# Fileset

[Hydride vapor phase epitaxy and characterization of highquality ScN epilayers.pdf](https://mdr.nims.go.jp/filesets/41c3245a-6800-461f-814c-d1706c891dbf/download)

## Creator

[Yuichi Oshima](https://orcid.org/0000-0001-8293-4891), Encarnación G. Víllora, [Kiyoshi Shimamura](https://orcid.org/0000-0001-6502-8731)

## Rights

This article may be downloaded for personal use only. Any other use requires prior permission of the author and AIP Publishing. This article appeared in Yuichi Oshima et al., J. Appl. Phys. 115, 153508 (2014) and may be found at https://doi.org/10.1063/1.4871656.[In Copyright](http://rightsstatements.org/vocab/InC/1.0/)

## Other metadata

[Hydride Vapor Phase Epitaxy and characterization of High-Quality ScN epilayers](https://mdr.nims.go.jp/datasets/85061408-6340-43dc-89cc-eaec2d9422a7)

## Fulltext

Hydride vapor phase epitaxy and characterization of high-quality ScN epilayersViewOnlineExportCitationCrossMarkRESEARCH ARTICLE |  APRIL 18 2014Hydride vapor phase epitaxy and characterization of high-quality ScN epilayers Yuichi Oshima; Encarnación G. Víllora; Kiyoshi ShimamuraJ. Appl. Phys. 115, 153508 (2014)https://doi.org/10.1063/1.4871656 30 January 2024 07:23:41https://pubs.aip.org/aip/jap/article/115/15/153508/138887/Hydride-vapor-phase-epitaxy-and-characterizationhttps://pubs.aip.org/aip/jap/article/115/15/153508/138887/Hydride-vapor-phase-epitaxy-and-characterization?pdfCoverIconEvent=citehttps://pubs.aip.org/aip/jap/article/115/15/153508/138887/Hydride-vapor-phase-epitaxy-and-characterization?pdfCoverIconEvent=crossmarkjavascript:;javascript:;javascript:;javascript:;https://doi.org/10.1063/1.4871656https://servedbyadbutler.com/redirect.spark?MID=176720&plid=2281479&setID=592934&channelID=0&CID=837667&banID=521597145&PID=0&textadID=0&tc=1&scheduleID=2201785&adSize=1640x440&data_keys=%7B%22%22%3A%22%22%7D&matches=%5B%22inurl%3A%5C%2Fjap%22%5D&mt=1706599421078384&spr=1&referrer=http%3A%2F%2Fpubs.aip.org%2Faip%2Fjap%2Farticle-pdf%2Fdoi%2F10.1063%2F1.4871656%2F15130852%2F153508_1_online.pdf&hc=701b572ff79847e48483417b98e8b3079f8ec979&location=Hydride vapor phase epitaxy and characterization of high-qualityScN epilayersYuichi Oshima,a) Encarnaci�on G. V�ıllora, and Kiyoshi ShimamuraEnvironment and Energy Materials Research Division, National Institute for Materials Science, 1-1 Namiki,Tsukuba, Ibaraki 305-0044, Japan(Received 7 March 2014; accepted 3 April 2014; published online 18 April 2014)The heteroepitaxial growth of ScN films was investigated on various substrates by hydride vaporphase epitaxy (HVPE). Single crystalline mirror-like ScN(100) and ScN(110) layers weresuccessfully deposited on r- and m-plane sapphire substrates, respectively. Homogeneousstoichiometric films (N/Sc ratio 1.01 6 0.10) up to 40 lm in thickness were deposited. Theirmosaicity drastically improved with increasing the film thickness. The band gap was determined byoptical methods to be 2.06 eV. Impurity concentrations including H, C, O, Si, and Cl wereinvestigated through energy dispersive X-ray spectrometry and secondary ion mass spectrometry.As a result, it was found that the presence of impurities was efficiently suppressed in comparisonwith that of HVPE-grown ScN films reported in the past, which was possible thanks to thehome-designed corrosion-free HVPE reactor. Room-temperature Hall measurements indicatedthat the residual free electron concentrations ranged between 1018–1020 cm�3, which was markedlylower than the reported values. The carrier mobility increased monotonically with the decreasingin carrier concentration, achieving the largest value ever reported, 284 cm2 V�1 s�1 atn¼ 3.7� 1018 cm�3. VC 2014 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4871656]I. INTRODUCTIONScN is a group III transition metal nitride semiconductor,which crystallizes in the cubic rock-salt crystal structure witha lattice constant of 4.505 Å.1 Similar to other transition metalnitrides, ScN exhibits excellent physical properties, such ashigh hardness and high thermodynamic stability, which arecomparable to those of TiN.1–3 ScN is reported to have adirect band gap of 2.1–3.2 eV, and an additional indirect bandgap of 0.9–1.3 eV has been predicted.3–5 Since ScN(111) lat-tice matches to wurtzite-GaN(0001) and cubic-GaN(111), ithas been used as a buffer layer or interlayer to reduce thethreading dislocation density in GaN epilayers.6–9 In addition,GaxSc1�xN alloys10–16 and the ScN/GaN hetero-junction17are being investigated to improve the performance ofGaN-based devices. Furthermore, among the transition metalnitrides, ScN possesses an anomalous high thermoelectricfactor,18,19 which make it potential candidate material forhigh temperature thermoelectric applications. ScN is