Characteristics of III-V Dry Etching In HBr-Based Discharges · 856 J. Electrochem. Soc., Vol. 139,...

856 J. Electrochem. Soc., Vol. 139, No. 3, March 1992 9 The Electrochemical Society, Inc.

Manuscript submitted March 20, 1991; revised manuscript received Oct. 11, 1991.

AT&T Bell Laboratories assisted in meeting the publica- tion costs of this article.

REFERENCES 1. C. A. Bollinger, D. Grube, S. A. Lytle, E. P. Martin,

J .A. Shimer, and H. R. Siddiqui, in "Proc. 7th Int. IEEE VLSI Multilevel Interconnect Conf.," IEEE, p. 21 (1990).

2. A. C. Adams, in "Plasma Deposited Thin Films," F. Jansen and J. Mort, Editors, CRC Press, New York (1986).

3. J. J. Hsieh, D. E. Ibbotson, J .A. Mucha, and D.L. Flamm, in "Proc. 6th Int. IEEE VLSI Multilevel In- terconnect Conf.," IEEE, p. 411 (1989).

4. C. S. Pal and C.-P. Chang, J. Appl. Phys., 68, 793 (1990). 5. D. C. H. Yu, D. P. Faveau, E. P. Martin, and A. S. Ma-

nocha, in "Proc. 7th Int. IEEE VLSI Multilevel Inter- connect Conf.," IEEE, p. 166 (1990).

6. N. Selamoglu, J. A. Mucha, D. E. Ibbotson, and D. L. Flamm, J. Vac. Sci. Technol, B7, 1345 (1989).

7. L. J. Olmer, E. R. Lory, W. W. Clauhs, and A. Harrus,

Abstract 255, p. 368, The Electrochemical Society Extended Abstracts, Vol. 87-1, Philadephia, PA (1987).

8. C.-P. Chang, C. S. Pal, and J. J. Hsieh, J. Appl. Phys., 67, 2119 (1990).

9. T.-C. D. Huo et al., To be published. 10. E. J. McInerney and S. C. Avanzino, IEEE Trans. Elec-

tron Dev., ED-34, 615 (1987). 11. T. N. Kennedy, J. Vac. Sci. Technol., 13, 1135 (1976). 12. S. Matsuo and M. Kiuchi, Jpn. J. Appl. Phys., 22, L210

(1983). 13. C. Chiang and D. B. Fraser, Abstract 179, p. 260, The

Electrochemical Society Extended Abstracts, Vol. 89-1, Los Angeles, CA (1989).

14. D. R. Denison, C. Chiang, and D. B. Fraser, Abstract 180, p . 261, ibid.

15. A.K. Sinha, H. J. Levinstein, and T. E. Smith, J. Appl. Phys., 49, 2423 (1978).

16. N. Lifshitz, G. Smolinsky, and J.M. Andrews, This Journal, 136, 1440 (1989).

17. W.-K. Chu, J. W. Mayer, and M.-A. Nicolet, "Backscat- tering Spectrometry," Academic Press, Orlando, FL (1978).

18. A. C. Adams, F. B. Alexander, C. D.Capio, and T. E. Smith, This Journal, 128, 1545 (1981).

Characteristics of III-V Dry Etching In HBr-Based Discharges S. J. Pearton, U. K. Chakrabarti, E. Lane, A. P. Perley, C. R. Abernathy, and W. S. Hobson

AT&T Bell Laboratories, Murray Hiff, New Jersey 07974

K. S. Jones University of Florida, Gainesville, Florida 32611

ABSTRACT

Electron cyclotron resonance (ECR) discharges of HBr/Ar, HBr/H2, or HBr/CH4 were used for dry etching of Ga-based (GaAs, GaSb, and A1GaAs) and In-based (InP, InAs, InSb, InGaAs, and InA1As) III-V semiconductors. The effects of varia- tions in pressure (1-20 mTorr), gas composition, and additional RF-induced bias on the sample were examined. At least -100 V of dc bias is required to initiate etching under all conditions, with the etch rates found to be fastest with CH4 addition, followed by H2 and then Ar. The etched surface morphologies are smooth with HBr/Ar and HBr/CH4 discharges, but tend to become increasingly rough as the microwave power into HBr/H2 discharges is increased, particularly with the In- based materials. The surfaces are chemically very clean after dry etching, with no evidence of Br-containing contamination remaining on any of the materials. The HBr/Ar and HBr/CH4 discharges produce the smallest changes in the electrical properties of the semiconductors, while HBr/H2 plasma exposure may cause dopant passivation and changes in barrier height under some conditions.

