CONTRIBUTED P A P E R
The Ammonothermal Crystal Growth of Gallium NitrideVA Technique on the Up Rise Progress for over a decade has resulted in fabrication of large, single, gallium nitride crystals with structural properties that make them suitable for production of optoelectronic devices. By Dirk Ehrentraut and Tsuguo Fukuda
ABSTRACT | Gallium nitride (GaN) is one of the most important wide band gap semiconductor materials in modern technology with even higher expectations for future applications it is ought to play a crucial role. Among this, the growth of lattice and thermally matched GaN substrates for the GaN device technology takes an essential piece. This paper is reporting on the achievements in the ammonothermal growth technique of GaN bulk crystals. Important features specific to the ammonothermal technique are focused on. Although only a few groups (currently G 10 worldwide) are directly involved in the development of the ammonothermal bulk crystal growth technology, partly due to the extreme technological challenges, tremendous progress over the last decade has recently resulted in the fabrication of 2 inch large, freestanding, single crystalline GaN with excellent structural perfection. KEYWORDS | Ammonothermal method; crystal growth; gallium nitride; mineralizer
I. INTRODUCTION The wide-band-gap semiconductor hexagonal gallium nitride (GaN, Eg ¼ 3:42 eV at 300 K) is essential in modern optoelectronics, high-frequency high-power electronics, fast-speed communication, etc., due to high electron mobility ( 1800 cm2 V s [1]), saturation velocity ð 19 106 cm s1 Þ, and breakdown voltage ðVBR ¼ 26 33 106 cm s1 Þ [2]. Manuscript received April 20, 2009; revised June 16, 2009 and August 4, 2009; accepted August 4, 2009. Date of publication November 17, 2009; date of current version June 18, 2010. The authors are with the WPI-Advanced Institute for Materials Research, Tohoku University, Sendai, Miyagi, 980-8577 Japan (e-mail:
[email protected];
[email protected]). Digital Object Identifier: 10.1109/JPROC.2009.2029878
1316
Proceedings of the IEEE | Vol. 98, No. 7, July 2010
Comparing the GaN device technology to some of the classic semiconductors like silicon (Si) or other group III-V materials like gallium arsenide (GaAs) and gallium phosphide (GaP), besides the physical properties, the most striking difference is that the availability and improvements in the quality of the bulk crystals has made possible the achievements in the development and industrialization of devices. Melt growth techniques such as the Czochralski and Bridgman methods can be applied to produce crystals large enough to manufacture 4 inch large wafer. However, the GaN-based device technology fully relies on the use of foreign substrates like sapphire ð Al2 O3 Þ and silicon carbide (4H- or 6H-SiC) due to the lack of free-standing and high quality GaN wafers. It is due to thermodynamic limitation of GaN, e.g., decomposition at 1150 K under atmospheric pressure [3], that it cannot be grown from the stoichiometric melt without extreme pressure and temperature of 9 6 GPa at 2220 C [4]. Hence, only vapor phase and solution techniques are appropriate to produce group-III nitride crystals in an economic way. GaN lattice-matched substrates for the group-III nitride (AlN, GaN, InN, and their alloys) device technology are currently being produced by the hydride vapor phase epitaxy (HVPE) solely. More recently, however, the ammonothermal growth of GaN has emerged as a powerful technique with great promises for the mass-production of large size GaN crystals [5]. The ammonothermal technique belongs to the wide family of solvothermal techniques, employing a polar solvent of inorganic or organic nature under subcritical or even supercritical conditions to dissolve and re-crystallize a polar material. The polar solvent, water ðH2 OÞ for the hydrothermal and ammonia ðNH3 Þ for the ammonothermal technique, forms metastable products with the solute 0018-9219/$26.00 Ó 2009 IEEE
Authorized licensed use limited to: TOHOKU UNIVERSITY. Downloaded on June 16,2010 at 01:58:55 UTC from IEEE Xplore. Restrictions apply.
Ehrentraut et al.: The Ammonothermal Crystal Growth of Gallium NitrideVA Technique on the Up Rise
(also called nutrient). The solubility of a solute can be amplified through use of a suited mineralizer (additive). If a closed system is utilized, and exchange of matter with ambient is impossible, often the solvent takes over a supercritical state, which also leads to improve the solubility. Merits of the solvothermal growth technology comprise the operation near the thermodynamic equilibrium to generate a high crystallinity; control of large crystal quantities over long process time (9 1000 crystals per growth cycle in case of the low temperature modification of quartz, SiO2 , [6]), thus enabling a high throughput; no need for vacuum technology; environmentally benign conditions for production and capability for recycling of the solution. The hydrothermal growth of SiO2 is already over 60 years in use for mass-production. Another example, zinc oxide (ZnO) is the first semiconducting crystal grown at industrial scale by a solvothermal route for the purpose of wafer production. The solvothermal growth of hydrothermal ZnO and ammonothermal GaN has recently been summarized [5]–[8]. This paper solely focuses on the recent progress in the ammonothermal growth of hexagonal GaN bulk crystals. Firstly, a brief history is given. Next, connected with the nature of crystal growth from solutions are two very specific features which will therefore be discussed: the effect of mineralizer, determining the chemistry of the solution, to improve the solubility, consequently, the chemistry of the solution in order to better understands the challenges inherent to this technique. Lastly, the quality of fabricated GaN crystals is compared and requirements for further improvement are given.
