2&L __ __ Fi!z applied surface science

ELSEVIER Applied Surface Science 82/83 (1994) 250-256

X-ray photoelectron spectroscopic and atomic force microscopic study of GaAs etching with a HCl solution

Zhen Song ‘, Satoshi Shogen, Masahiro Kawasaki, Ikuo Suemune Institute for Electronic Science, Hokkaido Uniuersity, Sapporo 060, Japan

Received 10 May 1994; accepted for publication 30 June 1994

Abstract

GaAs(100) substrates covered with native oxides have been etched with a HCl solution for lo-20 min and subsequently rinsed with water for several seconds. The surface of the GaAs substrate thus prepared has been examined with atomic force microscopy. The surface flatness is improved by this wet etching and the surface undulation remains within a f 1 monolayer fluctuation over a 1 pm X 1 pm surface area. X-ray photoelectron spectroscopy has been used to examine the chemical composition of the GaAs(100) surfaces after the HCl/H,O wet etching sequences. UV photoirradiation of the GaAs surface treated with the HCl solution has been performed to characterize the chlorides on the GaAs surface.

1. Introduction

Etching procedures of GaAs substrates have been

studied to improve the surface flatness with the

tendency toward miniaturization of VLSI chips. Pre- vious work, aimed at understanding the microchem- istry of halogen-GaAs systems, involved the use of XPS, LEED, AES, and SIMS to characterize the surface species. For example, Lu et al. [l] studied the GaAs(100) cleaning procedure with XPS. The sam- ple was first etched by solutions of H,SO, : H,O, or H,PO, : H,O,, and then rinsed in HCl solution. The HCl rinse removed the remaining oxides. We report here the study of surface cleaning at different stages of HCl etching and H,O rinsing of the GaAs(100)

samples with X-ray photoelectron spectroscopy (XPS) and atomic force microscopy (AFM). The AFM measurement which is complementary to con- ventional surface spectroscopies provides informa- tion on the flatness of the semiconductors. The wave- length effect of ultraviolet photoirradiation on the GaAs surface treated with the HCl solution has also been investigated to characterize the chlorides on the surface.

2. Experiment

’ On leave from Dalian Institute of Chemical Physics, Chinese

Academy of Sciences, Dalian 116023, People’s Republic of China.

An undoped GaAs(100) surface with a resistivity of 2 X lo7 fi. cm was used. The surface was origi- nally covered with native oxides or room-tempera- ture oxidized (RTO). The samples were etched in the HCl solution (36%), rinsed in water, and then di- rectly dried in flowing argon. AFM measurements were performed by a Digital Instrument NanoScope

0169-4332/94/$07.00 0 1994 Elsevier Science B.V. All rights reserved

SSDI 0169.4332(94)00226-6

Z. Song et al. /Applied Surface Science 82 / 83 (I 994) 250-256 251

III because it is applicable to highly resistive materi- tors [3]. XPS signal shapes were simulated with a als and has a high vertical resolution up to 2 : 1 mixed function of a Gaussian and a Lorentzian monoatomic steps on a semiconductor surface [2]. with a full width at half maximum of 1.2-1.6 eV.

In order to characterize the chemical species on the GaAs surface, angle-resolved XPS analysis was carried out in a Vacuum Generator ADES-400. The samples were evacuated in the XPS chamber for 2-4 h before analysis. The pressure in the chamber was typically 5 X lo-’ Torr. The polar emission angle 0 was defined as the take-off angle of photoelectron emission from a surface. The population of each element was obtained from the XPS signal area divided by the corresponding atomic sensitivity fac-

A pulse laser was used to irradiate the GaAs surface that was etched only with the HCl solution and was evacuated in the XPS chamber: an ArF excimer laser at 193 nm (Lambda Physics, EMG-101, 3 mJ/pulse, 10 Hz) or a frequency-tripled YAG laser at 355 nm (Quanta Ray, DCR-11, 8 or 16 mJ/pulse, 10 Hz). The laser power was measured just before the quartz window of the XPS vacuum chamber. The GaAs sample was irradiated for S-30 min on a surface area of 0.5 cm’.

