Morphology Control of Hot-Wall MOCVD Selective Area Grown

This paper presents growth of (0001)-oriented hexagonal GaN pyramids by hot-wall ... The presence of the {11̅02} facets near the apex of the pyramid ...
2 downloads 0 Views 2MB Size
Article pubs.acs.org/crystal

Morphology Control of Hot-Wall MOCVD Selective Area Grown Hexagonal GaN Pyramids Anders Lundskog,* Urban Forsberg, Per Olof Holtz, and Erik Janzén Department of Physics, Chemistry and Biology, Linköping University, SE-581 83 Linköping, Sweden ABSTRACT: Morphological variations of gallium polar (0001)-oriented hexagonal GaN pyramids grown by hot-wall metal organic chemical vapor deposition under various growth conditions are investigated. The stability of the semipolar {11̅02} and nonpolar {11̅00} facets is particularly discussed. The presence of the {11̅02} facets near the apex of the pyramid was found to be controllable by tuning the absolute flow rate of ammonia during the growth. Vertical nonpolar {110̅ 0} facets appeared in gallium-rich conditions, which automatically were created when the growth time was prolonged beyond pyramid completion. The result was attributed to a gallium passivation of the {11̅00} surface.



facets.13 However, the pyramids often possess bifacets, such as nonpolar14−17 {11̅00} and semipolar18,19 {11̅02} facets. Growth of gallium polar (0001)-oriented GaN nanowires has also been reported on (0001)-oriented GaN templates by the use of low continuous precursor flux20,21 and a pulsed method.22,23 Several reports have recently shown that SAG performed on nitrogen polar (0001̅) GaN substrates (or by the use of a nitridation step on a sapphire substrate) favors the formation of nitrogen polar (0001̅)-oriented GaN nanowires.24−27 Li et al. recently suggested a growth model explaining the growth mechanism of such nanowires through the influence of the polarity and hydrogen concentration in the carrier gas.27 However, to our knowledge, there are no reports explaining the formation mechanism of the {11̅02} and {11̅00} facets of hexagonal pyramids and nanowires. Previously, it was established that the crystal shape of SAG GaN depends on, for example, the growth temperature, reactor pressure, and TMGa and NH3 flow rates.28,29 We have investigated the morphological variations in relation to the growth temperature, growth time, and ammonia flow rate.

INTRODUCTION Gallium nitride (GaN)-based semiconductors are very attractive for use in blue and ultraviolet light-emitting diodes (LEDs) and laser diodes (LDs). However, GaN has a relatively high refractive index,1 resulting in a small critical angle at the interface with air. The angle can be increased by encapsulating the LED structure in a transparent plastic medium. The problem may also partially be resolved by the use of threedimensional (3D) structures, such as stripes,2 nanowires,3 and pyramids,4 where the photon incidence angle can be modified. 3D structures also enable growth of indium−gallium−nitride quantum wells (QWs) on non- and semipolar planes of GaN when using conventional (0001)-oriented sapphire or SiC substrates. QWs grown on non- or semipolar planes possess a vanishing or reduced quantum confined stark effect5 (QCSE), which is believed to be responsible for the reduction in emission efficiency of long-wavelength LEDs.6,7 In addition to LED and LD applications, the 3D structures have other applications, such as cold cathode field emitters8,9 and selective nucleation sites of InGaN quantum dots (QDs).10,11 Depending on the application type, the preferred geometry of the 3D structure varies. For example, sharp apexes are preferred for field emitter applications, whereas truncated apexes are preferred for utilization of site-controlled InGaN QDs with narrow emission lines.12 In this paper, the resulting morphology of the 3D structure is investigated under various growth conditions. A growth model and a process parameter roadmap are proposed to explain the morphology of the resulting 3D structure. Selective area growth (SAG) using a patterned SiN mask has been proven to be an efficient way to localize the nucleation position of 3D structures. Although there have been numerous achievements in metal organic chemical vapor deposition (MOCVD) SAG during the past decade, the understanding of the facet formation mechanism is far from complete. SAG performed on circular window openings and gallium polar (0001)-oriented GaN templates as substrates often results in hexagonal GaN pyramids with dominating semipolar {11̅01} © 2012 American Chemical Society