II . BRIEF HISTORY OF AMMONOTHERM AL CRYSTAL GROWTH OF BULK GaN The ammonothermal route has been used by Jacobs et al. as a pathway to synthesize crystals like aluminum nitride (AlN) from liquid phase using a transporting agent [9]. This knowledge had been applied to GaN where the synthesis of small-size, up to 25 (m large GaN crystals from a solution of supercritical NH3 and basic mineralizers like lithium and potassium amide (LiNH2 and KNH2 ) at T 550 C and p ¼ 400–500 MPa was reported in 1997 by Dwilicski et al. [10]. Purdy et al. have investigated the phase stability of selfnucleated GaN microcrystals grown from different acidic and neutral mineralizers like NH4 X and LiX (X ¼ Cl, Br, I) at T ¼ 470 C–510 C and p ¼ 206–276 MPa [11]. Often the cubic phase of GaN was obtained and even dominant in quite some cases where a large amount of LiX co-mineralizer had been applied. Purdy et al. already mentioned the use of platinum (Pt)-lined autoclave for up scaling the process.
Ketchum and Kolis produced small single crystals up to 0. 5 mm in size at T ¼ 400 C and p ¼ 240 MPa from the mixed mineralizer of KNH2 and potassium iodide (KI) [12]. Yet another chemistry has been applied by Demazeau et al. who employed hydrazine hydrochloride ðNH2 NH3 ClÞ and sodium azide ðNaN3 Þ to synthesize micrometer-large GaN crystals at T ¼ 400 C–800 C and p ¼ 100–200 MPa [13]. D’Evelyn et al. used the high-pressure ammonothermal method and an undisclosed mineralizer at T ¼ 600 C–1000 C and p ¼ 500–2000 MPa to grow millimeter-sized spontaneously nucleated GaN crystals as well as GaN seeded on GaN substrate crystals [14]. More recently, the uniform growth of GaN on an over1-inch HVPE seed at T ¼ 625 C–675 C and p ffi 214 MPa was reported by Hashimoto et al. [15]. The same group has demonstrated a 5 mm large well-facetted GaN crystal nucleated on a GaN HVPE seed [16]. Wang et al. [17] have demonstrated GaN single crystals of the size 10 mm 10 mm 1 mm which were grown on a GaN HVPE seed crystal under similar growth conditions of T ¼ 475 C–625 C and p ¼ 100–300 MPa. They realized an average growth rate up to 50 m per day. In both cases, basic mineralizers (NaNH2 þ NaI and KNH2 , respectively) were employed in a nickel (Ni)-based superalloy autoclave. However, it has been reported that using basic mineralizers Ni would contribute to impurities at the interface of the grown GaN crystal to the seed [18]. In addition, Ni was suspected to act as catalyst for GaN [19]. The acidic mineralizer NH4 X (X ¼ Cl, Br, I) is the choice of the research team around Tohoku University [20], [21]. The growth conditions are T ¼ 490 C–550 C and p 150 MPa. The first crack free ammonothermal GaN crystal on an HVPE GaN seed of 1 cm2 was grown around the year 2005 and later on 1 inch and 2 inch large HVPE GaN seed crystals were used to fabricate ammonothermally grown GaN. Fig. 1 shows some examples of crystals with thicknesses of the ammonothermally grown crystal fraction of 0.2 mm (1a) and 0.5 mm (1b, c). The first report on a 1 inch large GaN wafer sliced from a true bulk GaN crystal grown by the basic ammonothermal technique was published by Dwilicski et al. in 2008,
Fig. 1. GaN grown by the acidic ammonothermal technique and nucleated on HVPE seed crystals: (a) a 1 cm2 large crystal, (b) a 3 cm diameter large crystal, and (c) an earlier attempt on the growth of GaN on a 2-in large wafer crystal in the 10 cm inner diameter autoclave.
Vol. 98, No. 7, July 2010 | Proceedings of the IEEE
1317
Authorized licensed use limited to: TOHOKU UNIVERSITY. Downloaded on June 16,2010 at 01:58:55 UTC from IEEE Xplore. Restrictions apply.
Ehrentraut et al.: The Ammonothermal Crystal Growth of Gallium NitrideVA Technique on the Up Rise
followed by a paper in 2009 which shows a 1.5 inch large epitaxy-ready (epi-ready) GaN wafer [22], [23]. In Fig. 2 is shown a 2 inch large GaN grown by the Polish company and a 1 inch large epi-ready GaN wafer. Whereas the basic ammonothermal crystal growth technique is being investigated for about 15 years already, around 5 years were spent thus far to investigate the challenges in the acidic ammonothermal technique. Both numbers represent a far shorter time interval, not to mention monetary support and research personal involved, compared to the HVPE or other techniques which are able to produce bulk GaN crystals.