alsopromising as a host material of ferromagnetic semiconductorsdue to its high manganese solubility and an estimated highCurie temperature.20,21 AlxSc1�xN alloys shows an exception-ally large piezoelectric response,22,23 and meta-stable hexago-nal ScN is expected to have even greater piezoelectricconstant, comparable to that of ferroelectric perovskites.24ScN is thus a very attractive material, which has a greatpotential for various applications.The growth of ScN thin film has been investigatedmainly by molecular beam epitaxy (MBE), sputtering, andhydride vapor phase epitaxy (HVPE). The growth tempera-ture is below 900 �C in most of the cases. The typical growthrate is below 0.5 lm/h, and the film thickness is usuallybelow 1 lm except the case of HVPE. Although there are nolattice-matched substrate materials, the growth of singlecrystalline ScN films has been reported on sapphire, MgO,and Si.25–32 In general, heteroepitaxial growth on highlylattice-mismatched substrates leads to significant mosaicity,involving high density of crystal defect such as dislocations.There are a few reports on the structural quality of ScNfilms,27,30 and the best reported FWHM of X-ray rockingcurve is 0.35�, which is still far from the quality of conven-tional, high quality semiconductors, such as Si, GaAs, andGaN. On the other hand, in order to use semiconductors indevice applications, generally their conduction type and car-rier concentration need to be controlled precisely. In the caseof ScN, however, such control has not been realized yet.Nominally un-doped ScN usually shows strong n-type con-duction, with a high residual carrier concentration rangingbetween 1019 to 1022 cm�3.29–32 The plausible origins of thedonors are impurities and crystal defects such as nitrogenvacancies. However, their contribution has not been fullyclarified in detail yet. In order to exploit the full potential ofScN and to pave the road to practical use, it is essential toclarify and control the effect of crystal defects and impuritiesby the growth of high-quality and high-purity crystals.In this work, we have employed HVPE as the growthtechnique of ScN. HVPE is a kind of chemical vapor deposi-tion (CVD) technique, which was first reported by Tietjenand Amic in 1966 as a growth method of III-V semiconduc-tors such as GaAsP.33 Nowadays, HVPE is widely used forthe mass production of freestanding GaN single crystalwafers.34,35 The method is characterized by a fast growthrate and the high crystalline quality of resulting layers.Usually, the equilibrium constant of HVPE is far smallerthan those of MBE and metalorganic vapor phase epitaxy(MOVPE). Such small equilibrium constant allows one toa)Electronic mail: OSHIMA.Yuichi@nims.go.jp0021-8979/2014/115(15)/153508/8/$30.00 VC 2014 AIP Publishing LLC115, 153508-1JOURNAL OF APPLIED PHYSICS 115, 153508 (2014) 30 January 2024 07:23:41http://dx.doi.org/10.1063/1.4871656http://dx.doi.org/10.1063/1.4871656http://dx.doi.org/10.1063/1.4871656mailto:OSHIMA.Yuichi@nims.go.jphttp://crossmark.crossref.org/dialog/?doi=10.1063/1.4871656&domain=pdf&date_stamp=2014-04-18supply high concentration gaseous source materials to realizea high growth rate without parasitic gas phase reaction. Inaddition, such small equilibrium constant enables the crystalto grow under nearly thermal equilibrium condition, whichleads to the growth of high quality crystals.The first HVPE of ScN was reported by Dismukes et al.in 1970.31,32 They demonstrated the epitaxial growth of ScNon r-plane sapphire substrate within 850–930 �C. Besidesthis temperature range, all samples were polycrystalline. Thesingle crystalline growth was possible at growth rates below2.8 lm/h, and relatively thick films up to 20 lm wereobtained. Their nominally un-doped ScN films also showedn-type conductivity with high carrier concentrations in theorder of 1020–1021 cm�3, which are comparable to those ofScN films grown by MBE or sputtering. Their ScN filmscontained chlorine and oxygen impurities in high concentra-tions, i.e., [Cl]¼ 1.8� 1020 cm�3 and [O]¼ 4.4� 1019 cm�3,respectively. The high carrier concentration was attributed tothe presence of Cl ions, because their concentrations werecomparable. In spite of such high carrier concentration, thecarrier mobility was also relatively high, 151 cm2 V�1 s�1 atn¼ 1.1� 1020 cm�3, which is twice as high as that of silicon.In recent years, Edger et al. tried to grow ScN on6 H-SiC substrates by HVPE.36 Their ScN films containedaluminum and chlorine as impurities in much higher concen-trations, being even detectable by energy dispersive X-rayspectrometry (EDX). In this case, the Al impurity wasassumed to originate from the reaction between thescandium-chloride reactant and an alumina tube used to con-tain Sc metal