EXPERIMENTAL DETAILS

We have used a 2.0 μm thick gallium-terminated (0001) GaN epitaxial film grown on a Si-face (0001) 4H-SiC substrate starting with a 100 nm thick AlN nucleation layer as a template for the hexagonal GaN pyramids. The template was grown in a hot-wall MOCVD system as described elsewhere.30,31 To facilitate SAG, a 30 nm thick SiN layer was deposited by plasma-enhanced chemical vapor deposition (PECVD). Subsequently, photolithography was adopted to transfer the pattern onto the SiN-covered substrate, followed by inductively coupled plasma etching to open holes in the SiN mask layer. The pattern consisted of circles located in square arrays. The diameters of the openings were approximately 3.0 and 6.0 μm with center-to-center distances of approximately 6.0 and 9.5 μm, respectively. After chemical cleaning, the templates were placed into the same hot-wall MOCVD reactor as used for the template growth. The temperature was ramped Received: July 26, 2012 Revised: September 6, 2012 Published: October 1, 2012 5491

dx.doi.org/10.1021/cg301064p | Cryst. Growth Des. 2012, 12, 5491−5496

Crystal Growth & Design

Article

up in a constant flow rate of ammonia (NH3) in order to protect the GaN surface from decomposition. After the growth temperature was reached, trimethylgallium (TMGa) was introduced for growth of hexagonal GaN pyramids. A mixture of purified hydrogen (H2) and nitrogen (N2) was used as the carrier gas. The H2/N2 flow rate ratio was set to 1.7 for all the investigated samples. The total rector pressure was fixed at 50 mbar. The pyramid morphology was characterized by scanning electron microscopy (SEM) with an acceleration voltage of 5 kV. The size of the pyramids was determined by a MATLAB image toolbox script. The average volume of the pyramids was calculated using the formula V = (√3/2)tan(φ)(S3 − P3), where S and P are the average side lengths of the bottom and upper bases of the isosceles trapezoid-shaped {11̅01} facets, respectively, and φ is the oblique angle between the (0001) and a (11̅ 01) surface, which is approximately 62°. The average side lengths were calculated from 10−12 independent hexagonal pyramids; that is, each average is calculated from 60−72 measurements since every pyramid has six {110̅ 1} facets.

From Figure 1b, it is clear that the individual {110̅ 1} and {11̅02} facets are randomly distributed at the same pyramid. This type of irregularity is present on all pyramids that possess visible {11̅02} facets. Figure 1c shows a sketch of the relevant pyramid facets and their individual perfect cleavage termination. Figure 2a−f shows bird's-eye-view SEM images of GaN pyramids grown under different NH3 flow rates and growth



RESULTS AND DISCUSSION A cross-sectional SEM image of a pyramid grown with 2 slm of NH3 at 1060 °C is shown in Figure 1a. The planar-view SEM image of the same pyramid is shown in Figure 1b. The large arrow in Figure 1b shows the observation angle of Figure 1a. The oblique angles for the sidewall facets referring to the (0001) plane are about 62° and 43°, respectively, which corresponds to the {11̅01} and {11̅02} facets of wurtzite GaN. In Figure 1b, two {11̅02} facets are highlighted by white lines. Figure 2. Bird's-eye-view SEM pictures illustrating pyramid morphology changes of hexagonal GaN pyramids grown under various temperatures and NH3 flow rates.

temperatures. A TMGa flow rate of 3.35 sccm and a growth time of 60 min are employed, except for the sample shown in Figure 2b, where a growth time of 30 min was employed instead. The columns and rows of Figure 2 correspond to different growth temperatures and NH3 flow rates, respectively. GaN pyramids with six smooth {11̅01} facets are formed under all growth conditions. The pyramids grown with NH3 flow rates less than or equal to 0.75 and 2.0 slm at 1000 and 1060 °C, respectively, had {11̅02} facets near the apex of the pyramid. The {11̅02} facets are barely visible in Figure 2a,b but were confirmed to have the same oblique angle as the (0001) facet, as shown in Figure 1a. The {11̅02} facets were terminated when the NH3 flow rate was increased to 6 slm or more, as shown in Figure 2c,d. Vertical {11̅00} facets only appeared at low NH3 flow rates and temperatures, as shown in Figure 2a,c. Figure 2e,f shows pyramids grown at a NH3 flow rate of 10 slm at 1000 and 1060 °C, respectively. The growth rate drastically decreased along the [0001] direction, leaving a micrometer-sized (0001) facet at the apex of the pyramid. The reduced growth rate is most likely a combination of nanoparticle formations in the vapor phase32,33 and a TMGa dilution effect due to the maintained constant pressure in the reactor. Surface roughening of the (0001) facet was observed at 10 slm of NH3. The pyramids grown at 1000 °C were particularly rough, and large pitlike formations on the (0001) facet were formed, as shown in Figure 2e. Formation of the {11̅00} Facets. As shown in Figure 2a,c, vertical nonpolar {11̅00} facets appeared at reduced growth temperatures (1000 °C) and NH3 flow rates. To gain a further understanding of the {11̅00} facet formation, additional