III . BASIC FEATURES OF T HE AMMONOTHERMAL T ECHNIQUE A. Experimental Conditions The ammonothermal technique employs NH3 as solvent to fabricate GaN crystals, consequently, oxygen (O)-free conditions are essential. Aspects of the preparation of ammonothermal crystal growth can be found elsewhere [5], [7], [10]–[12], and [15]. The autoclave (Fig. 3) is loaded with GaN, mineralizer, a baffle plate to design a temperature gradient, and seed crystals mounted on a holder. This preparation is typically done in a glove box if the autoclave dimension does allow it. After sealing the autoclave, repeatedly flushing with high purity nitrogen ðN2 Þ gas is done prior loading with NH3 . Two or three heaters are assembled around the autoclave to create a suited temperature profile inside. A typical experiment contains the heating phase, dwell time during which the crystals are grown, and finally the cooling down of the autoclave. Unfortunately, greater details are typically subject to intellectual properties and are therefore not published. B. Basic Versus Acidic Mineralizer The use of basic mineralizers in supercritical ammonia causes a retrograde solubility of GaN [17], [22], [24], [32], [33]. This means that a lightly increased temperature in
Fig. 2. GaN grown by the basic ammonothermal method: (a) a 2 inch large square-shaped GaN crystal as grown and (b) a 1 inch large epiready wafer. Figures are courtesy of Ammono Sp. z o.o., Warsaw, Poland.
1318
Fig. 3. An autoclave used for research on ammonothermal growth of GaN using acidic mineralizer.
the GaN crystal growth zone of the autoclave is required to trigger nucleation on a given GaN seed crystal. Besides, growth temperature and pressure is very high, which is a real challenge for the alloy used to make the autoclave. Knowingly, the choice of suited alloys as autoclave material under such harsh conditions is limited, though ongoing developments in the steel industry are likely to deliver steadily improved alloys. From a standpoint of industry, the general trend in the growth technology from autoclaves is to reduce temperature and pressure of a process as much as possible to generate a more economic technology. We at Tohoku University work with the acidic mineralizer NH4 X and in particular with NH4 Cl to greatly enhance the solubility of GaN precursor. Latter consists of small HVPE-GaN crystals or pressed GaN powder [25]. A Pt-based inner liner is necessary to prevent the autoclave from corrosion. Typical dimensions for an autoclave are a few millimeter inner diameter (I.D.) and 15–40 cm inner length (I.L.) for research purposes, and 10 cm I.D. and several meters I.L. for production type autoclaves. Definitely, a goal consists in replacing the Pt-based liner, again to bring down the overall costs of the entire process, and work towards its realization is currently going on. Fascinatingly, the average growth rates are similar for basic and acidic ammonothermal conditions; around 30–80 m per day was already reported by several groups [17], [21], [25], [26]. This strongly points to similar growth kinetics despite the differences in the chemistry of the solution. Ehrentraut et al. have recently estimated the maximum stable growth rate being 170 m per day for ammonothermally grown GaN on a 2-in large GaN seed crystal [27]. Note that this value is the sum of two faces,
Proceedings of the IEEE | Vol. 98, No. 7, July 2010
Authorized licensed use limited to: TOHOKU UNIVERSITY. Downloaded on June 16,2010 at 01:58:55 UTC from IEEE Xplore. Restrictions apply.
Ehrentraut et al.: The Ammonothermal Crystal Growth of Gallium NitrideVA Technique on the Up Rise
i.e., around 85 m per day per face is averaged. Despite uncertainties, the experimental and theoretical results suggest that GaN crystals may be obtained at a reasonable growth rate by the ammonothermal technique. This is supported by the experimental finding that scaling up to larger autoclave sizes is realizing the better control over temperature gradients and hydrodynamic conditions, thus mass transport. For comparison, about 200 m and 500–600 m per day are typical growth rates for highquality crystals of ZnO and SiO2 , respectively [6].
C. Chemistry of the Solution A crucial point is the understanding of the chemistry of the solution during the whole cycle, which can be divided into the three cycle parts of system heating, crystal growth, and system cooling. Each cycle has certain characteristics in its temperature-pressure scheme; consequently, the chemical equilibrium differs. While the cycle part of cooling down to room temperature has no other meaning than reducing the thermal stress of the grown crystals and avoiding parasitic nucleation on them, the cycle part of heating determines the degree of supersaturation during the formation of stable nuclei (i.e. commencing crystal growth), which in term determines the quality of nucleation (island or step flow), hence the structural quality of the crystal to be grown. Once nucleation has been triggered, the crystal growth part of the cycle requires utmost stability of the environment, i.e., fluctuations in temperature and concentration has to be avoided to make sure no entrapment of solution and impurity molecules and consequently defect generation occurs. In reality, thus far, the chemistry of the solution in both cases for the acidic and basic ammonothermal environment remains unknown to a large amount. Because the relevant nitride chemistry requires very special p T conditions, consequently, establishing the suited experimental facilities fro direct monitoring is challenging. The reproducibility in crystal quality is purely based on a large number of systematic experimentsVthe Bclassic[ approach many of the industrially produced crystals have once started off. Our recent research on acidic mineralizers aimed at optimizing them to achieve a high yield of hexagonal GaN in order to vary the speed of the entire process from dissolving the GaN precursor until the re-crystallization on the GaN seed [21]. The employment of the mixed mineralizers like 80 mol% NH4 Cl= 20 mol% NH4 Br and 80 mol% NH4 Cl= 20 mol% NH4 I has been proven successful in terms of crystal yield and phase stability. The crystal yield, i.e., amount of dissolved and re-crystallized GaN feedstock increases with increasing acidity from NH4 Cl to NH4 I. Conversely, there is a limit in the acidity above which cubic phase GaN is growingly obtained, see Fig. 4. Acidic mineralizers such as NH4 Cl are likely to form ammonium chlorogallates being similar to ðNH4 Þ3 GaCln ,
Fig. 4. Comparison of yields of self-nucleated GaN synthesized under different acidities of the mineralizer during 96 h growth cycles. The appearance of purely hexagonal GaN (h) and clear evidence of cubic GaN (c) by means of x-ray diffraction is denoted; after [21].
