Hydrogen etching of GaN and its application to produce free-standing GaN thick films

(1)

Hydrogen etching of GaN and its application to produce

free-standing GaN thick ﬁlms

Yen-Hsien Yeh

n

, Kuei-Ming Chen, Yin-Hao Wu, Ying-Chia Hsu, Tzu-Yi Yu, Wei-I Lee

Department of Electrophysics, National Chiao Tung University, Hsinchu 30010, Taiwan

a r t i c l e

i n f o

Article history: Received 2 May 2011 Received in revised form 15 July 2011

Accepted 16 August 2011 Communicated by R. Bhat Available online 23 August 2011 Keywords:

A1. Hydrogen etching A1. Surface processes A1. Surface structure

A3. Hydride vapor phase epitaxy B1. Nitrides

B2. Semiconducting III–V materials

a b s t r a c t

This work investigates the morphology of GaN etched in hydrogen (H2) at different temperatures, the

activation energies of the rate-limiting steps of H2etching, and the overgrowth on a H2-etched GaN

template. The surfaces of GaN have different proﬁles after being etched in H2; they resemble a plane

decorated with columns and mooring posts in a low-temperature etching condition, and with deep cavities in a high-temperature etching condition. The etched proﬁles show that H2 etching has

controllable etching directions: vertical and lateral. In a low-temperature condition, H2etching has

both vertical and lateral etching directions; however, in a high-temperature condition, it has only the vertical etching direction. The activation energies of the rate-limiting steps under etching pressures of 100 and 700 Torr are estimated to be 3.22 and 3.77 eV, respectively. A thick GaN layer has been grown on a H2-etched GaN template, and it has self-separated from the underlying sapphire substrate.

1. Introduction

GaN is a major wide band gap semiconductor used in fabricat-ing green to ultraviolet optoelectronic devices. Various processes have been developed to study the properties of GaN; among them, etching is one of the most important steps. For example, the etching process is necessary for making a patterned structure on the surface of GaN to reduce stress and dislocation density during the manufacturing of devices or large-area free-standing GaN substrates [1–5]. Unfortunately, conventional wet etching is difﬁcult to be used due to the high chemical stability of GaN. The more effective methods are dry etching techniques such as reactive ion etching (RIE) [6] and inductively coupled plasma (ICP)[7]. These dry etching processes produce anisotropic etch-ing, i.e., GaN is etched in the vertical direction. However, the obtainable patterned structures are limited because of the single direction etching. It is valuable to develop an etching technique having both vertical and lateral etching directions. Additionally, an in situ maskless etching will also help to simplify the manufacturing processes of GaN substrates and devices.

In our previous work, the morphology of GaN etched in hydrogen (H2) under different pressures had been studied [8].

A model was developed to explain the mechanism of H2etching

under different pressures at 1050 1C. We found that H2 etching

begins at some weak areas such as dislocation sites, and all vicinity of dislocation sites has been etched to form cavities. In this work, we investigate the morphology of GaN etched in H2at

different temperatures, the activation energies of the rate-limit-ing steps of H2etching, and the overgrowth on a H2-etched GaN

template.

2. Experiment

Samples used in the experiments were MOCVD-grown c-plane GaN template layers on sapphire, and the thickness of the GaN layers was about 2

m

m. The ﬁrst experiment was temperature-varying etching. After ultrasonic cleaning, a sample was loaded into a home-made horizontal HVPE reactor, and then the reactor was heated to a desired temperature under a pressure of 700 Torr. In addition to nitrogen (N2), ammonia (NH3) was also introduced

into the reactor to avoid GaN decomposition during heating. After the temperature reached, NH3 was shut off, and H2 was

intro-duced to begin etching. The ﬂow rates of H2and N2were 1 and

4.9 slm, respectively; the etching time was 10 min. Six samples labeled A, B, C, D, E, and F were etched at the temperatures of 1050, 1060, 1070, 1080, 1090, and 1100 1C, respectively. The surface morphologies of the etched samples were observed through a high-resolution ﬁeld emission scanning electron micro-scope (FE-SEM, Hitachi s-4700i). The second experiment is the calculation of the activation energy of the reaction limiting GaN decomposition. Samples were weighed to within 0.1 mg using an Contents lists available atSciVerse ScienceDirect

journal homepage:www.elsevier.com/locate/jcrysgro

Journal of Crystal Growth

n

Corresponding author.