in the HVPE reactor. The [Al] was as high as1.5 at. % in their ScN film grown at 1000 �C. By decreasingthe temperature of the alumina tube (source temperature)from 1000 to 800 �C, the corrosion was suppressed and [Al]decreased. However, [Al] was still high (�0.5 at. %) and[Cl] increased from 1.9 to 2.7 at. % due to the diminished Cldesorption from the growth front.As stated above, ScN is a very attractive material, and itis necessary to realize high quality epilayers with well-controlled electrical conductivity for device applications.HVPE is, in principle, a promising growth method for ScN.However, this technique needs to be improved in order toavoid the incorporation of impurities into the films, which iscaused by the chamber corrosion with the highly reactivesource materials. In this work, we demonstrate the growth ofhigh-quality, high-purity ScN epilayers by using an origi-nally developed corrosion-resistant HVPE reactor. The crys-talline, optical, and electrical properties of the resulting ScNfilms are analyzed and discussed.II. EXPERIMENTAL METHODA. HVPE growth of ScNGenerally, quartz and alumina are the principal materi-als to construct a HVPE reactor. However, as was stated inthe former section, these materials are corroded by scandiumand scandium chloride, the source materials of ScN inHVPE.37,38 Therefore, we have designed our own home-made HVPE reactor avoiding the use of quartz and aluminain the high-temperature zone. The structure of the reactor isschematically shown in Fig. 1. A water-cooled quartz tube isemployed as the outer tube. Inside, a pBN-coated graphitetube is coaxially located. The graphite tube is inductivelyheated by an RF-coil around the quartz tube to create a hotzone inside. Between both quartz and graphite tubes, there isa pBN tube radiation shield, which prevents the overheatingof the inner quartz wall. Ar-gas is purged to avoid the expo-sure of the inner quartz wall to the corrosive materials.Inside the graphite tube, the inner flow channel, the substrateholder, and the Sc container are composed of corrosion-resistant materials, such as tungsten and SiC.ScN was grown in the home-made HVPE reactor atatmospheric pressure using scandium chloride and NH3(99.9999% pure) as source materials. No intentional dopingwas carried out. Scandium chloride was formed upstream inthe reactor by the reaction between solid Sc metal (99.99%pure) and HCl gas (99.999% pure). The partial pressures ofthe HCl (PHCl) and NH3 (PNH3) were varied between0.124–1.24 kPa and 6.2–19.3 kPa, respectively. The saturatedvapor pressure of ScCl3, one possible product formed by thereaction between Sc and HCl, reaches 100 kPa at 1037 �C.39Therefore, in order to prevent the condensation of ScCl3, thetemperature of the reactor was kept constant at a bit highervalue 1050 �C. Usually, nitrogen gas is used as a carrier gasin HVPE. However, N2 and Sc react directly with each otherto form ScN at high temperatures.40 Instead of N2, weemployed Ar as carrier gas.We have also investigated which substrates are appro-priate for the heteroepitaxial growth of ScN. Sapphire, MgO,Yttrium-stabilized zirconia (YSZ), GaN, and AlN were takenunder consideration. In this investigation, PHCl and PNH3were fixed to be 0.31 kPa and 10.6 kPa, respectively. All dep-ositions were carried out at 1050 �C for 5 min using the samegrowth recipe. The ScN films were directly grown on eachsubstrate without any buffer layer to compensate for latticemismatch.B. Characterization methodsThe crystal orientation of HVPE-grown ScN films wasinvestigated by X-ray diffraction x-2h scan and pole figuremeasurements. The surface morphology was observed bymeans of Nomarski microscopy and scanning electronmicroscopy (SEM). The growth rate was determined bycross-sectional SEM. The N/Sc composition ratio and con-centrations of impurities were estimated by Rutherford back-scattering spectrometry (RBS), EDX, and secondary ionFIG. 1. Schematic of the HVPE reactor used for ScN growth.153508-2 Oshima, V�ıllora, and Shimamura J. Appl. Phys. 115, 153508 (2014) 30 January 2024 07:23:41mass spectrometry (SIMS). The mosaicity (tilt and twist) inthe films was evaluated through X-ray rocking curve meas-urements in symmetric and skew-symmetric geometry.Transmittance and reflectance spectra were utilized to figureout the optical absorption coefficient and to determine theoptical band gap. Electrical properties (carrier concentrationand electron mobility) were investigated by Hall measure-ments by the van der Pauw method. All the measurementswere carried out at room temperature.III. RESULTS AND DISCUSSIONA. Chemistry of ScN growthFirst of all, we discuss the chemical reactions by whichScN is produced by HVPE. Dismukes et al. studied the vaporphase reactions