quite in analogy to ammonium hexafluorogallate, ðNH4 Þ3 GaF6 [28]. The ammonolysis of the latter compound tends to produce GaN at about 400 C. The pentaaminechlorogallium (III) dichloride ½GaðNH3 Þ5 Cl Cl2 has recently been suspected to be an effective precursor for the growth of GaN from a solution of NH3 NH4 Cl. This specie has been synthesized at 840 K [29] and is stable under the temperature conditions of acidic ammonothermal growth of GaN. The cationic ½GaðNH3 Þ5 Cl 2þ octahedra are surrounded by weakly-bonded distorted cubes of Cl anions, which relatively easy disconnect from the cationic octahedra. The said free Cl anions might straightforwardly form NH4 Cl, i.e., a mineralizer molecule again with the NHþ 4 derived from the autoprotolysis of NH3 , accordingly 2NH3 $ NHþ 4 þ NH2 . The KNH2 is the widely used mineralizer for the basic ammonothermal growth of GaN and is supposed to act as oxygen trap by forming KOH [30]. The latter is almost insoluble in NH3 . It has been postulated that following reaction would be favorable when KNH2 is used: KNH2 þ GaN þ 2NH3 $ KGaðNH2 Þ4 , whereas the use of NaNH2 as mineralizer would favor the reaction like 2NaNH2 þ GaNþ 2NH3 $ Na2 GaðNH2 Þ5 [17], [31]. Needless to say, knowledge of the solubility of a solute is a key figure in any solvothermal technology. Using a highpressure cell experimental setup we have recently reported the solubility of GaN under acidic conditions in the NH4 Cl NH3 solution [5]. In Fig. 5 is shown the solubility for different concentrations of NH4 Cl with respect to NH3 : molar ratio NH4 Cl/NH3 ¼ 0:0127, 0.032, and 0.127. Increasing the concentration of mineralizer in the solvent NH3 improves the solubility of GaN. Such behavior is generally known as regular solubility. Not shown in Fig. 5 is the data fitting by a simple first exponential function Vol. 98, No. 7, July 2010 | Proceedings of the IEEE
1319
Authorized licensed use limited to: TOHOKU UNIVERSITY. Downloaded on June 16,2010 at 01:58:55 UTC from IEEE Xplore. Restrictions apply.
Ehrentraut et al.: The Ammonothermal Crystal Growth of Gallium NitrideVA Technique on the Up Rise
Fig. 5. Regular solubility of GaN, and AlN for comparison, in solutions of NH3 with NH4 Cl. Molar ratio NH4 Cl=NH3 ¼ 0:0127 (h), 0.032 (r), and 0.127 ð Þ. The increase of mineralizer leads to improved solubility. The case of AlN shows that increasing the acidity of the mineralizer by adding 8 mol% NH4 I to NH4 Cl leads to reduced solubility in case of AlN.
(Arrhenius plot: Log solubility over T 1 ), which yields a strictly linear behavior of solubility over temperature. This clearly indicates the dominance of one chemical process in the dissolution of GaN under given conditions and, indeed, is a basic requirement to control the growth from a solution. The energy of formation was estimated to 15.9 kcal mol1 for the range T ¼ 673–820 K [5]. It needs to be said, from our experiments a minimum concentration of 0.3 mol% NH4 Cl is needed in practice to yield a reasonable growth rate of GaN on a GaN seed crystal and the minimum temperature should be no lower than 450 C. Wang et al. have reported on the retrograde solubility, i.e., a negative solubility coefficient of GaN in the KNH2 NH3 solution with 3:5 0:5 M KNH2 , see Fig. 6 [17]. They found that the solubility changes from 1–10% with lowering the temperature from 600 to 400 C. Dwilicski et al. have earlier shown that the retrograde solubility of GaN decreases with temperature but increases with rising pressure. About 3 mol% GaN would dissolve in a solution of KNH2 and NH3 , molar ratio KNH2 =NH3 ¼ 0:07, at T ¼ 400 C and p ¼ 300 MPa compared to about 2 mol% GaN at T ¼ 500 C and p ¼ 300 MPa [32].
