`re [2]. Vaporphase processing and properties of ZnO-based films has recently been reviewed by Pearton et al. [1]. Ohtomo and Tsukazaki [10] report on state-of-the-art in growth of thin films and superlattices based on ZnO by PLD. However, due to advantages like simple equipment, low temperature and ambient pressure, vicinity to the thermodynamic equilibrium, etc., growth from the liquid phase is attractive and has recently been demonstrated for different fluorides and oxides including ZnO [11e19]. Despite all efforts, fabrication of doped ZnO often remains a challenge where solvothermal (i.e. using a solvent at elevated temperature at which the solvent is in its liquid phase state) techniques may possess an advantage over vapor-phase techniques. By growth from a liquid phase the solute species are transported as metastable molecules and subsequently decomposed at a growing surface due to thermodynamic instability. The doping species is required to be incorporated in the correct valence state on either the zinc or the oxygen lattice site. In vapor growth technologies, additional activation of the doping species is therefore necessary. On the other hand, higher doping levels can be achieved by some vapor-phase growth techniques due to thermodynamically off-equilibrium deposition conditions. This review reports on the fabrication of ZnO crystals from non-aqueous solvents and, in greater detail, by the high-pressure hydrothermal technique in Section 3. However, it is regrettable that much of the initial work done in former Soviet Union laboratories is not available. The characteristics of ZnO wafers machined from hydrothermal ZnO are analyzed in Section 3. Section 4 comprises the solvothermal film growth from low-temperature aqueous solutions and by water-free liquid phase epitaxy (LPE) employing chloride solvents. The LPE films are characterized in detail. The constantly increasing field of ZnO-based nanocrystals [20] and nanostructures [21] will not be touched by this review although growth from solution is a major technology there as well.
282
Growth conditions
Results
Solute
Solvent
T ( C); DT (K)
Cooling rate (K h1)
Atm.
Seed
Crucible
Crystal size, growth rate, remarks
22 g ZnO
200 g PbF2
800e1150
1e10
Air, O2
None
Pt, 100 ml
1e5 cm Long in h11e20i direction; habit change of crystal: growth at below 1050 C yields drum-shaped crystals, platelets for the growth at 1150 C; not possible to chemically separate ZnO from the flux.
ZnO 20 mol% ZnO
PbF2 80 mol% PbF2
1000 800e1250
5.4 2e4
Air
None
Pt, 100 & 250 ml
117.3 g ZnO
92.4 g V2O5
900e1300
1.2
Air
None
Pt
42.9 g ZnO
37.3 g Zn3P2O8 4H2O þ 42.9 g V2O5
980e1330
1
5 5 3 mm3; 0.5 K cm1 DT 3 K cm1; Inclusions parallel (0001) planes; high DT of 3 K cm1 yields nucleation only on the bottom of the crucible; air jet was used; strong evaporation of PbF2 of up to 40 wt%. Pale green, clear crystals about 8 mm across; polycrystalline. Platelets, 20 mm across.
20e24 mol% V2O5 þ B2O3
1150
2e5
Air
Pt wire
Pt, 55 40 mm2
48 mol% MoO3 þ V2O5
Molten zone
0.5e1 mm h1 growth rate
None
None
ZnO: 76e80 mol%
52 mol% a b c
n.r. e Not reported. TSSG e top-seeded solution growth. TSFZ e traveling-solvent floating zone.
TSSGb; polycrystalline boule 22 mm diameter 4 mm; single crystals up to 10 5 2 mm3; 0.8% V and 0.7 % Mo contamination TSFZc; rod size 4 12; grains up to 2 2 1 mm3; 1.7% V contamination.
Refs.
[23]
[24] [25]
[26]
[27]
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
Table 1 ZnO crystals grown from non-aqueous solutions
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
283
Table 2 An overview over some conditions and results for the growth of SiO2 and ZnO from the presently largest autoclaves in production. Calculation of crystal weight, yield, and growth speed relates to a growth period of 100 days [34] Parameter
SiO2
ZnO
Autoclave I.D. Autoclave I.L. Volume Seed size Seeds per batch Weight of crystal Total yield per batch Growth speed (c-axis) 3-runs-per-year yield
0.65 m 14 m 4.6 m3 70 45 230 mm3 1400 1700 g 2300 kg 500e600 mm/day 6900 kg
0.2 ma 3 ma 0.2 m3 z50 mm diametera 112a 320 g; 20 mm thick 36 kg 200 mm/day 108 kg
a
Value may marginally differ.
2. General aspects of growth from solution In general, a solution consists of the solute and a solvent, which serves to transport the dissolved solute to the growth interface of the crystal. The chemical nature of the solvent, mainly determined by its bonding type, is a key point for chemical interaction with the solute. In some cases like for the growth of ZnO from aqueous solutions (Sections 3.2.1 and 4.1), enhancement of the low solubility by applying mineralizers is required. The ideal properties of a solvent to be applied in high-temperature solution growth have been summarized by Elwell and Scheel [22] and may as well apply to processes at low temperature: (a) high solubility for the solute, (b) appreciable change of solubility with temperature, (c) rather low melting point, (d) low vapor pressure to avoid changes in the solventesolute ratio, (e) the required crystal phase should be the only stable solid phase, (f) low dynamic viscosity in the range between 1 and 10 mPa s, (g) low reactivity with the material of the growth vessel, (h) appropriate physical density (i) the ease of separation of the grown crystal from the solvent by chemical or physical means, (j) availability in high purity at reasonable cost, and (k) low toxicity. General knowledge about solution growth is the fact that highest crystallinity can be achieved because the growth is conducted in the vicinity of the thermodynamic equilibrium. This is the point where neither growth nor dissolution of a given crystal occurs. If we increase the concentration of the solute relative to the concentration needed to saturate the solvent, i.e. to establish thermodynamic equilibrium, a supersaturation is built up. Finally, deposition of the solute on a seed crystal (¼crystal growth) is achieved upon reaching the threshold for supersaturation. Control of the supersaturation is usually made by managing the concentration of the solute species through the absolute temperature of the solution and mass transport. Control of the geometrical direction of the supersaturation to trigger crystal growth on the seed crystal can best be achieved by establishing a temperature gradient between seed crystal and its surrounding solution in the reactor. This is basically an engineering problem of designing the suitable reactor geometry. The ecological aspect of the discussed solvothermal syntheses is that this technology is environmentally benign. Solvents like water or alkaline-metal chlorides are easy to recycle. Growth temperatures are rather low and often air atmosphere is employed, which also contributes to keep production costs low as well as the ease of maintenance of the rather simple
284
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
equipment. A high throughput is made possible by using large growth vessels like those used for the production of hydrothermal quartz (Fig. 2). 3. Growth of ZnO crystal 3.1. Non-aqueous solutions Attempts have been made to grow ZnO crystals from the following molten salts: PbF2, Zn3P2O8 4H2O, V2O5, MoO3, B2O3, and mixtures thereof [23e27]. Table 1 comprises growth conditions and results. The growth temperatures are 800 C. The crystals grown from PbF2 are impossible to separate from the solidified flux [23]. Often the growth of platelets is reported [23,26], which is likely due to impurity effects caused by the solvent. This modification of the growth mechanism on some defined crystallographic directions by impurities is known from a variety of mainly solution-grown crystals. It was explained by a substantial change of the average binding forces operating along the surface between particles of the surface layer due to adsorbed impurities [28]. The pale green color of the crystals in Ref. [26] is likely due to impurities derived from the solvent. Also, inclusions parallel to (0001) planes have been reported [25]. Temperatures higher than 1300 C led to strong evaporation of PbF2. The use of a mixture of Zn3P2O8 4H2O and V2O5 by Wanklyn [26] gained 20 mm large platelets. It was reported that the mixtures did not adhere to the crystals. Separation of the liquid solution from the grown crystals was carried out by pouring it off the crystals. All crystals were singlephase ZnO. The resistivity was measured as 0.3 U cm. This is comparable to values obtained from pressure-melt grown ZnO, compare Fig. 19c, and hints to higher impurity levels. The top-seeded solution growth (TSSG) and traveling-solvent floating zone (TSFZ) techniques were applied to grow larger crystals from mixtures of ZnO with V2O5 þ B2O3 and MoO3 þ V2O5 [27]. TSSG comprises crystal growth on a seed immersed in a solution. In order to control the temperature field around the growing crystal, the seed crystal is simultaneously rotated and pulled from the solution at approximately at the rate at which the crystal is growing. In TSSG of ZnO a platinum wire served for seeding and polycrystalline aggregates with singlecrystalline regions of 10 5 2 mm3 were obtained. The crystal pulling and solution cooling rates were 0.5e1 mm h1 and 2e5 C, respectively. Crystal rotation rate was 20 rpm. Growth by TSFZ involves providing a feeding rod and a seed rod concentrically arranged to give a point of contact between both rods. At this contact point, the feeding rod is melted by an external heater. Successively, this molten zone is caused to travel toward the end of the feeding rod by moving the heater. After a first melting at a travel speed of the solution zone of 15e20 mm h1, the growth was carried out at rates of 0.5e1 mm h1. Feedstock and growing crystal were rotated at about 20 rpm in opposite directions. Grains up to 2 2 1 mm3 were obtained. In either case, however, a large amount of metal impurities up to 1.7% V was measured and contamination from the solvent was considered to be a serious issue [27]. Our own experiences with non-aqueous solution systems comprise the system PbOeZnO and employment of Zn3(PO4)2eLi2O. Whereas the first case showed the above-mentioned problem of ZnO separation from the PbO solvent after the growth and was therefore excluded from further investigation, the latter system was tried for a while. Following reaction path was used: Zn3 ðPO4 Þ2 þ 3Li2 O/3ZnO þ 2Li3 PO4 :
ð1Þ
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
285
The formation of phase-pure wurtzite ZnO at temperatures below 1000 C was confirmed by XRD measurements. However, the whole solution quickly solidified after completion of reaction (1). The melting point of Li3PO4 (1205 C [29]) is considerably higher than of Zn3(PO4)2 (900 C [29]), which caused the solidification. Attempts to seek a solvent for Li3PO4 were made among some phosphates like K4P2O7 and KPO3, but the experiments did not lead to satisfying results. The path with K4P2O7 would form KZnPO4, which is a competing phase to ZnO. The use of KPO3 would result K3PO4 þ KPO3, which would not solve the problem of lowering the melting point. In summary, the growth from non-aqueous solutions did not lead to the production of large ZnO crystals of low-impurity concentration and high crystallinity. Reasons why this route failed can be summarized as: (a) a proper solvent was not available, which would keep the solvent-derived impurity level sufficiently low. (b) The separation of grown ZnO from the solvent after the growth process has finished remained an unsatisfactory issue in most cases. (c) The hydrothermal growth of ZnO was tried at the same time and turned out to be more competitive in terms of crystal size and quality. Supporting is the fact that much knowledge on the hydrothermal growth technology has been collected during the development of the hydrothermal growth of quartz (a-SiO2). This dates back to Spezia in the year 1909 [30] who was first to report valuable results on seeded growth of quartz under hydrothermal conditions. He already applied a temperature gradient between the zone for dissolving the feedstock and the growth zone. However, the story of industrial growth of quartz dates back to the 1940s. In 1949, i.e. Buehler and Walker [31] reported the hydrothermal growth of twin-free quartz. In 1953, Walker [32] published the hydrothermal synthesis of quartz crystals weighing over 1 lb (453.6 metric gram), grown at the Bell Telephone Labs. in periods of less than two months. The same paper illustrates two crystals grown over 42 days to a weight of 540 g each. The temperaturedifference method and quartz nutrient was employed in autoclaves sizing 10 cm internal diameter and 122 cm internal length. Walker already pointed out in his paper that crystals grown from solutions are likely to be of more perfect quality than those grown from melts. A comprehensive review on the history of quartz crystal growth was recently given by Iwasaki and Iwasaki [33]. Nowadays, the hydrothermal growth of quartz is a routine. In 2004, about 1850 tons were produced worldwide with some 700 tons in Japan [34]. Table 2 compares the growth of quartz and ZnO from, to our knowledge, the largest autoclaves at present. The impressive amount of 2300 kg quartz from 1400 crystals can be produced over a 100-day growth run. The same figures for ZnO, 36 kg and about 100 crystals, are still not comparable with quartz, but really show the potential of the hydrothermal technology in terms of throughput. However, the driving force for scale-up of the hydrothermal growth of ZnO comes from the demand by industry. The hydrothermal growth of ZnO and characteristics of the crystal will be treated in greater detail in the following section. 3.2. Hydrothermal growth of ZnO 3.2.1. General characteristics of the method The term ‘‘hydrothermal’’ is derived from geology [35]. Hydrothermal growth comprises the use of aqueous solvents and mineralizers under elevated temperature and pressure in order to dissolve and recrystallize materials, which are barely soluble under ordinary conditions. Mineralizers are particularly important since they serve to establish a suitable solubility of the solute because most of the species are rather poorly soluable in water.
286
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
A large variety of crystals have been grown with a-quartz (SiO2) being the most prominent one and most of the technological developments in the hydrothermal technique are closely related with the development of the a-quartz technology. A detailed description on the hydrothermal technology with examples for many crystals is given by K. Byrappa [35]. Features of the hydrothermal method comprise: (a) use of a closed high-pressure growth vessel (autoclave); (b) use of a solvent; (c) use of solubility increasing mineralizers; (d) employment of a precursor; (e) employment of seed crystals; (f) a temperature gradient between the precursor-containing dissolution zone and the growth zone with the seed crystals; (g) DT z 0 at the interface between the growing crystal and the solution, which is why the concentration of structural crystal defects is smaller than for melt-grown crystals; (h) saturation of the solute while the seed crystal is already in contact with the undersaturated solution. The hydrothermal growth of ZnO requires the use of water in its supercritical state, which is achieved at its critical point with a critical temperature and pressure, Tc ¼ 374 K and pc ¼ 22.1 MPa, respectively. Fig. 1 shows the simplified peT diagram for water at constant volume. The existence region of supercritical water (SCW) covers the upper right hand side above the critical point. The peT region at which large hydrothermal ZnO crystals are grown is represented by the rectangle, which stretches between the pressure of 70e255 MPa at temperatures of 300e430 C, see also Table 3. SCW is characterized by an enhanced acidity, reduced density (0.05e0.2 g cm3) and lower polarity in comparison to water under normal pressure and temperature. In fact, SCW is an almost non-polar fluid due to reduced dielectric strength to values of 1e3 (water under ambient conditions shows 78) [36]. The diffusivity of SCW is strongly increased as well as the miscibility with gases. SCW exhibits reduced molecular ordering with less effective hydrogen bonding [37]. The enhanced acidity favors ionic processes, such as the dissolution of ZnO. However, the solubility of ZnO in SCW remains insufficient and makes use of mineralizers necessary. Mineralizers serve to increase the solubility of ZnO in SCW by forming metastable compounds between them and ZnO, which later decompose at a growing crystal face to deliver Zn2þ and O2 which are incorporated as the ZnO lattice. Typical mineralizers are LiOH, NaOH, KOH, Li2CO3, and H2O2, see also Table 3. The best solvents for ZnO are a mixture
Fig. 1. Schematic pressureetemperature diagram for water at the condition of constant volume.
Table 3 Growth conditions of hydrothermal ZnO crystals Growth conditions
Autoclave
Result
Refs.
Mineralizer
T ( C); DT (K)
P (MPa)
Seed
Filling (%)
Crucible lining
Baffle
Crystal size, growth rate
ZnO
KOH, LiOH, NaOH, NH4OH 3 M KOH þ 1 M LiOH þ 0.1 M H2O2 3 N NaOH þ 1 N KOH þ 0.5 N Li2CO3
230e300; 5e75
50e350
n.r.a
n.r.
n.r.
70e100
n.r.
Pt
Pt
Prior growth seeds were etched in HCl and NaOH solution. 10 mm B after 14 days
[38]
370e415; 10
different orientations (0001)
345e355; 10
n.r.
80
Pt
n.r.
n.r.
[42]
70e85
n.r.
n.r.
82e87; Ag best at 83
5%
(0001): 0.25e0.38 mm/day; [43] ð1011Þ: 0.12e0.25 mm/day (0001): 0.35 mm/day [44] (10-day growth run) and 0.3 mm/day (30-day growth run)
83e85 79e86.1 83
Ag
5%
Pt
n.r.
