Etch mechanism of In2O3 and SnO2 thin films in HBr

Etch mechanism of In2O3 and SnO2 thin films in HBr-based inductively coupled plasmas Kwang-Ho Kwon Department of Control and Instrumentation Engineering, Korea University, Chungnam 339-700, Korea

Alexander Efremov Department of Electronic Devices and Materials Technology, State University of Chemistry and Technology, 7 F. Engels St., 153000 Ivanovo, Russia

Moonkeun Kim, Nam Ki Min, and Jaehwa Jeong Department of Control and Instrumentation Engineering, Korea University, Chungnam 339-700, Korea

MunPyo Hong Department of Display and Semiconductor Physics, Korea University, Chungnam 339-700, Korea

Kwangsoo Kima兲 Department of Sogang Institute of Advanced Technology, Sogang University, Seoul 121-742, Korea

共Received 3 November 2009; accepted 28 December 2009; published 21 January 2010兲 The investigations of etch characteristics and mechanisms for both In2O3 and SnO2 thin films in the HBr-based inductively coupled plasmas were carried out. The etch rates were measured as functions of gas mixing ratio 共0%–100% Ar兲, input power 共400–700 W兲, and gas pressure 共4–10 mTorr兲 at fixed bias power 共200 W兲 and gas flow rate 关40 SCCM 共SCCM denotes cubic centimeter per minute at STP兲兴. Plasma parameters and composition were determined using a combination of plasma diagnostics by double Langmuir probe and global 共zero-dimensional兲 plasma model. The correlations between the behaviors of etch rates and fluxes of plasma active species allow one to infer both In2O3 and SnO2 etch mechanisms as the transitional regime of ion-assisted chemical reaction, which is controlled by neutral and charged fluxes together. © 2010 American Vacuum Society. 关DOI: 10.1116/1.3294712兴

I. INTRODUCTION Recently, many new materials have been involved in micro- and nanoelectronics technology aimed at developing new devices as well as improving conventional device characteristics. Among these materials, the transparent conductive metal oxides, including SnO2, In2O3, and their compositions 共In2O3兲x : 共SnO2兲y, have found numerous applications as an electrode materials in flat panel displays, solar cells, and organic light emitting diodes.1–3 That is why the development of an anisotropic dry etch process for SnO2 and In2O3 is an important task to be solved for obtaining an accurate pattern transfer as well as stable device parameters. Until now, there are only few papers reported on the etch characteristics of the In2O3 and SnO2 thin films for electronic device applications and the most of them are related to the “wet” etching technology. Particularly, Bradshaw and Huges4 studied the wet etch process for the 共In2O3兲4 : 共SnO2兲1 on the glass substrates with using the Zn powder and the aqueous solution of the HX 共X = F, Cl, or I兲 acids. The etch process was provided by the direct reduction of the oxides by the hydrogen formed in the reaction of Zn with HX. The highest etch rate of 15 nm/s was obtained for HCl while the best resolution of 2 – 3 ␮m was for HI where the rate was only about 2.5 nm/s. Similar results were obtained for the In2O3 : Sn films etched in the aqueous solution H3PO4 : H2O a兲

Electronic mail: [email protected]

226

J. Vac. Sci. Technol. A 28„2…, Mar/Apr 2010

= 1 : 3.5 Also, Ref. 6 reported optimal wet etch recipes and conditions for both In2O3 and SnO2 thin films, which are HCl: H2O : HNO3 mixed as 4:10:1 at 40 ° C and 47% HI at 60 ° C, respectively. As for the “dry” etch process, one of the earliest works in this field belongs to Braga et al.7 They used the CH2Cl3 with N2 as a carrier gas in a rf 13.56 MHz plasma system and reported the etch rate of 10 nm/min for SnO2. Later, Vaufrey et al.8 studied the etch characteristics of doped SnO2 thin films using the hydrogen-based plasma chemistry in the reactive ion etch process. Using the CH4 : H2 = 3 : 27 gas mixture, they obtained the maximum etch rate of about 22.5 nm/min with an average surface roughness of about 2.5 nm. Although the formation of the polymer film from CH4 was detected, the increasing etch rate with increasing input power and gas pressure allows one to infer the reaction-rate-limiting etch regime. From the data above, one can conclude that the existing dry etch-related studies report only the dependences of the etch rate and etch profile on operating conditions while the basic relationships between process parameters, plasma chemistry and etch kinetics were not analyzed. Therefore, the etch mechanisms are not clearly understood that obstructs the development and optimization of the dry etch process for In2O3 and SnO2. In this work, we carried out the modelbased analysis of etch mechanisms of In2O3 and SnO2 thin films in an inductively coupled HBr-based plasmas. The main goal was to investigate the dependencies of the In2O3 and SnO2 etch rates on the main operating parameters, such