Figure 1. (a) Cross-sectional SEM image of a GaN pyramid grown at 2 slm of NH3 at 1060 °C. (b) Planar-view image of the same pyramid. The white lines mark two of five visible {11̅02} facets, illustrating their irregularity. The white arrow illustrates the viewpoint of (a). (c) GaN pyramid sketch showing relevant facets and directions. 5492

dx.doi.org/10.1021/cg301064p | Cryst. Growth Des. 2012, 12, 5491−5496

Crystal Growth & Design

Article

is mass-transport-limited by gallium. This was also confirmed from the slope of Figure 3a. In Figure 3a, the volume increase per minute in the respective regimes was determined to be 2.05 and 0.40 μm3/min. Growth performed under identical conditions on an unmasked (0001)-oriented GaN template (which we have confirmed to be mass-transport-limited of gallium) resulted in a [0001] growth rate of 1.25 μm/h. The growth rate corresponds to a normalized SAG volume increase per time unit of 1.90 μm3/min, which is nearly identical to the value obtained from regime 1 in Figure 3a. In mass-transportlimited growth, the surface adatom concentration of gallium is practically zero since the reaction kinetics are much faster than the transport of the gallium adatoms to the surface. In regime 2, on the other hand, the GaN reaction rate limits the growth rate. The growth may, therefore, be considered as kinetic-limited in the sense that there are available gallium adatoms present on the surface, but all of them do not contribute to the growth. As a consequence, the gallium adatom concentration at the growth front is larger in regime 2 compared with regime 1. Regimes 1 and 2 have similar (if not identical) V/III ratios; however, regime 2 should still be considered as gallium-rich since the gallium species no longer are consumed at the growth front. SEM images are shown in Figure 3c,d of two pyramids where the growth ended in regimes 1 and 2, respectively. The vertical {11̅00} facets appeared only on the pyramid where the growth time was long enough to enter regime 2. A similar behavior was observed over a large range of V/III ratios (∼300−2000) and growth temperatures (925−1100 °C). The atomic structures of a nonpolar (11̅00) GaN surface were studied using first-principles density functional theory (DFT) calculations in the literature.34−36 The DFT calculations consistently showed that gallium−gallium dimers (surface becomes covered with metallic gallium) preferentially are formed on the (11̅00) surface in gallium-rich growth conditions. DFT calculations also suggest the formation of stoichiometric gallium−nitrogen dimers on the (11̅00) surface under stoichiometric and nitrogen-rich growth conditions. Experimentally, both types of surfaces have been reported.34,37,38 We believe that the gallium-enriched conditions in regime 2 promote the formation of gallium−gallium dimers, which passivates the {11̅00} surfaces. The dimer passivation reduces the ⟨11̅00⟩ growth rate, and according to the kinetic Wullf theory of crystal growth,39,40 the slowest-growing facets are visible post growth. In accordance with previous reported experimental results, the {11̅00} facets formed in gallium-rich conditions.14,19,24 It has also been shown that H2 carrier gas can etch the GaN via Ga−H and N−H formation at elevated temperature.19 This is apparently not the mechanism behind the {11̅00} facet formation in regime 2, as the H2 carrier and NH3 gas flows were kept constant during the growth experiment. Formation of the {11̅01} Facet. The absence of the {11̅00} and {11̅02} facets at the elevated growth temperature (1060 °C) in Figure 2d,f (this also applies to the conditions of Figure 2b when the growth time is prolonged) was concluded to be a consequence of increased gallium adatom desorption of the {11̅01} surfaces. This was concluded by systematic variations of the growth temperature and TMGa flow around 1060 °C. At a growth temperature of 1060 °C and a TMGa flow of 3.35 sccm, the adatom desorption rate equals the adsorption rate on the (11̅01) surface, resulting in a ⟨11̅01⟩ growth rate of 0 μm/h. A detailed discussion about the hightemperature kinetics will be published elsewhere.