Sintered ZnO, 1 B 3 mm3 n.r.
(0001) 20 mm2 0.5 mm 145e255 (0001), ð1011Þ (0001)
n.r.
1e2 M NaOH
387e430; 8e30
ZnO (grain size larger U.S. Standard Sieve Size No. 10) Hydrothermal recrystallized ZnO Pressed and sintered ZnO pellets
5.1 N, i.e. 5.45 molal KOH þ 0.7 molal LiOH
340e385 (best at 353); 14
5.1 M KOH 5.1 M KOH þ 2 M LiOH 6 M KOH þ 1 M LiOH
340e350; 10e15 55 (0001) 386e428; 25e30 116e169 Nutrient 365; 10 n.r. (0001) 7 B mm2
Pressed and sintered pellets, 1 B 3 mm3 Sintered powder 99.99% ZnO Sintered powder 99.99%, particle size < 3 mm
3 M KOH þ 1e2 M LiOH 360e380; z10e25
z100
(0001)
n.r.
4 M KOH þ 4 M NaOH þ Nutrient 355; 10 Li2CO3 3 M NaOH þ 1 M KOH þ 350e365; 15 1 M LiOH; 20 vol% N H4OH þ 475e490; 15 80 vol% 5 M KOH
n.r.
(0001)
80
Ag Ag, ID 35 mm 5% L 350 mm Pt n.r.
20
(0001)
75
Pt
136
70
n.r.
[40,41]
Average {0001} z 0.25 mm/day [45] 0.53e0.75 mm/day [46] 15 15 8 mm3; (0001): 0.45 mm/day; ð0001Þ: 0.22 mm/day; Perpendicular to c: 0.35 mm/day (0001) and ð1011Þ: [47] z0.2 mm/day (0001): averaged 0.25 mm/day for 30 days growth run Nitrogen doping up to 8 1018 atom cm3 confirmed by inert gas fusion analysis
[48] [49]
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
Precursor
(continued on next page) 287
288
Table 3 (continued ) Growth conditions
Autoclave
Result
Mineralizer
T ( C); DT (K)
P (MPa)
Seed
Filling (%)
Crucible lining
Baffle
Sintered powder
3e5 M KOH þ 20 wt% MgO (99.999%)
675e650; 25
102.5e 123.7
Single crystals used
45e60
Pt ID 14 mm, vol 16e 33 cm3
n.r.
Fine crystalline ZnO powder
5e5.4 M KOH þ 0.8e1.2 M LiOH
250e300; 20e80 15e140
(0001), ð1010Þ, ð1011Þ
n.r.
Ag, 500 ml
300e400
(0001)
n.r.
Pt
Sintered ZnO 3 M KOH þ 1 M LiOH poly-crystals Sintered ZnO, LiOH, NaOH, KOH 99.99%, granules, 1e6 mm Zn(OAc)2 2H2Ob, 80 ml KOH Co(OAc)2 4H2Oc Zn(OAc)2 2H2O
a b c d e f
DTPAd, EDTAe, TEPf þ KOH
n.r. e not reported. Zn(OAc)2 2H2O e zinc acetate. Co(OAc)2 4H2O e cobalt (II) acetate. DTPA e diethylenetriaminepentaacetic acid. EDTA e ethylenediaminetetraacetic acid. TEP e tetraethylenepentamine.
80e100
Crystal size, growth rate
Up to 5.5 mol% Mg in the ZnO crystal; ZnO film formed on the surface of the crystals during cooling to room temperature. 7e15% (0001): max. 2.053 mm/day; ð0001Þ: max. 1.097 mm/day; Both for Li-free solvent (5.15 M KOH) Pt 50 50 15 mm3
[50]
[51]
[52,53]
300e400; 10e15 100e 150
VP- and n.r. hydrothermal
ID 50 mm, 5e6% L 600 mm
n.r.
[54]
240
n.r.
n.r.
n.r.
n.r.
16 h growth time, precipitates
[55]
200
n.r.
None
n.r.
Teflon, 100 ml volume Teflon
Not used
Needles, 1 mm in (0001) from EDTA; duration 7e11 days
[39]
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
Precursor
Refs.
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
289
of KOH and LiOH. Pure KOH gives rise to a high growth rate which leads to low crystal quality and the growth process is hard to control. By contrast, LiOH alone is too weak to significantly increase the solubility of ZnO in SCW [38]. Diethylenetriaminepentaacetic acid (DTPA), ethylenediaminetetraacetic acid (EDTA) and tetraethylenepentamine (TEP) was employed to grow ZnO by a low-temperature metal-organic approach [39]. In this case, however, the crystal size did not exceed 1 mm. Larger ZnO crystals grown in this way have not yet been reported. However, the very low growth temperature is an interesting aspect there. In Table 3 are listed growth conditions and results obtained by several groups over more than four decades or so. The chemical reaction of ZnO and H2O to form ZnO H2O has already been investigated by Roth and Chall [56] and a phase diagram for the temperature interval 280e 320 K was reported as early as 1933 by Hu¨ttig and Mo¨ldner [57]. However, first growth of larger ZnO crystals was published in the 1960s by the group of Kolb and Laudise [43e45,58e60]. Disadvantages of the method are the inability to directly manipulate during a growth run as well as the lack of visual observation. Temperature changes driven by the heaters assembled around the autoclave have a long time constant; this prevents sharp temperature changes in the solution. Therefore, quasi steady-state conditions are applied to hydrothermal crystal production. 3.2.2. Technology for ZnO Research on the hydrothermal growth of ZnO was carried out mainly by using Morey and Tuttle type autoclaves [35,44,61,62]. The large capacity autoclaves used for the production of quartz (Fig. 2) and now for ZnO [63] are basically modified Bridgman autoclaves. A good overview over the different autoclave types and their specifications is given by Byrappa [35].
Fig. 2. (a) Quartz crystals being withdrawn from the autoclave of 0.65 m inner diameter and 14 m length; (b) crystal holder with about 25 cm large quartz crystals.
290
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
Laudise et al. [44] reported on six parameters to be important in the hydrothermal growth of ZnO: (1) a high concentration of alkaline mineralizers is required to achieve a proper solubility of ZnO [64] in SCW while keeping the dynamic viscosity of the solution sufficiently low. (2) The temperature difference DT between the dissolution and growth zones has to be such as DT 20 K and control of DT within 3 K is inevitable to suppress the tendency toward spontaneous nucleation and flawed growth in ZnO. This rate-limiting process was explained in terms of diffusion caused by a high concentration of mineralizers and their less effective mixing. (3) Wall nucleation, spontaneous nucleation and flawed growth were reduced by a two-step warm-up process. Firstly, the long autoclaves were slowly heated over a period of 24 h. Secondly, further heating was maintained by keeping the conditions of DT 25 K at 150 C and DT 20 K above 250 C. (4) Liþ containing mineralizers prevented the formation of growth hillocks in ZnO, i.e. free of flaws. (5) Seeds often possessed a strongly damaged surface as result of mechanical machining, see also Section 3.2.4. Good-quality ZnO was grown when a surface depth of 50e70 mm was etched away from (0001) oriented ZnO seeds. (6) Size of the precursor: very small particle size resulted in low growth rates and flawed growth. Best results were found when the precursor was larger than U.S. Standard Sieve Size No. 10. The use of hydrothermal grown lumps about 6 mm in size did not yield an improvement in either growth rate or quality. These experimental findings on the optimum precursor size were recently confirmed by Chen et al. [65] using numerical simulation to analyze the ammonothermal growth of GaN [66], which is very much inspired by the hydrothermal growth of ZnO. The flow of solution in the precursor stock and the temperature distribution were optimum for a precursor particle size of 3 mm. In this case, significant convective effects were seen in the precursor stock and the flow was highly three-dimensional. When the particle size was reduced to 0.6 mm, a very weak flow was obtained inside the precursor stock and the temperature distribution was controlled by thermal conduction (conduction mode). Dissolution of the precursor and mixing of the precursor-enriched solvent with the solvent above the precursor stock was therefore poor [65]. On the other hand, by increasing the particle size the surface/volume ratio becomes significantly smaller, i.e. the surface available for dissolution by the solvent decreases. Simultaneously, the temperature distribution is controlled by the solution flow (flow mode). A basic requirement for the growth from autoclaves is the need for a constant mass flow circulating from the feedstock to the seed crystal. This is established by proper heater arrangements around the autoclave. The inner part contains a baffle, which separates the feedstock (saturation zone; bottom of the autoclave) from the seed crystals (growth zone; upper fraction in the autoclave). The parameters for the baffle are the surface of the bore holes (opening typically around 5e15%) and their geometrical arrangement. If necessary, even several baffles could be used to get better control of the mass flow. Unfortunately, details on the inner construction of autoclaves ready to produce large ZnO crystals cannot be disclosed at present as they are the key to successful production of high-quality ZnO crystals and therefore intellectual properties of the ZnO-producing companies. The same is true for the growth process. The filling of an autoclave describes the amount of liquid solution which is introduced into the autoclave. This value spreads from 70 to 90 vol% as listed in Table 3. A high filling generates a higher internal pressure but provides a greater amount of solvent to dissolve more mineralizer and consequently more ZnO species. We worked with 65e80 vol% in case of our small autoclaves (16 mm inner diameter and 200 mm inner length) to achieve appreciable growth rates.
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
291
A closed Pt inner container [41,52] was found essential to prevent corrosion of the autoclave inner walls due to the effect of the relatively higher basicity of the growth solution than that used for artificial quartz [40e42,46,48,52,53]. In many former works Ag was used [44,45,47,61], which turned out to be less resistive against corrosion than Pt. In our work we used a Pt inner crucible throughout [52]. The volume between autoclave and Pt inner container was filled with a suitable amount of distilled water for pressure balancing to prevent the Pt inner container from serious deformation. The Pt inner container is filled up to 80 vol% filling with the already prepared homogeneous solution of water, mineralizer and ZnO feedstock. Seed holder and baffle are inside the Pt inner container and are both made of Pt. Fig. 3 shows the photograph of the Pt seed holder used for growth processes in autoclaves of 50 mm inner diameter and about 1 m inner length. However, this type of seed holder has the disadvantage of a relatively large surface area in relation to the volume of the inner Pt crucible. Often, parasitic nucleation can be found after the experiment. A general rule is therefore to decrease the size of the seed holder as much as possible to diminish the surface available for parasitic nucleation while the mechanical stability is still guaranteed. What we typically observe for small size autoclaves (16 mm inner diameter and 200 mm inner length) is the negative effect of seed holder and crystal on the mass flow. Sometimes crystals are not well-shaped, i.e. some facets were not fully developed due to effective shading and localized turbulence. This problem can be reduced by reducing the number of seed crystals; consequently, the seed holder would be smaller as well.
Fig. 3. Platinum seed holder as used for the autoclave of 50 mm inner diameter. The largest ZnO crystal (center) is about 25 mm across c-plane.
292
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
The above-mentioned trend gets less severe with larger autoclave size as shown in Fig. 3. The larger crystal (center) is about 1 inch in diameter across (0001) plane and has developed excellent facets like most of the other crystals from this experiment. The growth of 2 and 3 inch size crystals required to increase the inner diameter to 200 mm (length 3 m) and 300 mm, respectively. Fig. 4 illustrates the result from one growth run. Almost 100 specimens 2 inch in size have been grown on mainly (0001) and ð1010Þ oriented seeds. All crystals developed excellent facets and the average crystal thickness was about 1 cm in the h0001i direction. A ZnO crystal 3 inch in diameter and a wafer processed from it are shown in Fig. 5a and b, respectively. The crystal is about 1 cm in thickness. The light yellow coloration is due to impurities from the ð0001Þ face, see Section 3.2.3 on impurity analysis. To date, these are the largest ZnO single crystals grown by the hydrothermal method. The growth mechanism of ZnO under hydrothermal conditions has already been explained by Laudise et al. [44], Khodakovsky and Elkin [67] and Demianets and Kostomarov [51]. The 2 main zinc species in the solution are ZnOOH, Zn(OH)2 4 , ZnO2 and their concentrations de pend on the OH concentration (i.e. pH value) and temperature. By far the major quantity of the OH concentration derives from the basic mineralizers, but also the autoprotolysis of water contributes according to: 2H2 O4H3 Oþ þ OH
ð2Þ
The driving force to foster the growth process comes from the high concentration of 2 ZnOOH, Zn(OH)2 4 , ZnO2 in the solution, which builds-up the supersaturation necessary to trigger nucleation on the ZnO seed crystal. Possible reactions leading to the growth of ZnO comprise: ZnðOHÞ2 4 /ZnðOHÞ2 þ 2OH
ð3Þ
ZnðOHÞ2 /2Hþ þ ZnO2 2 :
ð4Þ
and
Fig. 4. Two-inch size ZnO crystals produced during one growth run.
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
293
Fig. 5. View down c-axis (a) of a 3-inch ZnO crystal, (b) 3 inch (0001) ZnO wafer with CMP finish. Scale bar is 80 mm.
From a comparison of the constant of reaction K: ð5Þ
K ¼ Kp =Kr with Kp and Kr the constant of reaction of products and reactants, respectively, we get: 2
K ¼ ½Hþ ½ZnO2 2 =½ZnðOHÞ2
ð6Þ
and 2
2 ½ZnO2 2 =½ZnðOHÞ2 ¼ Kr ½OH =Kw ;
ð7Þ
with Kw the ionic product of water, it was shown that an increase of OH in the solution yields an increase of ZnO2 2 [51]. The growth on the (0001) and ð0001Þ, i.e. Zn- and O-terminated faces or Znþ (surface) and O2þ (surface), respectively, likely involve: Znþ ðsurfaceÞ þ ZnO2 2 /2ZnOðcrystalÞ; ð0001Þ
ð8Þ
O2 ðsurfaceÞ þ ZnO2 2 þ 2H2 O/4OH þ ZnOðcrystalÞ; ð0001Þ:
ð9Þ
Znþ (surface) and O2þ (surface) describe those species that are provided by the crystal surface. The source for ZnO2 2 and H2O is the hydrothermal solution itself. In summary of reactions (2)e(9), during a ZnO growth experiment one might have the following flow of the Zn-containing species from the precursor to the crystal: 2
ZnOðprecursorÞ þ 4OH /ZnðOHÞ4 þ O2 /ZnðOHÞ2 þ 2OH þ /ZnO2 2 þ 2H /ZnOðcrystalÞ þ 2e
ð10Þ
The growth speed is two times faster on the (0001) face than on the ð0001Þ face, and has been explained by the ratio of formed ZnO units, ZnO (crystal), as obtained from formulae (8) and (9) [51]. This is close to the reported values of 2e3 [41,43,46,48,52]. Similarly, the growth speed is about two times faster on the (0001) face than on the ð1011Þ face [43]. The presence of Liþ slows down the growth speed in the h0001i direction and slightly increases it for the h1100i direction [44,51]. This is considered to be related to a decreased
294
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
positive surface charge that lowers the probability of incorporating Zn-containing negativelycharged species. A high growth rate of 2.053 and 1.097 mm/day for the (0001) face and ð0001Þ face, respectively, was reported by Demianets and Kostomarov [51], see Table 3. KOH was solely used as mineralizer but is known to reduce the crystal quality [38]. However, already from the old work of Laudise et al. [44] it is known that Liþ is needed to reduce the number of crystal defects (see also Section 3.2.1). This was confirmed by our work. We found that a molar ratio of 3:1 of KOH:LiOH delivers the best quality, i.e. highly transparent crystals with a very small X-ray rocking curve full-width half-maximum (FWHM, see Section 3.2.3). The above discussion shows the general dilemma with mineralizers: they are indispensable in establishing an effective solubility of the precursor. On the other hand, they are unwelcome for the process yield in crystal growth. This is intrinsic to the hydrothermal growth of ZnO and can only be compromised as shown above. Typical growth rates of high-quality ZnO crystals occur in the range up to a maximum of 0.3 mm/day (Table 3). This is comparatively slow compared with the pressurized melt growth of ZnO, where up to 10 mm h1 can be achieved [7]. Two-inch size (0001) ZnO produced by seeded chemical vapor transport (SCVT) grew at growth rates of <70 mm h1 [67]. Nevertheless, the high crystallinity and throughput in large autoclaves clearly indicates the favourable commercial potential of hydrothermal growth technology. Thus, the use of large-size autoclaves for ZnO like those used for the growth of quartz (Fig. 2, Table 2) would produce sufficient ZnO wafers for a variety of applications. 3.2.3. Characteristics of the material 3.2.3.1. Crystallinity, defects. The quality of (0001) substrates was investigated by rocking curve measurements using the (0002) reflection. An RINT-2000 (Rigaku) diffractometer with CuKa radiation was employed in combination with a four crystal Ge (220) channel monochromator, beam divergence 12 arcsec, scan speed 0.01 min1, step width 104 . The FWHM ranged between 19 and 30 arcsec after chemicalemechanical polishing (CMP). This is better than values reported for ZnO wafer machined from pressure-melt grown ZnO, 49 arcsec [7,69], and comparable to CVT ZnO, FWHM around 30 arcsec [70]. Other groups reported FWHMs of 43 and 37 arcsec for polished (0001) and ð0001Þ surfaces of hydrothermal ZnO, respectively [48], and 57 arcsec [49] for slightly N-doped hydrothermal ZnO. Contact-mode AFM measurements taken under air conditions reveal a root mean square (RMS) roughness of 0.285 nm (Fig. 6a) and 0.155 nm (Fig. 6b). The X-ray reciprocal space map using the (0002) symmetric reflection (Fig. 7) shows a highly-symmetric single peak with the FWHM from the u scan of 15 arcsec. This value was confirmed by Wenisch et al. [71] who reported about 12e15 arcsec by triple-axis ue2q scans. X-ray topography (CuKa radiation, 40 kV, 10 mA, detected by a film IX80; BergeBarrett geometry) was measured on a (0001) ZnO wafer 2 inch in size and 500 mm in thickness. The XRC FWHM from the (0002) reflection of the sample was 19 arcsec. Two thousand five hundred scans have been assembled to yield the contrast-enhanced image of Fig. 8. The (114) reflection at 2q ¼ 98.6 and u ¼ 49.3 was employed. The wafer appears very homogeneous over the entire area. Slight contrast effects are seen, which is presumed to be due to slightly different lattice parameters caused by fluctuations in the impurity concentration or stoichiometry [46]. The lower part in Fig. 8 shows a few lines with a surface of apparently irregular crystallographic orientation. Identification of the defect is thus impossible and more work is
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
295
Fig. 6. AFM images of ZnO wafers with different surface finish: (a) (0001) face after CMP, (b) ð0001Þ face after CMP, (c) (0001) after annealing at 1050 C and (d) (0001) after annealing at 1100 C. Annealing time was 120 min; oxygen flow rate was 200 ml min1.