III-V semiconductors may be dry etched using several different classes of discharges. The most common gas mixtures are those based on chlorine (e.g. C12, SIC14, BC13, and CC12F2), often diluted with Ar, He, or O (1-13). The etch rates obtained depend on the material itself and the exact plasma conditions, but in general the etch rates are quite high for Ga-based compounds such as GaAs, A1GaAs, and GaSb. The sole exception is when fluorine is present in the discharge, in this case AlGaAs displays a very slow etch rate due to formation of the relatively involatile A1F3 on the surface (14-16). The use of CC12F2 or the addition of fluorine-containing gases (usually SF6) to a C12-based discharge therefore provides an etch stop at an underlying A1GaAs when etching through GaAs (7, 12). The other class of gas mixtures used are based on CH4/H~, largely introduced to overcome the limitations of C12-based discharges for etching In-containing materials such as InP, InGaAs, and related compounds (17-21). In the latter cases, the removal of indium chlorides is difficult at conventional etch temperatures (<100~ (10) and preferential loss of the group V element (P, As, or Sb) will generally lead to surface morphologies which are somewhat rough. The CH4/H2 mixtures, on the other hand, will etch all of the In-based compounds in a slow but very smooth manner under ~he right conditions.

The third general class of gas mixtures are those based on Br2 or I2. Little information is available on the etching characteristics of III-V semiconductors, particularly under the ion-enhanced conditions necessary for production of small feature sizes (2, 22, 23). Ibbotson et al. (22) have reported extremely fast etch rates for GaAs (60 ~m- min -~) in high pressure (0.3 Torr) Br2 plasmas. Temperatures in ex- cess of 200~ were required to achieve etching of InP in similar plasmas. Takimoto et al. (23) fabricated etch-facet lasers using Br2/N2 and BrJAr reactive ion etching to ob- tain rates of 2 ~m 9 rain -~ for power densities of 0.78 W cm -2 at a pressure of - 4 mTorr. Also, HBr plasma etching has recently attracted attention for deep feature fabrica- tion in Si technology (24). It is of interest to see whether Br-based discharges will provide faster etching than CH~rI2 plasmas while retaining smooth surface morphologies.

In this paper we report on the etching characteristics of III-V semiconductors in electron cyclotron resonance (ECR) discharges of HBr/H2, HBr/CH4, and HBr/Ar. All of the common Ga-based (GaAs, Al0.3Ga0.7As, and GaSb) and In-based (InP, InAs, InSb, InGaAs, and InA1As) materials have been examined. The etch rates have been measured as a function of microwave power pressure and dc bias on the samples, while the surface morphology was examined

Downloaded 04 Apr 2011 to 128.227.135.101. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

J. Electrochem. Soc., Vol. 139, No. 3, March 1992 9 The

by scanning electron microscopy (SEM). Auger electron spectroscopy (AES) was used to monitor changes in surface composit ion as a result of the etching, while capaci- tance-voltage (C-V), current-voltage (I-V), and photolumi- nescence (PL) measurements were used to measure changes in the electrical and optical quality of the material. Finally, transmission electron microscopy (TEM) was used to check for lattice disruption in some of the samples due to the ion bombardment present during the dry etching.

Exper imenta l A variety of different types of material was used for

these experiments. For etch rate measurements, semiinsulating, (100), undoped GaAs, Fe-doped, (100), semiinsulating InP, undoped (p = 10 ~6 cm-~), (100) bulk GaSb, n-type (n = 101~ cm -~) bulk InAs and n-type (2 x 10 TM cm-~), bulk InSb wafers were patterned with Hoeschst Celanese 5209 E photoresist to yield a test pattern with various opening sizes. Epitaxial layers (~1 I~m thick) of n-type (n = l0 TM cm-3), Sn-doped A10.3Ga0.TAs were grown by metal organic molecular beam epitaxy (MOMBE) on semiinsulating substrates, while undoped (n -< 10 TM cm -3) epitaxial layers (~1 I~m thick) of In0.5~Ga0.47As and In0.~2A10.~As were grown by metal organic chemical vapor deposition (MOCVD) on Fe-doped, semiinsulating InP substrates. Some of these samples were lithographically patterned as described above. After etching, the resist was removed in acetone and the etched depth obtained from stylus profil- ometry measurements. The anisotropy and smoothness of the etching were examined by SEM. Unpatterned GaAs and InP were used for AES and PL measurements. The AES primary electron beam voltage was 10 kV at a beam current of 0.5 ~A, with a primary spot size of 5 ~m diam. Depth profiling was accomplished by sputtering the sample with a 3.5 keV Ar + ion beam at a rate of -30 A 9 min -~, rastered over 3 • 3 m m 2. The AES analysis was performed after removal from the etch chamber since we wish to know the state of the surface that would be presented for further processing, rather than the intrinsic changes caused by plasma exposure. The PL measurements were performed at 300 K using an Ar + ion laser as the excitation source.

For C-V and I-V measurements, n-type InP (n = 6 • 10 z5 cm -~) and GaAs (n = 10 ~ cm -~) bulk Si-doped wafers were contacted on the rear face with alloyed (400~ 20 s) AuGeNi/Au metallization to provide ohmic contact. After the dry etching treatments, TiPtAu (for GaAs) or Au (for InP) was evaporated onto the front face for Schottky con- tacts. Unpat terned sections of InSb were also prepared for cross-sectional TEM by chemical thinning and iodine ion milling. Multibeam bright-field imaging with seven beams included within the objective aperture was utilized to ob- tain lattice imaging.