in case of ZnO [6]. Moreover, the mineralizers and NH3 are a source for O and H and great effort is currently directed to improve this situation. The highest quality of ammonothermally grown GaN crystals was reported recently by Dwilicski et al. [33]: etch pit density (EPD; molten KOH at 400 C and 5 min; to determine defects including dislocations) of 5 103 cm3 , x-ray rocking curve (XRC) full-with half-maximum (FWHM) from (0002) reflection of 16 arcsec for the N polar face, and a curvature radius around 102 103 m. The structural quality is certainly outstanding, comparing to widely use free-standing GaN substrates produced by the hydride vapor phase epitaxy (HVPE) method. Here, using cathodoluminescence (CL), typical values for threading dislocation density and curvature radius are 5 106 cm2 , lowest around 2:5 106 cm2 , [34] and 10 m, respectively. Above paper by Dwilicski et al. [33] further reported on doping issues by their basic ammonothermal technique and detailed following carrier concentration ðnÞ and resistivity ðÞ: n-type doping (n ¼ 1018 cm3 , ¼ 103 102 cm), p-type doping (p ¼ 1018 cm3 , ¼ 101 102 cm), and semi-insulating ð ¼ 106 1012 cmÞ. Following impurities have been reported: O and Si up to 1019 cm3 [35]. Incorporation of impurities is obviously a critical point for the ammonothermal technique, at least at its current state. Table I gives an overview over impurities in bulk GaN crystals prepared by the basic and acidic ammonothermal technique and detected by SIMS. This is compared to the latest results achieved by Mitsubishi Chemical who uses the HVPE bulk technique to grow GaN crystals [36]. The main source for Cr, Fe, and Ni is the
IV. QUALITY OF GALLIUM NI T RI DE CRY ST AL S The crystals in Figs. 1 and 2, though produced under acidic and basic conditions, appear in a translucent brownish color which is due to defects. Nitrogen deficiency has been claimed to cause coloration [17]. However, metal impurities like Ni, Fe, Cr (see below) are also likely to contribute to coloration perhaps in a similar manner as Fe or Al does 1320
Fig. 6. Retrograde solubility of GaN in the KNH2 NH3 solution. The KNH2 concentration is about 3.5 0.5 M, T ffi 10 K cm1 , p ¼ 120 240 MPa [17]. Reprinted with permission. Ó 2006, Elsevier Publishers.
Proceedings of the IEEE | Vol. 98, No. 7, July 2010
Authorized licensed use limited to: TOHOKU UNIVERSITY. Downloaded on June 16,2010 at 01:58:55 UTC from IEEE Xplore. Restrictions apply.
Ehrentraut et al.: The Ammonothermal Crystal Growth of Gallium NitrideVA Technique on the Up Rise
Table 1 Impurities in Unintentionally Doped Gallium Nitride Crystals Prepared by the Basic and Acidic Ammonothermal Technique in Comparison to the HVPE Technique
autoclave alloy; Pt is probably derived from the inner liner in case of the acidic ammonothermal technique. Si and O are likely to come from the NH3 , the mineralizer, and the GaN feedstock. Comparing the results in Table 1, it gets evident that the purity of Mitsubishi’s crystals is much higher at this point. Hashimoto et al. have estimated by transmission electron microscope (TEM) observation a threading dislocation density of less than 106 cm2 for the Ga polar face and around 107 cm2 for the N polar face of GaN produced by means of basic mineralizer [16]. The same paper reports that the XRC FWHM monitored from the ð1010Þ reflection of m-plane wafers cut from a crystal have revealed a multiple grain structure and the narrowest FWHM of 220 arcsec. Yet another group disclosed the XRC FWHM recorded from the (0002) reflection of the Ga and N polar face being 535 arcsec and 859 arcsec, respectively [17]. The acidic ammonothermal growth has yielded homogeneous crystals with the XRC FWHM from the (0002) reflection with values around 108 and 339 arcsec for the Ga-polar and N-polar face, respectively. Not shown here are photoluminescence (PL) measurements from different GaN crystals grown under basic and acidic conditions. Using a 325 nm CW He-Cd laser we have measured the low-temperature (10 K) Pl signal from the near band edge region for a recently grown GaN crystal employing acidic mineralizer. In the near band gap region emission peaks were found at 3.27 eV, 3.378 eV, 3.451 eV 3.456 eV, 3.466 eV, and 3.472 eV, which were ascribed as donor-acceptor pair recombination (DAP), first ABO low phonon replica (LO), two-electron satellite (TES), elastic scattering of free exciton at a Si donor, acceptor bound exciton ðA0 XÞ, and donor bound exciton ðD0 XÞ, respectively, [37]. Red luminescence was observed peaking around 1.93 eV [27].
Wang et al. observed a weak broad band centered around 2.2 eV and a wide band edge peak at 3.46 eV from both polar faces with the exception of a strong donor acceptor related peak around 2.9 eV solely observed from the Ga polar face [17]. Hashimoto et al. noticed a dominant yellow luminescence around 2.4 eV for the Ga polar face whereas the emission from the band edge around 3.4 eV was more obvious from the N polar face [15]. The spectra were qualitatively homogeneous over the entire crystal face. Chichibu et al. have investigated thick, 4 m and 5 m for N-polar and Ga-polar face, respectively, GaN films nucleated on (0001) GaN HVPE seed crystals using spatially resolved CL [38]. They noticed that low threading dislocation density GaN can be grown by the ammonothermal method on Ga-polar GaN seed crystals due to dislocation bending during the facet growth. The N-polar film contained high density residual electrons likely due to incorporated O and Si impurities. We have undertaken TEM investigation at the interface of 1–10 m thick ammonothermal GaN films to the (0001) GaN HVPE seed crystal and have found similar effect of dislocation bending [39]. Depending on the growth conditions a 100–500 nm wide zone, corresponding to the area where first nucleation and subsequent growth takes place, with high dislocation density can usually be found. After that zone, low dislocation density of the order comparable to the HVPE substrate was observed. Dwilicski et al. have detailed a study on a GaN homoepitaxial layer deposited by MOCVD on top of a semi-insulating GaN substrate fabricated by the basic ammonothermal technique [23]. Fig. 7 shows spectra from micro-PL measurements (T ¼ 4:2 K at different excitation
Fig. 7. Low-temperature micro-PL spectra taken at different excitation powers from a homoepitaxial GaN film grown on semi-insulating ammonothermal GaN substrate. Emission from free exciton A and B (FX A and FX B , respectively), exciton A and B bound to neutral donors (D01 X A ; D 02 X A ; D 03 X A and D 0 X B , respectively), exciton bound to neutral acceptor ðA0 XÞ, and two-electron satellite (TES) is obtained [23]. Reprinted with permission. Ó 2009, Elsevier Publishers.