still required. Croxall et al. [46] reported on dislocations as the principal imperfections in their hydrothermal ZnO crystals. They lie in the basal plane and run parallel to or 20 inclined to the growth direction of h1010i. The point defects at the interface of the seed-grown crystal initiated dislocations. The etch pit density (EPD) was determined for (0001) and ð0001Þ wafers. Fig. 9 shows the result from etching with an aqueous solution of concentrated H3PO4 for 5 min at 25 C. The etching behavior is strikingly different as displayed by the shape of the etch figures. While those on the (0001) face clearly exhibit a 6-fold axis with pyramidal facets, those on the ð0001Þ face are less facetted. The EPD was about 300 cm2 after chemicalemechanical polishing and was further lowered to less than 80 cm2 by annealing [53]. The etch rate in the annealed wafers was low, which was related to the improved crystallinity. An aqueous solution of 0.7% HCl was applied for 5 min at 60 C. Sakagami et al. [40] report an EPD of 100 and 103 cm2 on (0001) and ð0001Þ faces, respectively, using an aqueous solution of H3PO4. Dislocations were responsible for the higher EPD on the ð0001Þ face.
296
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
Fig. 7. The (002) reflection from the X-ray reciprocal space mapping of a (0001) ZnO wafer.
Etching with 1% HCl yields an etch rate of 10 mm min1 according to another report [72] and a mixture of 1 ml:1 ml:10 ml of conc. H3PO4:conc. acetic acid:H2O etched at 1.5 mm min1. The use of 30% HNO3 aqueous solution produces hexagonal pyramids on the (0001) face of mechanically polished ZnO [73]. 3.2.3.2. Impurities. The incorporation of impurities and non-radiative recombination centers strongly depends on growth sectors [41,42,44,46,48], which are characteristic of the
Fig. 8. Transmission X-ray topography of a polished high-quality ZnO wafer of 2 inch in size. The image is contrast enhanced.
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
297
Fig. 9. Etch pits on the (0001) and ð0001Þ surface of a polished ZnO wafer.
hydrothermal growth of ZnO. Growth sectors are formed due to differences in growth rate and mechanism for faces with different crystallographic orientations (see also Section 3.2.2). Fig. 10 is the result from the observation of a (0001) wafer by fluorescence microscopy (lexc ¼ 365 nm) showing different intensities from the pyramidal, prismatic and basal sector. The intensities decrease from the prismatic to the pyramidal to the basal sector, which also was observed in the broad emission band at <2.8 eV in spectra derived by cathodoluminescence (CL) by Mass et al. [42]. It was pointed out that samples cut from the crystal volume directly above the (0001) seed did not show other than that found in the low-impurity basal sector. Results from XRC measurements on 2 inch wafers using the (0002) reflection support these findings. The FWHM is at 18e22 and 50 arcsec for that part from only the basal sector and for both prismatic and pyramidal sectors, respectively [63]. Consequently, this is used for the production of large-size ZnO wafers by Tokyo Denpa (TEW, Fig. 11). Little random strain was observed under crossed Nicols, which also supports the result from X-ray topography that the homogeneity of the wafer is very high.
Fig. 10. Different growth sectors as revealed by fluorescence microscopy using excitation with lexc ¼ 365 nm. The image shows the view through a (0001) wafer displaying pyramidal, prismatic and basal sector.
298
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
Fig. 11. Little random strain is revealed by optical observation under crossed Nicols of a 2-inch ZnO wafer after CMP.
Secondary ion mass spectrometry (SIMS) was measured to study the impurity distribution in the depth of a wafer (Fig. 12). The primary beam species was Csþ (5 kV, 350 nA); sputter speed 120e150 nm min1. As can be seen from Fig. 12 the impurity levels remain constant with increasing scan depth. The strong increase in intensity for all impurities measured over the last <100 nm is knowingly an artifact of SIMS. However, a more sensitive investigation on impurity concentrations in 2 inch ZnO from TEW has been made using inductively coupled plasma mass spectrometry (ICP-MS) [52,53]. In Fig. 13 is shown the concentration of Fe, Al, Li, K and its position dependence. A negative
Fig. 12. SIMS depth scans of impurities in a ZnO wafer with CMP surface finish.
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
299
Fig. 13. Concentration analysis of Fe, Al, Li, K as detected by ICP-MS in seven wafers cut from one crystal.
wafer number (3 to 1) indicates those specimens grown and cut from the ð0001Þ face and positive ones (1e4) refer to specimens from the (0001) face of the seed crystal. Increasing distance from the ð0001Þ face of the seed generates less impurity in the grown crystal. This is particularly obvious for Li with 12 and <1 ppm for wafer numbers 1 and 3, respectively. The concentration of K remains unchanged for both faces at <0.3 ppm. Both Fe and Al show higher concentrations in wafers grown on the ð0001Þ face of the seed crystal [52,53]. The concentration numbers are <11 and <1 ppm for Fe and <8 and <0.5 ppm for Al for the ð0001Þ face and (0001) face, respectively. Mass et al. [42] report about 1e10 ppm of Li, K and Na not above 1e2 ppm and Al, Fe, Si, C from 1 to 10 ppm. The result from the examination of the ZnO precursor chunks and 2 inch ZnO crystals by glow discharge mass spectrometry (GDMS) is summarized in Table 4. Argon was used as a discharge gas. The result for the precursor was measured at the end of the year 2001, that for the crystals in early 2005. The purity of the precursor has been improved during this period. The Pt concentration in the crystals, however, is clearly assigned to the inner liner made of Pt. A ZnO crystal grown by CVT using H2 and N2 as carrier gas [6] incorporates 0.03e0.1 ppm K and 0.2e1 ppm Na as recorded by AAS. Fe and Al were found at concentrations below the detection limit of 1 ppm and 7 ppm, respectively. The group of Triboulet et al. [70] uses Cl2 and C as transport agents in CVT and have detected about 0.053% of Cl and 0.05% of C when using graphitized ampoules. For comparison, pressure-melt grown ZnO [69] contains 4 ppm Fe, 8 ppm Pb and 2.5 ppm Cd. Hydrogen-related defects in hydrothermal, as-grown ZnO (1017 cm3 carrier concentration at room temperature, see also Fig. 19a) have been studied by infrared absorption spectroscopy [74]. A number of lines are located in the vicinity of the characteristic OeH stretch local vibrational modes and the line at 3577.3 cm1 was related to a defect containing one OeH bond primarily aligned with the c-axis of the crystal. The presence of a substitutional Ni atom in the defect was tentatively proposed. The same absorption line of 3577.3 cm1 at 12 K, and shifting to 3547 cm1 at 300 K, was characterized by electron paramagnetic resonance (EPR) in a hydrothermal ZnO sample [75]. This line, however, was assigned to an OH ion located on an oxygen site adjacent to an Liþ
300
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
Table 4 Impurity analysis of 2-inch size ZnO crystals and ZnO precursor chunks Element
Li Na K Rb Cs Mg Ca Ti V Cr Mn Fe Co Ni Cu Pb Cd Al Ga Si Ge As Sb Pt Lanthanides
Concentration (ppm wt)
Remarks
Crystal
Chunks
1.3 0.11 <0.1 <0.01 <0.05 0.04 <0.1 <0.005 <0.005 <0.05 <0.05 0.06 <0.01 <0.01 <0.01 0.07 <0.1 0.02 <0.05 0.18 <0.05 <0.05 <0.5 7.1 <0.05
19 0.13 6.5 e e 0.25 e e <0.005 0.15 e 3.7 e <0.01 <0.01 e <0.03 3.4 0.09 1.9 <0.05 <0.05 e <0.02 e
From mineralizer From mineralizer
From Pt liner Pm was not measured
ion on the zinc site. The concentration of the neutral complex LiOH was estimated to be 7.6 1017 cm3. Charge compensation by singly ionized acceptors was concluded to play a major role for hydrogen in ZnO. A band at 3546 cm1 at 300 K [64] has already been suspected to be caused by hydrogen located in a bond-centered position between oxygen and zinc. This band might contain some unresolved components from OH involving an acceptor on the zinc site. Furthermore, detection of Co2þ at 6005 cm1 and Ni2þ at 4240 cm1 was reported. Evidence of Fe3þ and Mn2þ was also found in the same hydrothermal ZnO sample [75]. Photo-induced EPR measurements revealed a signal due to neutral Li acceptors. The concentration of photoinduced neutral Li acceptors was approximately 1.3 1015 cm3. Recently, Demianets et al. [38] reported on the modification of the crystal habit upon incorporation of some bi- and tri-valent metal ions like Ni2þ, Cd2þ, Mn2þ, Co2þ, Fe2þ, Fe3þ and In3þ. The ions Ni2þ, Cd2þ, Mn2þ, Co2þ and In3þ substitute for the Zn site in the crystal lattice and Ni2þ, Mn2þ, Fe2þ and Fe3þ were said to cause crystal coloration. In the case of In3þ the formation of a point defect of the type Inþ Zn was found to be a shallow donor. This effect was recently used in combination with Li doping to achieve super-fast luminescence decay by donoreacceptor pair recombination [97]. Due to the relatively low growth temperature under 0 hydrothermal conditions, formation of Inþ Zn LiZn type associates does not happen. 3.2.3.3. Thermal properties. The linear thermal expansion coefficient DlL/L, specific heat cp and thermal diffusivity a of hydrothermal grown ZnO have been evaluated for samples
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
301
prepared from hydrothermal ZnO grown by TEW and for ZnO powder in case of cp; the thermal conductivity k was calculated from the measurement of a. The thermal expansion DL/L was measured with a dilatometer (RIGAKU Thermo plus TMA 8310) in comparison to the standard sample a-Al2O3 (sapphire). The sample size was 4 4 19 mm3 for both measurements in the h0001i and the h1120i directions. Fig. 14 shows DL/L versus temperature for the temperature interval 20 C T 1100 C. The following linear thermal expansion coefficients have been calculated: 6.06 106 K1 and 4.16 106 K1 at 25 C and 10.4 106 K1 and 4.87 106 K1 at 1000 C for the h1120i and h0001i directions, respectively. Cermet Inc. [76] gives the value of DL/L ¼ 2.9 106 K1 (no temperature range specified, but obviously for h0001i direction at RT) for their wafers were prepared from melt-grown ZnO ingots. Based on first principle theory, Reeber and Wang [77] have calculated DL/L for the h1120i and h0001i direction, DL/L ¼ 4.867 106 K1 and 2.911 106 K1 (298 K) and 9.403 106 K1 and 5.439 106 K1 (1400 K), respectively. These results are quite consistent with our measured values. For comparison with hexagonal GaN, the linear thermal expansion coefficient at RT along the a-axis and the c-axis is 5 106 K1 and 4.5 106 K1, respectively [78]. Particularly the difference in h0001i direction between ZnO and GaN is quite similar. The specific heat of ZnO powder (99.999% purity) has been measured using a Perkin Elmer DSC-7. Fig. 15 shows the temperature dependence of cp for the range 20e300 C. The specific heat of ZnO is 0.492 and 0.603 J g1 K1 at 20 and 300 C, respectively. Melt-grown ZnO from Cermet [76] is quoted as 0.523 J g1 K1 (apparently at a temperature related to the RT region e no temperature was specified in the reference). The thermal diffusivity was measured using the laser flash method as described by Waseda et al. [79]. A (0001) and a ð1120Þ oriented, polished and Au-coated ZnO disks of 1 mm in thickness and 10 mm in diameter were exposed to a laser pulse (l ¼ 694 nm) and the emitted radiation was measured by an IR sensor. The following formula was used to calculate a [80]:
Fig. 14. Thermal expansion along a- and c-axis of hydrothermal ZnO from TEW for the temperature range of 20e1000 C.
302
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
Fig. 15. Heat capacity of ZnO powder for the temperature range of 20e300 C.
a ¼ 1:38L2 =P2 t1=2 ;
ð11Þ
where L is the thickness of the sample and t1/2 the time required for the temperature response to reach 50% of its maximum value. At 25 C we obtain 1.76 105 and 2.05 105 m2 s1 for the a-axis and the c-axis, respectively. The thermal conductivity k can be calculated accordingly to: k ¼ arcp ;
ð12Þ
where r is the density of ZnO. The values for k are 49.1 and 57.2 W m1 K1 for the a- and c-axis, respectively. Melt-grown ZnO from Cermet [76] is referenced to as k ¼ 130 W m1 K1 (no temperature specified), respectively. For comparison k of sapphire, which is widely used as a substrate material in GaN epitaxy, is 46.06 W K1 m1 at 273 K [82] which is a little lower and will be even lower at 298 K. 3.2.3.4. Optical properties. The optical transmittance of (0001) ZnO wafers from TEW (polished sample of 0.5 mm thickness, no visible scratches on surface) [63], hydrothermal ZnO from SPC Goodwill (GW; polished sample of 0.5 mm thickness, no visible scratches on surface; purchased in the year of 2004) and a thin, 1% In-doped hydrothermal ZnO crystal (platelet of 3 mm in diameter and 0.2 in thickness with surfaces as grown) grown by TEW is shown in Fig. 16. The sample from TEW shows a transmittance of 80% at l ¼ 410 nm and 87% at l ¼ 700 nm. The sample from GW, which appeared a touch more yellowish than the TEW specimen, shows a slightly higher absorption in the range 390 nm l 500 nm. This could be assigned to a slightly higher impurity level in this GW sample. In-doping lowers the transmittance to about 60%. The measurement has been performed on about 200 mm thin crystals. The as grown (0001) surfaces were used and loss due to surface scattering might marginally distort the result. The refractive index for the ordinary (no) and extraordinary (ne) beam of hydrothermal ZnO from TEW and GW is shown in Fig. 17. The samples are the same as for above measurement of
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
303
Fig. 16. Optical transmittance of 0.5 mm thick hydrothermal grown (0001) ZnO wafers from TEW and GW. The 1% In-doped ZnO sample was a 0.1 mm thin crystal with natural {0001} surfaces.
optical transmittance. The laser beam was coupled into the sample by a rutile prism. In contrast to the optical transmittance identical values were obtained for each sample, which can be fitted by a second-order exponential decay. The photoluminescence (PL) signal was obtained by excitation using a continuous-wave HeeCd laser (lexc ¼ 325 nm, Pout ¼ 1.6 mW) at 4 K. The signal was detected by a CCD camera (Princeton Instruments Inc.) after dispersion with a 30 cm triple grating monochromator. The PL spectrum of a TEW ZnO sample is shown in Fig. 18. The broad emission band from 1.7 to 2.8 eV peaks around 2.3 eV. The nature of this broad emission involves donoreacceptor pair recombination due to Li [83,84]. Copper impurities at levels lower 1 ppm have been assigned to emissions around 2.4 eV [85e87] and Fe3þ shows well-defined lines at 1.7e 1.8 eV [87]. The latter has not been observed in our samples.
