CrystEngComm

PAPER

Cite this: CrystEngComm, 2017, 19,

2359

Received 9th March 2017,Accepted 27th March 2017

DOI: 10.1039/c7ce00479f

rsc.li/crystengcomm

On-axis Si-face 4H-SiC epitaxial growth withenhanced polytype stability by controlling micro-steps during the H2 etching process

Hyunwoo Kim,a Hunhee Lee,b Young Seok Kim,a Suhyeong Lee,a Hongjeon Kang,a

Jaeyeong Heo *c and Hyeong Joon Kim*a

In situ H2 etching and homoepitaxial growth of 4H-SiC were performed on on-axis Si-face substrates using

low-pressure chemical vapor deposition. We investigated the effect of various etching temperatures and

durations on the surface of the on-axis substrates and used optimized etching characteristics to enhance

the polytype stability of the epitaxial layer. It was found that the micro-steps at the etched surface of the

on-axis substrates play a vital role in enhancing the stability of the 4H-SiC polytype. Homoepitaxial 4H-SiC

with up to 99% stability was successfully grown with spread-out micro-steps without step-bunching during

the etching process.

Introduction

Silicon carbide (SiC) is a wide bandgap semiconductorpossessing properties such as high decomposition tempera-ture, large breakdown electric field, good thermal conductiv-ity, and high saturation velocity. Because of these uniqueproperties, SiC has been actively studied as the material ofchoice for high temperature, high voltage, and high frequencyapplications.1–3 Different stacking sequences of Si–C bilayerslead to different polytypes of SiC. As a result, numerous stud-ies on polytypes such as 3C-SiC, 4H-SiC, and 6H-SiC havebeen reported.4–6 Such polytypes exhibit different characteris-tics,7 and it is important to control the formation of polytypesin the crystal growth of SiC.8

Currently, for enhancing the polytype stability of SiChomoepitaxy, an off-axis substrate with several degrees off-cut toward the [112̄0] direction is commonly used.9,10 How-ever, even though the polytype stability is enhanced in thiscase, numerous other issues exist. First, the basal plane dislo-cations (BPDs) are transferred into the epilayer from the sub-strate. BPDs are known as “killer defects,” which, when pres-ent within the epilayer, largely degrade the forward voltage ofbipolar devices.11,12 Second, the number of wafers obtained

decreases when the substrate is cut along the off-cut direc-tion in the SiC ingot. In recent years, there has been analarming increase in the amount of discarded SiC ingots asthe sizes of wafers have increased. Third, anisotropies of theepilayers can appear along the step-flow direction because ofthe off-cut angle. These include not only structural aniso-tropies13 but also anisotropies in the channel mobility or thethreshold voltage in trench metal-oxide-semiconductor field-effect transistors because of the dependence of the metal-oxide-semiconductor interface on the face polarity.14–16

Hence, in order to solve these issues, it is essential tostudy epitaxy in nominally on-axis substrates. However, be-cause of the low step density in nominally on-axis substrates,there is a possibility of creating unintended polytypes on epi-taxial layers.17–19 This polytype instability has been the big-gest problem faced in the study of on-axis epitaxy. Currently,hydrogen etching prior to the deposition of the epilayer isconsidered to be a crucial step for achieving homoepitaxialgrowth in epitaxy with nominally on-axis substrates.20–23

Hassan et al. reported that hydrogen etching in a Si-rich envi-ronment increases the polytype stability of the epilayer.20

Kojima et al. reported the enhanced morphology and poly-type stability of the epilayer on a C-face substrate after etch-ing when compared to a Si-face substrate.21

However, it is not clear which characteristics of the etchedsubstrate actually enhance the polytype stability of the epi-layer. Moreover, most of the on-axis epitaxy studies usesource materials that have C/Si ratios less than 1.0.20–22 If asource with a low C/Si ratio is used, the surface diffusionlength becomes longer, which is advantageous for improvingthe polytype stability on on-axis substrates.24 However, a ma-jor disadvantage in such low C/Si ratio sources is that it

CrystEngComm, 2017, 19, 2359–2366 | 2359This journal is © The Royal Society of Chemistry 2017

aDepartment of Materials Science and Engineering and, Inter-university

Semiconductor Research Center (ISRC), Seoul National University, San 56-1,

Sillim-dong, Gwanak-gu, Seoul 151-742, Republic of Korea.