samples with varying growth times were grown. The average pyramid volume versus the growth time emerging from SiN opening diameters of 6.0 μm (T = 1000 °C, NH3 flow rate = 6 slm, TMGa = 3.35 sccm) is shown in Figure 3a. In this plot,

Figure 3. Average GaN pyramid volume versus growth time of the pyramids. (b) The average pyramid volume versus TMGa flow for regime 1. The growth of the GaN pyramids was stopped in (c) regime 1 and (d) regime 2.

two pyramid growth regimes were observed. The existence of two regimes can be explained by the anisotropic growth rates of the pyramids. In our experiments, the [0001] growth rate is much larger than the ⟨11̅01⟩ growth rate. At the initial stages of the growth, there are plenty of (0001) facet areas exposed to the vapor phase due to the truncated profile of the pyramids. At this stage, the growth occurs simultaneously in the [0001] and ⟨11̅01⟩ directions. Accordingly, the volume increase per time unit is relatively large (regime 1). As growth proceeds, the (0001) facet areas diminish; thus, the pyramids can only expand in the slowly growing ⟨11̅00⟩ and ⟨11̅01⟩ directions, resulting in a relatively low volume increase per time unit (regime 2). The transition time (denoted as ttrans in Figure 3a) between the growth regimes is mainly dependent on the SiN aperture radius due to the cubic relation between the pyramid volume and SiN aperture radius. The growth in regime 1 was further analyzed by five additional growth experiments with varying TMGa flows (T = 1035 °C, NH3 flow rate = 6 slm, growth time = 45 min). The total volume of the pyramids versus the TMGa flow is shown in Figure 3b. To avoid coalescence of the pyramids, the 4.7 sccm TMGa data point in Figure 3b was linearly extrapolated from a run with the growth time set to 30 min. In Figure 3b, the pyramid volume increases linearly with the TMGa flow; hence, we make the important conclusion that the growth in regime 1 5493

dx.doi.org/10.1021/cg301064p | Cryst. Growth Des. 2012, 12, 5491−5496

Crystal Growth & Design

Article

At reduced growth temperatures, on the other hand, the adsorption rate on the (11̅01) surface is much larger than the desorption rate; thus, the pyramid expands in the ⟨11̅01⟩ direction. Observe that the growth rate in the ⟨11̅01⟩ direction also contributes to a lateral component, leading to lateral overgrowth of the SiN mask. However, the ⟨11̅01⟩ growth rate is generally low in comparison to the growth rates in the [0001] and ⟨11̅02⟩ directions, independent of the growth temperature and TMGa and NH3 flow rates. The results are explained by a hydrogen passivation layer stabilizing the (11̅01) surface. Northrup and co-workers41 performed DFT calculations on a nitrogen-terminated (0001̅) surface and concluded that hydrogen exhibits a strong affinity to the (0001)̅ surface. During MOCVD growth, the (0001̅) surface will be covered by 0.75 monolayers of hydrogen.42 The result explains the lower growth rate typically obtained on the (0001̅) surface compared to the (0001) surface in MOCVD growth experiments. Feenstra et al.43 and Akiyama et al.44 have independently also proposed that a similar hydrogen passivation layer forms on the (11̅01) surface where the topmost nitrogen atoms are terminated by hydrogen atoms with a large energy gain of roughly 4 eV per N−H bond.45 Formation of the {11̅02} Facet. As shown in Figure 2a,b, {11̅02} appeared in reduced NH3 flows. To gain a further understanding of the {11̅02} facet formation, additional samples with varying NH3 and TMGa flow rates were grown in the mass-transport-limited regime. The V/III ratio used here is defined by the input flow ratio between NH3 and TMGa, although the effective local V/III ratio on the surface depends on, for example, the mask geometry and the presence of various facets. To avoid complications in the definition of the V/III ratio, we limit the discussion to the mass-transported regime, where (1) all gallium precursors are consumed and (2) the absolute areas of the {11̅02} and {11̅00} surfaces are relatively small compared with the (0001) and {11̅01} surfaces. By doing so, we believe that the V/III ratio used here has a similar meaning as the conventionally used V/III ratio in planar epitaxy. The results from some selected samples grown at 1000 °C are summarized in Table 1. Bird's-eye-view SEM images of