Fig. 1. AFM image of a GaAs surface Lower panel: after cleaning with organic solvents. Upper panel: after cleaning with organic solvents,

HCl etching (20 min) and H,O rinse (10 s). X- and Y-axes are in units of nm. The Z-axis is 5 times enlarged.

252 Z. Song el al. /Applied Surface Science 82 /R3 (1994) 250-256

3. Results

3.1. AFM observation

AFM characterizes atomic-level flatness of the GaAs surfaces before and after etching with the HCl solution and subsequent rinse with water. The lower pane1 of Fig. 1 shows the AFM image of a RTO GaAs surface in water after cleaning with organic solvents, In order to minimize extrinsic acoustic noise coming from the environment, the AFM image was measured in water. The surface had an undula- tion of l-2 nm. This GaAs surface was flattened when treated by sequentially (a) cleaning in organic solvents. (b) etching in a solution of 4H,SO,:

H,O, : H,O, (c) immersing in HCl solution, and (d) rinsing in H,O for several seconds. The AFM image of the upper pane1 of Fig. 1 shows a drastic improve- ment of the overall flatness of the etched surface, which was measured in air. The remaining slowly varying undulation on the surface was less than 0.25 nm over a 1 pm X 1 Frn surface area and therefore

most of the surface undulation remained within a + 1 monolayer fluctuation.

3.2. XPS measurement

3.2.1. Untreated GaAs

0 Is, Ga3d, Ga 2p, As 3d and As2p XPS peaks were measured for the RTO GaAs sample before the wet HCl etching. For example, Fig. 2a shows rather wide 0 1s signals that correspond to a mixture of Ga,O, and As,O,. The Ga3d peak at a binding energy of 21.0 eV was assigned to the oxide and that at 19.6 eV to the bulk GaAs element. Complete reversal of the relative Ga 3d XPS intensity of Ga,O, to the bulk GaAs element between high and low 8 was observed, that is, [GaAs]/[Ga,O,] were 0.4 at 0 = 30”, 0.7 at 45”, and 1.2 at 70”. This demonstrates the presence of the oxide overlayer on the bulk GaAs. The XPS sensitivity to depth is related to the escape depth A, of the emitted electron [4]. The Ga3d and As3d transitions involve large kinetic energies atd sample comparatively thick layers, A, = 30-45 A. The 3d signals thus sampled both the overlayer and bulk. The thickness 0% the overlayered oxides was estimated to be 28 k 6 A from the angu- lar dependence of the relative intensity of the bulk

I ” ” 17 ” I ” 9 I ” ” /

a d A

b 4 I,, , I, I ,,,.I ,,,,I.,, I,

535 530 525 535’ 530 525

Binding Energy I eV

Fig. 2. XPS spectra of 0 Is from GaAs samples, 0 = 70” and 20”.

(a) GaAs substrate with native oxides. (b) HCl solution etching for

10 min. (c) H,O rinse after wet HCI solution etching (10 min) and

H,O rinse (10 s).

and oxide signals [5]. As 2p and Ga 2p signals for the surface oxides were also observed at 1327 and 1119.7 eV, respectively. The ratio [Ga],,/[As],, at 13 = 70” was 4.4 + 0.4. The Ga2p and As 2p transitions in- volving small kinetic energies are rather surface sensitive and the A, are lo-15 and 6-10 A, respec- tively. The 2p signals sampled only the overlayer oxides, As,O, and Ga,O,.

3.2.2. GaAs treated with HCl

When a RTO GaAs surface was etched with the HCl solution for 10 min, the 0 1s signal intensity decreased by a factor of 4 as shown in Fig. 2b. Fig. 3a shows the Cl 2p and As 3s spectra at 0 = 70” and 20” for this GaAs sample. The relative intensity of Cl 2p with respect to As 3s increased with decreasing

0 due to an overlayer of Cl containing species on the bulk GaAs. When the Ga2p XPS spectra of this GaAs sample were measured, a complete reversal of the relative intensity of the overlayer peak and the bulk GaAs peak between high and low 0 was ob- served. Because this ratio was unity at 0 = 20”, the thickness of the overlayer was estimated to be 2.0 & 0.5 A [5]. Since the observed As2p signal had only one peak at 1324 eV that corresponds to the bulk GaAs, arsenic chlorides and oxides were not present on the surface. The chemical species of the overlayer

Z. Song et al. /Applied Su face Science 82 /83 (1994) 250-256 253

on the GaAs surface etched with the HCl solution were a mixture of gallium oxides and chlorides. The [Ga],,/[As],, ratio at 8 = 70” was 1.0.