E-mail address: [email protected] (Y.-H. Yeh).

(2)

analytical balance before etching, and then they performed H2

etching at different temperatures by HVPE. After H2etching, each

sample was reweighed to determine the mass loss. The etching rate was calculated from the mass loss divided by the etching area and time. Two activation energies at etching pressures of 100 and 700 Torr have been estimated. The third experiment was the overgrowth on a H2 etched sample by HVPE. The H2-etching

condition was a two-step etching that was described in our previous study[8]. After etching, a 320

m

m-thick GaN layer was

sequentially grown on the sample at 1050 1C under the high pressure of 700 Torr for 120 min, and the V/III ratio was 40.

3. Result and discussion

Fig. 1 shows SEM images of the six samples etched at the temperatures of 1050–1100 1C. It can be found that the etching temperature has remarkable inﬂuence on the morphologies of GaN surfaces. The surface resembles a plane decorated with mooring posts or columns at the low etching temperature of 1050 1C, as shown in Fig. 1(a), and there are a lot of undercut shapes revealing the lateral etching ability of H2 etching. As a

higher etching temperature is used, the surface becomes ﬂatter. Deep cavities begin to appear at the high temperature of 1090 1C, and the density of them becomes larger at the highest tempera-ture of 1100 1C.

We can compare the effects of temperature with those of pressure. By comparing our previous study [8], it can be found that the effects of temperature are similar to those of pressure but with an opposite tendency, i.e., in the pressure-varying experi-ment, the etched surfaces are decorated with bollard-like posts at high pressures and with deep cavities at low pressures.

The mechanism of the formation of the morphologies at these temperatures can be explained by a model, which is similar to the one we proposed in our previous study. It has been reported that H2enhances GaN decomposition; H and N atoms combine to form

NH3[9–13]. As H mostly reacts with N, we presume that during

H2etching, a facet formed with N atoms (N-terminated) will be

unstable whereas a facet formed with Ga atoms (Ga-terminated) will be stable. The as-grown surface of a MOCVD-grown c-plane GaN ﬁlm is Ga-polarity (Ga-face) and the reverse side is N-polarity (N-face); the inclined {10–11}, {11–22} facets are mainly N-termi-nated; however, the nonpolar (a-plane and m-plane) facets contain equal number of Ga and N atoms. Therefore, we suppose that during H2etching, the relative stabilities of the facets follow the order:

Ga-face4nonpolar face4inclined face4N-face. At 1050 1C under 700 Torr, the etching mechanism is proposed to be as follows. As H2arrives at the surface, it is difﬁcult to decompose most of GaN on

the surface due to the high stability of Ga-polarity. Thus, etching occurs at some weak areas such as dislocation sites to form a vertical cavity, as shown in Fig. 2(b) (the dotted line indicates a dislocation). As H2etch GaN downwardly, the products, NH3and N2,

Fig. 1. SEM images of the samples etched at different temperatures in plan views (insets), tilted views (left column) and cross-section views (right column): (a) 1050, (b) 1060, (c) 1070, (d) 1080, (e) 1090, and (f) 1100 1C. All the scale bars are 1mm.

Fig. 2. Stages of H2etching at low temperature: (a) unetched template, the top

surface is Ga-face and the dotted line indicates the dislocation, (b) H2etching at

the dislocation site, (c) lateral etching at the bottom of the cavity, and then N-face facet is exposed, (d) etching in the upward direction due to the instability of N-face, and (e) the resulting etching proﬁle.