Figure 4. (a−e) Bird's-eye view of samples A−E of Table 1. (f) A planar-view SEM of sample A, where the darker areas near the (0001) surface are the {11̅02} surfaces. In (a) and (f), one of the six {11̅02} facets is marked with a cyan line.

also had a slightly lower V/III ratio than sample B, which clearly indicates that the {110̅ 2} facet formation does not have an obvious relation to the V/III ratio. Instead, the absolute flow rate (or partial pressure) of NH3 during the growth is the critical parameter. Strong similarities were found on samples grown at 1035 and 1060 °C. However, elevated growth temperature required a slightly higher NH3 flow rate to remove the {11̅02} facets. An approximate NH3 flow of 2 and 4 slm was required at 1035 and 1060 °C, respectively. Stabilization of the {110̅ 2} facets at reduced NH3 partial pressures and growth temperatures was also recently reported by Choi et al.21. No differences were observed between pyramids emerging from 3.0 and 6.0 μm apertures in the SiN mask. Since the slowest-growing facets39,40 remain visible post growth, it is relevant to determine if the {11̅02} facets were removed by either a {11̅01} growth rate increase or a {11̅02} growth rate decrease during the NH3-rich conditions. Simply adjusting the NH3 flow rate and measuring the average base length of the pyramids while keeping the process pressure constant leads to incorrect conclusions about the {11̅01} facet growth rate as the TMGa precursors become diluted at elevated NH3 flows. In Figure 5, the average pyramid {110̅ 1} facet

Table 1. {11̅02} Facet Visibility of Hexagonal GaN Pyramids Grown at 1000 °C under Various TMGa and NH3 Flow Rates sample

NH3 flow rate (slm)

TMGa flow rate (sccm)

V/III ratio

{110̅ 2} facet visible

A B C D E

0.8 2.0 1.25 6.0 2.0

1.35 3.35 3.35 3.35 4.70

600 600 375 1800 425

yes no yes no no

Figure 5. TMGa partial pressure normalized bottom facet pyramid length versus the NH3 flow rate.

the samples are also shown in Figure 4a−e. In Figure 4f, a planar-view SEM image of a pyramid from sample A is shown. A NH3 flow rate of 0.8 slm (V/III ratio = 600) and a TMGa flow rate of 1.35 sccm resulted in pyramids with {110̅ 2} facets near the apex. However, no {11̅02} facets were observed on sample B, which was grown with an identical V/III ratio, but with slightly higher NH3 and TMGa flow rates of 2 slm and 3.35 sccm, respectively. When the TMGa flow rate was locked at 3.35 sccm while varying the NH3 flow rate (samples B−D) {11̅02}, facets only appeared on sample C, which was grown with the lowest NH3 flow rate of 1.25 slm. Samples C and E

length normalized to the partial pressure of TMGa versus the NH3 flow rate while keeping all other growth parameters fixed (T = 1035 °C, TMGa flow rate = 4.7 sccm, process pressure = 50 mbar) is shown. The TMGa partial pressure was calculated using Daltons law, which states that the total pressure exerted by the mixture of gases equals the sum of the partial pressures of individual gases. The near independent relation between the NH3 flow and the compensated average pyramid facet length in Figure 5 suggests that the ⟨11̅01⟩ growth rate is independent of the NH3 flow. Thus, the appearance of the {11̅02} facets at low 5494

dx.doi.org/10.1021/cg301064p | Cryst. Growth Des. 2012, 12, 5491−5496

Crystal Growth & Design

Article

NH3 flow must accordingly be caused by a ⟨110̅ 2⟩ growth rate reduction. To our knowledge, there are no DFT studies published regarding the reconstruction of the (11̅02) surface; thus, quantitative comparisons is difficult. However, our experiments show that the {11̅02} facets are more stable than the {11̅01} facets at reduced NH3 flows and growth temperatures. We, therefore, assume that the {11̅02} surfaces are stabilized in relatively low partial pressure reactants that originate from decomposed NH3, although this must be confirmed by proper DFT calculations.