E-mail: thinfilm@snu.ac.krbDiffusion Technology Team, Memory Division, Samsung Electronics Co. Ltd.,

Hwaseong-si, Gyeonggi-do, 445-701, Republic of Koreac Department of Materials Science and Engineering and Optoelectronics

Convergence Research Center, Chonnam National University, Gwangju 61186,

Republic of Korea. E-mail: jheo@jnu.ac.kr

2360 | CrystEngComm, 2017, 19, 2359–2366 This journal is © The Royal Society of Chemistry 2017

becomes difficult to control the background doping concen-tration because of the increased carbon vacancy.25

In this study, we investigated the etching characteristics ofon-axis Si-face substrates and report the effect of these char-acteristics on the polytype stability of the epilayer grownusing a bisIJtrimethylsilyl)methane (BTMSM) source with ahigh C/Si ratio of 3.5. By understanding the correlation be-tween the etching characteristics and the epilayer, we con-trolled the micro-steps of the etched substrates and improvedthe polytype stability of 4H-SiC up to 99%.

Experimental

The SiC epitaxial growth experiments were performed in alow pressure chemical vapor deposition reactor. A horizontalcold wall system, with a SiC-coated graphite susceptor heatedinductively using a RF generator, was used. The substratetemperature was measured using an optical pyrometer, whichwas calibrated using the melting temperature of Si (1418 °C).Research grade on-axis 4H-SiC (0001) substrates were pur-chased from Cree Inc. Unlike other off-axis wafers that areavailable in product grade, only research grade is availablefor on-axis substrates. 8 × 8 mm2 specimens cut from a 100mm diameter 4H-SiC Si-face on-axis wafer polished by chemi-cal mechanical polishing (CMP) were used for the experi-ments. The nominal off-cut angle of the substrates measuredby the supplier was 0–0.04 degrees. BisIJtrimethylsilyl)methane(C7H20Si2) was used as the precursor for epitaxial growth. Thesubstrates were cleaned with acetone followed by methanol, amixture of H2SO4/H2O2 in a 4 : 1 ratio at 130 °C and diluteHF, and then rinsed with deionized water. This was carriedout to eliminate any surface contamination and native oxidelayer. Prior to epitaxial growth, the substrates were etchedwith H2 gas at a pressure of 180 Torr to prepare them for epi-taxial growth and eliminate any substrate damage that mayhave occurred during the polishing step. The etching temper-ature was varied between 1300 °C and 1600 °C to investigateits effects on the etching characteristics of the on-axis sub-strates. Unlike our conventional process for off-axis sub-strates26 or general etching conditions for SiC epitaxy,18,23,24

where an etching duration of 10–30 min above 1500 °C ismaintained, the etching duration was varied from 10 min to120 min at 1400 °C for the on-axis substrate used here. Theepitaxial growth temperature was 1550 °C and the growthpressure was 180 Torr. The input flow rate of the diluent H2

gas was kept at 3000 sccm. The BTMSM source was bubbledthrough H2 gas at a flow rate of 10 sccm, in a hand-madesource container. The source container was immersed in awater bath controlled by a thermostat device with a circula-tion pump (LAUDA E100, temperature control: ±0.02 °C). Thebubbler temperature was maintained at 24 °C. The vaporpressure of the BTMSM source at 24 °C was estimated to beapproximately 13.55 Torr by using the Clausius–Clapeyronequation. The actual flow rate of BTMSM with a H2 bubblinggas flow rate of 10 sccm was calculated to be 0.82 sccm.

The surface morphology of the H2-etched substrates andepitaxial layers was observed using optical microscopy withNomarski differential interference contrast (DIC). The surfaceroughness and step structure of the etched substrates werecharacterized by atomic force microscopy (AFM) in the tap-ping mode. In order to analyze the dislocations on the sur-face, the substrates were etched in molten KOH at 550 °C for8 min 30 s. Polytype identification of the layers wasperformed by micro-Raman spectroscopy using an Ar laserbeam (∼532 nm) with a spot size of a few μm.2 The polytypestability of the epilayers was determined using electron back-scatter diffraction (EBSD) mapping and by measuring the 4Hpolytype ratio in the image maps. The epitaxial layers wereanalyzed using scanning electron microscopy (SEM) and thecross-sectional images were used to investigate the thicknessof the epilayers.