involved in ScN HVPE.32 They presumedthat ScCl2 is the predominant scandium chloride speciessince only a dihalide is generated in the case of chemicalreaction between Cl2 and Y, being chemically similar to Sc.In our case, however, the predominant scandium chloridespecies would be (ScCl3)x, since the molar ratio of the sup-plied HCl and the decrease of Sc metal was 3:1. According toPatrikeev et al., ScCl3 and Sc2Cl6 co-exist in thermal equilib-rium,39 and approximately 74% should be Sc2Cl6 at our reac-tor temperature 1050 �C. Therefore, we presume thatfollowing reactions are taking place simultaneously:Sc2Cl6 þ 2NH3 ¼ 2ScNþ 6HCl;ScCl3 þ NH3 ¼ ScNþ 3HCl:B. Investigation of substrates for heretoepitaxyof ScNTable I summarizes the results of ScN growth on varioussubstrates. Films grown on MgO, YSZ, GaN were peeled offand broke into pieces after deposition. This behavior isattributed to a substrate etching process at the interface,probably related to the presence of HCl gas in the growthatmosphere.Mechanically stable films were successfully grown onsapphire and AlN. X-ray x-2h scan (not shown) revealedthat the ScN films grown on c- and a-plane sapphire and AlNincluded domains with mixed out-of-plane orientations of(100) and (111). ScN films with single out-of-plane orienta-tion were obtained on r- and m-plane sapphires, i.e.,(100)ScN//(10-12)sap and (110)ScN//(10-10)sap, respectively.Figures 2(a) and 2(b) are the X-ray pole figures of theScN(100) and the ScN(110) films. The pole figure of theScN(100) film shows (111) poles with four-fold symmetry,and that of the ScN(110) film shows (200) poles withtwo-fold symmetry as expected for single in-plane orienta-tion films. This proves that single crystalline films wereachieved on r- and m-plane sapphire. From the /-scan pro-files of the epilayers and the substrates (not shown), follow-ing epitaxial relationships were elucidated:ScN on r-plane sapphire : ð100ÞScN==ð10� 12Þsap and ½001�ScN==½11� 20�sap;ScN on m-plane sapphire : ð110ÞScN==ð10� 10Þsap and ½001�ScN==½11� 20�sap:In the following, the characterization of these singlecrystalline films is described.C. Growth rate of ScNFigure 3 shows the growth rate of ScN as a function of(a) PHCl, and (b) PNH3. The results of ScN(100) and ScN(110)were identical within the thickness accuracy estimated bySEM. The growth rate first increases with increasing PHCl,reaches maximum around PHCl¼ 0.5 kPa, and then starts todecrease. The dependence on PNH3 is similar. The decrease ingrowth rate probably results from a gas-phase reactionupstream in the reactor that consumes scandium chloride andNH3 before they arrive at the substrate.The next sections present and discuss the morphological,structural, optical, and electrical properties of ScN films grownunder representative growth condition, PHCl¼ 0.31 kPa andTABLE I. Results of HVPE growth of ScN on various substrates.Substrate ResultSapphire c-plane (0001) Mixed orientations, (100) and (111)a-plane (11–20)m-plane (10–10) (110) single crystalliner-plane (10–12) (100) single crystallineMgO (100) Chemically unstable(110)YSZa (100)GaN, c-plane (0001)AlN, c-plane (0001) Mixed orientations, (100) and (111)aY2O3 (10 mol. %)—ZrO2 (90 mol. %)FIG. 2. X-ray pole figures of ScN films grown on (a) r-plane sapphire and(b) m-plane sapphire.153508-3 Oshima, V�ıllora, and Shimamura J. Appl. Phys. 115, 153508 (2014) 30 January 2024 07:23:41PNH3¼ 10.6 kPa, corresponding to a growth rate of �5 lm/h.The growth time was varied from 5 min to 8 h, thus the thick-ness ranged approximately between 0.4� 40 lm.D. Surface morphology of HVPE-grown ScNThe surface of all the samples was mirror-like for humaneyes. Figures 4 and 5 show surface Nomarski microscopeimages and SEM images (bird’s eye view) of the HVPE-grown ScN films. On the surface can be observed hillocks,whose averaged diameter increases with the layer thickness.As it will be seen in Sec. III E, the crystalline quality of theScN films improves with increasing film thickness. So, iftheses hillocks are reflecting the sub-grain structure of themosaic films, this increase in hillock size, i.e., decrease inthe density of hillocks, agrees well with the improvement ofthe structural quality.In SEM images of Fig. 5, ScN(100) surface appearssmoother than ScN(110). This can be explained by the differ-ence in the surface diffusion barrier, which arises from thenumber of dangling bonds. In the case of ScN(100) surface,each atom will have only a single dangling bond, whereaseach atom of ScN(110) surface will have two danglingbonds. Therefore, the surface diffusion barrier of ScN(100)should be lower than that of ScN(110). An analogous argu-ment is found in the comparison of MBE-grown ScN(100)and ScN(111) surfaces.3 Our films, both