Fig. 17. Refractive indices for the ordinary (no) and extraordinary (ne) beam of light of hydrothermal grown ZnO from TEW and GW.
304
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
Fig. 18. PL spectrums from a TEW wafer for the range 1.7e3.6 eV and 3.18e3.43 eV (inset) at a temperature of 4 K.
The luminescence in the region near the band edge (see inset of Fig. 18) is clearly pronounced with peaks at 3.233, 3.307 (with a shoulder at 3.33 eV), 3.356, and around 3.37 eV, which can be assigned due to a donoreacceptor pair (DAP) transition, two-electron satellite transitions (TES), neutral acceptor bound exciton (I9) recombination, and a free A-exciton recombination (FEA), respectively [5,88,89]. At lower energy the phonon replicas (LO phonon energy ¼ 72 meV) appear: 1LO at 3.161 eV and 2LO at 3.09 eV. Hydrothermal ZnO reported in other literature reveal similar PL and CL spectra [42,49,53]. This hints to comparable impurity and defect concentrations in hydrothermal ZnO grown by the various groups. Our picture of PL from hydrothermal ZnO over the entire energy range of about 2e3.5 eV is quite consistent with samples grown by SCVT by Eagle Picher [5,87]. Interestingly, the broad yellow band peaking around 2.3 eV also appears in the SCVT sample although the growth conditions are very different. This might lead to the assumption that the levels of active impurities in our hydrothermal ZnO are comparable to those of SCVT ZnO. Hence, this suggests that the hydrothermal growth technique is an economic way to yield a large quantity of high-quality, large-sized ZnO crystals (Table 3). 3.2.3.5. Electrical properties. There is some concern of the impact of surface properties on the reliability of results from measurements of electrical properties using surface contacts. It is known that the ZnO surface does adsorb molecules, such as CO2, CO, O2, H2, etc., as revealed in the work by Go¨pel and Esser [90,91]. The formation of a highly conductive layer on the surface of highly resistive ZnO crystals was reported by Markevich et al. [92] and Schmidt et al. [93]. This was explained by adsorbed oxygen atoms on the surface of a ZnO crystal, which led to the capture of electrons. This results in a negative surface charge and a depletion layer with reduced conductivity. The thickness of this layer on undoped ZnO was given as <1 mm [92]. Temperature-dependent Hall-effect technique with Van der Pauw geometry was used to examine the carrier concentration (N, Fig. 19a), carrier mobility (mH, Fig. 19b), and electrical resistivity (R, Fig. 19c) from a 10 10 mm2 specimen cut from a high-quality TEW ZnO crystal. The surface was polished to an RMS roughness of 0.2 nm. The Ti/Au contacts were produced by thermal evaporation. A melt-grown sample was measured for comparison.
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
305
Fig. 19. Comparison of the temperature dependence of (a) carrier concentration, (b) carrier mobility and (c) resistivity of hydrothermal TEW ZnO and pressure-melt grown ZnO (all measured at p z 0.1 Pa).
306
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
The carrier concentration is very much lower in the hydrothermal sample than in the melt grown one and decreased with increasing 1/T from N ¼ 2 1016 cm3 at 500 K (103/T ¼ 20) to N ¼ 4 1013 at 100 K. Polyakov and co-worker [94] obtained very similar results of N ¼ 1.3e4.6 1013 and N ¼ 6.4 1011 at 300 K and 77 K, respectively, for measurements on four samples of TD ZnO as purchased. In both cases, the N value is clearly the effect of low-impurity, point defect, and dislocation concentrations, quite in good agreement with our results from XRD and impurity analysis. The above species are known to produce electrically-active centers [95]. The slight hysteresis slope at 103/T < 4 was obtained from the measurement during heating up and cooling down and is likely the effect related to surface conductivity [92,93]. For surface conductivity see also the discussion in Section 3.2.4.2. SCVT grown ZnO showed a slightly lower carrier concentration at 103/T ¼ 20, N ¼ 3 1014 [95]. In the same paper, annealing the SCVT sample at 950 C in an He atmosphere lowers N to about 1014, which was ascribed to a reduction of most of the shallowest center, or at least the result of the Fermi level dropping below the energy of this center. The Hall mobility peaks at 100 K, mH ¼ 530 cm2 V1 s1 and drops down to about 40 cm2 V1 s1 at 580 K. The higher mobility in comparison to the melt-grown sample, that peaks at 480 cm2 V1 s1 at 80 K and 430 cm2 V1 s1 at 100 K, is due to the lower impurity concentration as indicated by the lower carrier concentration (Fig. 19a). A formerly hydrothermal grown sample [95] had already showed very similar results of mH for the measured temperature range of 200e400 K. The mH lowered from about 300 to 100 cm2 V1 s1. SCVT grown ZnO shows higher mH up to almost 2000 cm2 V1 s1 at 40 K. Compared to GaN, one would find a lower mobility because of a higher effective mass and larger optical phonon scattering parameter [95]. The electrical resistivity of the hydrothermal ZnO sample (Fig. 19c) is about two orders of magnitude higher than the sample grown from the melt [76], with the minimum of 20 U cm at 60 K and 0.1 U cm at 200 K, respectively. Hydrothermal ZnO from GW shows a higher electrical resistivity of 500e1000 U cm [81] than the TEW material. One could speculate that the Li concentration must be higher in the GW crystal. However, no impurity data were available. Other results on TEW ZnO [94] reported on a large variation of R between 96 and 5 105 U cm, which was speculated to come from the Li concentration in the samples. It is possible that different growth sectors were present in the specimens and therefore Li was incorporated in quite different concentrations there. The uniformity of R, N and mH over a 2 inch wafer from TEW was measured (Fig. 20) and gives following values for R ¼ 380 U cm 15%, N ¼ 8 1013 cm3 20%, mH ¼ 200 cm2 V1 s1 10%, respectively [53]. These uniformities are quite satisfactory.
3.2.4. Machining of ZnO 3.2.4.1. Cutting and polishing. We used a diamond blade saw for cutting ZnO. The damage induced by cutting is removed by a lapping process. A subsequent mechanical polishing is followed by the chemicalemechanical polishing (CMP), both serving for removal of the damage caused by lapping. The polishing was done with a silica-based alkaline (KOH) slurry of pH ¼ 10.5 firstly, followed by a second step using a pH of 8e9. The grain size of the polishing powder has been reduced successively from 10 to <2 mm. The typical FWHM of the XRC using (0002) reflection is between 25 and 40 arcsec for both (0001) and ð0001Þ face of a scratch-free ZnO wafer.
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
307
Fig. 20. Uniformity of resistivity, carrier concentration and mobility across a 2 inch epi-ready ZnO wafer provided by TEW.
Generally, mechanical machining causes surface damage in the ZnO wafer. The surface damage in hydrothermal ZnO substrates as supplied has already been reported by Wenisch et al. [71] using double-axis (0002) HRXRD ue2q rocking curve measurements. A shoulder on the low angle side of the main peak was related to a larger lattice constant. The same group of Yao et al. [96] later confirmed by (0002) HRXRD ue2q scans that a damaged surface layer under tensile strain exists up to 15 mm deep into the volume of a CMP ZnO wafer. In the low temperature (10 K) PL spectrum relative intense first order phonon-assisted luminescence was clearly found at 3.294 eV and a highly intense emission from the ionized donor bound exciton at 3.370 eV was revealed, both of which were assigned to crystal defects. We found evidence for damage in the surface layer by comparing the results obtained by radioluminescence (RL) measurements of a ZnO hydrothermal substrate and an LPE-grown ZnO film processed on this substrate (Fig. 38a) [18,97]. The LPE film is considered as a highly-crystalline, naturally grown ZnO face. The low temperature (80 K) RL measurement shows a five times higher intensity of the excitonic emission from the LPE film. The same emission from the machined wafer, by contrast, almost disappears at room temperature but is clearly visible for the film. Moreover, the ratio of the intensity from the excitonic and green-yellow emission was about 0.1 and 2 for LPE film and the substrate, respectively. The damaged layer in ð0001Þ bulk crystals may be removed by etching with CF3COOH [96], however, surface roughening is then observed. This caused the milky appearance of the treated surface. 3.2.4.2. Effects from annealing. Generally, the annealing aims to improve the crystallinity of the machined ZnO wafers. Particularly important is the control of the surface morphology of wafers to be employed in subsequent epitaxial growth. Here, monatomic steps are desirable to control the epitaxial growth process at an atomic level. We used a fused silica tube to define atmosphere control. This was placed in a horizontal furnace. An O2 flow of 200 ml min1, at temperatures from 600 to 1100 C and annealing times between 10 and 900 min were applied. Neither temperature nor process duration affected the
308
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
surface morphology. We found that the annealing time needed to be decreased with increasing temperatures in order to achieve comparable results. Fig. 6c and d shows the results of annealing for 120 min at 1050 and 1100 C. The 1 1 mm2 AFM scan on the (0001) face of a ZnO wafer cut with 0.5 miscut angle toward h1010i displays the beginning of the formation of monatomic steps (Fig. 6c). The RMS roughness decreased to 0.096 nm. Increasing the annealing temperature to 1100 C for 120 min yields steps of 1 nm in height equivalent to 2 unit cells (lattice parameter c ¼ 0.52066 nm) with the interstep distance varying from 120 to 150 nm (Fig. 6d). The RMS roughness was 0.1713 nm as a result of the formed steps. Fig. 21 is an SEM image taken from the ð1010Þ face. The steps of 50 nm in width appeared upon annealing at 1050 C for 120 min and O2 flow of 200 ml min1. Apparently, the formation of steps on the m-plane requires a little less thermal activation than on the c-plane. See also the AFM image of a (0001) face after annealing under similar conditions (Fig. 6c). The evolution of the morphology of the ð0001Þ face with the annealing parameter temperature was reported by Ko et al. [98]. The samples were directly exposed to a high purity (99.999%) oxygen stream for 60 min. It was found that a 1000 C annealing temperature is sufficient to improve the crystallinity as proven by the low (0002) rocking curve FWHM of 17.2 arcsec. The RMS roughness was reduced from 3.1 nm (surface quality as supplied) to 0.3 nm for the sample annealed at 1000 C. Atomically flat surfaces with a step height and width of 0.5 and 280 nm, respectively, were obtained for the annealing temperature. The annealing of the (0001) face at 1000 C for 1e5 h in an oxygen atmosphere was reported by Cho et al. [96]. Steps were obtained by annealing for 1 h. They are aligned in the h1010i direction. It was shown that the step height is a function of the annealing time and rose linearly, i.e. 2 nm at 1 h to 10 nm for the 5 h anneal. Annealing for 3 h yielded the best crystallinity as indicated by the FWHM from the rocking curves of (0002), (0004), (0006), and ð1011Þ reflections of 12.6, 12.24, 12.96, and 16.56 arcsec, respectively. Annealing for longer than 3 h deteriorated the crystalline quality.
Fig. 21. SEM image of the m-plane showing steps of about 50 nm in width obtained by annealing at 1050 C for 120 min; oxygen flow rate was 200 ml min1.
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
309
The above paper [96] also reports on annealing under N2, O2, and high vacuum conditions (5 109 Torr) at 950 C for 4 h. Desorption or formation of pits at the surface occurs for either an N2 or an O2 atmosphere, respectively. Clear steps were obtained by annealing under high vacuum giving an RMS roughness of 0.25 nm. Surface reconstruction of a 3 3 pattern was observed by RHEED by annealing over 30 min. Annealing over 3 h slowly decreased the RHEED intensity. Using an oxygen RF plasma and applying a chamber pressure of 5 105 Torr did not lead to an observation of a reconstruction pattern. It has to be noted here that results from annealing ZnO strongly depends on the experimental conditions governed by the setup of the equipment. This is confirmed by the slightly different observations as shown above. Fig. 22 shows that annealing may cause the formation of voids as observed on the surface of a polished and annealed (0001) sample. The size of the void typically is less than 1 mm but sometimes extends to more than 10 mm and exhibits a layer consisting of sub-micrometer size needle crystals (region B in Fig. 22). It could be speculated that crystal defects like a high local impurity concentration might be the cause that initiates the formation of voids. However, no further experiments have been conducted so far. The surface surrounding the voids appears with the typical step formation (region A of Fig. 22). Scratches caused by improperly polishing do heal out in parts as demonstrated by Fig. 23. Hexagonal pits of some hundreds of nanometers and smaller irregular voids are visible in the trace of the scratch (located by the darker bar in the center of Fig. 23). The voids in the upper
Fig. 22. The SEM image of the (0001) plane after annealing displays a large void. The bright part consists of recrystallized ZnO (B). The surface around the void appears with typical step pattern (A).
310
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
Fig. 23. SEM image of the (0001) plane of a ZnO wafer after annealing at 1100 C shows the healing effect of annealing. Large hexagonal pits, smaller voids and steps were formed. The trace of the scratch induced by polishing has been visualized by the darkened bar.
part of the image do not relate to this polishing effect, but rather to the above-mentioned formation of voids. The impact of annealing on electrical properties of the ZnO surface was recently reported by Schmidt et al. [93]. Upon annealing at 650 K in a vacuum of <0.1 Pa, a semi-insulating (i.e. highly resistive due to Li doping) specimen showed a high surface conductivity. This was explained by the accumulation of electrons in the potential well near the surface. Thus, an electrically conducting layer was formed. Subsequent annealing in air recovered the originally high resistivity. Fig. 24 shows the Al concentration profile of a specimen before and after annealing. The SIMS data reveal a slight decrease in the Al concentration from 6 1016 to 3 1016 cm3.
Fig. 24. SIMS depth scans demonstrating the effect of annealing on the Al concentration in a ZnO wafer.