(3)

are produced; they dissipate slowly and then occupy the bottom of the cavity at the low temperature of 1050 1C under 700 Torr. H2

cannot ﬂow into the bottom of the cavity due to the occupancy of NH3and N2; therefore, the bottom of the cavity becomes a high V/III

ratio environment. It is proposed from the previous researches that the inclined faces are stable in a condition of high V/III ratio[3,5,14]. Hence, we infer that the nonpolar sidewalls become unstable in such an environment. As a result, the lateral etching begins at the bottom of the cavity, as shown inFig. 2(c). After the sidewalls are etched for some distance, the N-face facet is exposed, and then it is etched due to its instability; therefore, etching begins in the upward direction, as shown in Fig. 2(d). The resulting etching proﬁle is shown inFig. 2(e), and it has an undercut shape. A lot of cavities with this undercut proﬁle expand and merge to make the morphol-ogy resemble a surface decorated with columns or mooring posts, as shown inFig. 1(a).

The formation of the cavities at a low etching pressure in our previous study is proposed as follows. As the pressure become lower, NH3and N2dissipate more rapidly, and the bottom of the

cavity becomes a H2environment; the sidewalls become stable,

and deep cavities appear. We suggest the formation of the cavities at high etching temperatures is also due to a similar mechanism. At high temperature, the dissipation rate of NH3and N2increases,

so the bottom of the cavity becomes a H2environment; therefore,

the sidewalls become stable, and deep cavities appear as a result. To estimate the activation energies of the rate-limiting steps of H2 etching under pressures of 100 and 700 Torr, we performed

experiments to obtain the etching rates at different temperatures. Fig. 3shows the Arrhenius plot of the results, and we estimate the activation energies at the etching pressures of 100 and 700 Torr to be about 3.22 and 3.77 eV, and the pre-exponential factors of them are 3.34 1027 and 2.62 1029molecules/cm2s, respec-tively. These values indicate that the rate-limiting steps for H2

etching at 1030–1055 1C under pressures of 100 and 700 Torr are the same, i.e., formation and the desorption of N2[9]. However,

the different activation energies may have some implications. In Ref.[9], the activation energy of the etching condition under pure H2at 40 and 76 Torr is 3.4 eV, and that under pure N2at 76 and

150 Torr is 3.62 eV; the activation energy under pure N2is larger.

In our results, the activation energy under the high pressure of 700 Torr is larger than that of 100 Torr (the same H2/N2ratio), so

we infer this result agrees with our hypothetical model that the bottom of the cavities is relatively a high V/III ration environment (without H2) at high pressure and a H2 environment at low

pressure [8]. The sidewalls are unstable at high pressure, and then the lateral etching begins as the mechanism discussed in Ref.[8].

To demonstrate the application of H2 etching, we performed

an overgrowth experiment. A sample was ﬁrst etched using two-step etching to produce porous caves structure that is shown in

our previous study[8], and then a 320

m

m-thick GaN layer was grown on the sample. The overgrown GaN thick layer self-separated from the underlying sapphire substrate; however, it also cracked. The growth area is 10 mm 10 mm, and the largest piece is shown inFig. 4. It is supposed that the porous caves make the interface of the GaN layer and the sapphire substrate become weak, so the thick ﬁlm separates from the substrate due to the stress that rises during cooling. Nevertheless, the etching and growth parameters need to be optimized to avoid cracks. This experiment demonstrates one possible application of H2etching,

and we believe that H2etching is also helpful for the reduction of

threading dislocation density during the overgrowth process; however, the dislocation density of the self-separated sample does not reduce greatly. Fig. 5 shows the AFM image of the surface of the free-standing GaN thick ﬁlm after an EDP experi-ment. The dislocation density is estimated to be about 2 107_cm 2_{, and this value is just a little lower than that of a}

thick ﬁlm, which is grown without H2etching. We consider that

the growth condition needs to be optimized to reduce the dislocation density.