ScN(100) andScN(110), exhibit a much smoother surface compared to theHVPE-grown ScN films reported by Dismukes et al. andEdger et al.32,36 This fact can also be explained from theview point of surface diffusion barrier. The ScN films ofDismukes et al. and Edger et al. had surfaces of ScN(321)and ScN(111), respectively. On the both planes, each atomhas three dangling bonds and hence the surface diffusion bar-rier should be higher than those of ScN(100) and ScN(110).In addition, our growth temperature is substantially higherthan those of the previous reports. Therefore, it is plausiblethat such a high growth temperature favors the surface diffu-sion, resulting thus in a smoother surface.E. Structural propertiesFigures 6(a) and 6(b) show the thickness dependences oftilt and twist angles of our HVPE-grown ScN films. In the caseof ScN(100) films, we use the FWHMs of (200) and (131)rocking curves as the measure of the out-of-plane and in-planemosaicity, i.e., tilt and twist angles, respectively. Analogously,we employed the FWHMs of (220) and (-240) reflections astilt and twist angles for ScN(110) films. Both (131) and (-240)are inclined from the film surface by approximately 72�.The tilt and twist angles monotonically decreased withincreasing film thickness. A similar quality improvement hasFIG. 3. Growth rate of ScN as a func-tion of (a) HCl partial pressure, (b)NH3 partial pressure.FIG. 4. Nomarski microscope surface images of: ScN (100) films with an approximate thickness of (a) 10 lm, (b) 20 lm, (c) 40 lm ScN (110) films with an ap-proximate thickness of (d) 10 lm, (e) 20 lm, (f) 40 lm.153508-4 Oshima, V�ıllora, and Shimamura J. Appl. Phys. 115, 153508 (2014) 30 January 2024 07:23:41been observed in the case of GaN films. The interpretationgiven to this phenomenon is that dislocations with oppositeBurgers vector gradually approach each other due to theattractive interaction between them, and then react with eachother to make a dislocation loop during thick film growth.41The same mechanism would be possible in the case of ScN.In order to confirm this, it will be necessary to perform thedirect observation of dislocations by cross-sectional TEM.We have also found that ScN(110) films have a betterquality than ScN(100) ones when their thickness is the same.The lattice mismatch between m-plane sapphire and twoperiods of ScN(110) along [1–10]ScN is 2%, while thatbetween r-plane sapphire and three periods of ScN (100)along [010]ScN is 14%. The difference in the structuralquality is likely to be originated from this difference in lat-tice match.F. Optical propertiesFigure 7(a) shows the transmittance (T), the reflectance(R), and the absorption coefficient (a) of a ScN(110) film(3.7 lm thick) as a function of photon energy (hv). Both Tand R show a clear oscillation due to the multiple reflectionson both sides. a was estimated from T and R by the methoddescribed elsewhere.42 a shows a steep increase around2.1 eV. No threshold was observed around 0.9–1.3 eV, wherethe existence of the indirect transition is predicted. It shouldbe noted that the apparent a saturation above 2.1 eV is anFIG. 6. Thickness dependence of XRC-FWHM of (a) ScN (100) and (b) ScN(110).FIG. 5. SEM images (bird’s eye-view) of: ScN (100) films with an approximate thickness of (a) 10 lm, (b) 20 lm, (c) 40 lm ScN (110) films with an approxi-mate thickness of (d) 10 lm, (e) 20 lm, (f) 40 lm.FIG. 7. (a) Transmittance and reflec-tance spectra of ScN (110), (b)Absorption coefficient in (h�a)2 vs h�plot.153508-5 Oshima, V�ıllora, and Shimamura J. Appl. Phys. 115, 153508 (2014) 30 January 2024 07:23:41experimental artifact caused by the decrease in transmittancesignal below the detection limit of our instrument for the rel-atively large film thickness. Therefore, a should be greater inreality. Figure 7(b) is the absorption coefficient spectrum in(hva)2 vs hv plot. The linear range correlates clearly with thedirect transition, and the band gap energy was determined tobe 2.06 eV by a linear fit. The data for ScN(100) films (notshown) show no significant difference with the results ofScN(110) films.The direct band gap of ScN is reported to lie in the rangeof 2.1–3.2 eV. Our band gap corresponds with the minimumreported values. Moram et al. have demonstrated that oxygencontamination leads to an increase in the direct band gapenergy of ScN.43 Therefore, our small energy gap is indica-tive of a low oxygen contamination in our samples. Indeed,the oxygen concentration is found by SIMS to be in the orderof 1017 cm�3, as it will be seen in Sec. III G.G. Composition and impurity concentrationWe have investigated the N/Sc composition ratio of 1.2-lm-thick ScN(110) film by RBS using Hþ and 4Heþ as inci-dent ions. Nitrogen sensitive measurements are