Results and discussionEtching characteristics of the on-axis substrate

Prior to epilayer deposition, it is essential to etch the sub-strates with hydrogen to remove the damage due topolishing.27,28 In order to optimize the experimental condi-tions for hydrogen etching of on-axis substrates, we variedthe etching temperature from 1300 °C to 1600 °C. Fig. 1(a)–(d)illustrate the optical images of the specimens hydrogen-etched at 1300 °C, 1400 °C, 1500 °C, and 1600 °C, respec-tively, for 10 min. The inset in Fig. 1(a) is an image of thebare SiC wafer with the same scale. Though no special fea-ture is observed at 1300 °C (Fig. 1(a)), it can be seen that hex-agonal etch pits appear at 1400 °C, like in region 1 inFig. 1(b). The inset in Fig. 1(b) is a magnified image of the

Fig. 1 Optical images of the surface of the substrates etched attemperatures of (a) 1300 °C, (b) 1400 °C, (c) 1500 °C, and (d) 1600 °Cfor 10 min. The surface of the bare wafer is shown in the inset in (a)with the same scale. Regions marked 1 and 2 in (b) are an etch pit (1)and the periphery of the etch pit (2) of the etched substrate. Regionsmarked 3 and 4 in (c) are the micro-steps (3) and the bunched step (4)of the etched substrate. The inset in (b) is a magnified image of theetch pit.

CrystEngCommPaper

CrystEngComm, 2017, 19, 2359–2366 | 2361This journal is © The Royal Society of Chemistry 2017

etch pit. In addition, wavy step-like structures appear at1500 °C (Fig. 1(c)) and 1600 °C (Fig. 1(d)). Degradation of theoff-axis substrates is frequently observed because of step-bunching, which is the direct result of high-temperaturehydrogen etching. However, the behavior of the on-axissubstrates appears to be quite different from the case of theoff-axis substrates. The evolution of the etch pits occurred ata relatively low temperature of 1400 °C and step-bunchingstarted at and above 1500 °C. In contrast, no significantchange was observed after etching at 1500 °C for 4° off-axisor at 1600 °C for 8° off-axis substrates in our system. Thisresult implies that the on-axis substrates degraded even atlow temperatures because of etching when compared to theoff-axis substrates. There have been several reports statingthat step-bunching increases at smaller tilt angles of thesubstrates.29–31 The exact mechanism for this observation isnot well understood, but the step-bunching phenomenon inthe low angle off-axis substrates may be related to thestepped structure of the vicinal faces with low inclinationangles for the ideal (0001) plane.

To investigate the surface of the specimens microscopi-cally, 5 × 5 μm2 AFM images and root-mean-square (RMS)roughness of the substrates etched at temperatures rangingfrom 1300 °C to 1600 °C and of the bare wafer (unetched) areshown in Fig. 2. For the specimens etched at temperaturesbetween 1400 °C and 1600 °C, the morphology may varydepending on where the AFM measurements are performed.Fig. 2 shows a general area without the specific morphologies(etch pits for 1400 °C, bunched steps for 1500 °C and 1600°C). The AFM images of such specific regions are shown inFig. 3. The RMS roughness of the bare wafer is 1.02 nm, andit is slightly larger than those of other commercial productgrade off-axis substrates (<0.75 nm). Scratches due to roughmechanical polishing were not observed. Though the macro-scopic morphologies degraded because of etching at hightemperatures (Fig. 1), the microscopic RMS roughness valuesof the etched substrates are lower than that of the bare wafer,irrespective of the temperature. This implies that any damageto the wafer because of polishing is effectively removed dur-ing the hydrogen etching process. From the AFM images, weobserve that uniform steps are formed on the surface at etch-ing temperatures of 1500 °C and 1600 °C. Such uniform stepswere unexpected, as they did not exist prior to the high tem-

perature etching process. In general, while steps of severalnanometers are created because of step-bunching at hightemperature,32–34 in our case, the on-axis substrates exhibitedregular steps with uniform height. The heights of thesemicro-steps were 1 nm, which is consistent with the unit-cellheight of 4H-SiC. Regular micro-steps of unit-cell height havebeen reported in several studies for both on-axis 4H- and6H-SiC.20,30,35,36 There are several explanations for the originof these micro-steps. Some researchers suggested that themicro-steps originate from the etching of screw dislocations.Hassan et al. investigated the etched surface of 4H-SiC andreported that regular micro-steps form from elementaryscrew dislocations and then spread over the entire surface.30

Another explanation is that micro-steps arise from the

Fig. 2 AFM images (5 × 5 μm2) and RMS values of the bare wafer and specimens that were etched at temperatures from 1300 °C to 1600 °C for10 min.