0734-2101/2010/28„2…/226/6/$30.00

©2010 American Vacuum Society

226

227

Kwon et al.: Etch mechanism of In2O3 and SnO2 thin films in HBr-based inductively coupled plasmas

as gas mixing ratio, input power, and gas pressure, as well as to find the correlations between the behaviors of the etch rates and fluxes of active species in order to determine the limiting stage of the etch process. II. EXPERIMENT AND MODELING DETAILS A. Experiment

In2O3 and SnO2 thin films were deposited on Si 共001兲 substrate using the dc-magnetron sputtering of In2O3 共purity, 99.995%兲 and SnO2 共purity, 99.99%兲 targets, respectively. The process was held in Ar at gas pressure of 5 mTorr and input power of 80 W. The thickness of the In2O3 and SnO2 films was about 200 nm, respectively. More deposition detail can be found in Ref. 9. The experiments were performed in the planar inductively coupled plasma reactor described in Ref. 10. The plasma was excited in the cylindrical quartz chamber 共r = 16 cm, l = 12.8 cm兲 at fixed gas flow rate 共q兲 of 40 SCCM 共SCCM denotes cubic centimeter per minute at STP兲 and bias power 共Wdc兲 of 200 W while the gas mixing ratio, input power, and gas pressure were independently adjusted in the ranges of 0%–100% Ar, 400–700 W, and 4–10 mTorr, respectively. The etched samples had the size of about 2 ⫻ 2 cm2. The much smaller size of the etched samples compared to the area of the bottom electrode allows one to neglect both loading effect and the disturbance of plasma parameters by the etch products. The temperature of etched samples was stabilized at 18⫾ 2 ° C. The etched depths were measured using a surface profiler 共Alpha-step 500, Tencor兲. For this purpose, we developed the line striping of the photoresist 共PR兲共AZ1512, positive兲 with a line width/spacing ratio of 2 / 2 ␮m. The initial thickness of the PR layer was approximately 1.2 ␮m. Plasma diagnostics was realized with double Langmuir probes 共DLP2000, Plasmart Inc.兲 which were installed through the chamber wall-side view port. The probes were placed at 4 cm above the bottom electrode and centered in the radial direction. Similar with Refs. 10 and 11, the electron temperature 共Te兲, the ion current density 共Jis兲, and the total positive ion density 共n+兲 were derived from the original I-V traces using the software supplied by the equipment manufacturer. The calculations involved Johnson and Malter’s double probes theory12 as well as the Allen–Boyd– Reynolds approximation for Jis.13 B. Zero-dimensional „global… plasma model

To obtain the data on the densities and fluxes of plasma active species, we used a simplified zero-dimensional model with a Maxwellian electron energy distribution function 共EEDF兲 and with the experimental data on Te and n+ as input parameters.10,11 The modeling algorithm used the fivecomponent approximation for neutral ground-state species 共HBr/ H / Br/ H2 / Br2兲 and was based on the simultaneous solution of following equations: 共1兲 the equations of chemical kinetics for both neutral and charged species in a steady-state 共dn / dt = 0兲 approximation and 共2兲 the quasineutrality condiJVST A - Vacuum, Surfaces, and Films