possiblewhen Hþ is utilized, while 4Heþ enables high depth resolu-tion measurements. Figure 8(a) shows the RBS spectrumobtained with Hþ as incident ion. The spectrum was accu-rately reproduced by assuming the existence of only Sc, N,Al, and O, i.e., no impurity was observed by RBS. The esti-mated N/Sc ratio from this spectrum is 1.01 6 0.10.Therefore, no significant deviation from the stoichiometricratio (N/Sc¼ 1) was observed. Figure 8(b) is the depth pro-file of the composition calculated from the RBS spectrum(not shown) obtained with 4Heþ as incident ion. The N/Scratio was confirmed to be stoichiometric along the depthdirection within the RBS accuracy.No element except Sc and N was detected by EDX. Inorder to obtain more detailed information about potentialimpurities (namely H, Cl, C, O, Si, and W), SIMS was car-ried out. H and Cl are contained in HCl and NH3 gases. C,O, Si, and W are constituents of the HVPE reactor materials.As a result, all the elements except W were detected. Figure9 shows the depth profile of the SIMS measurement. Thepresent [C] and [O] are approximately 7.9� 1017 cm�3 and8.8� 1017 cm�3, respectively. Even though these concentra-tions are significant, these are remarkably lower than thosereported by Dismukes et al., with [C]¼ 1.7� 1019 cm�3 and[O]¼ 4.4� 1019 cm�3.32 In the case of Si, there is no reporton ScN, and our value [Si]¼ 1.6� 1016 cm�3 is comparableto that of HVPE-grown GaN. Regarding [H] and [Cl], reli-able standards to determine absolute concentration were notavailable. We tentatively used the relative sensitivity factors(RSF) of GaN, since these are expected to be similar to thoseof ScN. Consequently, the actual [H] and [Cl] in Fig. 9 maydiffer from the true values by a factor of 2–3. However, evenif the true value for [Cl] was 10 times larger than our tenta-tive one ([Cl]¼ 1.0� 1018 cm�3), this is still considerablysmaller than that reported by Dismukes et al., with[Cl]¼ 1.8� 1020 cm�3.32 These results indicate that the cor-rosion of the reactor materials is efficiently suppressed evenat the high growth temperature of 1050 �C, which is substan-tially higher than those of previous HVPE reports. In addi-tion, our high growth temperature could enhance thedesorption of chlorine from the growth surface and favor alow [Cl] in our ScN films.H. Electrical propertiesThe carrier concentration dependence of the electronmobility is shown in Fig. 10. The represented data corre-spond to nominally un-doped samples with various thick-nesses (between 0.4–40 lm), using the same growth recipebut different deposition times. The carrier concentrationvaries within two order of magnitude, no systematical corre-lation is found between this and the thickness or crystallineFIG. 8. (a) RBS spectrum of ScN(100), (b) depth profile of atomic ratioof Sc, N, Al, O.FIG. 9. SIMS depth profile of ScN(100). Note that concentrations of H andCl are tentative ones calibrated with the RSFs of GaN. Therefore, these val-ues may differ from the true ones by a factor of 2–3.153508-6 Oshima, V�ıllora, and Shimamura J. Appl. Phys. 115, 153508 (2014) 30 January 2024 07:23:41quality. Clearly, some unknown and uncontrolled factor isdominating the residual carrier concentration. It appears alsothat there is no significant difference between ScN(100) andScN(110). The data reported in the literature29,32 are showntogether for the sake of comparison. It is seen that the meas-ured carrier concentrations are substantially lower than thereported ones, and that the electron mobility monotonicallyincreases with decreasing carrier concentration. This can bereasonably explained by assuming a decrease of carrier scat-tering on ionized donors.In order to investigate the origin of the residual carrierconcentration, we compared the SIMS results between twosamples with largely different carrier concentrations (Table II).Here, we denote the samples with high (n¼ 8.1� 1019 cm�3)and low (n¼ 1.0� 1019 cm�3) carrier concentration as samplesA and B, respectively. Again, it should be reminded that [H]and [Cl] are tentative values calibrated with the RSFs of GaN,and therefore these could be different from their true values bya factor of 2–3. Among the impurities listed in Table II, C andO are the only elements whose concentration is higher in sam-ple A. However, the variances between sample A and B are toosmall to explain the difference in carrier concentration. On theother hand, [Si] and [Cl] are larger in sample B. However, theyshould act as donor in ScN, and cannot be the cause of the car-rier concentration difference. [H] is also greater in sample B,but its contribution to the electrical properties of ScN is stillunknown. However, also in this case, the difference is toosmall to account for the carrier