Vol. 98, No. 7, July 2010 | Proceedings of the IEEE
1321
Authorized licensed use limited to: TOHOKU UNIVERSITY. Downloaded on June 16,2010 at 01:58:55 UTC from IEEE Xplore. Restrictions apply.
Ehrentraut et al.: The Ammonothermal Crystal Growth of Gallium NitrideVA Technique on the Up Rise
powers of 0.01–2 mW). The well-resolved excitonic structure was revealed with the A0 X peak at 3.4663 eV appearing as narrow band with the FWHM of 0.3 meV. The emission quality was reported to be highly uniform over the entire film outlining that ammonothermally grown GaN enables the growth of high-quality (low strain) homoepitaxial films.
V. CONCLUSION AND PROSPECTIVES FOR AMMONOTHERMAL GaN CRYSTALS The ammonothermal growth of large size GaN crystals was briefly reviewed. Basic and acidic mineralizers can be employed for solubility enhancement, leading to a retrograde or regular solubility of GaN, respectively, under ammonothermal conditions. Remarkable improvements in crystal size and quality have been obtained
REFERENCES [1] M. Sakai, H. Ishikawa, T. Egawa, T. Jimbo, M. Umeno, T. Shibata, K. Asai, S. Sumiya, Y. Kuraoka, M. Tanaka, and O. Oda, BGrowth of high-quality GaN films on epitaxial AlN/ sapphire templates by MOCVD,[ J. Cryst. Growth, vol. 244, no. 1, pp. 6–11, Sep. 2002. [2] S. Adachi, Properties of Group-IV, III-V and II-VI Semiconductors. Chichester: John Wiley & Sons, 2005. [3] J. Karpinski, J. Jun, and S. Porowski, BEquilibrium pressure of N2 over GaN and high pressure solution growth of GaN,[ J. Cryst. Growth, vol. 66, no. 1, pp. 1–10, Jan.–Feb. 1984. [4] W. Utsumi, H. Saitoh, H. Kaneko, T. Watanuki, K. Aoki, and O. Shimomura, BCongruent melting of gallium nitride at 6 GPa and its application to single-crystal growth,[ Nat. Mater., vol. 2, no. 11, pp. 735–738, Nov. 2003. [5] T. Fukuda and D. Ehrentraut, BProspects for the ammonothermal growth of large GaN crystal,[ J. Cryst. Growth, vol. 305, no. 2, pp. 304–310, Jul. 2007. [6] D. Ehrentraut, H. Sato, Y. Kagamitani, H. Sato, A. Yoshikawa, and T. Fukuda, BSolvothermal growth of ZnO,[ Prog. Cryst. Growth Char. Mater., vol. 52, no. 4, pp. 280–335, Apr. 2006. [7] B. Wang and M. J. Callahan, BAmmonothermalsynthesis of III-nitride crystals,[ Cryst. Growth Des., vol. 6, no. 6, pp. 1227–1246, Jun. 2006. [8] D. Ehrentraut and Z. Sitar, BAdvances in bulk crystal growth of AlN and GaN,[ MRS Bulletin, vol. 33, no. 4, pp. 259–265, Apr. 2009. [9] H. Jacobs and D. Schmidt, BHigh-pressure ammonolysis in solid-state chemistry,[ in Current Topics in Materials Science, vol. 8, E. Kaldis, Ed. Amsterdam: North-Holland Publishing Co., 1982, pp. 381–427. [10] R. Dwilicski, J. Doradzicski, J. Garczycski, L. Sierzputowski, J. M. Baranowski, and M. Kamicska, BExciton photo-luminescence of GaN bulk crystals grown by the AMMONO method,[ Mater. Sci. Eng. B, vol. 50, no. 1–3, pp. 46–49, Dec. 1997. [11] A. P. Purdy, R. J. Jouet, and C. F. George, BAmmonothermal recrystallization of gallium
1322
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
recently for GaN crystal synthesized from the ammonothermal technique with both basic and acidic mineralizers. This has yielded 2 inch large GaN crystals by the basic ammonothermal technique of higher structural quality than the currently used HVPE technique. A critical point is the incorporation of impurities (Si, O, Cl, Fe, Ni, and Cr) from the solution. This must greatly be improved to compete with GaN prepared using the HVPE technique. GaN substrates must be price competitive, at least to their HVPE rivals. A way to achieve this goal might be the fabrication of very large size GaN crystals. The great potential the ammonothermal technique holds to massproduce GaN bulk crystal in an economic fashion similar to the present production of quartz crystals was shown. High-power electronics and optoelectronics will likely be the target for the ammonothermal GaN wafers due to their high structural quality. h