The dry etching was performed in the hybrid ECR-radio frequency (RF, 13.56 MHz) Plasma Therm SL 720 system described in detail previously (25, 26). In brief, this utilizes a 2.45 GHz microwave excitation source (Wavemat Model 300) of the resonant cavity design, with additional RF bias superimposed at the sample position. The electronic grade HBr, H=, CH~, and Ar were introduced into the microwave source via mass flow controllers. Typical flow rates were 15 standard cubic centimeters per min (sccm) for HBr, and 5 sccm for the other gases. These ratios were chosen specifically to compare the efficiency of the etching with CHJI-I= mixture~ in which the CH~:H= ratio is typically ~3:1. We also used total flows of one-half and three times these values but the same general trends were observed re- gardless of the absolute flow rates. We also found that neat HBr discharges were difficult to initiate at the low end of our pressure range (1-5 mTorr), while initiation of the three mixtures listed above was straightforward. Prior to loading into the etching system, the samples were dipped in 1HF:I H=O (for the Ga-based materials) or 10 HCI:10 HF:50 H~O:IHzO~ (for the In-based materials) in order to remove their native oxides. The samples were then dried in filtered N= and transferred via the load-lock into the Plasma Thcrm system.

Electrochemical Society, Inc 857

200

150

100

50

E 0 E

o< 150 LU I-- < lO0 rv -r" o 50 I-- LU

150

1 O0

50

• HBr/CH 4

[ ] AIGaAs V GaSh

HBr/H 2 o ~ s

-II I AIGaAs 9 GaSb

HBr/Ar -A--~s

9 AIGaAs x GaSb

HBr/CH 4 9 [nGaAs

- ~ InAIAs

HBr/H 2 - 9 InGaAs

9 ]nAIAs

HSr/Ar - %7 lnGaAs

9 ]nAIAs

HBr/CH 4 O TnP [] InAs V InSb HBr/H 2 9 InP 9 InAs 9 InSb HBr/Ar 2, InP 9 lnAs

0 x InSb o

150 W MICROWAVE / ) 1 mTorr / t ~ :

, - ' \ / / /

.," ?,J. . - "z: / . . 1 ; f / f

...:'2:1 Sz/.

I I I 100 200 300 400

DC BIAS (V)

Fig. 1. Etch rate of the nine different semiconductors in 1 mTorr, 150 W (microwave) discharges of 15 HBr/5 CH4, 15 HBr/5 H2, or 15 HBr/S Ar, as a function of the dc bias on the sample.

Results and Discussion Figure i shows the etching rate of the eight materials in-

vestigated as a function of the dc bias at fixed pressure (1 mTorr) and microwave power (150 W) for the 15 HBr/5 H2, 15 HBr/5 CH4, and 15 HBr/5 Ar discharges. The etching rates are all generally slow, and have the same trend for each different semiconductor in that the CH4 additive produces faster rates than H2, which in turn is faster than Ar addition. The role of the CH4 is primarily to aid removal of group III species (X = Ga, In, or A1) by formation of volatile species of the form CH,(X)~, where n = 1-3 and m is probably unity (17-21). These species appear to be more readily removed than the group III bromides. For the latter, A1Br3 has a boiling point of 263~ and a 1 mm vapor pressure at 81~ whereas GaBr3 boils at 279~ and InBr and InBr2 boil at 662 and 632~ respectively (InBr3 sub- limes) (1). These values, while not always a reliable indica- tor of etch rates since it is the volatility under ion- enhanced conditions that is important, are all higher than their C1 substituted counterparts. We stress however that the vapor pressures are to be used as qualitative guides only. The role of the H~ is to enhance removal of the group V species as PH3, AsH~, or SbH~ (17-21), and these appear to be somewhat more volatile than their counterparts PBrs, PBr3, AsBr3, and SbBr3 under our conditions. This trend is indeed borne out by a consideration of the melting points: PBr~ (106~ and PBr3 (173~ v s . PH3 (-88~ AsBr3 (221~ v s . AsH3 (-55~ and SbBr~ (280~ v s . SbH3 (-17~ (1). The vapor pressures for these species show a similar hierarchy. The predominant role of the Ar is to enhance the ion bombardment component of the etching, and as the dc bias, and hence the energy of the incident ions, is increased, the etch rate also increases, as expected. Of the Ga-based materials, the etch rates increase in the order A1GaAs < GaAs < GaSb, whereas for the In-based semiconductors it increases in the order InA1As < InP < InAs < InSb-InGaAs. Rather high dc biases are needed to initiate etching for all of the materials relative to C12-based mixtures where even 10-20 V will cause etching of the Ga- based semiconductor (-80 V for In-based under similar flow rates, 20 sccm, and pressure 1 mTorr) (13) and CHJtt2



Fig. 2. SEM micrographs of features etched into GaAs (top and bottom left), InAs (top right), and InP (bottom right) with 15 HBr/5 CH4, I mTorr, -250 Vdc, 150 W (microwave) discharges.

mixtures where the In-based materials will etch at -40 V (InAlAs requires >-120 V) and Ga-based materials at >-80 V (27). The requirement for high dc biases is certainly a dis- advantage in terms of utility for etching electronic devices because of the introduction of damage to the semiconductor and possible mask erosion during longer plasma exposures.