truncated pyramids and needle-like apexes are formed at short and long growth times, respectively, since R0001 ≫ R11̅01 = 0. Regions III and IV (low growth temperature, and low/high NH3 flow rates) show vertical {11̅00} facets at prolonged growth times as the growth enters a kinetically limited regime. However, region III shows {11̅02} facets, whereas region IV does not. Pyramids that show visible {11̅02} facets tend to be rather irregular, particularly near the apex, as shown in Figure 1b. The same type of irregularity (but located at the bottom of the pyramid) applies to the pyramids that show {11̅00} facets, as shown in Figures 2a,c and 3d. Therefore, region II is the easiest region to obtain highly uniform pyramids. Observe that the (0001) truncation does not disappear in regions III and IV when the growth time is prolonged due to the self-limited facet mechanism.46−48



SUMMARY AND CONCLUSIONS In summary, the presence of the {11̅01}, {11̅02}, and {11̅00} facets on gallium polar (0001)-oriented hexagonal GaN pyramids grown by hot-wall MOCVD was investigated. The results demonstrated that it is possible to control the morphology of the pyramids by adjusting process parameters, such as the growth temperature and NH3 and TMGa flow rates. In Figure 6a, the summarized growth characteristics of hexagonal GaN pyramids are shown. The viewpoint of the



AUTHOR INFORMATION

Corresponding Author

*Phone: +46 13 282799. Fax: +46 70 3076500. E-mail: anlun@ ifm.liu.se. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by the NANO-N consortium funded by the Swedish Foundation for Strategic Research.



REFERENCES

(1) Ejder, E. Phys. Status Solidi A 1971, 6, 445−448. (2) Scholz, F.; Wunderer, T.; Neubert, B.; Feneberg, M.; Thonke, K. MRS Bull. 2009, 34, 328−333. (3) Kim, H. M.; Cho, Y. H.; Lee, H.; Kim, S. I.; Ryu, S. R.; Kim, D. Y.; Kang, T. W.; Chung, K. S. Nano Lett. 2004, 4, 1059−1062. (4) Choi, J. H.; Zoulkarneev, A.; Kim, S. I.; Baik, C. W.; Yang, M. H.; Park, S. S.; Suh, H.; Kim, U. J.; Son, H. B.; Lee, J. S.; Kim, M.; Kim, J. K.; Kim, K. Nat. Photonics 2011, 5, 763−769. (5) Ryou, J. H.; Yoder, P. D.; Liu, J.; Lochnerm, Z.; Hyunsoo, K.; Choi, S.; Kim, H. J.; Dupuis, R. D. IEEE J. Sel. Top. Quantum Electron. 2009, 15, 1080−1091. (6) Bernardini, F.; Fiorentini, V. Appl. Phys. Lett. 2011, 80, 4145. (7) Takeuchi, T.; Sota, S.; Katsuragawa, M.; Komori, M.; Takeuchi, H.; Amano, H.; Akasaki, I. Jpn. J. Appl. Phys. 1997, 36, L382. (8) Kozawa, T.; Suzuki, M.; Taga, Y.; Gotoh, Y.; Ishikawa, J. J. Vac. Sci. Technol., B 1998, 16, 833. (9) Chattopadhyay, S.; Chen, L. I.; Chen, K. H. Crit. Rev. Solid State Mater. Sci. 2006, 31, 15−53. (10) Hsu, C. W.; Lundskog, A.; Karlsson, K. F.; Forsberg, U.; Janzén, E.; Holtz, P. O. Nano Lett. 2011, 11, 2415−2418. (11) Edwards, P. R.; Martin, R. W.; Watson, I. M.; Liu, C.; Taylor, R. A.; Rice, J. H.; Na, J. H.; Robinson, J. W.; Smith, J. D. Appl. Phys. Lett. 2004, 85, 4281. (12) Lundskog, A.; Palisaitis, J.; Hsu, C. W.; Eriksson, M.; Karlsson, K. F.; Hultman, L.; Persson, P. O. Å.; Forsberg, U.; Holtz, P. O.; Janzén, E. Nanotechnology 2012, 23, 305708. (13) Hiramatsu, K. J. Phys.: Condens. Matter 2001, 13, 6961. (14) Lundin, W. V.; Zavarin, E. E.; Rozhavskaya, M. M.; Troshkov, S. I.; Tsatsulnikov, A. F. Tech. Phys. Lett. 2011, 37, 735−738. (15) Akasaka, T.; Kobayashi, Y.; Ando, S.; Kobayashi, N. Appl. Phys. Lett. 1997, 71, 2196. (16) Sun, Q.; Yerino, C. D.; Leung, B.; Han, J.; Coltrin, M. E. J. Appl. Phys. 2011, 110, 053517. (17) Akasaka, T.; Kobayashi, Y.; Ando, S.; Kobayashi, N.; Kumagai, M. J. Cryst. Growth 1998, 189−190, 72−77. (18) Fang, Z. L.; Lin, Y. X.; Kang, J. Y. Appl. Phys. Lett. 2011, 98, 061911.