nitride with acidic mineralizers,[ Cryst. Growth Des., vol. 2, no. 2, pp. 141–145, Mar. 2001. D. R. Ketchum and J. W. Kolis, BCrystal growth of gallium nitride in supercritical ammonia,[ J. Cryst. Growth, vol. 222, no. 3, pp. 431–434, Jan. 2001. G. Demazeau, G. Goglio, A. Denis, and A. Largeteau, BSolvothermal synthesis: A new route for preparing nitrides,[ J. Phys.: Condens. Matter, vol. 14, no. 44, pp. 11 085–11 088, Nov. 2002. M. P. D’Evelyn, K. J. Narang, D.-S. Park, H. C. Hong, M. Barber, S. A. Tysoe, J. Leman, J. Balch, V. L. Lou, S. F. LeBoeuf, Y. Gao, J. A. Teetsov, P. J. Codella, P. R. Tavernier, D. R. Clarke, and R. J. Molnar, BGrowth and characterization of bulk GaN crystals at high pressure and high temperature,[ Mat. Res. Soc. Symp. Proc., vol. 798, pp. Y2.3.1–Y.2.3.6, 2004. T. Hashimoto, K. Fujito, M. Saito, J. S. Speck, and S. Nakamura, BAmmonothermal growth of GaN on an over-1-inch seed crystal,[ Jpn. J. Appl. Phys., vol. 44, no. 50–52, pp. L1570–L1572, Dec. 2005. T. Hashimoto, F. Wu, J. S. Speck, and S. Nakamura, BAmmonothermal growth of bulk GaN,[ J. Cryst. Growth, vol. 310, no. 17, pp. 3907–3910, Aug. 2008. B. Wang, M. J. Callahan, K. D. Rakes, L. O. Bouthillette, S.-Q. Wang, D. F. Bliss, and J. W. Kolis, BAmmonothermal growth of GaN crystals in alkaline solutions,[ J. Cryst. Growth, vol. 287, no. 2, pp. 376–380, Jan. 2006. B. Raghothamachar, J. Bai, M. Dudley, R. Dalmau, D. Zhuang, Z. Herro, R. Schlesser, Z. Sitar, B. Wang, M. Callahan, K. Rakes, P. Konkapaka, and M. Spencer, BCharacterization of bulk grown GaN and AlN single crystal materials,[ J. Cryst. Growth, vol. 287, no. 2, pp. 349–353, Jan. 2006. M. J. Callahan, BTrend of research for ZnO, GaN, AlN crystal growth,[ presented at the Annual Symposium on Mutual Exchange Between Industry and AcademyVFuturistic Crystals and Their Applications, Sendai, Miyagi, Nov. 26, 2003. A. Yoshikawa, E. Ohshima, T. Fukuda, H. Tsuji, and K. Oshima, BCrystal growth of GaN by ammonothermal method,[ J. Cryst.
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
Growth, vol. 260, no. 1–2, pp. 67–72, Jan. 2004. D. Ehrentraut, N. Hoshino, Y. Kagamitani, A. Yoshikawa, T. Fukuda, H. Itoh, and S. Kawabata, BTemperature effect of ammonium halogenides as mineralizers on phase stability of gallium nitride synthesized under acidic ammonothermal conditions,[ J. Mater. Chem., vol. 17, no. 9, pp. 886–893, Mar. 2007. R. Dwilicski, R. Doradzicski, J. Garczycski, L. P. Sierzputowski, A. Puchalski, Y. Kanbara, K. Yagi, H. Minakuchi, and H. Hayashi, BExcellent crystallinity of truly bulk ammonothermal GaN,[ J. Cryst. Growth, vol. 310, no. 17, pp. 3911–3916, Aug. 2008. R. Dwilinski, R. Doradzinski, J. Garczynski, L. P. Sierzputowski, M. Zajac, and M. Rudzinski, BHomoepitaxy on bulk ammonothermal GaN,[ J. Cryst. Growth, vol. 311, no. 10, pp. 3058–3062, May 2009. T. Hashimoto, K. Fujito, F. Wu, B. A. Haskell, P. T. Fini, J. S. Speck, and S. Nakamura, BAmmonothermal growth of GaN utilizing negative temperature dependence of solubility in basic ammonia,[ in Mater. Res. Soc. Symp. Proc., 2005, vol. 831, pp. E2.8.1–E.2.8.6. Y. Kagamitani, D. Ehrentraut, A. Yoshikawa, N. Hoshino, T. Fukuda, S. Kawabata, and K. Inaba, BAmmonothermal epitaxy of thick GaN film usingNH4 Cl mineralizer,[ Jpn. J. Appl. Phys., vol. 45, no. 5A, pp. 4018–4020, May 2006. T. Hashimoto, F. Wu, J. S. Speck, and S. Nakamura, BGrowth of bulk GaN crystals by the basic ammonothermal method,[ Jpn. J. Appl. Phys., vol. 46, no. 36–40, pp. L889–L891, Oct. 2007. D. Ehrentraut, Y. Kagamitani, T. Fukuda, F. Orito, S. Kawabata, K. Katano, and S. Terada, BRecent developments in the acidic ammonothermal growth of GaN crystal,[ J. Cryst. Growth, vol. 310, no. 17, pp. 3902–3906, Aug. 2008. M. Roos, J. Wittrock, G. Meyer, S. Fritz, and J. Stra¨hle, BThe courses of the ammonolyses of the ammonium hexafluorometalates of aluminum, gallium, and indium, ðNH4 Þ3 MF6 ðM ¼ Al; Ga; InÞ,[ Z. Anorg. Allg. Chem., vol. 626, no. 5, pp. 1179–1185, May 2000.
Proceedings of the IEEE | Vol. 98, No. 7, July 2010
Authorized licensed use limited to: TOHOKU UNIVERSITY. Downloaded on June 16,2010 at 01:58:55 UTC from IEEE Xplore. Restrictions apply.