Features etched into GaAs (top and bottom left), InAs (top right) and InP (bottom right) with 15 HBr/5 CH4 at 1 mTorr pressure, -250 V dc bias and 150 W of microwave power are shown in Fig. 2. The etching is quite anisotropic and the surfaces of all three materials are smooth. Similar results were obtained with HBr/I-t2 and HBr/Ar discharges, provided the pressure was kept below -10 mTorr, the microwave power below -500 W (particularly for HBr/H2 dis-

charges for reasons to be discussed later), and the de bias kept to -<-300 V. Above any of these values, surface roughness or sloped sidewalls were observed for all of the materials investigated.

In Fig. 3 we show SEM micrographs from InSb etched in HBr/Ar discharges at 1 mTorr pressure, 150 W of microwave power, and -250 V dc bias (top left) or -350 V bias (bottom left). While the surface morphology is quite smooth in the former case, some degree of surface roughening is evident at the higher bias. This is most likely a result of the preferential sputtering of Sb, leaving the surface In-rich. This is a common occurrence with In-based binary semiconductors (28). The right hand side of Fig. 3 shows SEM micrographs of features etched into InGaAs with HBr/H2 discharges with -250 V dc bias at a pressure of

Fig. 3. SEM micrographs of features etched into InSb in 1 mTorr, -250 V dc (top left) or -350 V dc (bottom left), 150 W (micra- wave) 15 HBr/5 Ar discharges. At. right are micrographs of InGaAs etched in 15 HBr/5 H2, 150W, -250 V dc plasmas at 20 mTorr (tap right) or I mTarr (bottom right).


J. Electrochem. Soc., Vol. 139, No. 3, March 1992 9 The Electrochemical Society, Inc. 859

Fig. 4. SEM micrographs of features etched into InAIAs with 15 HBr/5 Ar, 1 mTorr, 150W discharges with either - 2 5 0 V (top left) or - 4 0 0 V (bottom left) bias. At right are micrographs of InP etched in 15 HBr/5 H~ discharges at - 4 0 0 V dc and 150 W (microwave) and either 1 mTorr (bottom right) or 20 mTorr (top right).

20 mTorr (top) or 1 mTorr (bottom). In the latter case the etching is anisotropic, but at the higher pressure there is some undercut evident on the sidewalls. This indicates an increase in the chemical etch component of the gas mixture at the higher pressure, as expected.

Figure 4 shows further examples of the characteristics of the HBr-based dry etching. The SEM micrographs at left show features etched into InA1As with HBr/Ar discharges at 1 mTorr pressure 150 W microwave power and either -250 (top) or -400 V (bottom) dc bias. In both cases the etched surface morphologies are smooth and indicate that ternary materials are more resistant to preferential sputtering of the group V element than binaries such as InAs, 400 InSb, and InP. This is evident from a comparison of the InA1As results with those from InSb in the previous figure 300 and also from the micrograph at the bottom of Fig. 4, which shows InP etched in a 1 mTorr, 150 W (microwave), 200 -400 V dc HBr/I-I2 discharge. In this case there are In droplets evident on the surface due to the preferential loss of P. With a combination of high bias (-400 V) and high 100 pressure (20 mTorr), the HBr/H2 etching produces sputtering of the resist mask and undercutt ing of the feature side- ~- 0 walls leading to the sidewall profile shown at the top right E of Fig. 4.

The dependence of semiconductor etching rates in o<~ 300 HBr/H~ and HBr/CH4 discharges (150 W microwave power, -200 V dc bias) are shown in Fig. 5. For GaAs and A1GaAs, < 200 cc the etch rates show little change with pressure in the range 1-20 rnTorr, indicating that the etching is not re- ~ 100 actant-limited for these materials under these conditions, m For GaSb, InGaAs, InP, InAs, and InSb, the etching rates increase at higher pressures as more active Br, H, and CH3 0 species are supplied to the semiconductor surfaces. By contrast, the etch rate of InA1As decreases at the higher 300 pressures, even though the dc bias is left constant. This may be due to an increase in coverage of the surface with gas molecules which impede release of volatile etch prod- 200 ucts, particularly those of the A1.

The dependence of etching rate for each of the materials 100 on applied microwave power is shown in Fig. 6. In all cases the etching rates increase as the respective discharges be- 0 come more dissociated at the higher microwave powers. 0 Based on preliminary measurements of the average electron densities 4 cm downstream from the baseplate of the microwave source using a plasma density monitor, the typical electron density increased from ~10 ~ to 8 • 10 ~ cm -3 in a 15 HBr/5 Ar, 1 mTorr discharge when the micro-

wave power was increased from 100 to 700 W. As more active species are incident on the surface at higher microwave powers, the etching rates also increase.