3.2.3. G&s treated with HCI and H,O Fig. 3b shows a decrease of the C12p XPS inten-

sity when a RTO GaAs sample was etched with the HCl solution for 10 min and sequentially rinsed with

water for 10 s. Fig. 2c shows the 0 1s XPS signal. These results indicate that almost all the oxides and

chlorides were removed from the surface by the short rinse with water after the HCl etching. Since gallium oxides are not soluble in water, the chemical species of the overlayer are gallium oxichlorides as water soluble species. When we measured the Ga2p and As 2p XPS spectra of the GaAs surface after the

HCl/H,O treatment, the [Ga],,/[As],, ratio at B = 70” was 0.6 f 0.1 and neither arsenic oxides nor arsenic chlorides were detected.

3.3. Laser irradiation effect

3.3.1. Laser irradiation at 193 nm The GaAs sample etched only with the HCl solu-

tion was set into the XPS chamber and was photoir- radiated at either 193 or 35.5 nm. The C12p and As 3s XPS intensities were measured as a function of total number of photons irradiating the surface. Fig. 4 shows that [Cl],,/[As],, at 0 = 20” decreased with

As3s

! ,,,I,,,,,,,,,, 210 205 200 195

Binding Energy / eV

Fig. 3. XPS spectra of As3s and C12p from GaAs surfaces. (a)

and (a’) HCI etching (10 min) and no H,O rinse, (b) HCI etching (10 min) and H,O rinse (10 s).

r=” 0

2 0.8-

. Np

G ti

0.6- 0 0

193 Ilm 0

0.4- 0 0 I I I 1 0 I I

0 2 4 6 8

Photon number I 10”

Fig. 4. Photoirradiation effect on [Cl],, /[As],, at 193 and 355

nm. The laser intensity is 0.3 MW/cm’ at 193 nm and 2

MW/cm’ at 355 nm. 0 = 20”.

total number of photons at 193 nm. The 193 nm irradiation induced desorption of the chlorides to

vacuum. No change of [Ga],,/[As],, was observed with the 193 nm photoirradiation.

3.3.2. Laser irradiation at 3.55 nm Fig. 4 also shows the effect of 355 nm laser

irradiation on the GaAs(100) surface etched with the HCl solution. At 2 MW/cm2 and a total photon number up to 7 X 1019 photons, [Cl],,/[As],, was about unity and no change was observed upon pho- toirradiation at 355 nm. These results indicate no

desorption of the chlorides. When the laser peak power was increased to 4 MW/cm2 and the total photon number was in the range of (1.5 - 2.4) X 102’ photons, the [Cl]2,/[As]3, ratio decreased to 0.5 f 0.1. This change is due to thermal desorption of the chlorides by the 355 nm laser irradiation. The ratio

[Ga12,/[~12, increased by a factor of two after laser irradiation at 4 MW/cm2. Because arsenic atoms easily desorb by heating the GaAs substrate, this is a typical thermal phenomenon.

For 355 nm laser irradiation, the temperature rise of the GaAs substrate is estimated by [6],

AT=21(1 -R) (r/pc*) 1’2

(K)1’2 ’

254 Z. Song et al. /Applied Surface Science 82 / 83 (I 994) 250-256

where I is the laser intensity and T is the laser pulse duration. The surface reflectivity R at the laser wavelength, specific heat c, density p, and thermal conductivity K are obtained from Refs. [7,8]. Thus, the temperature rise AT is estimated to be 140°C for 2 MW/cm2 laser irradiation, and 290°C for 4 MW/cm2. The surface temperature increased from room temperature (300 K) to 440 K for 2 MW/cm’ and to 590 K for 4 MW/cm’.