Ehrentraut et al.: The Ammonothermal Crystal Growth of Gallium NitrideVA Technique on the Up Rise
[29] H. Yamane, Y. Mikawa, and C. Yokoyama, BPentaamminechlorogallium (III) dichloride,[ Acta Cryst. E, vol. 63, pt. 2, pp. i59–i61, Feb. 2007. [30] D. Peters, BAmmonothermal synthesis of aluminum nitride,[ J. Cryst. Growth, vol. 104, no. 2, pp. 411–418, Jul. 1990. [31] H. Jacobs and B. No¨cker, BRedetermination of structure and properties of the isotypic sodium tetraamido metalates of aluminum and gallium,[ Z. Anorg. Allg. Chem., vol. 619, no. 2, pp. 381–386, Feb. 1993. [32] R. T. Dwilinski, R. M. Doradzinski, J. Garczynski, L. P. Sierzputowski, and Y. Kanbara, BBulk monocrystalline gallium nitride,[ U.S. Patent 6 656 615, Dec. 2, 2003. [33] R. Dwilicski, R. Doradzicski, J. Garczycski, L. P. Sierzputowski, A. Puchalski, Y. Kanbara, K. Yagi, H. Minakuchi, and H. Hayashi, BBulk
ammonothermal GaN,[ J. Cryst. Growth, vol. 311, no. 10, pp. 3015–3018, May 2009. [34] K. Fujito, K. Kiyomi, T. Mochizuki, H. Oota, H. Namita, S. Nagao, and I. Fujimura, BHigh-quality nonpolar m-plane GaN substrates grown by HVPE,[ Phys. Stat. Solidi (a), vol. 205, no. 5, pp. 1056–1059, May 2008. [35] R. Dwilicski, R. Doradzicski, J. Garczycski, L. P. Sierzputowski, A. Puchalski, K. Yagi, and Y. Kanbara, BExcellent crystallinity of truly bulk ammonothermal GaN,[ presented at the 5th International Workshop on Bulk Nitride Semiconductors (IWBNS-V), Itaparica, Bahia, Brazil, Sep. 24–28, 2007. Workshop Program and Abstracts. [36] K. Fujito, S. Kubo, and I. Fujimura, BDevelopment of bulk GaN crystals and nonpolar/semipolar substrates by HVPE,[ MRS Bulletin, vol. 34, no. 5, pp. 313–317, May 2009.
[37] K. Fujii, G. Fujimoto, T. Goto, T. Yao, Y. Kagamitani, N. Hoshino, D. Ehrentraut, and T. Fukuda, BNear band-edge 3.357 eV emission of Ga-face (0001) GaN grown by ammonothermal method,[ Phys. Stat. Solidi (a), vol. 204, no. 12, pp. 4266–4271, Dec. 2007. [38] S. F. Chichibu, T. Onuma, T. Hashimoto, K. Fujito, F. Wu, J. S. Speck, and S. Nakamura, BImpacts of dislocation bending and impurity incorporation on the local cathodoluminescence spectra of GaN grown by ammobnothermal method,[ Appl. Phys. Lett., vol. 91, no. 25, pp. 25 1911-1–25 1911-3, Dec. 2007. [39] D. Ehrentraut and Y. Kagamitani, Evolution of Dislocations in Acidic Ammonothermal GaN. Unpublished.
ABOUT THE AUTHORS Dirk Ehrentraut received a diploma in crystallography from the Humboldt-University of Berlin, Germany in 1995 and the Ph.D. degree from the Swiss Federal Institute of Technology (EPFL), Lausanne, Switzerland in 2003. He was with the Leibniz-Institute of Crystal Growth, Berlin from 1995–1997 and with the Institute of Micro- and Optoelectronics and the Institute of Applied Optics, both at EPFL, from 1997–2003. He held positions as Visiting Professor and Research Professor at the Tohoku University, Sendai, Japan. In 2009, he was appointed Associate Professor of the newly established World Premier International Research Center InitiativeVAdvanced Institute for Materials Research (WPI-AIMR), Tohoku University. His research focus is on crystal growth technology and crystal chemistry of the wide band gap materials ZnO and Group III nitrides from bulk down to nano size, and related hybrid materials. Prof. Ehrentraut is a member of the Materials Research Society.
Tsuguo Fukuda graduated from the Faculty of Science, the University of Tokyo, Japan in 1964 and completed his PhD on ferroelectric crystals in 1971 from the University of Tokyo. After 24 years of working at Tokyo Shibaura Electric Co. (Toshiba Corporation) including 5 years working at the Optoelectronic Joint Research Lab. in Japan, he became Professor of the Institute of Materials Research, Tohoku University, Sendai, Japan in 1987. In 2002 he became Research Professor of the Institute of Multidisciplinary Research for Advanced Materials, Tohoku University. In 2009, he was appointed Adjunct Professor of the newly established World Premier International Research Center InitiativeVAdvanced Institute for Materials Research (WPI-AIMR), Tohoku University. His research interests are crystal growth technology of bulk single crystals for optical, scintillator, and semiconductor applications. He has co-authored over 600 papers. Prof. Fukuda is a member of the Japan Society for Applied Physics.
Vol. 98, No. 7, July 2010 | Proceedings of the IEEE
1323
Authorized licensed use limited to: TOHOKU UNIVERSITY. Downloaded on June 16,2010 at 01:58:55 UTC from IEEE Xplore. Restrictions apply.