The benefits of operating at high microwave powers must, however, be balanced against the effect on the re- sultant etched surface morphology. The Ga-based materials appear relatively insensitive to the introduction of surface roughness at high microwave powers. The left-hand side of Fig. 7 shows features etched into GaAs with HBr/I-I2 discharges (1 mTorr, -200 V dc) at either 500 (top) or

HBr/CH 4 HBr/H 2 150 W MICROWAVE 0 GaAs 9 GaAs -200 V DC , ~ [:3 AIGaAs 9 AIGaAs ~ . . . ~

o . . L .___ . O. . . . . . . . - o . . . . . . . . .c.- . . . . . . . "(

. . . . . . . . . . . . . . . . . ,

HBr/CH 4 HBr/H 2 0 InGaAs 9 InGaAs A InAIAs 9 InAIAs

C- . . . . . .a- . . . . . . . . ~ . . . . . . . . .

e

HBr/CH 4 HBr/H 2 O InP 9 InP ~.'q ~ ...[ [3 ]nAs 9 InAs . . . " ~.I"- . . . . -~ ~7 lnSb 9 . ~ . ~

I I I 5 10 15 20

PRESSURE (mTorr)

Fig. 5. Etch rates of the nine different semiconductors in 150 W (microwave) - 2 0 0 V dc discharges of 15 HBr/5 CH4 or 15 HBr/5 H2 as a function of pressure.


860 J. Electrochem. Soc., Vol.

200

150

100

50

E 0 E

o<< 150

n -

T o UJ

I _ HBr/CH 4 - 2 0 0 V DO ]

1 rnmorr /

O GaAs ..~

V GaSb - ~ ,,~7~ ~ ~

HBr/H 2 ~ . . ~

- 9 AIGaAs

x

HBr/Ar O" ~ " ~ "

9 AIGaAs x GaSb

HBr/CH 4 0 InGaAs

- A lnAIAs

I HSr/H2 100 - 9 lnGaAs

9 ]nAIAs

HBr/Ar 50 - ~ s

9 ]nAIAs

o HBr/CH 4

O InP 150 D lnAs

~7 ]nSb

HBr/H 2

100 9 [nP ~ InAs [nSb

5 0 HBr/Ar

9 InAs 0 x ]nSb

lOO

~jT~ ~ ~ s ~

I I I q 200 300 400 500 600

MICROWAVE POWER (W) 700

Fig. 6. Etch rates of the different materials in 1 mTorr, - 2 0 0 V dc discharges of 15 HBr/5 CH4, 15 HBr/5 H2, or 15 HBr/5 Ar as a function of microwave power.

150 W (bottom). In both cases the surfaces are quite smooth. By contrast, the In-based materials, with the exception of InA1As, generally displayed rougher surfaces at high power. The right-hand side of Fig. 7 shows InGaAs surfaces after etching in HBr/H2 discharges (1 mTorr, -200 V dc) at either 150 (top) or 500 W (bottom). Based on our past experience (21, 26) and that of others (24-32), in the effects of hydrogen-containing plasmas on In-containing compounds, we ascribe this surface roughening to the preferential loss of As from the material. This was con- firmed by AES analysis, which showed enrichment of In

139, No . 3, M a r c h 1992 9 The E lec t rochemica l Society, Inc.

and A1 to a depth of -160 A in the sample of Fig. 7. This roughening is evident from short plasma exposures for high microwave powers and not, we believe, due to heat- ing of the surface and subsequent evaporation of the group V species. Similarly, thermal bonding of the samples di- rectly to the water-cooled cathode did not prevent the surface roughening. We estimate the temperature of our sample to be -<60~ at all times for thermally unbonded sections and -<40~ for those bonded to the cathode, based on fluoro-optic probe measurements.

Further examples of the effects of high microwave power levels on the surfaces of binary materials, InP and InAs, are shown in the SEM micrographs of Fig. 8. In both cases the samples were etched in 20 mTorr, -200 V, 15 HBr/5 H2 discharges with 500 W of microwave power. The combination of undercut, due to the higher pressure, and roughening, due to the preferential loss of P or As, leads to very poor etch characteristics. Similar results were obtained with both HBr/CH4 and HBr/Ar discharges.

Figure 9 shows AES surface scans and depth profiles from GaAs samples etched in 1 mTorr, -200 V, 150 W (microwave) 15 HBr/5 H~ or HBr/5 CH4 discharges for 15 min, as well as comparable data from an unetched GaAs control sample. Under these conditions there are no significant changes caused to the GaAs surface by the dry etching. For these moderate bias and microwave power levels there is no preferential loss of As, and there are no Br-containing species remaining on the surface. This is somewhat sur- prising given the relatively low etch rates with the HBr- based mixtures, in that one might expect to find Br-containing compounds remaining on the surface. However these compounds might only be found for very low bias conditions, for which we have no AES data. Moreover, the CH4-containing plasma does not add significant carbon contamination to the GaAs surface. Indeed, the HBr-based etching is quite clean relative to some gas mixtures (e.g., CF~Br or CC12F2 with additives) which are proven to poly- mer deposition.