4. Discussion

4.1. Overlayer species on GaAs afrer HC1 etching

Based on our XPS measurement, a mixture of

GaCl,, GaO,,, Ga(OH), and GaCl,O, are probable candidates as the overlayer species on the GaAs after etching only with the HCl solution. Because the 0 1s signal intensity became very weak with the sequen- tial rinse with water most of the oxide species are GaCl.0,. Su et al. [9] investigated the thermal etch- ing of a GaAs substrate by Cl, at steady state and reported the temperature ranges over which the des- orption of the etching products occurred, i.e. GaCl, (350-700 K), GaCl (600-950 K), Ga (> 850 K), AsCl, (350-500 K), As, (600-950 K) and As, (500-850 K>. The 355 nm laser irradiation at 2 MW/cm’ increased the GaAs surface temperature to 440 K, which is in the range of the desorption

temperature of GaCl,. However, no decrease of the Cl signals was observed at 2 MW/cm’. This indi- cates that volatile GaCl, species had already des- orbed from the GaAs surface after the 2-4 h evacu- ated in the XPS chamber. GaCl and GaGl, formed by the etching of the GaAs surface are most probable candidates as chemically adsorbed species. When the 355 nm irradiation at 4 MW/cm’ increased the GaAs surface temperature to 590 K, GaCl may be thermally desorbed from the surface. Actually, at 4 MW/cm’ the [Cl],,/[As],, ratio decreased by a factor of two. The strong laser irradiation at 355 nm induced thermal desorption of the surface chlorides.

Comparing the laser irradiation results at the two different wavelengths, we conclude that desorption of the chlorides by the 193 nm laser irradiation is due to direct photocleavage of the Ga-Cl bond and

the Ga atom is left on the surface, because (a) even with weak laser intensity (0.3 MW/cm2) photoirra- diation at 193 nm was much more effective in promoting the desorption of the chlorides than that at 355 nm, and (b) the [Ga],,/[As],, ratio did not change with the 193 nm laser irradiation. The Ga-Cl

bond cleavage is induced by direct photoabsorption of the adsorbed chlorides at 193 nm. The threshold wavelengths of gas phase photoabsorption are 270 nm for GaCl and 230 nm for GaCl, [lO,ll]. Al- though GaCl has A’II,-X ‘z+ and B3n,-X ‘c, transitions at 350 nm, these intensities are very weak because of spin violence in the transitions [12]. The wavelength effect observed in the present experiment is in reasonable agreement with the reported gas phase absorption spectra.

After the HCl etching and subsequent Hz0 rinse, the amount of each species on the surface shows that the surface stoichiometry changes a great deal. The

ratio [Ga],,/[As],, of the untreated RTO GaAs sam- ple was about 4. After immersion in the HCl solution for 10 min, the [Ga],,/[As],, ratio became 1.0. When rinsed shortly with water after the HCl etch-

ing, the ]Ga],,/]As],, ratio became 0.6. The HCl/H,O etching makes the GaAs surface As-rich. These results indicate the following etching mecha- nism. When the RTO GaAs is immersed in the HCl solution, the thick overlayer of oxides reacts with HCl and most of the oxides dissolve into the solu- tion, leaving a mixture of gallium chlorides and

oxichlorides on the surface. When the GaAs sub- strate is subsequently dipped in water for several seconds, the surface species (gallium chlorides and oxichlorides) dissolves, leaving an As-rich surface.

4.2. Etching mechanism

The wet etching mechanism may be compared with the dry etching mechanism. Chloride species on GaAs surfaces were measured with XPS, UPS, TDS for the Cl,/GaAs system [9,13,14]. Freedman and Stinespring [13] studied gas phase halogenation of Ga-rich surfaces of GaAs(100) with beams of atomic chlorine under vacuum conditions. Their XPS analy- sis indicates that chlorine atoms produce a product layer even at low temperatures, 130-273 K. At 323 K or above, a stable reaction layer was not observed

Z. Song et al. /Applied Surface Science 82 /83 (1994) 250-256 255

GaClxOy GaClxOv GaClxO, 1 Cl Cl Cl I

,&a,? C!a,H Ga H As As As ‘it3

Ga Ga Ga

_

I

,HHHHHHH!

As As As AS AS As As I Ga Ga Ga Ga Ga Ga

Fig. 5. Schematic diagram of the wet HCl/H,O etching proce-

dures.

and an As-rich GaAs surface was formed. The fact that the chlorinated reaction layer decreased notice- ably at 323 K is due to the increasing volatility of GaCl,. The dry etching product has been investi- gated by the use of modulated molecular beam scat- tering, angular distribution and time-of-flight distri- bution [9,14-161. GaCl, and GaCl begin to desorb at 350 and 600 K, respectively [15]. Saito and Kondo [17] studied the gas phase etching of GaAs with a HCl gas beam. They found the presence of gallium chlorides (GaCl, and GaCl,) under high HCl flux conditions and temperatures > 670 K.