concentration divergence evenif the true values of [H] were 10 times larger than our tentativevalues. These results suggest that the dominant donor speciesin our ScN films are not related to the elements listed in TableII. According to Smith et al., nitrogen vacancies can also act asdonors.3 In order to clarify the actual origin of the residual car-rier concentration, further investigations are required.IV. SUMMARYSingle crystalline ScN(100) and ScN(110) films are suc-cessfully grown on r- and m-plane sapphire substrates byHVPE. The growth rate has a peak for the partial pressuresof both HCl and NH3, and the maximum growth rate was5 lm/h. All the single crystalline samples had mirror-like sur-face with hillocks, which were visible by Nomarski micro-scope and SEM. The hillock size increased with increasingthe film thickness. The surface of ScN(100) appeared to besmoother than that of ScN(110), which can be originatedfrom the difference in surface diffusion barrier. The tilt andtwist angles for both ScN(100) and ScN(110) drasticallyimproved with increasing the film thickness. The structuralquality of ScN(110) was better than that of ScN(100) prob-ably due to the better lattice matching to sapphire substrates.The optical band gap was estimated to be 2.06 eV throughtransmittance and reflectance measurements. The N/Sc com-position ratio was 1.01 6 0.10, i.e., stoichiometric within theaccuracy of RBS. EDX and SIMS revealed that impurity con-centrations were successfully reduced compared to thereported values. This was achieved by a home-made corro-sion-resistant HVPE reactor and a high growth temperature.Our nominally un-doped samples show n-type conductivitysimilar to the reported results. However, the residual carrierconcentration was markedly lower, ranging between1018-1020 cm�3. The carrier mobility increased with decreas-ing the carrier concentration, and reached 284 cm2 V�1 s�1 atn¼ 3.7� 1018 cm�3, which is the highest value ever reported.It was found that the electrical properties (carrier concentra-tion and mobility) are not directly affected by the structuralquality. Additionally, there is no clear evidence of a correla-tion between the impurity concentrations of H, C, O, Si, andCl and the electrical properties. Further work is required toclarify the dominant origin of the residual donors in order toenable the conductivity control.ACKNOWLEDGMENTSThis work was partly supported by a Grant-in-Aid forYoung Scientists (B) No. 25790050 from Japan Society forthe Promotion of Science (JSPS). Part of this work was con-ducted for implementing the Ministry of Education, Culture,Sports, Science and Technology (MEXT) Elements StrategyInitiative to Form Core Center of Japan. The authors wish tothank Dr. Takaaki Mano for his great help in measuring X-ray pole figures.1M. A. Moram, Z. H. Barber, C. J. Humphreys, T. B. Joyce, and P. R.Chalker, J. Appl. Phys. 100, 023514 (2006).2D. Gall, I. Petrov, N. Hellgren, L. Hultman, J. E. Sundgren, and J. E.Greene, J. Appl. Phys. 84, 6034 (1998).FIG. 10. Carrier concentration dependence of electron mobility.TABLE II. Comparison of impurity concentrations (cm�3) in ScN filmswith largely different free electron concentrations.ElementSample An¼ 8.1� 1019 (cm�3)Sample Bn¼ 1.0� 1019 (cm�3)H (1.6� 1018)a (2.3� 1018)aC 7.9� 1017 6.2� 1017O 8.8� 1017 2.9� 1017Si 1.6� 1016 5.4� 1016Cl (1.0� 1018)a (1.7� 1018)aaTentative values, calibrated with the RSFs of GaN.153508-7 Oshima, V�ıllora, and Shimamura J. Appl. Phys. 115, 153508 (2014) 30 January 2024 07:23:41http://dx.doi.org/10.1063/1.2217106http://dx.doi.org/10.1063/1.3689133A. R. Smith, H. A. AL-Brithen, D. C. Ingram, and D. Gall, J. Appl. Phys.90, 1809 (2001).4H. A. Al-Brithen, A. R. Smith, and D. Gall, Phys. Rev. B 70, 045303 (2004).5X. Bai and M. E. Kordesch, Appl. Surf. Sci. 175–176, 499 (2001).6M. A. Moram, M. J. Kappers, Z. H. Barber, and C. J. Humphreys, J. Cryst.Growth 298, 268 (2007).7M. J. Kappers, M. A. Moram, Y. Zhang, M. E. Vickers, Z. H. Barber, andC. J. Humphreys, Physica B 401–402, 296 (2007).8M. A. Moram, Y. Zhang, M. J. Kappers, Z. H. Barber, and C. J.Humphreys, Appl. Phys. Lett. 91, 152101 (2007).9M. A. Moram, M. J. Kappers, T. B. Joyce, P. R. Chalker, Z. H. Barber,and C. J. Humphreys, J. Cryst. Growth 308, 302 (2007).10M. E. Little and M. E. Kordesch, Appl. Phys. Lett. 78, 2891 (2001).11C. Constantin, H. A. Al-Brithen, M. B. Haider, D. Ingram, and A. R.Smith, Phys. Rev. B 70, 193309 (2004).12C. Constantin, M. B. Haider, D. Ingram, A. R. Smith, N. Sandler, K. Sun,and P. Ordej�on, J. Appl. Phys. 98, 123501 (2005).13A. B. Fredj, Y. Oussaifi, N. Bouarissa, and M. Said, Phys. Status Solidi B243, 2780 (2006).14S. Zerroug, F. A. Sahraoui, and N. Bouarissa, J. Appl. Phys. 103, 