Similar data for InP samples etched under the same conditions are shown in Fig. 10. Once again, the etching is re- markably clean, with no Br-containing species detected by the Auger analysis. In the case of HBr/H2 dry etching, however, there is a slight deficiency of P present to a depth of -100 /L This again is most likely due to preferential removal of P by active hydrogen from the discharge. This could be countered by altering the HBr/H2 ratio of the gas

Fig. 7. SEM micrographs of GaAs etched in 15 HBr/5 H2, 1 mTorr, - 2 0 0 V dc discharges at either 500 W (top left) or 150 W (bottom left) microwave power. At right are features etched into InGaAs in similar plasmas at either 5 0 0 W (bottom right) or 150 W (top right).



Fig. 8. SEM micrographs of features etched into InP (top, left and right) or InAs (bottom, left and right) in 15 HBr/5 H2 20mTorr, 500 W (microwave), -200 V dc discharges.

mixture to reduce the amount of free hydrogen in the plasma.

Changes in the optical quality of the semiconductors as a result of HBr-based dry etching were examined by room temperature PL measurements. Figure 11 shows PL spectra from GaAs and InP samples before and after a 15 min exposure to 1 mTorr, -250 V, 150 W discharges of 15 HBr/

7

6 5 4

3

2

1 -

0

GaAs CONTROL

G Ga

I I I

7 ~, 6 - HBr/H2

~ 4

C 2 - As

0

7 i

6 -

4

3

2 -

200

HBr/CH 4

Ga I I I

600 1000 1400 1800

KINETIC ENERGY (eV)

7

6

5 ,,,.,.,.-N ~ G a

4 % A s

3

2 )

1 /C

0

7 _ /v~,~m,-V-Ga

6 ~

s ~ 3 -

2 - O

0 I

7

5

4

:hi o 0 150 300

DEPTH (A)

Fig. 9. AES surface and depth profiles of GaAs samples before and after etching in 1 mTorr, -200 V dc, 150 W (microwave) discharges of 15 HBr/5 H2 or 15 HBr/5 CH4.

5 H2 or 15 HBr/5 CH4. In the case of the GaAs, the dry etching reduces the luminescent efficiency of the material for both mixtures. The decrease however is less for the H2 addition to the HBr. The most l ikely explanation for this is that there is a greater relative concentration of atomic hydrogen available in this case to passivate some of the deep levels created by ion bombardment of the GaAs near-surface region. This is a c o m m o n feature of experiments in-

2

1

0

7

G 6 -

5 -

4

~ 2

0

7

6 -

5 -

4 -

2 - P

1 -

0 I 200

InP CONTROL

I I I I I

HBr/H 2

O

I I I I

O

680 I 1400 1000

KINETIC ENERGY {eV)

HBr/CH 4

P

I 1800

~4 s ~2

1

0

7 6 ~ I n

5 ~ p 4

3

0 150 300

DEPTH (~)

Fig. 10. AES surface and depth profiles of InP samples etched under the same conditions as in Fig. 9.


862 J. Electrochem. Sac., Vol. 139, No. 3, March 1992 9 The Electrochemical Society, Inc.

4

b

3 E 8

~ 2

Z w 1 t--" 7

d a .

o

InP 4

GaAS CONTROL

. . . . . HBr/H 2 : ~ -- '-- HBr/CH 4

!..'".," \ ,/ \- / ,& \

~ ' ~ CONTROL . . . . . HBr/H 2 _. _ Br/CH 4

b

3 E

g .~. 2

N Z m 1 Z . d O_

1.3 1.4 1.5 1.6 1.7 ENERGY (eV)

Fig. 11. Room temperature PL spectra from GaAs and InP samples etched in 1 mTorr, - 2 5 0 V de, 150 W (microwave) discharges of 15 HBr/5 Hz or 15 HBr/5 CH4 for 15 min.

lo 2 f I I / / l! /-"

10-51 ~ ~ 0.0 0.2 0.4 0.6

V F (V)

. . . . CONTROL HBr/CH 4

-- . - - HBdH 2

I 0.8 1.0

l

. . . . . . . . . . . . . . .

I I I I -0.8 -0.6 -0.4 -0.2 0.0

<=L --40 r r

-60 -

-80

-100 -1.0

VR (V)

Fig. 12. Forward and reverse I-V characteristics from Au contacted, n-lnP Schottky diodes after etching for 15 min prior to contact deposition in 1 mTorr, - 2 5 0 V dc, 150W (microwave) discharges of 15 HBr/5 H2 or 15 HBr/5 CH4.

volving hydrogen ion bombardment of semiconductor (33-35). By contrast, in the case of InP, the degradation of PL intensity is far greater for HBr/H2 etching than by comparable HBr/CH4 exposure. We ascribe this difference to the fact that the former mixture causes In enrichment of the surface due to enhanced removal of P by atomic hydrogen, and, as we have reported previously, this will result in significant decreases in PL efficiency from InP (26). The remaining nonstoichiometric surface region is rich in radi- ative centers, and therefore has low PL efficiency.