227

tions for densities 共ne + nBr− = n+, where n+ ⬇ nHBr+ + nAr+兲 and fluxes 共⌫e ⬇ ⌫HBr+ + ⌫Ar+兲 of charged species. The list of processes taken into account by the model can be found in Ref. 11. The rate coefficients of heterogeneous recombination for H and Br atoms were estimated as kS = 关共⌳2 / D兲 + 共2r / ␥␷T兲兴,14 where D is the effective diffusion coefficient,15,16 ␥ is the recombination probability, ⌳−2 = 共2.405/ r兲2 + 共␲ / l兲2 is the effective diffusion length,16 and ␷T = 共8kBT / ␲m兲1/2. Since the recombination probabilities for both Br and H atoms in the HBr plasma are not known, we used the values recommended for pure Br2 and H2 gases 关␥Br ⬇ 0.2 共Ref. 17兲 and ␥H ⬇ 0.05 共Ref. 18兲兴. For positive ions, we used kS = ␷ / dc, where dc = 0.5rl / 共rhl + lhr兲.15 The ion Bohm velocities ␷ as well as the parameters hl and hr determined by the ion mean free path ␭i are given by the low pressure ␭i ⱕ 共Ti / Te兲共r , l兲 diffusion theory.15,16 For negative ions, we applied kS = 0 due to the presence of negative floating potential on the reactor walls.15 Based on the analysis of the HCl plasma chemistry,19,20 we ignored the influence of the dissociative attachment to the vibrationally excited HBr on the kinetics of negative ions. More modeling details can be found in Ref. 11. III. RESULTS AND DISCUSSION When analyzing the etch mechanism in chemically active plasmas, the question of primary importance is the dominant desorption pathway for reaction products which determines the limiting stage of whole etch process as well as the dependencies of the etch rate on the operating parameters. While the direct data on saturated vapor pressures for the expected reaction products are not available, we can refer to the melting points Tmp for corresponding compounds21 assuming a qualitative correlation of these parameters with the volatility. From Ref. 21, it can be understood that the indium bromides can be related to low-volatile compounds 共Tmp = 290 ° C for InBr and 420 ° C for InBr3兲. In fact, this predetermines the negligible role of thermally activated 共spontaneous兲 desorption at surface temperatures below 100 ° C as well as allows one to infer the ion-stimulated desorption as the main desorption mechanism for indium bromides. At the same time, the volatility of tin bromides depends strongly on their stoichiometry and is much higher for the saturated molecules 共Tmp = 29.1 ° C for SnBr4 and Tmp = 215 ° C for SnBr2兲. In such situation, the limiting stage of the SnO2 etch process depends on the fractional ratios between various types of reaction products. As for the hydrides, the volatility of InH3 共Tmp ⬍ 80 ° C兲 is much lower than that for SnH4 which is a gas even at room temperature 共Tmp = −146 ° C兲. Figure 1 shows the effects of main operating parameters on the behaviors of In2O3 and SnO2 etch rates as well as on the negative dc bias voltage on the substrate. From Fig. 1共a兲, it can be seen that as the Ar mixing ratio increases from 0% to 100% at p = 6 mTorr and W = 700 W, the In2O3 and SnO2 etch rates decrease monotonically in the ranges of 49.7–6.5 and 188.9–33.4 nm/min, respectively, which is also accompanied by a decreasing dc bias voltage 共−Udc = 293– 155 V兲. An increase in input power from 400 to 700

228


0.0

ne -3 11

10

2.5 2.0

0.2

0.4

0.6

0.8

1.0

50

a)

1.5

0

0.0

0.2

400

1.0

11

200 100

450

500

550

600

650

700

10

b)

2.0

0

400

500

600

700

Input power [W]

Input power [W] 3.2

200

8

300

7

SnO2

3.0

100 50

100

In2O3

-Udc [V]

200

6

2.8

8

10

0

6 5 4

2.6

c)

4

Te [eV]

150

10

-3

400

-3

2.2

c)

4

5

6

7

8

9

10

[cm ]

In2O3

2.4

10

40

2.6

3

Density, 10

80

10

Density [cm ]

SnO2

Te [eV]

120

-Udc [V]

Etch rate [nm/min]

0.8

2.8

300

b)

Etch rate [nm/min]

0.6

3.0

160

0

0.4

10

10

Ar fraction in HBr/Ar


0

Density [cm ]

100

In2O3

Te [eV]

150

a)

0

3.0

200

n-

Te n+

250

SnO2

100 50

10

3.5

-Udc [V]

Etch rate [nm/min]

150

12

300

200

228

2

Pressure [mTorr]

Pressure [mTorr]

FIG. 1. SnO2 and In2O3 etch rates as well as the negative dc bias voltage as functions of HBr/Ar mixing ratio 共a兲, input power 共b兲, and pressure 共c兲. The conditions are q = 40 SCCM, p = 6 mTorr, W = 700 W, and Wdc = 200 W 共a兲; 25% Ar, q = 40 SCCM, p = 6 mTorr, and Wdc = 200 W 共b兲; 25% Ar, q = 40 SCCM, W = 700 W, and Wdc = 200 W 共c兲. The lines are to guide the eye only.