063510(2008).15M. A. Moram, Y. Zhang, T. B. Joyce, D. Holec, P. R. Chalker, P. H.Mayrhofer, M. J. Kappers, and C. J. Humphreys, J. Appl. Phys 106,113533 (2009).16S. M. Knoll, S. Zhang, T. B. Joyce, M. J. Kappers, C. J. Humphreys, andM. A. Moram, Phys. Status Solidi A 209, 33 (2012).17F. Perjeru, X. Bai, M. I. Ortiz-Libreros, R. Higgins, and M. E. Kordesch,Appl. Surf. Sci. 175–176, 490 (2001).18S. Kerdsongpanya, N. V. Nong, N. Pryds, A. �Zukauskait_e, J. Jensen, J.Birch, J. Lu, L. Hultman, G. Wingqvist, and P. Eklund, Appl. Phys. Lett.99, 232113 (2011).19P. V. Burmistrova, J. Maassen, T. Favaloro, B. Saha, S. Salamat, Y. RuiKoh, M. S. Lundstrom, A. Shakouri, and T. D. Sands, J. Appl. Phys. 113,153704 (2013).20H. A. Al-Brithen, H. Yang, and A. R. Smith, J. Appl. Phys. 96, 3787(2004).21A. Herwadkar and W. R. L. Lambrecht, Phys. Rev. B 72, 235207(2005).22M. Akiyama, K. Kano, and A. Teshigahara, Appl. Phys. Lett. 95, 162107(2009).23M. Akiyama, K. Umeda, A. Honda, and T. Nagase, Appl. Phys. Lett. 102,021915 (2013).24V. Ranjan, L. Bellaiche, and E. J. Walter, Phys. Rev. Lett. 90, 257602(2003).25H. Al-Brithen and A. R. Smith, Appl. Phys. Lett. 77, 2485 (2000).26H. Al-Brithen, E. M. Trifan, D. C. Ingram, A. R. Smith, and D. Gall,J. Cryst. Growth 242, 345 (2002).27M. A. Moram, T. B. Joyce, P. R. Chalker, Z. H. Barber, and C. J.Humphreys, Appl. Surf. Sci. 252, 8385 (2006).28M. A. Moram, S. V. Novikov, A. J. Kent, C. N€orenberg, C. T. Foxon, andC. J. Humphreys, J. Cryst. Growth 310, 2746 (2008).29J. M. Gregoire, S. D. Kirby, G. E. Scopelianos, F. H. Lee, and R. B.Dover, J. Appl. Phys. 104, 074913 (2008).30T. Ohgaki, K. Watanabe, Y. Adachi, I. Sakaguchi, S. Hishita, N. Ohashi,and H. Haneda, J. Appl. Phys. 114, 093704 (2013).31J. P. Dismukes, W. M. Yim, J. J. Tietjen, and R. E. Novak, RCA Rev. 31,680 (1970).32J. P. Dismukes, W. M. Yim, and V. S. Ban, J. Cryst. Growth 13/14, 365(1972).33J. J. Tietjen and J. A. Amic, J. Electrochem. Soc. 113, 724 (1966).34Y. Oshima, T. Eri, M. Shibata, H. Sunakawa, K. Kobayashi, T. Ichihashi,and A. Usui, Jpn. J. Appl. Phys., Part 2 42, L1 (2003).35K. Motoki, T. Okahisa, N. Matsumoto, M. Matsushima, H. Kimura, H.Kasai, K. Takemoto, K. Uematsu, T. Hirano, M. Nakayama, S. Nakahata,M. Ueno, D. Hara, Y. Kumagai, A. Koukitu, and H. Seki, Jpn. J. Appl.Phys., Part 2 40, L140 (2001).36J. H. Edgar, T. Bohnen, and P. R. Hageman, J. Cryst. Growth 310, 1075(2008).37K. J. Hubbard and D. G. Schlom, J. Mater. Res. 11, 2757 (1996).38L. D. Polyachenok, K. Nazarov, and O. G. Polyachenok, Z. Fiz. Khim 52,1767 (1978).39Y. B. Patrikeev, V. A. Morozova, G. P. Dudchik, O. G. Polyachenok, andG. I. Novikov, Russian J. Phys. Chem. 47, 266 (1973).40W. Lengauer, J. Solid State Chem. 76, 412 (1988).41K. Fujito, S. Kubo, H. Nagaoka, T. Mochizuki, H. Namita, and S. Nagao,J. Cryst. Growth 311, 3011 (2009).42D. Gall, I. Petrov, L. D. Madsen, J. E. Sundgren and J. E. Greene, J. Vac.Sci. Technol. A 16, 2411 (1998).43M. A. Moram, Z. H. Barber, and C. J. Humphreys, Thin Solid Films 516,8569 (2008).153508-8 Oshima, V�ıllora, and Shimamura J. Appl. Phys. 115, 153508 (2014) 30 January 2024 07:23:41http://dx.doi.org/10.1063/1.1388161http://dx.doi.org/10.1103/PhysRevB.70.045303http://dx.doi.org/10.1016/S0169-4332(01)00165-9http://dx.doi.org/10.1016/j.jcrysgro.2006.10.027http://dx.doi.org/10.1016/j.jcrysgro.2006.10.027http://dx.doi.org/10.1016/j.physb.2007.08.170http://dx.doi.org/10.1063/1.2794009http://dx.doi.org/10.1016/j.jcrysgro.2007.09.009http://dx.doi.org/10.1063/1.1370548http://dx.doi.org/10.1103/PhysRevB.70.193309http://dx.doi.org/10.1063/1.2140889http://dx.doi.org/10.1002/pssb.200541493http://dx.doi.org/10.1063/1.2884580http://dx.doi.org/10.1063/1.3268466http://dx.doi.org/10.1002/pssa.201100158http://dx.doi.org/10.1016/S0169-4332(01)00079-4http://dx.doi.org/10.1063/1.3665945http://dx.doi.org/10.1063/1.4801886http://dx.doi.org/10.1063/1.1788842http://dx.doi.org/10.1103/PhysRevB.72.235207http://dx.doi.org/10.1063/1.3251072http://dx.doi.org/10.1063/1.4788728http://dx.doi.org/10.1103/PhysRevLett.90.257602http://dx.doi.org/10.1063/1.1318227http://dx.doi.org/10.1016/S0022-0248(02)01447-1http://dx.doi.org/10.1016/j.apsusc.2005.11.069http://dx.doi.org/10.1016/j.jcrysgro.2008.01.045http://dx.doi.org/10.1063/1.2996006http://dx.doi.org/10.1063/1.4820391http://dx.doi.org/10.1016/0022-0248(72)90185-6http://dx.doi.org/10.1149/1.2424100http://dx.doi.org/10.1143/JJAP.42.L1http://dx.doi.org/10.1143/JJAP.40.L140http://dx.doi.org/10.1143/JJAP.40.L140http://dx.doi.org/10.1016/j.jcrysgro.2007.12.053http://dx.doi.org/10.1557/JMR.1996.0350http://dx.doi.org/10.1016/0022-4596(88)90236-8http://dx.doi.org/10.1016/j.jcrysgro.2009.01.046http://dx.doi.org/10.1116/1.581360http://dx.doi.org/10.1116/1.581360http://dx.doi.org/10.1016/j.tsf.2008.05.050