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
311
However, the Al concentration in the annealed sample slightly increases toward the surface and exceeds the value from the untreated sample in the volume up to 200 nm deep into the sample. Again, this might rather be related to the SIMS measurement than the true properties of the sample. Polyakov et al. [94] have reported on the out-diffusion of Li at temperatures as low as 500 C. Mainly the surface near region was affected. This effect indeed might be useful to tune the resistivity of the ZnO surface. 4. Solvothermal film growth of ZnO 4.1. Growth from aqueous solvents The limited choice of solvents for ZnO has already been discussed in an earlier section of this paper and fully applies to the growth of ZnO films. In thin film technologies, however, we are faced with an additional challenge which is related to the substrate. The film growth process has to be such as to avoid substrate corrosion, i.e. dissolution or any kind of ion exchange with the solution should be minimized in order to obtain a defined interface between substrate and grown film. This is often the limiting aspect in film growth technology. Typically, ZnO film growth is based on chemical reactions to produce a ZnO solution which consists of a solvent, precursor and a complexing agent. The latter through pH (concentration of H3Oþ) control [39,89,100] is responsible for the crystal habit. The majority of growth techniques are based on aqueous solvents including waterealcohol mixtures. Spin-coating is a non-aqueous technique using 2-methoxyethanol, which will be evaporated after the coating process. Similar to water, 2-methoxyethanol contains an OH group, which may be incorporated in the ZnO film. The potential of alkaline-metal chlorides as water-free solvents will be treated in detail in the following Section 4.2. Common to all water-based growth techniques is the low process temperature, even <100 C in air under atmospheric pressure. The removal of growth solution from the fabricated film is trouble-free. The growth of ZnO films has been demonstrated by a variety of methods: chemical bath deposition (CBD), selective ionic-layer absorption and reaction (SILAR), electroless deposition (ED), liquid flow deposition (LFD). A comprehensive description is given by Niesen and De Guire [100]. CBD produces a solid film in a single immersion through control of the kinetics of the formation of ZnO. SILAR is characterized by the use of alternating aqueous solutions containing a zinc salt and a hydrolyzing solution (water or water/ammonia). ED uses oxidizing or hydrolyzing agents to form ZnO and is similar to CBD. Films grown are mostly without crystallographic texture. LFD involves the flow of the solution past the substrate at a controlled rate and reactants are replenished continuously. This process can be maintained over longer time span to deliver thick films. The above methods are used in ceramics technology but might be adaptable to the growth of single-crystalline film by choice of a proper single-crystalline substrate. Table 5 gives an overview over achievements in the growth of poly- and single-crystalline ZnO film from aqueous solvents including 2-methoxyethanol. The application of glass-like substrates, plastics or silicon wafers causes the growth of islands and polycrystalline ZnO films composed of grains or columns. The best crystallinity was obtained by using either single-crystalline (0001) ScAlMgO4 (SCAM) or a (111) MgAl2O4 substrate. SCAM is isostructural with YbFe2O4, having a space ˚ and c ¼ 25.195 A ˚ . The lattice misfit in the a plane group R3m with lattice constants a ¼ 3.246 A
312
Table 5 Growth conditions of solvothermal ZnO films Growth conditions
Growth vessel Precursor/complexant
H2 O
0.05 mol l1 Zn(NO3)2 þ 0.15 mol l1 DMABb
50
0.1 mol l1 Zn(NO3)2 þ 0.1 mol l1 DMABb
60
0.005 mol l1 Zinc carboxylate salts (acetate, formate) þ 0.005 mol l1 HMTd 2-Methoxyethanol 0.75 M Zn(OAc)2 2H2Oe þ MEAf
H2 O
RT
80
RT
H2 O
0.025e0.1 mol l1 Zn(NO3)2 6H2O þ NH4OH
H2 O
(1) Zn(NO3)2 6H2O þ 80 NH4OH þ NH3; (2) Zn(NO3)2 6H2O þ NH3 þ 90 Na(cit) nH2Og Zn(NO3)2, Zn(OAc)2 100 2H2Oe, urea, HMTd
H2 O
6.2
90
2-Methoxyethanol 0.75 M solution of Zn(OAc)2 2H2Oe þ MEAf 2-Methoxyethanol 0.35 M Zn(OAc)2 2H2Oe, MEAf
pH
5.0
Substrate
Silica glass
/
(0001) ScAlMgO4; (0001) a-Al2O3
/
Corning 1737 glass
150e200 7e11 (111) MgAl2O4
(111) MgAl2O4
10.9 n.r.
Glass covered by F:SnO2
Refs.
Method; film thickness; growth rate; film morphology
Corning 7059 glass; n.ra Growth of polycrystalline films with surface catalyzed with hexagonal grains z 0.3 mm; SnCl2 and PdCl2 0.6 mm h1 Corning 1737 glass; Boron-containing films up to 1 mm; surface catalyzed with 0.42e0.58 mm h1 SnCl2 and PdCl2 TOFc glass; Au coated Open system, volume Nanocolumnar films up to 3 mm; 100 cm3 growth time 1 h
n.r.
7.5
Result
[101]
[102]
[103]
Post-growth annealing at above 500 C; 100 mm thickness (several dippings required) / Spin-coating at 5000 rpm for 30 s; post-growth annealing at max. 850 C; 13e25 nm thick films. / Spin-coating at 3000 rpm for 30 s; post-growth annealing at max. 700 C; nano-crystalline films mainly oriented with c-axis perpendicular to substrate surface, grain size 80 nm. Teflon-lined stainless Film thickness z 15 mm; Island growth steel autoclave, and coalescence vol. 45 ml Epitaxial lateral overgrowth; (1) thin ZnO film and (2) growth from window region of film (1).
[104,105]
Pyrex reactor, open system, vol. 80 ml
[111]
Open system
Microwave activation. Films of rods mainly oriented with c-axis perpendicular to substrate surface, few mm thick.
[106,107]
[108]
[19,109]
[110]
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
T ( C)
Solvent
91 mM Zn(OAc)2 2H2Oe, 4.5 mM NaOH
RT (?)
6.7
(100) Si, aminederivatized
H2O þ ethanol
0.6 M Zn(OAc)2 2H2Oe þ InCl3
400 5
n.r.
Soda lime glass
H2 O
0.1 M Zn(NO3)2 xH2O þ HMTd
95
n.r.
ZnO-coated Si, PETh, a-Al2O3
H2 O
0.025 M Zn(NO3)2 xH2O þ 0.025 M HMTd or DTi
90
n.r.
ZnO-coated (100) Si, PDMSj
a b c d e f g h i j
n.r. e Not reported. DMAB e dimethylamineborane. TOF e fluorine-doped tin oxide. HMT e hexamethylenetetramine. Zn(OAc)2 2H2O e zinc acetate. MEA e monoethanolamine. Na(cit) nH2O e sodium citrate. PET e polyethyleneterephthalate. DT e diethylenetriamine. PDMS e polydimethylsiloxane.
Open system
Solegel synthesis; Island growth, [112] substrate coverage 70%, 3e5 nm thickness Open system, 200 ml Chemical spray pyrolysis of In:ZnO, [113] solution, spray rate nano-crystalline films, grain of 10 ml min1 size 34.7 nm, film thickness of 0.54 mm. Preferential c-axis orientation Reaction flask with Aligned nanowires, film thickness [114] reflux condenser, 3 mm, growth time 10e20 h. 40 ml solution Open crystallizing Aligned nanowires, film thickness [115] dish 3 mm, growth time 0.5e6 h.
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
H2 O
313
314
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
of (0001) SCAM to (0001) ZnO is 0.09% [117]. SCAM is composed of alternating layers of wurtzite-type (0001) (Mg, Al)Ox and rock-salt type (111) ScOy, which is the reason for the excellent (0001) cleavage. Recently, the first vapor-grown blue LED has been demonstrated using (0001) SCAM as substrate [3]. It has also been used as substrate in ZnO film growth by spin-coating CBD and successive thermal annealing at a maximum of 850 C [106]. The (0001) ZnO films were epitaxially grown in-plane with the (0001) SCAM substrate as revealed by XRD and
HRTEM. The crystallographic orientation relationship between ZnO and SCAM is ð0001ÞZnO ð0001ÞSCAM and ½2110ZnO k½2110SCAM . The following remarkable results have been achieved by the group of Lange from UCSB. Thick epitaxial films of about 15 mm were grown on (111) MgAl2O4 (spinel, space group Fd3m) at 200 C [19]. The growth solution was composed of different concentrations of Zn(NO3)2 6H2O, NH4OH, NH3, and Na-citrate (Na3C6H5O7 2H2O) at pH between 7 and 11 [19,110,111]. Fig. 25 shows the SEM image of a sample grown by two-cycle lateral epitaxial overgrowth (LEO), with the second film not completely coalesced [110]. This second film displays no etch pits on the planar surface, which is the result of growth from window regions lying above the wing regions of the first film. A photo resist (PR) channel stamping technique has been used to produce growth windows [118]. The crystallographic relationship between ZnO and MgAl2O4 was determined by XRD. The out-of-plane orientation is h111ispinel h0001iZnO , the in-plane orientation is indicated to ½1120ZnO k½112spinel and ½0110ZnO k½110spinel , respectively. The reduction of the lattice mismatch of ZnO on (111) MgAl2O4 from 13.6% to 1.6% was explained by a 30 rotation of oxygen planes in the wurtzite structure, which converts the Mg and Al sites in the spinel structure to tetrahedrally coordinated Zn sites in ZnO. Hþ ions are necessary for charge balance at the interface between the ZnO and MgAl2O4. The same group [110] recently reported on lateral epitaxial overgrowth by applying a stamped mask to a previously grown ZnO film. The growth of coalesced ZnO films was achieved at temperatures 90 C and required the addition of sodium citrate to the solution. The film growth started with island growth which eventually coalesced to a closed film [19].
Fig. 25. SEM image showing lateral epitaxial overgrowth of ZnO at 90 C (courtesy of Jin Hyeok Kim, David Andeen, and F.F. Lange).
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
315
It was shown by Chaparro [119] that Zn(OH)2 is the predominant zinc species in aqueous solutions containing ammonia as a complexing agent at pH values between 7.6 and 11.4. Ammonia plays an outstanding role as a complexing agent, since it forms metastable complex ions that can result in the complete dissolution of zinc, establish an appropriate basic pH that decomposes the oxygen delivering precursor, and also possesses a sufficiently high stability in the solution. 4.2. Liquid phase epitaxy from chloride solution Liquid phase epitaxy (LPE) from chloride solution has recently been proven successful for the film growth of some oxides as shown, e.g. by Ehrentraut et al. [15] and Romanyuk et al. [17]. Basically, in LPE, a solution consisting of solvent and solute is used to deposit a film on a single-crystalline substrate so as to preserve the crystallographic information resulting from the substrate. Consequently, a ZnO substrate would be the ideal choice to fabricate ZnO films with zero lattice and thermal misfits. LPE from chloride solution is basically water-free and, consequently, hydrogen plays no role as electrically-active shallow donor (18e35 meV [121]) impurity in ZnO [120]. 4.2.1. Technology Different LPE technologies have been developed for a broad spectrum of semiconductors and oxides. We applied a dipping technique [18] as described by Levinstein et al. [122]. The schematic setup is shown in Fig. 26. Experiments were carried out using industrial scale
Fig. 26. Schematic of the LPE setup as employed for ZnO film growth.
316
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
LPE furnaces with three heating zones with automatic temperature control giving a temperature deviation of 1 K. A lift and rotation unit is mounted above the furnace. The liquid solution inside the crucible is separated into growth and saturation zones above and beneath the baffle, respectively. We used LiCl (melting Point Tm ¼ 605 C) and a eutectic mixture of 35 mol% NaCle65 mol% CsCl (Tm ¼ 486 C) as the low-temperature solvent. The highest solubility of ZnO at 650 C was obtained in LiCl, 9.14 102 g with respect to 1 mol LiCl. The solubility of ZnO in NaCleCsCl is about 5.3 103 g mol1. Due to the comparatively high solubility in LiCl, we have chosen LiCl as solvent for ZnO for all LPE films grown and discussed in the remainder of the paper. A K2CO3 pellet on the bottom inside the crucible served as an oxygen source to form ZnO with zinc from the zinc source ZnCl2, which was thoroughly mixed with the solute. The following reaction appears in the crucible: Solvent þ ZnCl2 þ K2 CO3 /Solvent þ ZnO þ 2KCl þ CO2 [:
ð13Þ
This reaction results in the formation of KCl that forms a solid solution with all of the applied solvents. The melting point of the solution is slightly lower for LiCl, NaCl, and CsCl when used as solvents. A large quantity of the resulting ZnO occurs as crystallites 1e10 mm in size along the c-axis and is deposited at the bottom of the crucible. This was used to control the supersaturation of the solution at the beginning of each experiment. The choice of K2CO3 as an oxygen source has been made due to its easy availability and high purity. Fig. 27 compares the results from powder XRD measurements of the deposit as the product of the reaction of ZnCl2 with Na2CO3, K2CO3, and Na2O. The latter was available only in 97% purity. Nevertheless the formation of XRD phase-pure ZnO was confirmed. All chemicals in our experiments have been used in purities 99.99%. We used crucibles of high-density alumina and zirconia ceramic. The alumina crucibles never showed any corrosion, i.e. neither a weight change nor coloration. Zirconia, by contrast, is slowly attacked by the ZnCl2/LiCl solution. This sometimes caused cracking of the crucible after multiple uses.
Fig. 27. Powder XRD of deposited ZnO microcrystals grown from different oxygen sources: (a) Li2CO3, (b) Na2CO3, (c) K2CO3, (d) Li2O and (e) Na2O.
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
317
The substrate was attached to a platinum wire (0.5 mm diameter), which was mounted on an alumina rod. The substrate was immersed into the liquid solution after a dwell time of 2 h and the rotation speed was slowly increased up to 5e10 rpm. The vertical temperature gradient in the solution between the bottom and the substrate was set to zero. The experiments were performed under air at atmospheric pressure and at a constant temperature. The growth was finished by separating the substrate from the solution. The adhesive solvent was removed from the film by rinsing with distilled water. After that 2-propanol was used to remove water from the sample. The state of the oxygen source, i.e. as a powder of micrometer particle size or as a polycrystalline pellet, strongly affects the duration of the above chemical reaction [123]. The powder reacted within few minutes whereas the pressed and sintered pellet serves well over 24 h. This can be explained in terms of surface size of the oxygen source, which is very large for the powder and minimized in case of the pellet. The alkaline-metal chlorides possess some advantageous properties, which are (a) low melting points from 605 C for LiCl to 801 C for NaCl; (b) complete dissociation to monatomic ions when molten [124], which results in a low dynamic viscosity; (c) a high solubility in water, therefore cleaning of the grown film is quite easy; (d) availability in high purity at reasonable price. The ambiguous growth nature of ZnO along the h0001i direction requires different growth conditions for either the face (0001) or the ð0001Þ. Films grown on ð0001Þ generally were of poor quality and much lower thickness. In good agreement to Suscavage et al. [48] we observed the ratio of growth speed of ð0001Þ : ð0001Þ ¼ ð2:8e3Þ : 1. That is the reason why we focused exclusively on the growth on (0001) ZnO. In addition, the (0001) face is considered to be superior to ð0001Þ with respect to p-type doping [125]. 4.2.2. Undoped ZnO film The concentration of ZnO determines the growth mechanism [18], which evolves from island growth at csol ZnO < 2 mmol ZnO to columnar growth (free-standing columns at csol ZnO ¼ 2.5 0.5 mmol ZnO, to porous columnar film at 3.5 0.5 mmol ZnO, to closed cosol lumnar film at csol ZnO ¼ 5 mmol ZnO) thence to step-flow growth at cZnO ¼ 12 0.5 mmol ZnO. The SEM images (Fig. 28) show: (a) the film grown from 10 mmol ZnO displays a columnar growth mechanism; (b) 13 mmol ZnO results in large steps, which are aligned parallel to each other; (c) step bunching is obtained for 16.5 mmol ZnO in the solution. High structural quality, single-crystalline films, X-ray full-width half-maximum (FWHM) of 26e31 arcsec for (0002) reflection, with steps propagating over macroscopic dimensions were fabricated in the concentration range 13 csol ZnO 15 mmol ZnO. The X-ray reciprocal space map using the (101) reflection (Fig. 29) shows the high crystallinity of the film (b) grown from 13 mmol ZnO concentration (a). The FWHMs for 2q/u and u are 22 arcsec for both and 12 and 15 arcsec for substrate and film, respectively. The morphological conditions of the substrate are apparently conserved [18]. The steps themselves possess a very flat surface with no contrast being resolved by AFM (Fig. 30, scan area 5 5 mm2). The RMS roughness of 0.218 nm was calculated. The line scan reveals an interstep distance of 2 0.5 mm and a step height of 0.5 nm, which corresponds to the height of one unit cell. The thickness of the film is about 1 mm as estimated from weight difference of the substrate before and after growth. Typically, the film thickness varies with growth time. The growth rate for film growth on (0001) ranges between 0.12 mm h1 Vh0001i 0.25 mm h1. Best qualities were grown at the lowest value of Vh0001i.
318
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
Fig. 28. SEM images of ZnO films grown under different csol ZnO. (a) Columnar growth mechanism is observed for the film sol grown from csol ZnO ¼ 1 mol%. (b) The film grown from cZnO ¼ 1.3 mol% displays large steps. (c) The film grown from csol ZnO ¼ 1.65 mol% is showing the effect of high supersaturation.