Forward and reverse I-V characteristics from n-type InP before and after a 15 rain etch in 1 mTorr -250 V, 150 W discharges of 15 HBr/5 H2 or 15 HBr/5 CH4 are shown in Fig. 12. As was the case with the PL measurements, more degradation of the characteristics is observed with the HBr/I-I2 mixture. Indeed, the I-V curves are purely ohmic in this case due to the In-enrichment of the surface. How- ever, for the HBr/CH4 dry etching, the Au contact is still rectifying, even though the reverse current is increased relative to that of the control sample. The Schottky barrier height (qb~) and diode ideality factor (n) were extracted from the forward-bias characteristic, according to the rela- tion (36) [ (ev)]

, = A**T exp e x p 1 \ k T / -~-T-

where J is the current density, A** the effective Rich- ardson constant, T the measurement temperature, e the electronic charge, and k is Boltzmann's constant. The control sample displayed a r value of 0.45 eV and n = 1.09, while the sample etched in HBr/CH4 yielded qbB = 0.42 eV and n = 1.09. The predominant cause of the increased current in the latter case is therefore the slightly smaller barrier height as a result of the dry etching. The fact that the ideality factor is unchanged from that of the control sample may again be a result of a significant amount of hydrogen passivation of deep levels created by the ion bombardment of the surface since even the HBr/CH4 mixture

will obviously have a substantial concentration of hydrogen available for this purpose.

Changes in the near-surface carrier concentration in n-type GaAs etched in the three different types of discharge (15 rain exposures, 1 mTorr, -150 V dc bias, and 200 W of microwave power) are shown in Fig. 13. Larger decreases in near-surface carrier density are observed with H2 addition relative to CH4 addition, whereas Ar addition

E o v

Z o_ EE I- z iJ.i o z o o m o z o t~

1018

1017

n - GaAs ECR - RF 1 mTorr -150 V DC 200 W MICROWAVE

INITIAL

H B r / A ~ HBr/CH 4

HBr/H 2

1016 I i i I 0.0 0.2 0.4 0.6 0.8 1.0

DEPTH (~trn)

Fig. 13. Carrier profiles in n-type GaAs etched for 15min in 1 mTorr, - 1 5 0 V dc, 200 W (microwave) discharges of 15 HBr/5 CH4, 15 HBr/5 H2, or 15 HBr/5 Ar.



Fig. 14. TEM cross section of InSb etched in 1 mTorr - 2 5 0 V dc, 400 W (microwave) discharges of 15 HBr/5 H2 (top and bottom left) or 15 HBr/5 CH4 (top and bottom right). The micrographs at bottom are lattice images.

causes only a minor change. The greater effect present in the first two cases appears to be a result of the more significant hydrogen passivation of the Si donors in the material, as one would expect due to the relatively greater amount of hydrogen available (33). These results are con- sistent with both the PL and I-V measurements described earlier.

TEM cross-sectional micrographs from InSb samples etched in either HBr/H2 (top and bottom left) or HBr/CH4 (top and bottom right) discharges at the same pressure (1 mTorr), dc bias (-250 V), and microwave power (400 W) are shown in Fig. 14. The low-resolution images at top show that, as expected, the surface morphology is much smoother with the HBr/CH4 mixture under these condi-

Fig. 15. Lattice image of InSb etched in 1 mTorr, - 2 5 0 V dc, 400 W (microwave) 15 HBr/5 CH4 discharge for 15 min.



tions. The lattice images at bot tom show that there is no percept ible d isorder visible at this resolution in either sample. Usually we do not observe defects visible by TEM in dry etched samples unless the dc bias during the p lasma exposure was ->350 V. The oxide thickness is 30-40/k in the case of the HBr/CH, etched sample, consist- ent with the value found by AES analysis. An expanded view of the lattice-image of this InSb sample is shown in Fig. 15. Again, there is no visible disorder introduced by the dry etching.

Summary and Conclusions The etch rates of III-V semiconductors in HBr-based dis-

charges are quite slow under the conditions needed for smooth anisotropic etching. Relat ively high self-biases are needed to initiate etching with any of the three mixtures (HBr/Ar, HBr/H2, and HBr/CH4) we investigated. At high dc biases, the b inary In-based materials show surface roughening due to preferential sputter ing of the group V element, and at high microwave power levels the surfaces of all the In-based semiconductors (except InA1As) become rough. For these materials, the HBr/CH4 mixture provides smooth etching over the widest range of conditions. Changes in the electrical and optical propert ies of the semiconductors are generally not as severe as might be expected at the high (-250 V) dc biases used in many of our exper iments because of the passivation of deep-levels by hydrogen from the discharges.

Acknowledgments The authors acknowledge the technical support of J. R.

Lothian, B. Tseng, and R. A. Keane, and the continued en- couragement of S. S. Pei, W. C. Dautremont-Smith, L. A. Heimbrook, D. V. Lang, W. L. Brodsky, and V. D. Mattera.

Manuscr ipt submi t ted Sept. 5, 1991; revised manuscr ip t received Oct. 28, 1991.

AT&T Bell Laboratories assisted in meeting the publica- tion costs of this article.

REFERENCES 1. R. H. Burton, R. A. Gottscho, and G. Smolinsky, in

"Dry Etching for Microelectroncs," R.A. Powell, Editor, Elsevier, Amsterdam (1984).