FIG. 2. Measured electron temperature and total positive ion density 共symbols兲 as well as model-predicted electron and negative ion densities 共dashed lines兲 as functions of HBr/Ar mixing ratio 共a兲, input power 共b兲, and pressure 共c兲. The conditions are q = 40 SCCM, p = 6 mTorr, W = 700 W, and Wdc = 200 W 共a兲; 25% Ar, q = 40 SCCM, p = 6 mTorr, and Wdc = 200 W 共b兲; 25% Ar, q = 40 SCCM, W = 700 W, and Wdc = 200 W 共c兲. The solid lines are to guide the eye only.

W for the 75% HBr+ 25% Ar gas mixture at p = 6 mTorr 关Fig. 1共b兲兴 results in increasing etch rates 共by 1.8 times for In2O3 and by 2.1 times for SnO2兲, but in decreasing −Udc in the range of 342–275 V. An increase in gas pressure causes a decrease in the etch rates for both materials 共56.5–47.0 nm/ min for In2O3 and 182.7–140.6 nm/min for In2O3 in 75% HBr+ 25% Ar gas mixture at W = 700 W兲 while the negative dc bias voltage keeps a nearly constant value of about 275 V. From the data above, one can see that higher etch rates of SnO2 compared to In2O3 in the HBr-rich plasmas are in reasonable agreement with the volatilities of expected reaction products. This also can be taken as an indirect proof that the chemical etch pathway dominates over physical one in the HBr-rich plasmas. The similar relative changes in both etch rates with variations of operating parameters indicate the same limiting stage for both etch processes, at least for the given set of experimental conditions. Probably, this corresponds to the domination of SnBr2 among the SnO2 etch

products. Also, the differences in the In2O3 and SnO2 etch rates in pure Ar plasma are in good agreement with the differences in the sputtering yields 关0.96 and 0.57 atom/ion for SnO2 and In2O3, respectively, for Ar+ ions at 600 eV 共Ref. 22兲兴. In order to get the additional information on the In2O3 and SnO2 etch mechanisms, the data on densities and fluxes of plasma active species are needed. From plasma diagnostics by Langmuir probes, it was found that an increase in Ar mixing ratio results in increasing electron temperature 共Te = 2.88– 3.42 eV for 0%–100% Ar at p = 6 mTorr and W = 700 W兲 and total positive ion density 共n+ = 6.22 ⫻ 1010 – 6.06⫻ 1011 cm−3兲关Fig. 2共a兲兴 that provides the similar trends for both ion saturated current density Jis = 1.5– 15.7 mA/ cm2 and total flux of positive ions ⌫+ ⬇ hl兺n+,j␷ j = 6.94⫻ 1015 – 1.06⫻ 1017 cm−2 s−1. The reasons are the decreasing electron energy losses in the medium part of the EEDF and the increasing total ionization rate. Accordingly, the model-predicted electron density also increases

J. Vac. Sci. Technol. A, Vol. 28, No. 2, Mar/Apr 2010


JVST A - Vacuum, Surfaces, and Films

229

Br

-3

Density [cm ]

HBr

H2

13

10

Br2

H

a)

12

10

0.0

0.2

0.4

0.6

0.8

1.0


H2

-3

Density [cm ]

HBr

13

10

Br

Br2

H b)

12

10

400

500

600

700

Input power [W]