Figure 6. (a) Cross-sectional sketch showing the MOCVD growth characteristics of gallium polar (0001)-oriented GaN pyramids. (b) Process parameter roadmap for various GaN geometries. tshort and tlong represent short or long growth times, respectively.

pyramid in Figure 6a is the same as that shown in Figure 1a. In Figure 6a, the growth rates of the respective Miller index, denoted as (hkil) facets, are denoted as Rhkil. Facets with small Rhkil are visible post growth. Observe that R11̅00 automatically reduces significantly when the growth proceeds to the kinetically limited regime. On the basis of this information, a process parameter roadmap for different 3D structures is constructed and shown in Figure 6b. At short growth times (tshort < ttrans, where ttrans denotes the transition time between the mass transport and kinetic-limited growth regimes), region I (high growth temperature and low NH3 flow rates) results in pyramids with visible {110̅ 2} facets near the apex. The {110̅ 2} facets are terminated when the growth time is prolonged (tlong > ttrans) since R110̅ 1 = 0, while R110̅ 2 > 0. In region II (high growth temperature and high NH3 flow rates), (0001)5495

dx.doi.org/10.1021/cg301064p | Cryst. Growth Des. 2012, 12, 5491−5496

Crystal Growth & Design

Article

(19) Li, S. F.; Fuendling, S.; Wang, X.; Merzsch, S.; Al-Suleiman, M. A. M.; Wei, J. D.; Wehmann, H.-H.; Waag, A.; Bergbauer, W.; Strassburg, M. Cryst. Growth Des. 2011, 11, 1573−1577. (20) Chen, X. J.; Gayral, B.; Sam-Giao, D.; Bougerol, C.; Durand, C.; Eymery, J. Appl. Phys. Lett. 2011, 99, 251910. (21) Choi, K.; Arita, M.; Arakawa, Y. J. Cryst. Growth 2012, 357, 58− 61. (22) Hersee, S. D.; Sun, X. Y.; Wang, X. Nano Lett. 2006, 6, 1808− 1811. (23) Tang, T. Y.; Shiao, W. Y.; Lin, C. H.; Shen, K. C.; Huang, J. J.; Ting, S. Y.; Liu, T. C.; Yang, C. C.; Yao, C. L.; Yeh, J. H.; Hsu, T. C.; Chen, W. C.; Hsu, H. C.; Chen, L. I. J. Appl. Phys. 2009, 105, 023501. (24) Bergbauer, W.; Strassburg, M.; Kölper, Ch.; Linder, N.; Roder, C.; Lähnemann, J.; Trampert, A.; Fündling, S.; Li, S. F.; Wehmann, H. H.; Waag, A. J. Cryst. Growth 2011, 315, 164−167. (25) Li, S. F.; Fündling, S.; Wang, X.; Erenburg, M.; Al-Suleiman, M. A. M.; Wei, J. D.; Bergbauer, W.; Strassburg, M.; Wehmann, H. H.; Waag, A. Phys. Status Solidi C 2011, 7−8, 2318−2320. (26) W Bergbauer, W.; Strassburg, M.; Kolper, C.; Linder, N.; Roder, C.; Lahnemann, J.; Trampert, A.; Fundling, S.; Li, S. F.; Wehmann, H. H.; Waag, A. Nanotechnology 2010, 21, 