In the wet etching, native oxides on GaAs react with HCI to form overlayers of GaCl,O, and GaCl.. GaCl, species are formed at the very surface of the GaAs substrate, and are more likely formed at a defect (step) site than at a flat site. The role of HCl/H,O etching is to remove Ga atoms on kinks and steps as GaCl, to produce an As-rich surface.

This phenomenon of a Ga-rich surface becoming As-rich after HCl treatment was reported by Lu et al. [l], assuming that the elementary arsenic is insoluble in the HCl solution. Formation of the As-rich surface has been found in Cl,/GaAs dry etching processes [9,13,14,16]. Schematics of the HCl/H,O etching processes are shown in Fig. 5. Nooney et al. [181 reported the interaction of gaseous HCl with Ga-rich and As-rich GaAs(100) surfaces. Their HREELS data showed that when the surface is dosed with H atoms they bond to both As and Ga, but when HCl reacts with the surface the H atoms bond only to As, on both the Ga-rich and As-rich surfaces. A reaction model is proposed in which HCl adds across Ga-As backbonds at Ga dimer vacancies, with H bonding to As and Cl bonding to Ga.

Acknowledgement

This work was supported by a Grant-in-Aid from the Ministry of Education, Science and Culture of Japan (1.S.) and was partly defrayed a Grant-in-aid on Priority-Area-Research “Photochemical Dynam- ics” (M.K.).

References

[II

121

[31

[41

151

b1

[71

k31

[91

Z.H. Lu, C. Lagarde, E. Sacher, J.F. Currie and A. YeIon, J.

Vat. Sci. Technol. A 7 (1989) 646.

M. Suzuki, Y. Kudoh, Y. Homma and R. Kaneko, AppI.

Phys. Lett. 58 (1991) 2225.

C.D. Wagner, W.M. Riggs, L.E. Davis, J.F. Moulder and

GE. Muilenberg, Handbook of X-ray Photoelectron Spec-

troscopy (Perkin-Elmer, Eden Praire, MN, 1979).

S. Tanuma, C.J. Powell and D.R. Penn, Surf. Interface Anal.

17 (1991) 927.

Z. Song, S. Shogen, M. Kawasaki and 1. Suemune, to be

published.

J.F. Osmundsen, C.C. AbeIe and J.G. Eden, J. Appl. Phys.

57 (1985) 2921.

SM. Sze, Physics of Semiconductor Devices (Wiley, New

York, 1981) p. 850.

H.R. Philippe and H. Ehrenreich, Phys. Rev. 129 (1962)

1550; J.R. Chelikowsky and M.L. Cohen, Phys. Rev. B 14

(1976) 556. C. Su, M. Xi. Z.-G. Dai, M.F. Vernon and B.E. Bent, Surf.

Sci. 282 (1993) 357.

256 Z. Song et al. /Applied Surface Science 82 / 83 (1994) 250-256

[lo] F.K. Levine and J.G. Winans, Phys. Rev. 84 (1951) 431.

[ll] W.M. Wenk, Phys. Acta 14 (1941) 355.

[12] V.M. Donnely and R.F. Karlicek, .I. Appl. Phys. 53 (1982)

6399.

[13] A. Freedman and C.D. Stinespring, J. Phys. Chem. 96 (1992)

2253.

[14] L.A. DeLouise, J. Appl. Phys. 70 (1991) 1718.

[lS] P. Bond, D.N. Brier, .I. Fletcher, P.A. Gorry and M.E.

Pemple, Chem. Phys. Lett. 208 (1993) 269.

[16] A. Ludviksson, M. Xu and R.M. Martin, Surf. Sci. 277

(1992) 282.

[17] J. Saito and K. Kondo, J. Appl. Phys. 67 (1990) 6274.

[18] M. Nooney, V. Lieberman, M. Xu, A. Ludvisksson and R.M.

Martin, Surf. Sci. 302 (1994) 192.