Fig. 29. X-ray reciprocal space maps using (101) reflection from (a) the substrate and (b) the homoepitaxial ZnO film grown on this particular substrate.
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
319
Fig. 30. AFM image and line scan revealing monatomic steps on the surface of an LPE-grown undoped ZnO film.
The films do appear completely transparent, colorless, and highly reflective. Step distance and height strongly vary with csol ZnO. Fig. 31 shows the SIMS profile from a Ga-doped, 1.7 mm thick ZnO film on a hydrothermal ZnO substrate. The normalized intensities of the species Li, Ga, Mg, Al, Na, and Si are shown. The Na and Li counts are increased in the film up to two orders of magnitude in comparison with the ZnO host. Typical for the proposed LPE growth of ZnO, cations from the solvent would be detected in the ZnO film by SIMS. The intentionally undoped ZnO films are therefore slightly doped with alkaline-metal ions (order of 100 ppm wt for Li). Mainly the starting chlorides are made responsible for that, whereas the source for Al and Si might rather be the ceramic crucibles [18]. 4.2.3. Solid solutions and doping 4.2.3.1. MgOeZnO. Magnesium is widely used for band gap engineering of ZnO to achieve a UV shift. This shift increases with increasing x in MgxZn1xO. However, MgO crystallizes in the rock-salt structure (coordination number of nearest neighbors is 6, by contrast to 4 in case of wurtzite structure), which is likely the reason for the increasing tendency for phase segregation into MgO and ZnO with increasing x in MgxZn1xO. Phase segregation for x 0.36 was reported for the growth of thermodynamically metastable films by PLD by Ohtomo
320
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
Fig. 31. SIMS profile from a Ga-doped, 1.7 mm thick homoepitaxial ZnO film. The hatched area comprises the crosssection of the LPE film.
et al. [117]. The thermodynamic solubility limit of x z 0.04 at 1600 C [126] has been accepted for a long time. Recently, however, x z 0.06 was reported by our group for films grown by LPE on a (0001) ZnO substrate [127]. The source for Mg was MgCl2 (99.9% purity) which was employed at csol Mg 20 mol%. The growth occurred at 640 C using a growth time 24 h. A micrograph of a film taken by NDIM is shown as Fig. 34a. Growth steps and the formation of a 1011g facet at the rim of the (0001) substrate is characteristic [127]. Based on structural analysis using 2q/ueu X-ray reciprocal space maps of the symmetric (002) and asymmetric ð105Þ reflection from grown MgxZn1xO/ZnO heterostructures and X-ray powder diffraction from solid phases deposited at the bottom of the crucible, the solubility limit has now been verified between 0.06 x 0.1 [128]. The in-plane strain in films increases with increasing x as indicated by broadening of the u peak to 176 arcsec for x ¼ 0.06 (Fig. 32a). Using the asymmetric ð105Þ reflection (Fig. 32b) the MgxZn1xO film grown from csol Mg ¼ 10 mol% is characterized by strain relief as the peak from the film is slightly lefthand shifted with respect to the peak from the ZnO substrate. Recently, Wang et al. [50] have reported the hydrothermal growth of ZnO bulk crystals containing up to 5.5 mol% Mg (see also Table 3). This result strongly confirms our findings. In fact, the similarity of both results is striking despite the different chemistry of the solutions. SIMS measurements from the region near the interface of MgxZn1xO/ZnO heterostructures have been carried out and do reveal a highly uniform Mg content within the doped films (Fig. 33). The transition of the Mg concentration between substrate and film is relatively sharp [128]. More growth results are reported in Table 6. Hall measurement on the Mg0.06Zn0.94O film made under the same conditions as for the hydrothermal ZnO (Section 3.2.3) revealed a mobility of mH ¼ 20 cm2 V1 s1 at 560 K and N z 1017. Lorenz et al. [129] reported mH about 20 cm2 V1 s1 at 300 K and N z 1017 cm3 for PLD films grown on sapphire substrates and Ogata [130] gave mH ¼ 100 cm2 V1 s1 at 300 K and N ¼ mid 1017 cm3 for MBE grown films on a-plane sapphire. 4.2.3.2. CdOeZnO. It has been shown that the system CdOeZnO can be employed to achieve a red shift in the excitonic emission from CdxZn1xO films grown by PLD, MBE, MOVPE, and
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
321
Fig. 32. 2q/u over u X-ray reciprocal space maps of MgxZn1xO/ZnO heterostructures grown from solutions with different csol Mg using (a) the symmetric (002) reflection. The broadening of the u peak indicates increasing strain in the film with increasing x. (b) Asymmetric (105) reflection: the MgxZn1xO film grown from csol Mg ¼ 10 mol% is characterized by strain relieve as the peak from the film is slightly left-hand shifted with respect to the peak from the ZnO substrate.
RPE-MOCVD [8,131e135]. Concentrations of x 0.697 in the films have been reported, which is far beyond the thermodynamic solubility limit of x ¼ 0.02 as reported by Makino et al. [8]. The latter value, however, is likely the limit for ZnO films fabricated by solvothermal methods. This might be supported by the fact that the effective ionic radius is considerably 2þ IV IV ˚ ˚ larger for Cd2þ, rCd , rZn 2þ ¼ 0:78 A than for Zn 2þ ¼ 0:60 A [136]. Moreover, as for MgO, CdO crystallizes in the rock-salt structure. Consequently, the solubility range is expected to be narrower than for the system MgOeZnO. We used anhydrous CdCl2 (99.998% purity) as a Cd source. A typical LPE-grown Cd-doped ZnO film grown from csol Cd ¼ 1 mol% is shown in Fig. 34b. Defects of unknown nature are visible as dark spots. Films 1.1 and 3.1 mm thick have been grown at Vh0001i z 200 nm h1 (Table 6). The growth speed under similar growth conditions was about a factor of two higher than for the other dopants except for Cu. SIMS measurements do not reveal any significant change of the Cd content between substrate and film. We speculate that the formation of CdO is favored over ZnO. In this case CdO would precipitate from the solution.
322
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335 1.E+22 undoped 1.0 mol% 3.0 mol%
Mg concentration [atm/cm3]
1.E+21
1.E+20
1.E+19
1.E+18
1.E+17
1.E+16
0
1000
2000
3000
4000
5000
Depth [nm] Fig. 33. SIMS profile from the region near interface of Mg-doped films.
4.2.3.3. Ga2O3eZnO. The Ga dopant in ZnO is very effective in enhancing the electrical IV conductivity [137]. The effective ionic radius of Ga3þ is smaller than for Zn2þ, rGa 3þ ¼ 0:47 ˚ [136]. A We used anhydrous GaCl3 (>99.999% purity) as the Ga source. The films appeared with strongly pronounced growth steps on the (0001) face and large, parallel aligned steps are typical (Fig. 34c). The film was grown from csol Ga ¼ 5 mol% and was 1.7 mm thick. The XRC FWHM of 28 arcsec was measured (Table 6). In contrast to all the other doping attempts, the formation of steps of a few nm in height has been observed for the growth on the ð0001Þ face. This indicates a change of the solubility of Ga-doped ZnO in LiCl in comparison to the system ZnOeLiCl.
Table 6 LPE growth of doped ZnO films on the (0001) face of hydrothermal ZnO substrate System
csol (mol%)
cfilm (mol%)
XRC FWHM (arcsec)
Thickness (nm)
Vh0001i (nm h1)
Remarks
MgO CdO
30 0.1 1 5
6
15e176 41
300e1700 1100 3100 1700
17e109 214 192 109
530 800 1500 1900 1500 5200
33 50 92 117 94 325
Increased tendency for facet formation Some inclusions; growth about two times faster Some inclusions; formation of growth steps on ð0001Þ face Extremely low growth speed
Ga2O3 In2O3 GeO2 Sb2O5 CuO
0.2 2 0.525 3.3 5 5
28 24 35 22 31 17.3
Pronounced faceting exhibits 1010g faces Fast growth
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
323
Fig. 34. NDIM micrographs taken from films grown with different dopant concentration in the initial solution: (a) 6 mol% Mg, (b) 1 mol% Cd, (c) 5 mol% Ga, (d) 3.3 mol% Ge.
The concentration of Ga in the ZnO films has not been estimated quantitatively, the SIMS profile, however, shows an increase in the counts of about two orders of magnitudes for the film with respect to the substrate (Fig. 31). 4.2.3.4. In2O3eZnO. Indium in ZnO acts similarly to Ga. Doping ZnO with In led to an increase in the n-type conductivity of the ZnO films, which may be used in applications, i.e. as a transparent conductor. The use of In-doped ZnO as a scintillator [138] has recently become attractive, particularly in the light of reduced self-absorption effects [97,139]. The effective IV ˚ ionic radius of In3þ is slightly larger than for Zn2þ, rIn 3þ ¼ 0:62 A [136]. InCl3 (99.999% purity) has been employed for use as the In source. The XRC FWHM using the (0002) reflection (Fig. 35b) gives 24 and 35 arcsec for the films of thicknesses 530 and 800 nm grown from csol In ¼ 0.1 and 2 mol%, respectively (Table 6). The growth speed is more than a factor of two lower than for the comparable Ga-doped film. 4.2.3.5. GeO2eZnO. Only a few reports attempted to change the electronic structure of ZnO by doping with Ge [140]. Ge-doped ZnO ceramic was obtained by solid-state reaction and some films have been grown by PLD by Fan et al. [141]. The occupation of the Zn site should IV ˚ decrease the lattice parameters (rGe 4þ ¼ 0:39 A [136]) while it would lead to an increase when present as an interstitial. Both cases, however, yield lattice distortion. The solubility of Ge in ZnO is rather limited to about 0.7 mol% [140]. We tried to incorporate Ge4þ into ZnO by LPE.
324
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
Fig. 35. XRC measurement reveals the broadening of the (0002) reflection peaks as result of increased doping concentration from (a) 0.5 to 3 mol% Ge; film thickness 1.5 and 1.9 mm, respectively. (b) 0.1 and 2 mol% In was doped; film thickness 0.53 and 0.80 mm, respectively.
The source for Ge4þ was GeI4 (99.999% purity). We worked with csol Ge ¼ 0.525 and 3.3 mol% at constant growth temperature of 640 C. Fig. 34d shows the NDIM image of the film grown from csol Ge ¼ 3.3 mol%, which reveals a pronounced formation of 1010g facets. Vh0001i was between 92 and 117 nm h1 (Table 6). The FWHM increased from 22 arcsec for 0.525 mol% doping to 31 arcsec for 3.3 mol% doping (Fig. 35a). 4.2.3.6. Sb2O5eZnO. Antimony (Sb) might be a candidate as p-type dopant in ZnO if oxygen VI ˚ substitutes for Sb on the oxygen site. The effective ionic radius of Sb5þ is rSb 5þ ¼ 0:62 A and of 2 IV ˚ O rO2 ¼ 1:38 A [136]. The source for Sb was ZnSb (99.999% purity). We worked with csol Sb ¼ 5 mol% at a constant growth temperature of 640 C. A film of 1.5 mm has been grown on the (0001) face, Vh0001i ¼ 94 nm h1 (Table 6). The surface appears rather rough in comparison to the
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
325
Mg0.06Zn0.94O film of Fig. 34a. The XRC FWHM from the (0002) reflection was about 30 arcsec. The sign of a shoulder formation at the lower angle has been found, which might be due to the phase separation. The SIMS measurement implies significant incorporation of Sb as the count intensity increases by a factor of about 103. The interface between the substrate and the film was hardly detectable as the count intensity increased slowly. 4.2.3.7. CuOeZnO. The source for Cu was anhydrous CuCl2 of (99.99% purity). We worked with csol Cu ¼ 5 mol% at constant growth temperature of 640 C. The growth rate Vh0001i was 1 325 nm h (Table 6). The XRC measurement using the (0002) reflection gives an FWHM of 17.3 arcsec. This was the lowest value for an FWHM we have ever measured from an LPE-grown ZnO-based even though the doped film was grown from a higher supersaturation, csol ZnO ¼ 17.5 mmol. SIMS revealed that Cu had been easily incorporated into the ZnO film. The count intensity raised two orders of magnitude and a clear interface between substrate and film was observed. This led us to conclude that Cu-doping of LPE ZnO films is easy to achieve. 4.2.4. Photo- and radioluminescence PL was measured on undoped and doped ZnO LPE films at 4 K under similar conditions as for the bulk ZnO, see Section 3.2.3. The signal from an undoped ZnO film grown on the hydrothermal ZnO substrate from TEW is shown in Fig. 36a. The dominant peak signal lies at 3.3601 eV with a shoulder peak at 3.364 eV [18]. This is attributed to neutral donor bound excitation (D0X). More emission peaks at 3.372, 3.374, 3.378, 3.39, and 3.42 eV have been measured. Among them, 3.378 eV is assigned to the A-free exciton (FEA) from the ground state, 3.39 eV due to B-free exciton (FEB), and 3.42 eV due to n ¼ 2 state (or 1st excited state) of FEA [88,125]. The latter has not been measured from the substrate. It was reported [142] that FEA emission structure becomes clearly visible at higher TG, i.e. 630 C. Apparently, our result is comparable to the ZnO films grown on ScAlMgO4 [142], which appear with similar splitting of exciton ground states. From reflectance measurements we determined the binding energies for FEA and FEB of 58 and 55 meV, respectively. The substitution of Zn for Mg in the ZnO lattice shifts the excitonic emission to higher energies [143] as shown in Fig. 36b. The two peaks at 3.357 and 3.425 eV are related to the ZnO substrate and the Mg0.06Zn0.94O film, respectively [127]. The inset in Fig. 36b shows the PL emission from 1.8 to 3.7 eV. The broad band peaks around 2.4 eV. The peak from the substrate lies at 3.359 eV before being used for the epitaxial growth, see the dashed line. It has been shown by Meyer et al. [5] that the intensity of the emission peak around 3.36 eV decreases upon annealing at 600 C. By contrast, the bound exciton line at 3.357 eV remains unaffected. Wang et al. [50] found the excitonic peak at 3.414 and 3.4468 eV for their MgxZn1xO hydrothermal crystals with x ¼ 0.03 and 0.055, respectively. The Cd-doped film (csol Cd ¼ 1 mol%) exhibits PL emission peaks only from the ð0001Þ face at 3.36, 3.346, 3.295, 3.225, and around 3 eV. The latter is possibly due to the slightly increased Cd content. It was shown that a Cd0.04Zn0.96O film grown by MOCVD shows a characteristic peak at 2.91 eV [132]. We assume that the Cd concentration in the (0001) film is below the detection limit but is well above it in the ð0001Þ film of lower crystallinity. The situation is different when the film was doped with Ga as shown in Fig. 37a. The bound exciton emission spectrum is considerably changed and the dominant peak becomes the I8 at 3.358 eV, which is not observed from undoped ZnO. A slight red shift by about 0.5 meV was
326
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
Fig. 36. Low-temperature PL spectra from the region near the band edge of (a) the undoped ZnO substrate (dashed line) and the LPE-grown film (straight line) and (b) the undoped ZnO substrate (dashed line) and an MgxZn1xO film (straight line). The peaks at 3.357 eV and 3.425 eV are due to emission from the ZnO substrate and the MgxZn1xO film, respectively. The inset shows the PL spectrum of the MgxZn1xO film over the energy range from 1.8 to 3.7 eV. Intensity is Log scale for all graphs.
explained by electroneelectron interactions [5] causing an increased metallic conduction behavior. Also in good agreement with findings from other groups [5,143] is the greatly weakened emission at 3.369 eV. The TES peak is at 3.32 eV, which is in good accordance with Ref. [5]. The effect of In-doping is shown in the PL measurement taken at 12 K (Fig. 37b). The dominant line is the recombination I9 [5] at 3.359 eV and the TES peaks around 3.309 eV. Not shown here are the PL measurements from LPE films doped with Ge, Sb and Cu recorded at 12 K. In the Ge-doped film the major peak was around 3.357 eV, similar to that in the hydrothermal ZnO substrate which was used in this experiment. The emission peak at 3.372 eV was weakening with increasing csol Ge from 0.525 to 3.3 mol%. The reason remains unknown at present.