305201. (27) Chen, X. J.; Hwang, J. S.; Perillat-Merceroz, G.; Landis, S.; Martin, B.; Dang, D. L. S.; Eymery, J.; Durand, C. J. Cryst. Growth 2011, 322, 15−22. (28) Hiramatsu, K; Nishiyama, K.; Motogaito, A.; Miyake, H.; Iyechika, Y.; Maeda, T. Phys. Status Solidi A 1999, 176, 535−543. (29) Akasaka, T.; Kobayashi, Y.; Ando, S.; Kobayashi, N.; Kumagai, M. J. Cryst. Growth 1998, 189−190, 72−77. (30) Forsberg, U.; Lundskog, A.; Kakanakova-Georgieva, A.; Ciechonski, R.; Janzén, E. J. Cryst. Growth 2009, 311, 3007−3010. (31) Kakanakova-Georgieva, A.; Ciechonski, R.; Forsberg, U.; Lundskog, A.; Janzén, E. Cryst. Growth Des. 2009, 9, 880−884. (32) Creighton, J. R.; Wang, G. T.; Breiland, W. G.; Coltrin, M. E. J. Cryst. Growth 2004, 261, 204−213. (33) Dauelsberg, M.; Protzmann, C. M. H.; Boyd, A. R.; Thrush, E. J.; Käppeler, J.; Heuken, M.; Talalaev, R. A.; Yakovlev, E. V.; Kondratyev, A. V. J. Cryst. Growth 2007, 298, 418−424. (34) Lee, C. D.; Feenstra, R. M.; Northrup, J. E.; Lymperakis, L.; Neugebauer, J. Appl. Phys. Lett. 2003, 82, 1793. (35) Northrup, J. E.; Neugebauer, J. Phys. Rev. B 1996, 53, R10477− R10480. (36) Segev, D.; Van de Walle, C. G. Europhys. Lett. 2006, 76, 305. (37) Waltereit, P.; Brandt, O.; Ramsteiner, M.; Trampert, A.; Grahn, H. T.; Menniger, J.; Reiche, M.; Uecker, R.; Reiche, P.; Ploog, K. H. Phys. Status Solidi A 2000, 180, 133−138. (38) Bertelli, M.; Loptien, P.; Wenderoth, M.; Rizzi, A.; Ulbrich, R. G.; Righi, M. C.; Ferretti, A.; Martin-Samos, L.; Bertoni, C. M.; Catellani, A. Phys. Rev. B 2009, 80, 115324. (39) Du, D.; Srolovitz, D. J.; Coltrin, M. E.; Mitchell, C. C. Phys. Rev. Lett. 2005, 95, 155503. (40) Sun, Q.; Yerino, C. D.; Ko, T. S.; Cho, Y. S.; Lee, U. H.; Han, J.; Coltrin, M. E. J. Appl. Phys. 2008, 104, 093523. (41) Northrup, J. E.; Neugebauer, J. Appl. Phys. Lett. 2004, 85, 3429. (42) Sung, M. M.; Ahn, J.; Bykov, V.; Rabalais, J. W.; Koleske, D. D.; Wickenden, A. E. Phys. Rev. B 1996, 54, 652. (43) Feenstra, R. M.; Dong, Y.; Lee, C. D.; Northrup, J. E. J. Vac. Sci. Technol., B 2005, 23, 1174. (44) Akiyama, T.; Ammi, D.; Nakamura, K.; Ito, T. Phys. Rev. B 2010, 81, 245317. (45) Akiyama, T.; Yamashita, T.; Nakamura, K.; Ito, T. J. Cryst. Growth 2011, 318, 79−83. (46) Seigo, A.; Honda, T.; Kobayashi, N. Jpn. J. Appl. Phys. 1993, 32, L104−L106. (47) Kumakura, K.; Nakakoshi, K.; Motohisa, M.; Fukui, T.; Hasegawa, H. Jpn. J. Appl. Phys. 1995, 34, 4387−4389. (48) Kitamura, S.; Hiramatsu, K.; Sawaki, N. Jpn. J. Appl. Phys. 1995, 34, L1184−L1186.

5496

dx.doi.org/10.1021/cg301064p | Cryst. Growth Des. 2012, 12, 5491−5496