Fig. 3 AFM images of (a) the etch pit (20 × 20 μm2, region 1 in Fig. 1)and (b) the peripheral part of the etch pit (5 × 5 μm2, region 2 in Fig. 1)of the substrate etched at 1400 °C for 10 min. 5 × 5 μm2 AFM imagesof (c) the micro-steps (region 3 in Fig. 1) and (d) the bunched step (re-gion 4 in Fig. 1) of the substrate etched at 1500 °C for 10 min.

CrystEngComm Paper

2362 | CrystEngComm, 2017, 19, 2359–2366 This journal is © The Royal Society of Chemistry 2017

unintentional miscut of the wafers, which is related to thestacking sequences of Si–C bilayers. Nakajima et al. claimedthat the anisotropic etching rate of vicinal surfaces leads tothe step structures of the 6H-SiC surface.35 Interestingly, thegeneration of steps because of the etching of screw disloca-tions was also reported in the same study.

In our case, the etching of threading screw dislocations(TSDs) appears to influence the formation of the micro-steps.In order to understand the origin of the micro-steps, it isimportant to note that etch pits first appeared at 1400 °C.The AFM image of the etch pit at 1400 °C is shown inFig. 3(a) (region 1 in Fig. 1(b)). From Fig. 3(a), it can be seenthat uniform steps were created at the center of the etch pit,while no micro-steps were observed at the periphery of theetch pit in Fig. 3(b) (region 2 in Fig. 1(b)). The steps in theetch pit have a height of 1 nm, as shown in the profile withan arrow in Fig. 3(a), and this is consistent with the latticeparameter (c) of 4H-SiC. In addition, the density of the etchpit is approximately 5 × 103 cm−2, determined from the opti-cal microscopy image. The average concentration of TSDs onthe substrate was determined to be 3.8 × 103 cm−2 using mol-ten KOH etching in six different areas, which is similar to thedensity of the etch pits. Here, the density of the measuredTSDs is slightly higher than those of other commerciallyavailable off-axis wafers (<1 × 103 cm−2). When the height ofthe steps and the concentration of the etch pits are consid-ered, it appears that this etch pit arises because of selectiveetching of TSDs.

On closer observation of the specimen surface, it can beseen that etching at a high temperature (1500 °C), as shownin Fig. 3(c) (region 3 in Fig. 1(c)), results in micro-stepscreated throughout the entire specimen, unlike etching atlow temperature. Notably, the direction of the steps was al-ways perpendicular to the <11̄00> direction group, and thestep height corresponds to the strength of the Burgers vectorof the screw dislocations (1 nm). Other than these regularmicro-steps, bunched steps were observed in some parts ofthe specimen, as representatively shown in Fig. 3(d) (region 4in Fig. 1(c)). These bunched steps had a step height of severalnanometers to tens of nanometers. The bunched steps canbe formed from the merging of micro-steps because of thehigh etching temperature, or step-bunching of anunintentional off-cut.36 Their exact origin is unclear cur-rently, but it is evident that more bunched steps with a largerheight form at increased etching temperatures. Notably, themicro-steps disappear at the peripheral part of a bunchedstep, as seen in Fig. 3(d). It is important to understand howthese micro-steps created during the etching process affectthe epitaxial growth, as discussed later.

Dependence of epilayer characteristics on etching temperature

Fig. 4(a)–(d) illustrate the top-view optical micrographs ofthe epilayers grown for 1 h at a deposition temperature of1550 °C after hydrogen etching at various temperatures be-tween 1300 °C and 1600 °C for 10 min, respectively. Notably,

large hillocks can be seen in the optical microscopy imagesof the grown specimen (dotted lines and triangles). The hill-ocks in Fig. 4 may look concave compared to the surround-ings, though they are actually convex. This is because thepseudo-3D appearance of the DIC-imaged specimens is af-fected by the translation of the objective Nomarski prism.The growth mechanism of the SiC layers can be classifiedinto three types. Hillocks are formed from the spiral growthwith TSDs in the center, and they are common for variouson-axis substrates (not only in SiC).37–39 The second mecha-nism is step-flow growth through the steps of the substrate.This step-flow growth may occur at micro-steps generatedduring the etching process. The third mechanism is growthby nucleation on the terrace and it shows a high probabilityof growth of other polytypes.40 Other regions observed inFig. 4, except hillocks, may form following the second and/or third mechanisms. These two mechanisms are indistin-guishable in the optical microscopy images.