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
327
Fig. 37. Low-temperature PL spectra from the region near the band edge of (a) a 5% Ga-doped and (b) a 0.1% In-doped ZnO film. Intensity is Log scale for all graphs.
Sb-doping reveals peaks at 3.357 and 3.372 eV and the DAP peaks at 3.31 eV. At lower energies there is a peak at 3.265 eV with replicas repeating at 72e75 meV, i.e. at 3.193, 3.12 and 3.045 eV. The Cu-doped film showed the peaks at 3.371, 3.358, 3.31 and 3.235 eV. RL measurements at temperatures from 80 K up to 295 K have been obtained following X-ray exposure at lexc ¼ 0.154 nm, using a spectrofluorometer (199S, Edinburgh Instruments) equipped with a steady-state X-ray excitation source (35 kV) and combined with a single grating monochromator and with the photomultiplier XP2233 used in the photon counting mode. Fig. 38a details the spectrum from the bulk ZnO (dotted line) and a typical LPE-grown film (straight line) of 2 mm in thickness. Luminescence from the band edge region is almost suppressed in the bulk sample and the ratio of the arbitrary intensity to the broad band emission from the bulk sample peaking around 2.1 eV is only about 0.1. The LPE film shows contrary strong emission with a maximum around 3.2 eV. The ratio of intensity from this emission to
328
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
Fig. 38. Room-temperature RL spectrum of (a) an undoped LPE ZnO film and a hydrothermal grown bulk sample of ZnO. The integral under the curve between 3.1 eV and 3.18 eV contributed to the measurement of the decay time (b,c). Room-temperature RL decays from (b) a undoped ZnO and (c) an In-doped ZnO sample. Intensity is Log scale for graphs (b) and (c).
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
329
the broad band peak around 2.3 eV behaves like 2:1. The observed effect is attributed to a damaged surface layer due to machining [18,97] and, by contrast, is not observed by XRC where substrate and film show comparable FWHM. ZnO holds great expectations as scintillator material, particularly its very short luminescence decay time t in the sub-nanosecond range is of interest for applications in time-resolved devices like time-of-flight positron emission tomography (TOF-PET). Such TOF-PET devices in medical scanning applications could for example enable the detection of the initial state of a cancer due to the high spatial resolution available. Up to now, barium fluoride (BaF2) is the hottest candidate as a super-fast TOF-PET device. Very fast luminescent decay of BaF2 of 0.6 ns was reported [144]. Unfortunately, the density of BaF2 (4.88 g cm3) is rather low which may hinder its widespread application. High-density material, however, is required to provide high radiation stopping power so that high detection efficiency can be achieved. Consequently, the thickness of the detector can be lowered by using a high-density scintillator material, which is now a strong requirement from industry. Moreover, while BaF2 emits around 200 nm, the exciton wavelength of ZnO is around 400 nm which can be easily detected by a conventional photomultiplier. For Ga-doped ZnO powder the FWHM of the output pulse of 0.21 ns at 295 K was reported by Derenzo et al. [145] and about 0.65 ns FWHM upon exposure to a 241Am radiation source was demonstrated for an In-doped ZnO melt-grown crystal by Simpson et al. [138]. Doping with the group-V metals gallium and indium serves to decrease self-absorption at higher energies beyond 3.36 eV [97]. Luminescence decay time was measured for the energy region 3.1e3.18 eV, i.e. 400 lem 390 nm, which refers to the integral intensity as marked by the shaded area in Fig. 38a. An undoped, epi-ready ZnO wafer and the In-doped as grown platelet were exposed to a pulsed laser beam (160 fs pulse duration, lexc ¼ 260 nm, 1 kHz repetition rate, excitation energy < 1 mJ/pulse, 109 W cm2 peak intensity). The decay spectrum of the undoped and the In-doped sample is shown in Fig. 38b and c, respectively. Two decay components have been assigned for both samples: a fast one with 23 and 40 ps decay time and a slow one with 100 and 647 ps decay time for undoped and In-doped ZnO, respectively. The measured curves are fitted by a second exponential decay with offset: I(t) ¼ 2.2 e[t/0.023 ns] þ 0.057 e[t/0.1 ns] þ 0.00024 in the case of an undoped ZnO wafer and I(t) ¼ 0.315 e[t/0.04ns] þ 0.92 e[t/0.647ns] þ 5 e5 for the In-doped ZnO platelet. The rather noisy spectrum obtained from the undoped, epi-ready ZnO wafer for t > 0.2 ns is speculated to be due to a damaged surface layer as already discussed before. Consequently, emission quenching is likely, which causes the very weak signal and non-radiative energy transfer is strongly pronounced [97]. The better understanding of the scintillator properties of ZnO and the effects of doping and machining have been emphasized in our group since recently and more output may be expected in the forthcoming years. LPE is being used in this context as a tool of fast screening to obtain ZnO-based, thermodynamically stable phases with surfaces free of any mechanical damage due to machining. Also, the role of DAP recombination in the luminescence from LPE-grown ZnO films is now being investigated. 5. Summary and future trends Solvothermal technologies for ZnO crystals and films were reviewed. Among all the solvents investigated supercritical water turned out to be superior in terms of crystal perfection, process control and scalability. The hydrothermal technique produces the largest ZnO single crystals
330
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
with excellent structural properties. Two-inch size is now available from mass production and 3 inch size has already been demonstrated in this paper. It would be expected that the trend to increase ZnO crystal size and output will accelerate with increasing demand from industry. This would certainly contribute to a significantly lower price for ZnO wafers. However, critical issues remain like the quality of the surface of wafers. Manipulation of effects connected with surface conductivity is necessary. Doping of hydrothermal ZnO could open more application windows such as its role as a scintillator. In particular reproducible p-type doping is expected to boost ZnO technology. The fabrication of ZnO films from aqueous and chloride solutions may gain competitiveness to the growth from vapor phase for special applications in ZnO thin film technology. This is demonstrated for the case of homoepitaxy by LPE, where highly-crystalline ZnO films can be fabricated from alkaline-metal chloride solutions at temperatures <650 C. A hydrothermal and alkaline-metal free route produces thick ZnO films on foreign substrates as low as 90 C. Moreover, solution-grown ZnO films have the potential for in-situ doping since the high crystallinity of the film is preserved. Still, there are many open questions asking for continuation in both fields, the growth of large bulk crystals and films from liquid solutions.
Acknowledgements The authors are grateful to T. Ono and K. Maeda (Tokyo Denpa Co. Ltd., Tokyo) for provision of ZnO substrates, E. Ohshima and J.M. Ko (both formerly with Tohoku Univ.) for experiments, M. Miyamoto (Mitsubishi Gas, Tokyo) for some LPE experiments and SIMS measurements, B.-P. Zhang and N.T. Binh (Photodynamics Research Center/RIKEN, Sendai) for assistance in PL measurements, M. Nikl (Institute of Physics, AS CR, Prague) for RL investigation, P. Kiesel and O. Schmidt (Palo Alto Research Center Inc., Palo Alto) for Hall measurements and discussion, K. Sugiyama (Univ. of Tokyo) for measuring X-ray topography and T. Yao (Tohoku Univ.) for using the AFM. D.E. expresses his appreciation to F.F. Lange (Univ. of California, Santa Barbara) for discussion and providing data and thanks F. Orito (Mitsubishi Chemical Corp., Tokyo) for support. Finally, the great help of J.B. Mullin in the final revision of the manuscript is deeply appreciated. We gratefully acknowledge funding by following organizations: The Japanese Industrial Technology Research Grant Program in 03A26014a from New Energy and Industrial Development Organization (NEDO); The Special Coordination Fund by the Ministry of Education, Culture, Sports, Science; The technology program ‘‘Development of Growth Method of Semiconductor Crystals for Next Generation Solid-State Lighting’’. Particular gratitude is due to Mitsubishi Chemical Corp. and Tokyo Denpa Co. Ltd. for support over the years.
References [1] S.J. Pearton, D.P. Norton, K. Ip, Y.W. Heo, T. Steiner, Prog. Mat. Sci. 50 (2005) 293. [2] R. Triboulet, J. Perrie`re, Prog. Cryst. Growth Char. Mater. 47 (2005) 65. [3] A. Tsukazaki, A. Ohtomo, T. Onuma, M. Ohtani, T. Makino, M. Sumiya, K. Ohtani, S.F. Chichibu, S. Fuke, Y. Segawa, H. Ohno, H. Koinuma, M. Kawasaki, Nat. Mater. 4 (2005) 42. [4] A. Dadgar, A. Krtschil, A. Diez, F. Bertram, J. Bla¨sing, J. Christen, A. Krost, in: Third International Conference on Materials for Advanced Technologies, Singapore 2005, paper N-6-OR12.
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
331
[5] B.K. Meyer, H. Alver, D.M. Hofmann, W. Kriegseis, D. Forster, F. Bertram, J. Christen, A. Hoffmann, M. Straßburg, M. Dworzak, U. Haboeck, A.V. Rodina, Phys. Status Solidi B 241 (2004) 231. [6] R. Helbig, J. Cryst. Growth 15 (1972) 25. [7] J. Nause, B. Nemeth, Semicond. Sci. Technol. 20 (2005) S45. [8] T. Makino, Y. Segawa, M. Kawasaki, A. Ohtomo, R. Shiroki, K. Tamura, T. Yasuda, H. Koinuma, Appl. Phys. Lett. 78 (2001) 1237. [9] A. Ohtomo, M. Kawasaki, Y. Sakurai, I. Ohkubo, R. Shiroki, Y. Yoshida, T. Yasuda, Y. Segawa, H. Koinuma, Mater. Sci. Eng. B 56 (1998) 263. [10] A. Ohtomo, A. Tsukazaki, Semicond. Sci. Technol. 20 (2005) S1. [11] C. Klemenz, H.J. Scheel, J. Cryst. Growth 129 (1993) 421. [12] T. Kawaguchi, D.-H. Yoon, M. Minakata, Y. Okada, M. Imaeda, T. Fukuda, J. Cryst. Growth 152 (1995) 87. [13] P. Rogin, J. Hulliger, J. Cryst. Growth 179 (1997) 551. [14] C. Klemenz, J. Cryst. Growth 237e239 (2002) 714. [15] D. Ehrentraut, M. Pollnau, S. Ku¨ck, Appl. Phys. B 75 (2002) 59. [16] D. Ehrentraut, Y.E. Romanyuk, M. Pollnau, J. Ceram. Process. Res. 5 (2004) 256. [17] Y.E. Romanyuk, I. Utke, D. Ehrentraut, V. Apostolopoulos, M. Pollnau, S. Garc´ia-Revilla, R. Valiente, J. Cryst. Growth 269 (2004) 377. [18] D. Ehrentraut, H. Sato, M. Miyamoto, T. Fukuda, M. Nikl, K. Maeda, I. Niikura, J. Cryst. Growth 287 (2006) 367. [19] D. Andeen, L. Loeffler, N. Padture, F.F. Lange, J. Cryst. Growth 259 (2003) 103. [20] Z.L. Wang, Mater. Today 6 (2004) 26. [21] Z.R. Tian, J.A. Voigt, J. Liu, B. McKennzie, M.J. McDermott, M.A. Rodriguez, H. Konishi, H. Xu, Nat. Mater. 2 (2003) 821. [22] D. Elwell, H.J. Scheel, Crystal Growth from High-Temperature Solutions, Academic Press, London, New York, San Francisco, 1975. [23] J.W. Nielsen, E.F. Dearborn, J. Phys. Chem. 64 (1960) 1762. [24] W. Kleber, R. Mlodoch, Kristall Techn. 1 (1966) 249. [25] A.B. Chase, J. Osmer, J. Am. Ceram. Soc. 50 (1967) 325. [26] B.M. Wanklyn, J. Cryst. Growth 7 (1970) 107. [27] K. Oka, H. Shibata, S. Kashiwaya, J. Cryst. Growth 237e239 (2002) 509. [28] A.A. Chernov, Modern Crystallography III Crystal Growth, Springer-Verlag, Berlin Heidelberg, 1984 p. 161. [29] D.R. Linde (Ed.), Handbook of Chemistry and Physics, 72nd ed. RCR Press, Boca Raton, 1991. [30] G. Spezia, Atti Accad. Sci. Torino 44 (1909) 95. [31] E. Buehler, A.C. Walker, Sci. Monthly 69 (1949) 148. [32] A.C. Walker, J. Am. Ceram. Soc. 36 (1953) 250. [33] F. Iwasaki, H. Iwasaki, J. Cryst. Growth 237e239 (2002) 820. [34] M. Mikawa, private communication. [35] K. Byrappa, Hydrothermal growth of crystals, in: D.T.J. Hurle (Ed.), Handbook of Crystal Growth, Bulk Crystal Growth, 2a Basic Techniques, North-Holland, Amsterdam, 1994, p. 467. [36] C. Haar, J.S. Gallagher, G.S. Kell, NBS/NRC Steam Tables, Hemisphere Publishing, Washington DC, 1984. [37] J.D. Taylor, J.I. Steinfeld, J.W. Tester, Ind. Eng. Chem. Res. 40 (2001) 67. [38] L.N. Demianets, D.V. Kostomarov, I.P. Kuz’mina, S.V. Pushko, Crystallogr. Rep. 47 (Suppl. 1) (2002) S86. [39] L. DiLeo, D. Romano, L. Schaeffer, B. Gersten, C. Foster, M.C. Gelabert, J. Cryst. Growth 271 (2004) 65. [40] N. Sakagami, M. Yamashita, T. Sekiguchi, S. Miyashita, K. Obara, T. Shishido, J. Cryst. Growth 229 (2001) 98. [41] T. Sekiguchi, S. Miyashita, K. Obara, T. Shishido, N. Sakagami, J. Cryst. Growth 214/215 (2000) 72. [42] J. Mass, M. Avella, J. Jime´nez, M. Callahan, E. Grant, K. Rakes, D. Bliss, B. Wang, Mater. Res. Soc. Symp. Proc. 878E (2005) Y1.7.1. [43] R.A. Laudise, A.A. Ballman, J. Phys. Chem. 64 (1960) 688. [44] R.A. Laudise, E.D. Kolb, A.J. Caporaso, J. Am. Ceram. Soc. 47 (1964) 9. [45] E.D. Kolb, A.S. Coriell, R.A. Laudise, A.R. Hutson, Mater. Res. Bull. 2 (1967) 1099. [46] D.F. Croxall, R.C. Ward, C.A. Wallace, R.C. Kell, J. Cryst. Growth 22 (1974) 117. [47] N. Sakagami, K. Shibayama, Jpn. J. Appl. Phys. 20 (1981) 201. [48] M. Suscavage, M. Harris, D. Bliss, P. Yip, S.-Q. Wang, D. Schwall, L. Bouthillette, J. Bailey, M. Callahan, D.C. Look, D.C. Reynolds, R.L. Jones, C.W. Litton, MRS Internet J. Nitride Semicond. Res. 4S1 (1999) G3.40. [49] B. Wang, M.J. Callahan, L.O. Bouthillette, Chunchuan Xu, M.J. Suscavage, J. Cryst. Growth 287 (2006) 381. [50] B. Wang, M.J. Callahan, L.O. Bouthillette, Cryst. Growth Des. 6 (2006) 1256. [51] L.N. Demianets, D.V. Kostomarov, Ann. Chim. Sci. Mat. 26 (2001) 193.
332 [52] [53] [54] [55] [56] [57] [58] [59] [60] [61] [62] [63] [64] [65] [66] [67] [69] [70] [71] [72] [73] [74] [75] [76] [77] [78] [79] [80] [81] [82] [83] [84] [85] [86] [87] [88] [89] [90] [91] [92] [93] [94] [95] [96] [97] [98] [100]
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335 E. Ohshima, H. Ogino, I. Niikura, K. Maeda, M. Sato, M. Ito, T. Fukuda, J. Cryst. Growth 260 (2004) 166. K. Maeda, M. Sato, I. Niikura, T. Fukuda, Semicond. Sci. Technol. 20 (2005) S49. L. Yin, L. Zhang, F. Li, M. Yu, Mater. Res. Bull. 40 (2005) 2219. M. Bouloudenine, N. Viart, S. Colis, A. Dinia, Chem. Phys. Lett. 397 (2004) 73. W.A. Roth, P. Chall, Z. Elektrochem. 34 (1928) 185. And references therein. G.F. Hu¨ttig, H. Mo¨ldner, Z. Anorg. Allg. Chem. 211 (1933) 368. A.J. Caporaso, E.D. Kolb, R.A. Laudise, US patent 3,201,209 (1965). E.D. Kolb, R.A. Laudise, J. Am. Ceram. Soc. 48 (1965) 342. E.D. Kolb, R.A. Laudise, J. Am. Ceram. Soc. 49 (1966) 302. G.W. Morey, Paul Niggli, J. Am. Chem. Soc. 35 (1913) 1086. O.F. Tuttle, Am. J. Sci. 246 (1948) 628. Tokyo Denpa Co. Ltd.