HBr -3

from 2.06⫻ 1010 to 6.06⫻ 1011 cm−3 for 0%–100% Ar while the density of Br decreases monotonically in the range of 共4.16– 2.60兲 ⫻ 1010 cm−3 for 0%–80% Ar 共following the rate of dissociative attachment兲 that corresponds to nBr− / ne = 2.0– 0.2. An increase in input power in the range of 400–700 W 关Fig. 2共b兲兴 causes a weak increase in Te 共2.75–2.95 eV for 25% Ar and p = 6 mTorr兲 and a twofold increase in total positive ion density 关n+ = 共3.73– 7.33兲 ⫻ 1010 cm−3兴, ion saturated current density 共Jis = 0.94– 1.85 mA/ cm2兲, and ion flux 关⌫+ = 共8.16– 8.58兲 ⫻ 1015 cm−2 s−1兴. The total ionization rate increases faster than the attachment rate because the last is retarded by decreasing attachment rate coefficient 共due to decreasing fraction of low energy electrons in the EEDF which are responsible for this process兲 and the HBr density 共due to increasing dissociation frequency and the dissociation degree兲. That is why the model-predicted electron density increases with increasing input power by more than 3.5 times 关ne = 共8.48⫻ 109 – 3.28兲 ⫻ 1010 cm−3 for 400–700 W兴 that results in decreasing nBr− / ne 共3.4–1.2兲 in spite of increasing nBr− 关共2.88– 4.04兲 ⫻ 1010 cm−3兴. An increase in gas pressure from 4 to 10 mTorr 关Fig. 2共c兲兴 results in deceasing Te 共3.05–2.80 eV for 25% Ar and W = 700 W兲 due to a well-known effect of increasing both electron collision frequency and electron energy loss while the total density of positive keeps a near-to-constant values of about 7.35⫻ 1010 cm−3 that results from the similar behavior of the total ionization rate. At the same time, both Jis and ⌫+ show a decreasing tendencies 关1.41– 1.34 mA/ cm2 and 共8.81– 8.38兲 ⫻ 1015 cm−2 s−1, respectively兴 because of decreasing ion mean free path and Bohm velocity. Also, an increase in the rate of dissociative attachment with increasing gas pressure results in decreasing model-predicted electron density 关ne = 共4.09– 2.41兲 ⫻ 1010 cm−3 for 4–10 mTorr兴 as well as in increasing both nBr− 关共3.30– 4.94兲 ⫻ 1010 cm−3兴 and nBr− / ne 共0.81–2.05兲. As for the kinetics of neutral species, the dominant formation mechanism for Br atoms is the direct electron-impact dissociation HBr+ e → H + Br+ e 共R1兲 while the contribution of dissociative attachment HBr+ e → H + Br− 共R2兲 is much lower due to the lower cross section 关k1 / k2 = 4.6– 8.7 and k2 / 共k1 + k2兲 = 0.17– 0.10 for 0%–100% Ar at p = 6 mTorr and W = 700 W兴. Although the overall HBr decomposition rate in R1 and R2 represents a near-to-equal source of H and Br atoms, the Br atom density is noticeably higher 关nBr / nH = 11.1– 4.0 for 0%–80% Ar, see Fig. 3共a兲兴. The reason is the fast decay of H atoms in atom-molecular reactions H + HBr → H2 + Br 共R3 : k3 = 6.5⫻ 10−12 cm3 / s兲 and H + Br2 → HBr + Br 共R4 : k4 = 6.0⫻ 10−11 cm3 / s兲 while the reverse reactions H2 + Br→ H + HBr 共R5兲 and HBr+ Br→ H + Br2 共R6兲 have the much lower rate coefficients 共k5 = 4.5⫻ 10−23 cm3 / s and k6 = 2.5⫻ 10−39 cm3 / s兲. Also, since R3 and R4 additionally produce H2 and HBr, we obtain a relatively high densities of H2 compared to those for Br2 共nBr2 / nH2 = 0.54– 0.28 for 0%– 80% Ar兲 as well as the relatively low dissociation degree of HBr 关nHBr / 共nH + nBr兲 = 2.01– 0.81 for 0%–80% Ar兴. As it can be seen from Fig. 3共a兲, an increase in Ar mixing ratio causes

Density [cm ]

229

Br H2

13

10

Br2 H c)

12

10

4

6

8

10

Pressure [mTorr]

FIG. 3. Model-predicted densities of neutral species as functions of HBr/Ar mixing ratio 共a兲, input power 共b兲, and pressure 共c兲. The conditions are q = 40 SCCM, p = 6 mTorr, W = 700 W, and Wdc = 200 W 共a兲; 25% Ar, q = 40 SCCM, p = 6 mTorr, and Wdc = 200 W 共b兲; 25% Ar, q = 40 SCCM, W = 700 W, and Wdc = 200 W 共c兲.