We can observe the polytype through EBSD mapping ofthe surface of the epilayers grown after etching at tempera-tures of 1400 °C and 1500 °C, as shown in Fig. 5. The redarea in the mapping image shows an area with a 4H-SiC poly-type, whereas the black area shows an area with other poly-types. The polytype of the black area was not identified fromthe EBSD mapping, but it was revealed to be 3C-SiC byRaman analysis as discussed in the following paragraph. Allthe hillocks seen in the optical image are 4H-SiC (dotted re-gions). This can be attributed to the fact that the screw dislo-cation acts as a growth site in the on-axis substrate and itprovides a step with a 1c (unit cell height) Burgers vector,transferring the 4H polytype stacking sequence to the layer.In other areas, we can observe a mixture of 4H-SiC and otherpolytypes. They are probably grown by step-flow modeforming 4H and by nucleation forming other polytypes.

Raman analysis confirms the polytypes of the depositedlayer. Fig. 6 shows the results of Raman analysis on several

Fig. 4 Optical images of the surface of epitaxial layers grown for 1 hat 1550 °C. Specimens are etched before the growth process at (a)1300 °C, (b) 1400 °C, (c) 1500 °C, and (d) 1600 °C for 10 min. Trianglesand dotted lines indicate the hillocks grown by screw dislocation.

CrystEngCommPaper

CrystEngComm, 2017, 19, 2359–2366 | 2363This journal is © The Royal Society of Chemistry 2017

regions of the epitaxial layers grown at 1550 °C for 1 h. Etch-ing was performed at 1500 °C for 10 min before growth. Asin the case of EBSD mapping, shown in Fig. 5(b), it is seenthat only the 4H-SiC-related peaks were detected in the hill-ocks (4Hh in Fig. 6(a)). Regions other than the hillocks canbe divided into 3C-SiC (3C1 and 3C2) and 4H-SiC (4H1 and4H2) domains using Raman spectroscopy. As shown inFig. 6(b), the FLO peak of 3C-SiC at 971 cm−1 is observed inthe 3C domains. On the other hand, only the FLO peak of4H-SiC at 964 cm−1 is observed in the 4H domains. Notably,some of the 4H domains have a small FLO peak of 3C-SiC at971 cm−1 because of 3C inclusion, as shown in a representa-tive 4H1-2 region in Fig. 6(c). The intensity of the FLO peakof 3C attributed to the inclusions is weaker than that in the3C domains such as 3C1 and 3C2. Generation of differentpolytypes within the same crystal bulk and specific mecha-nism, which is very similar to our case, is reported for pro-teins by Aquilano et al.41

The role of screw dislocations and micro-steps is more evi-dent when we look at the specimen at the beginning ofgrowth. The optical micrographs of the surface of the speci-

men deposited at 1550 °C for 10 min after etching at 1400 °Cand 1500 °C for 10 min are shown in Fig. 7(a) and (b), respec-tively. In the case of the specimen etched at 1400 °C, we ob-served hillocks created at the early stage of deposition andwavy steps appear in other regions. A similarity was found be-tween these wavy structures and the etched surfaces at hightemperatures of 1500 °C or 1600 °C (Fig. 1(c) and (d)). Thecross-sectional SEM image in Fig. 7(c) shows that the epitax-ial layer was not deposited on other parts except the hillocks.Fig. 7(d) is a magnified image of the hillock. It suggests thatonly the hillocks are selectively formed, while step-bunchingdue to high temperature occurs at other parts. On the otherhand, in the case of the specimen etched at 1500 °C, the sur-face image in Fig. 7(b) indicates that growth occurred moreuniformly throughout the entire surface. The presence of acontinuous ∼200 nm-thick epilayer is confirmed by thecross-sectional SEM image in Fig. 7(e). Here, it is difficult toseparate the hillocks and the regions where step-flow growthoccurred based on the optical and SEM images.

Fig. 6 Raman analysis results for epitaxial layers grown for 1 h at 1550 °C. The specimen was etched for 10 min at 1500 °C before the growthprocess. (a) Optical microscopy image with hillock, 3C, and 4H regions. (b) Raman spectra of the hillock, 3C, and 4H regions in the epitaxial layers.(c) Magnified Raman spectra of the 4H1-2 region.

Fig. 5 (a) and (b) EBSD maps of epitaxial layers grown for 1 h at 1550°C and (c) inverse pole figure triangle. Specimens are etched beforethe growth process at (a) 1400 °C and (b) 1500 °C for 10 min. Dottedlines indicate the hillocks. All the hillocks are 4H-SiC.