4. Summary

In conclusion, different morphologies appear on the GaN surface etched in H2 at different temperatures under a high

pressure of 700 Torr. It is considered that H2 etching begins in

certain weak areas such as dislocation sites. In addition to the vertical downward direction, H2 etching also has lateral and

Fig. 3. The Arrhenius plot of the etching rates at pressures of 100 and 700 Torr.

Fig. 4. The free-standing GaN thick ﬁlm.

Fig. 5. AFM image of the self-separated sample after an EDP experiment. Y.-H. Yeh et al. / Journal of Crystal Growth 333 (2011) 16–19

(4)

upward directions at the low temperature of 1050 1C, and these etching directions make the surface proﬁle resemble a plane clustered with columns or mooring posts. On the other hand, H2

etching merely has the vertical direction at high temperatures that make the surface decorate with cavities. The activation energies of the rate-limiting steps at pressures of 100 and 700 Torr are estimated to be 3.22 and 3.77 eV, respectively. A 320

m

m-thick GaN layer has been grown on a H2-etched GaN

template, and it has self-separated from the underlying sapphire substrate.

Acknowledgment

This work was ﬁnancially supported by National Science Council of Taiwan (Contract no. NSC 98-2221-E-009–026). References

[1] A. Sakai, H. Sunakawa, A. Usui, Applied Physics Letters 71 (1997) 2259. [2] O.-H. Nam, M.D. Bremser, T.S. Zheleva, R.F. Davis, Applied Physics Letters 71

(1997) 2638.

[3] K. Hiramatsu, K. Nishiyama, A. Motogaito, H. Miyake, Y. Iyechika, T. Maeda, Physica Status Solidi A 176 (1999) 535.

[4] I. Kidoguchi, A. lshibashi, G. Sugahara, Y. Ban, Applied Physics Letters 76 (2000) 3768.

[5] K. Hiramatsu, K. Nishiyama, M. Onishi, H. Mizutani, M. Narukawa, A. Motogaito, H. Miyake, Y. Iyechika, T. Maeda, Journal of Crystal Growth 221 (2000) 316.

[6] M.E. Lin, Z.F. Fan, Z. Ma, L.H. Allen, H. Morkoc-, Applied Physics Letters 64 (1994) 887.

[7] R.J. Shul, G.B. McClellan, S.A. Casalnuovo, D.J. Rieger, S.J. Pearton, C. Constantine, C. Barratt, Applied Physics Letters 69 (1996) 1119. [8] Y.H. Yeh, K.M. Chen, Y.H. Wu, Y.C. Hsu, W.I. Lee, Journal of Crystal Growth 314

(2011) 9.

[9] D.D. Koleske, A.E. Wickenden, R.L. Henry, J.C. Culbertson, M.E. Twigg, Journal of Crystal Growth 223 (2001) 466.

[10] T.H. Myers, B.L. VanMil, J.J. Holbert, C.Y. Peng, C.D. Stinespring, J. Alam, J.A. Freitas Jr., V.A. Dmitriev, A. Pechnikov, Y. Shapovalova, V. Ivantsov, Journal of Crystal Growth 246 (2002) 244.

[11] N. Kobayahi, Y. Kobayahi, Applied Surface Science 159–160 (2000) 398. [12] M.A. Mastro, O.M. Kryliouk, M.D. Reed, T.J. Anderson, A.J. Shapiro, Physica

Status Solidi A 188 (2001) 467.

[13] M. Mayumi, F. Satoh, Y. Kumagai, K. Takemoto, A. Koukito, Japanese Journal of Applied Physics 39 (Part 2) (2000) L707.

[14] Hai-Ping Liua, Jenq-Dar Tsayb, Wen-Yueh Liub, Yih-Der Guob, Jung Tsung Hsub, In-Gann Chen, Journal of Crystal Growth 260 (2004) 79.