. C.H. Seager, S.M. Myers, J. Appl. Phys. 94 (2003) 2888. Q.-S. Chen, V. Prasad, W.R. Hy, J. Cryst. Growth 258 (2003) 181. R. Dwilin´ski, J. Doradzin´ski, J. Garczyn´ski, L. Sierzputowski, J.M. Baranowski, M. Kamin´ska, Mater. Sci. Eng. B50 (1997) 46. I.L. Khodakovsky, A.E. Elkin, Geochemistry 10 (1975) 1490. J. Nause, in: The Third International Workshop on ZnO and Related Materials, Sendai, 2004, paper session V. J.-M. Ntep, S. Said Hassani, A. Lusson, A. Tromson-Carli, D. Ballutaud, G. Didier, R. Triboulet, J. Cryst. Growth 207 (1999) 30. H. Wenisch, V. Kirchner, S.K. Hong, Y.F. Chen, H.J. Ko, T. Yao, J. Cryst. Growth 227e228 (2001) 944. M.J. Vellekoop, C.C.O. Visser, P.M. Sarro, A. Venema, Sens. Actuators A 23 (1990) 1027. C.J. Youn, T.S. Jeong, M.S. Han, J.H. Kim, J. Cryst. Growth 261 (2004) 526. E.V. Lavrov, Physica B 340e342 (2003) 195. L.E. Halliburton, L. Wang, L. Bai, N.Y. Garces, N.C. Giles, M.J. Callahan, B. Wang, J. Appl. Phys. 96 (2004) 7168. Cermet Inc. . R.R. Reeber, K. Wang, Mater. Res. Soc. Symp. 622 (2000) T6.35.1. S. Adachi, Properties of Group-IV, IIIeV and IIeVI Semiconductors, Wiley & Sons, Chichester, 2005. Y. Waseda, K. Watanabe, H. Ohta, An-Pang Tsai, A. Inoue, T. Masumoto, Z. Naturforsch. 48a (1993) 784. W.J. Parker, R.J. Jenkins, C.P. Butler, G.L. Abbott, J. Appl. Phys. 32 (1963) 1979. SPC Goodwill. . Roditi International Corp. Ltd. . O.F. Schirmer, D. Zwingel, Solid State Commun. 8 (1970) 1559. D. Zwingel, J. Lumin. 5 (1972) 385. N.Y. Garces, L. Wang, L. Bai, N.C. Giles, L.E. Halliburton, G. Cantwell, Appl. Phys. Lett. 81 (2002) 622. R. Dingle, Phys. Rev. Lett. 23 (1969) 579. N.C. Giles, N.Y. Garces, L. Wang, L.E. Halliburton, in: Proceedings of SPIE, vol. 5359, 2004, p. 267. D.C. Reynolds, D.C. Look, B. Jogai, C.W. Litton, G. Cantwell, C.W. Harsch, Phys. Rev. B 60 (1999) 2340. D.C. Reynolds, D.C. Look, B. Jogai, C.W. Litton, T.C. Collins, W.C. Harsch, G. Cantwell, Phys. Rev. B 57 (1999) 12151. W. Go¨pel, Surf. Sci. 62 (1977) 165. P. Esser, W. Go¨pel, Surf. Sci. 97 (1980) 309. I.V. Markevich, V.I. Kushnirenko, L.V. Borkovska, B.M. Bulakh, Solid State Commun. 136 (2005) 475. See also references therein. O. Schmidt, P. Kiesel, C.G. Van de Walle, N.M. Johnson, J. Nause, G.H. Do¨hler, Jpn. J. Appl. Phys. 44 (2005) 7271. A.Y. Polyakov, N.B. Smirnov, A.V. Govorkov, E.A. Kozhukhova, S.J. Pearton, D.P. Norton, A. Osinsky, Amir Dabiran, J. Electron. Mater. 35 (2006) 663. D.C. Look, Mater. Sci. Eng. B 80 (2001) 383. M.W. Cho, C. Harada, H. Suzuki, T. Minegishi, T. Yao, H. Ko, K. Maeda, I. Nikura, Superlattice Microst. 38 (2005) 349. D. Ehrentraut, H. Sato, Y. Kagamitani, A. Yoshikawa, T. Fukuda, J. Pejchal, K. Polak, M. Nikl, H. Odaka, K. Hatanaka, H. Fukumura, J. Mater. Chem. 16 (2006) 3369. H.-J. Ko, M.-S. Han, Y.-S. Park, Y.-S. Yu, B.-I. Kim, S.S. Kim, J.-H. Kim, J. Cryst. Growth 269 (2004) 493. T.P. Niesen, M.R. De Guire, J. Electroceram. 6 (2001) 169.
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335 [101] [102] [103] [104] [105] [106] [107] [108] [109] [110] [111] [112] [113] [114] [115] [117] [118] [119] [120] [121] [122] [123] [124] [125] [126] [127] [128] [129] [130] [131] [132] [133] [134] [135] [136] [137] [138] [139] [140] [141] [142] [143]
333
H.Y. Xu, H. Wang, Y.C. Zhang, S. Wang, M.K. Zhu, H. Yan, Cryst. Res. Technol. 38 (2003) 429. M. Izaki, T. Omi, J. Electrochem. Soc. 144 (1997) L3. M. Izaki, J. Katayama, J. Electrochem. Soc. 147 (2000) 210. K. Govender, D.S. Boyle, P. O’Brien, D. Blinks, D. West, D. Coleman, Adv. Mater. 14 (2002) 1221. M. Ohyama, H. Kozuka, T. Yoko, S. Sakka, J. Ceram. Soc. Jpn. 104 (1996) 296. M. Ohyama, H. Kozuka, T. Yoko, Thin Solid Films 306 (1997) 78. B. Wessler, A. Steinecker, W. Mader, J. Cryst. Growth 242 (2002) 283. B. Wessler, F.F. Lange, W. Mader, J. Mater. Res. 17 (2002) 1644. J.-H. Lee, K.-H. Ko, B.-O. Park, J. Cryst. Growth 247 (2003) 119. D. Andeen, J.H. Kim, F.F. Lange, G.K.L. Goh, S. Tripathy, Adv. Funct. Mater. 16 (2006) 799. D. Andeen, F.F. Lange, Low Temperature, Hydrothermal Lateral Epitaxial Overgrowth of Zinc Oxide Abstract No. Y1.3, Materials Research Society, San Francisco, CA, 2005 Abstract Viewer. A.M. Peiro´, C. Domingo, J. Peral, X. Dome`nech, E. Vigil, M.A. Herna´ndez-Fenollosa, M. Mollar, B. Marı´, J.A. Ayllo´n, Thin Solid Films 483 (2005) 79. N.I. Kovtyukhova, E.V. Buzaneva, C.C. Waraksa, B.R. Martin, T.E. Mallouk, Chem. Mater. 12 (2000) 383. P.M.R. Kumar, C.S. Kartha, K.P. Vijayakumar, T. Abe, Y. Kashiwaba, F. Singh, D.K. Avasthi, Semicond. Sci. Technol. 20 (2005) 120. Q. Li, V. Kumar, Y. Li, H. Zhang, T.J. Marks, R.P.H. Chang, Chem. Mater. 17 (2005) 1001. A. Ohtomo, K. Tamura, K. Saikusa, K. Takahashi, T. Makino, Y. Segawa, H. Koinuma, M. Kawasaki, Appl. Phys. Lett. 75 (1999) 2635. P.M. Moran, F.F. Lange, Appl. Phys. Lett. 74 (1999) 1332. A.M. Chaparro, Chem. Mater. 17 (2005) 4118. C.G. Van de Walle, Phys. Rev. Lett. 85 (2000) 1012. D.C. Look, B.B. Claflin, G. Cantwell, S.-J. Park, G.M. Renlund, in: Third International Workshop on ZnO and Related Materials, Sendai, Japan, 2004, Abstracts, p. 2. H.J. Levinstein, S. Licht, R.W. Landorf, S.L. Blank, Appl. Phys. Lett. 19 (1971) 486. H. Sato, D. Ehrentraut, M. Miyamoto, Kyoung Jin Kim, O. Schmidt, P. Kiesel, T. Fukuda, Thermodynamic analysis of liquid phase epitaxy of ZnO and characteristics of Li:Mg0.06Zn0.94O/ZnO heterostructure, in press. Y. Marcus, Introduction to Liquid State Chemistry, John Wiley & Sons, London, New York, Sydney, Toronto, 1977 And references therein. B.P. Zhang, K. Wakatsuki, N.T. Binh, N. Usami, Y. Segawa, Thin Solid Films 449 (2004) 12. E.R. Segnit, A.E. Holland, J. Am. Ceram. Soc. 48 (1965) 409. H. Sato, D. Ehrentraut, T. Fukuda, Jpn. J. Appl. Phys. 45 (1A) (2006) 190. H. Sato, D. Ehrentraut, T. Fukuda, in: Third Conference on Crystal Growth and Crystal Technology, Beijing, 2005, paper P109. M. Lorenz, E.M. Kaidashev, H. von Wenckstern, V. Riede, C. Bundesmann, D. Spemann, G. Benndorf, H. Hochmuth, A. Rahm, H.-C. Semmelhack, M. Grundmann, Solid-State Electron. 47 (2003) 2205. K. Ogata, K. Koike, T. Tanite, T. Komuro, F. Yan, S. Sasa, M. Inoue, M. Yano, J. Cryst. Growth 251 (2003) 623. K. Sakurai, T. Kubo, D. Kajita, T. Tanabe, H. Takasu, Sz. Fujita, Sg. Fujita, Jpn. J. Appl. Phys. 39 (2000) L1146. A. Nakamura, J. Ishihara, S. Shigemori, K. Yamamoto, T. Aoki, H. Gotoh, J. Temmyo, Jpn. J. Appl. Phys. 44 (2005) L4. S. Shigemori, A. Nakamura, J. Ishihara, T. Aoki, J. Temmyo, Jpn. J. Appl. Phys. 43 (2005) L1088. A. Nakamura, J. Ishihara, S. Shigemori, K. Yamamoto, T. Aoki, H. Gotoh, J. Temmyo, Jpn. J. Appl. Phys. 43 (2004) L1452. Th. Gruber, C. Kirchner, R. Kling, F. Reuss, A. Waag, F. Bertram, D. Forster, J. Christen, M. Schreck, Appl. Phys. Lett. 83 (2003) 3290. R.D. Shannon, Acta Cryst. A32 (1976) 751. S. Kohiki, M. Nishitani, T. Wada, J. Appl. Phys. 75 (1994) 2069. P.J. Simpson, R. Tjossem, A.W. Hunt, K.G. Lynn, V. Munne´, Nucl. Instrum. Methods A 505 (2003) 82. M. Nikl, private communication. Y.S. Yu, G.Y. Kim, B.H. Min, S.C. Kim, J. Eur. Ceram. Soc. 24 (2004) 1865. D.H. Fan, Z.Y. Ning, M.F. Jiang, Appl. Surf. Sci. 245 (2005) 414. T. Koida, S.F. Chichibu, A. Uedono, A. Tsukazaki, M. Kawasaki, T. Sota, Y. Segawa, H. Koinuma, Appl. Phys. Lett. 82 (2003) 532. H.J. Ko, Y.F. Chen, S.K. Hong, H. Wenisch, T. Yao, D.C. Look, Appl. Phys. Lett. 77 (2000) 3761.
334
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
[144] K. Kamada, T. Nawata, Y. Inui, H. Yanagi, H. Sato, A. Yoshikawa, M. Nikl, T. Fukuda, Nucl. Instrum. Methods A 537 (2005) 159. [145] S.E. Derenzo, M.J. Weber, M.K. Klintenberg, Nucl. Instrum. Methods A 486 (2002) 214.
Dr. Dirk Ehrentraut was born in Germany in 1968. He was educated at Humboldt University of Berlin, Germany where he obtained a Diploma in crystallography in 1995. From 1995 to 1997 he was Researcher at the Institute of Crystal Growth, Berlin. In 1997 he joined the Institute of Micro- and Optoelectronics at the Swiss Federal Institute of Technology Lausanne (EPFL), Switzerland. From 2000 to 2003 he worked at the Institute of Applied Optics of the EPFL from where he obtained a Doctorate in Science for his thesis on novel coherent light sources. Since end of 2003 he is invited as Visiting Associated Professor at the Institute of Multidisciplinary Research for Advanced Materials (IMRAM), Tohoku University, Sendai, Japan, where his focus is on crystal chemistry and growth of wide band gap materials (ZnO, III-Nitrides) and sesquioxides fabricated from liquid solution. His field of competence includes the growth of bulk crystals and single-crystalline films of a large variety of oxides and semiconductors applying vapor phase and liquid phase (melt, solution) technologies.
Hideto Sato was born in Japan in 1973. He graduated from the Ritsumeikan University in Shiga, Japan and afterwards joined Murata Manufacturing Co., Ltd. where his research was focused on piezoelectric bulk crystals. From 2004 on he worked on the growth of ZnO film within the frame of a joined project between Murata Manufacturing Co., Ltd. and IMRAM.
Yuji Kagamitani was born in 1977 in Japan. He obtained a Masters in Science (MSc.) in 2002 from the Department of Chemistry, Faculty of Science, Tohoku University, Sendai, Japan. He is currently a PhD student at IMRAM, Tohoku University, preparing a PhD thesis entitled ‘‘Study on Crystal Growth of Zinc Oxide with Metal Ion Doping and Its Scintillation Properties’’. His research interest is on the solvothermal growth of ZnO and GaN crystals.
Dr. Hiroki Sato was born in 1974 in Yamagata/Japan. He graduated with a MSc. from the Department of Metallurgy, Graduate School of Engineering, Tohoku University in 1997 and obtained a Doctorate from Tohoku University in 2005. Between 1997 and 2003 he was employed by NEC Tokin Corp. where he worked on CdMnTe/CdMnHgTe solid solution and rutile for optical applications. In 2003 he joined Fukuda X’tal Lab. Inc. to continue his research on fluorides and scintillator materials.
D. Ehrentraut et al. / Progress in Crystal Growth and Characterization of Materials 52 (2006) 280e335
335
Prof. Akira Yoshikawa was born in Japan in 1970. He obtained a MSc. and Doctorate in Science from the University of Tokyo in 1996 and 1999, respectively. He became Research Associate of the Fukuda laboratory at the Institute of Materials Research (IMR), Tohoku University, Sendai in 1997. In 2002 he was assigned Research Associate at IMRAM, Tohoku University, Sendai. Since October 2003 he is an Associate Professor at IMRAM and temporary a Visiting Associate Professor of the University Claude Bernard Lyon 1, France. His research interests are on the melt growth of single-crystalline oxides and fluorides for scintillator and laser applications and the solvothermal growth of wide band gap semiconductors ZnO and GaN.
Prof. Tsuguo Fukuda was born in Japan in 1939. In 1964 he graduated from the Faculty of Science, University of Tokyo, Japan. During 24 years of working in Tokyo Shibaura Electric Co. (named later Toshiba Corp.) including 5 years working in the Optoelectronic Joint Research Lab. in Japan, in 1971 he obtained a PhD from the University of Tokyo for his thesis on ferroelectric crystals. He became Professor of IMR, Tohoku University in 1987. In 2002, he became Research Professor of IMRAM, Tohoku University. His research interests are crystal growth technology of bulk single crystals for optical applications. Presently, one of his main research topic is hydrothermal and ammonothermal growth of ZnO and GaN. He is author or co-author of more than 600 papers. He is the founder and president of Fukuda X’tal Lab. Inc.