Fig. 7 Optical and cross-sectional SEM images of epitaxial layersgrown for 10 min at 1550 °C. Specimens are etched before the growthprocess at 1400 °C ((a), (c) and (d)) and 1500 °C ((b) and (e)) for 10 min.Dotted lines and arrows in (a) indicate the hillocks. The epitaxial layerwas not deposited on other parts except the hillocks in (c). A magnifiedimage of the hillock is shown in (d). (e) The cross-sectional SEM imageof (b).

CrystEngComm Paper

2364 | CrystEngComm, 2017, 19, 2359–2366 This journal is © The Royal Society of Chemistry 2017

The initial growth of the epilayer is highly influenced bythe surface conditions after etching. When the specimen isetched at 1400 °C for 10 min, etch pits are generated andmicro-steps only exist in the etch pits. This observation ex-plains why selective deposition proceeded only at a screw dis-location, resulting in the formation of a hillock: micro-stepsinside the etch pit and steps provided by the screw disloca-tion act as the growth sites. On the other hand, in the case ofetching at 1500 °C for 10 min, the micro-steps are formedover a large area during etching. Therefore, micro-steps be-come the growth sites and the SiC layer is deposited on theentire specimen. Fig. 4 also suggests the selective depositionof hillocks for specimens etched at low temperatures (epitax-ial layers were grown for 1 h). It can be seen that the size ofhillocks in Fig. 4(a) and (b) is larger than that inFig. 4(c) and (d). Micro-steps created by etching can act onthe epilayers as follows: 1) the micro-steps provide a growthsite for step-flow growth on an on-axis substrate that initiallydid not have a step site; 2) as the stacking sequence of 4H-SiC (ABAC) is revealed on the side walls of the micro-steps, itenhances the polytype stability of the deposited layer. Theschematic illustration of the micro-steps exhibiting the stack-ing sequence of 4H-SiC is shown in Fig. 8.

Although micro-steps are present on the entire specimenwhen etched at high temperature, they cannot explain thegrowth of other polytypes such as 3C-SiC. Low polytype stabil-

ity at high-temperature-etched specimens may be affected bythe bunched steps observed from the AFM image of the spec-imen etched at high temperature (Fig. 3(d)). Micro-stepsdisappeared in the peripheral part of a bunched step so step-flow growth should be weakened. In such a case, other poly-types of 3C-SiC can nucleate. There was a similar report thatthe growth of 3C is promoted by step-bunching, which gener-ates a longer terrace.40

Improvement of polytype stability according to changes inetching conditions

If micro-steps with a height of 1 nm can provide sites for epi-layer growth and improve the stability of the 4H polytype, itis important to make sure that the micro-steps are present inthe specimen without step-bunching. At an etching tempera-ture of 1400 °C, etch pits appeared with selective etching ofTSDs; however, when etched for 10 min, micro-steps wereformed only around the etch pit part. On the other hand,though micro-steps were spread throughout the specimenat high etching temperature, step-bunching occurred.Based on this observation, we attempted to realize micro-steps spread throughout the specimen without step-bunching, by increasing the etching duration at a tempera-ture of 1400 °C.

Fig. 9(a) shows the optical image of the surface of thespecimen obtained using a Nomarski microscope in whichwe increased the etching duration to 90 min at a temperatureof 1400 °C. It is shown that the etch pit disappeared com-pared to the case of 10 min-etching (Fig. 1(b)). In addition, itshows a better surface morphology than the specimen etchedfor 10 min at 1500 °C (Fig. 1(c)). Formation of uniform stepsthroughout the surface was observed from the AFM image inFig. 9(b). For the specimen etched for 90 min at 1400 °C, wefound that there are regions with step-bunching, as shown inFig. 9(c) and the bunched step had a maximum height ofabout 3 nm. However, this step-bunching was very rare com-pared to the etching at high temperature.

Fig. 10 shows the EBSD mapping of the epilayers depos-ited for 1 h after etching for 10–120 min at a temperature of1400 °C. The stability of the 4H-SiC polytype improved withincreasing etching duration and almost 99% stability of the

Fig. 9 (a) Optical image and 5 × 5 μm2 AFM images of (b) micro-steps and (c) the bunched step of a substrate etched at 1400 °C for 90 min.

Fig. 8 Schematic diagram of the generated micro-steps. The stackingsequence of 4H-SiC (ABAC) is revealed on the side walls of the micro-steps.

CrystEngCommPaper

CrystEngComm, 2017, 19, 2359–2366 | 2365This journal is © The Royal Society of Chemistry 2017

4H-SiC polytype was achieved at 120 min. As the etching du-ration is increased, the micro-steps spread throughout thespecimen and the micro-steps would provide sites for step-flow growth. As a result, the polytype stability is enhanced.The temperature ramp from 1400 °C for etching to 1550 °Cfor growth took 30 s and it is believed that the surface degra-dation occurring because of step-bunching is insignificant.