2. V. M. Donnelly, D. L. Flamm, C. W. Tu, and D. E. Ib- botson, This Journal, 129, 2533 (1982).

3. S. Schwartz and W. Beinvogel, ibid., 130, 310 (1983). 4. H. Yamada, H. Ito, and H. Inada, J. Vac. Sci. Technol. B,

3, 884 (1985). 5. A. Scherer, H. G. Craighead, and E. D. Beebe, ibid., 5,

1599 (1987). 6. R. A. Gottscho, G. Smolinsky, and R. H. Burton, J.

Appl. Phys., 53, 5908 (1982). 7. E. L. Hu and R. E. Howard, Appl. Phys. Lett., 37, 1022

(1980). 8. R. E. Klinger and J. E. Greene, ibid., 38, 620 (1981).

9. D. L. Flamm, V. M. Donnelly, and D. E. Ibbotson, "VLSI Electronics Microstructure Science," N. Ein- spruch, Editors, Academic, New York (1984).

10. S. C. McNevin, J. Vac. Sci. Technol. B, 4, 1216 (1986). 11. S. W. Pang, G. A. Lincoln, R. W. McClelland, P. D. De-

Groff, M. W. Geis and W. J. Piacentini, ibid., 1, 1334 (1983).

12. C. B. Cooper III, S. Salimian, and H.F . MacMillan, Appl. Phys. Lett., 51, 2225 (1987).

13. S. J. Pearton, U. K. Chakrabarti , W.S. Hobson, and A. P. Kinsella, J. Vac. Sci. Technol. B, 8, 607 (1990).

14. K. L. Seaward, N. J. Moll, and W. F. Stickle, ibid., 6, 1645 (1988).

15. K. L. Seaward, N. J. Moll, D. Jo Coulman, and W. F. Stickle, J. Appl. Phys., 61, 2358 (1987).

16. A. Seabaugh, J. Vac. Sci. Technol. B, 6, 77 (1988). 17. U. Niggebrugge, M. Klug, and G. Garus, Inst. Phys.

Conf. Set., 79, 367 (1985). 18. R. Cheung, S. Thomas, S. P. Beaumont, C. Doughty~V.

Law, and C. D. W. Wilkinson, Electron. Lett., 23, 857 (1987).

19. T. R. Hayes, M. Dreisbach, P. M. Thomas, W. C. Dau- t remont-Smith, and L.A. Heimbrook, J. Vac. Sci. Technol. B, 7, 1130 (1989).

20. T. Matsui, H. Sugimoto, T. Ohnishi, and H. Ogata, Elec- tron. Lett., 24, 798 (1988).

21. S. J. Pearton, U. K. Chakrabarti , and F. A. Baiocchi, Appl. Phys. Lett., 55, 1633 (1989).

22. D. E. Ibbotson, D. L. Flamm, and V. M. Donnelly, J. Appl. Phys., 54, 5974 (1985).

23. K. Takimoto, K. Ohnaka, and J. Shibata, Appl. Phys. Lett., 54, 1947 (1989).

24. G. S. Oehrlein, Y. Zhang, G. M. W. Kroesen, E. de Fr~- sart, and T. D. Bestwick, ibid., 58, 2252 (1991).

25. C. Constantine, D. Johnson, S. J. Pearton, U. K. Chak- rabarti , A.B. Emerson, W.S. Hobson, and A .P . Kinsella, J. Vac. Sci. Technol. B, 8, 596 (1990).

26. S. J. Pearton, U. K. Chakrabarti , A .P . Kinsella, D. Johnson, and C. Constantine, Appl. Phys. Left., 56, 1424 (1990).

27. S. J. Pearton, U. K. Chakrabarti , A. Katz, A. P. Perley, W.S. Hobson, and C. Constantine, J. Vac. Sci. Technol. B, 9, 1421 (1991).

28. W. C. Dautremont-Smith, R. A. Gottscho, and R.J . Schutz, in "Semicond. Mat. and Proc. Techn. Hand- book," G. E. McGuire, Editor, Noyes, New Jersey (1988).

29. R. P. H. Chang, C. C. Chang, and S. Darak, J. Vac. Sci. Technol., 20, 45 (1982).

30. J. H. Thomas, G. Kaganowicz, and J .W. Robinson, This Journal, 135, 1201 (1988).

31. S. J. Fonash, Solid State Technol., 28, 150 (1985). 32. F. Proix, O. M'hamedi, and C. A. Schemee, J. Vac. Sci.

Technol. A, 6, 199 (1988). 33. S. J. Pearton, J. W. Corbett, and T. S. Shi, Appl. Phys.

A, 43, 153 (1987). 34. J. S. Wang, S. J. Fonash, and S. Ashok, IEEE Electron.

Dev. Lett., EDL-4, 432 (1983). 35. R. Singh and S. Ashok, Appl. Phys. Lett., 47,426 (1985). 36. S. M. Sze, "Physics of Semiconductor Devices," Aca-

demic Press, New York, (1981).


Characteristics of III-V Dry Etching In HBr-Based Discharges · 856 J. Electrochem. Soc., Vol. 139,...

Documents

Transcript of Characteristics of III-V Dry Etching In HBr-Based Discharges · 856 J. Electrochem. Soc., Vol. 139,...