Our results demonstrate that it is possible to improvepolytype stability using precursors, even with high C/Si ratios,such as BTMSM, in on-axis Si-face epitaxy which will providemore control on background doping of the grown epilayers.As mentioned earlier, if C/Si becomes higher, the surface dif-fusion length becomes smaller. Therefore, it is difficult foradatoms to reach a step because of the short surface diffu-sion length in the on-axis substrate with low step densitycompared to an off-axis substrate. Thus, it is highly disadvan-tageous for step-flow growth. That is, it is difficult to main-tain the polytype stability, as nucleation of 3C-SiC is likely tooccur. However, by increasing the etching duration at a lowtemperature and realizing micro-steps spread throughoutwithout step-bunching, we could provide a site for step-flowgrowth despite this short surface diffusion length, and there-fore, we could increase the stability of 4H-SiC.

Conclusions

A homoepitaxial layer of 4H-SiC was successfully grown onthe on-axis Si-face substrate with improved polytype stabilityusing a BTMSM precursor with a high C/Si ratio of 3.5. Forthe on-axis substrates, micro-steps were generated on the sur-face of the substrates after H2 etching because of selectiveetching of TSDs. These micro-steps at the etched surface con-tribute to the growth of the 4H polytype because they providesites for step-flow growth at the side walls. However, for ahigh etching temperature (above 1500 °C), step-bunching oc-curred and micro-steps disappeared around a bunched step.Since other polytypes such as 3C easily nucleate at bunchedsteps with a larger terrace, the formation of regular micro-steps without step-bunching on an etched surface is vital for

enhancing the polytype stability. To spread out the micro-steps without step-bunching during the etching process, anincrease in the etching duration at a temperature lower thanthe conventional etching temperature was required. Improvedpolytype stability of 4H-SiC up to 99% was finally achieved.

Acknowledgements

This work was supported by the Research Center for StrategicMaterials of POSCO and by Basic Science Research Programthrough the National Research Foundation of Korea (NRF)funded by the Ministry of Science, ICT & Future Planning(NRF-2015R1C1A1A02036616).

References

1 R. J. Trew, J.-B. Yan and P. M. Mock, Proc. IEEE, 1991, 79,598–620.

2 B. J. Baliga, Proc. IEEE, 1994, 82, 1112–1122.3 S. G. Müller, M. Brady, A. Burk, H. M. Hobgood, J. Jenny, R.

Leonard, D. Malta, A. Powell, J. Sumakeris and V. Tsvetkov,Superlattices Microstruct., 2006, 40, 195–200.

4 M. Bosi, C. Ferrari, D. Nilsson and P. Ward, CrystEngComm,2016, 18, 7478–7486.

5 N. Hoshino, I. Kamata, Y. Tokuda, E. Makino, N. Sugiyama,J. Kojima and H. Tsuchida, Appl. Phys. Express, 2014, 7,065502.

6 S. Lin, Z. Chen, P. Liang, Y. Ba and S. Liu, CrystEngComm,2011, 13, 2709–2713.

7 R. C. Marshall, Silicon carbide–1973: proceedings, Univ ofSouth Carolina Pr, 1974.

8 N. Tsavdaris, K. Ariyawong, J. M. Dedulle, E. Sarigiannidouand D. Chaussende, Cryst. Growth Des., 2015, 15, 156–163.

9 T. Kimoto, A. Itoh and H. Matsunami, Phys. Status Solidi B,1997, 202, 247–262.

10 H.-S. Kong, J. T. Glass and R. F. Davis, J. Appl. Physiol.,1988, 64, 2672–2679.

11 P. Bergman, H. Lendenmann, P. Å. Nilsson, U. Lindefelt andP. Skytt, Mater. Sci. Forum, 2001, 353, 299–302.

12 P. Å. Persson, L. Hultman, H. Jacobson, J. Bergman, E.Janzén, J. Molina-Aldareguia, W. Clegg and T. Tuomi, Appl.Phys. Lett., 2002, 80, 4852–4854.

13 T. Kimoto and H. Matsunami, J. Appl. Phys., 1994, 76,7322–7327.

14 S. Harada, S. Ito, M. Kato, A. Takatsuka, K. Kojima, K.Fukuda and H. Okumura, Mater. Sci. Forum, 2010, 645,999–1004.

15 Y. Ueoka, K. Shingu, H. Yano, T. Hatayama and T. Fuyuki,Jpn. J. Appl. Phys., 2012, 51, 110201.

16 K. Fukuda, M. Kato, K. Kojima and J. Senzaki, Appl. Phys.Lett., 2004, 84, 2088–2090.

17 W. Muench and I. Pfaffeneder, Thin Solid Films, 1976, 31,39–51.

18 V. Jennings, A. Sommer and H. Chang, J. Electrochem. Soc.,1966, 113, 728–731.

Fig. 10 EBSD maps of epitaxial layers grown for 1 h at 1550 °C afteretching with increasing etching duration. Specimens are etched beforethe growth process at 1400 °C for (a) 10 min, (b) 30 min, (c) 60 min,(d) 90 min, and (e) 120 min.

CrystEngComm Paper

2366 | CrystEngComm, 2017, 19, 2359–2366 This journal is © The Royal Society of Chemistry 2017

19 S. Nakamura, T. Kimoto and H. Matsunami, J. Cryst. Growth,2003, 256, 341–346.

20 J. Hassan, J. Bergman, A. Henry and E. Janzén, J. Cryst.Growth, 2008, 310, 4424–4429.

21 K. Kojima, H. Okumura, S. Kuroda and K. Arai, J. Cryst.Growth, 2004, 269, 367–376.

22 P. Bergman, J. Palisaitis, A. Henry, P. J. McNally, S. Andersonand E. Janzén, Mater. Sci. Forum, 2010, 645, 83–88.

23 S. Leone, F. C. Beyer, H. Pedersen, O. Kordina, A. Henry andE. Janzen, Cryst. Growth Des., 2010, 10, 5334–5340.

24 T. Kimoto and H. Matsunami, J. Appl. Phys., 1995, 78,3132–3137.

25 D. Larkin, Phys. Status Solidi B, 1997, 202, 305–320.26 H. S. Seo, H. G. Song, J. H. Moon, J. H. Yim, M. S. Oh, J. H.

Lee, Y. J. Choi and H. J. Kim, Mater. Sci. Forum, 2009, 600,151–154.

27 V. Ramachandran, M. Brady, A. Smith, R. Feenstra and D.Greve, J. Electron. Mater., 1998, 27, 308–312.

28 C. Hallin, F. Owman, P. Mårtensson, A. Ellison, A.Konstantinov, O. Kordina and E. Janzén, J. Cryst. Growth,1997, 181, 241–253.

29 K. Kościewicz, W. Strupiński, D. Teklinska, K. Mazur, M.Tokarczyk, G. Kowalski and A. R. Olszyna, Mater. Sci. Forum,2011, 679, 95–98.

30 J. Hassan, J. Bergman, A. Henry and E. Janzén, J. Cryst.Growth, 2008, 310, 4430–4437.

31 K. Kojima, K. Masumoto, S. Ito, A. Nagata and H. Okumura,ECS J. Solid State Sci. Technol., 2013, 2, N3012–N3017.

32 T. Kimoto, A. Itoh, H. Matsunami and T. Okano, J. Appl.Phys., 1997, 81, 3494–3500.

33 N. Ohtani, M. Katsuno, T. Aigo, T. Fujimoto, H. Tsuge, H.Yashiro and M. Kanaya, J. Cryst. Growth, 2000, 210,613–622.

34 Y. Ishida, T. Takahashi, H. Okumura, K. Arai and S.Yoshida, Mater. Sci. Forum, 2009, 600, 473–476.

35 A. Nakajima, H. Yokoya, Y. Furukawa and H. Yonezu, J. Appl.Phys., 2005, 97, 4919.

36 S. Nie, C. Lee, R. M. Feenstra, Y. Ke, R. Devaty, W. Choyke,C. Inoki, T. Kuan and G. Gu, Surf. Sci., 2008, 602, 2936–2942.

37 J. Bergman, J. Hassan, L. Lilja, A. Salemi, A. Hallen and E.Janzén, Meeting Abstracts, 2013, p. 1905.

38 S.-I. Nakamura, T. Kimoto and H. Matsunami, Jpn. J. Appl.Phys., 2003, 42, L846.

39 P. Smereka, Phys. D, 2000, 138, 282–301.40 A. Konstantinov, C. Hallin, B. Pecz, O. Kordina and E.

Janzen, J. Cryst. Growth, 1997, 178, 495–504.41 D. Aquilano, S. Veesler, J. P. Astier and L. Pastero, J. Cryst.

Growth, 2003, 247, 541–550.

CrystEngCommPaper