the monotonic decrease in the densities of all HBr-related species, except the H atoms. The nonmonotonic behavior of the H atom density results from rapid decrease in the rates of R3 and R4. An increase in input power in the range of 400– 700 W 关Fig. 3共b兲兴 is accompanied by decreasing densities of the molecular plasma components 关nHBr = 共4.26– 3.15兲 ⫻ 1013 cm−3, nH2 = 共1.75– 1.42兲 ⫻ 1013 cm−3 and nBr2 = 共1.25⫻ 1013兲 – 共6.55⫻ 1012兲 cm−3 for 25% Ar and p = 6 mTorr兴, but by increasing densities of H and Br atoms because of increasing electron-impact dissociation rates. Accordingly, an increasing HBr dissociation degree nHBr / 共nH + nBr兲 = 5.38– 1.61 takes place. An increase in gas pressure 关Fig. 3共c兲兴 causes increasing densities for HBr, H, Br, and Br2 while the density of H atoms decreases. The last effect is associated with increasing rates of R3 and R4. In order to analyze SnO2 and In2O3 etch mechanisms, one should refer to the relationships between plasma composition


J. Vac. Sci. Technol. A, Vol. 28, No. 2, Mar/Apr 2010

230

3

5 *H

2

4

3

*Br

3

1 a)

0

0.0

0.2

0.4

-2

6

1/2

*Br+ *H

0.6

0.8

2

1/2

4

miH Jis, 10 [eV mAcm ]

17

-2 -1

Neutral flux, 10 [cm s ]

7

1

1.0

-2

4


2

2

0

1

*Br + *H

1

*Br

*H

400

b)

500

600

700

3

1/2

3

1/2

Neutral flux, 10

17

-2 -1

[cm s ]


0

Input power [W]

-2

*Br + *H

3

3

2.5

1/2

4


*H

2

1

*Br

4

1/2

-2 -1

[cm s ]

3.0

17

and the etch rate given by the theory of free surface sites 共Langmuir–Hinshelwood theory兲.23–27 In our case, when two types of chemically active species with the fluxes of ⌫Br and ⌫H can be involved in the etch process, the overall rate of the steady-state chemical reaction 共in fact, the flux of reaction products leaving the surface兲 can be expressed as ␥Br⌫Br + ␥H⌫H, where ␥ = 共1 − ␪兲s0␦ is the reaction probability for the given type of active species, ␪ is the fraction of the surface covered by the reaction products, s0 is the sticking probability for active species, and ␦Br is the stoichiometry coefficient for reaction products. The variations of physical etch pathway, including both sputtering of native surface and lowvolatile reaction products, follows the parameter Y s⌫+, where Y s is the sputtering yield for low-volatile reaction products 共in other words, the yield of ion-stimulated desorption兲. Assuming that Y s is proportional to the momentum transferred from the incident ion to the etched surface,24,25 and ⌫+ ⬃ Jis, the relative behavior of the physical etch pathway can be simply characterized by mi␧1/2Jis, where mi is the effective ion mass, and ␧ is the incident ion energy determined by the sum of floating potential and the negative dc bias voltage −Udc applied to the substrate. From Fig. 4共a兲, it can be seen that although the condition nBr Ⰷ nH takes place, the fluxes of these species ⌫ ⬇ 0.25n冑8kBT / ␲m are quite close because the lower density of H atoms is overcompensated by the higher thermal velocity due to the lower mass. The shift in the HBr/Ar mixing ratio toward Ar-rich plasmas suppresses the chemical reaction and accelerates the physical etch pathway 共mi␧1/2Jis = 2430– 6340 eV1/2 mA cm−2 for 0%–100% Ar at p = 6 mTorr and W = 700 W兲 because a decrease in mi␧1/2 共mi = 105– 40 and ␧ = 311– 176 eV for 0%–100% Ar兲 is overcompensated by increasing Jis. From the comparison of Figs. 1共a兲 and 4共a兲, it can be easily seen that the behaviors of both SnO2 and In2O3 etch rates are in disagreement with mi␧1/2Jis, ⌫H, and ⌫Br + ⌫H, but follows ⌫Br only. The opposite behaviors of the measured etch rates and ion energy flux mi␧1/2Jis as well as the much lower etch rates in pure Ar plasma compared with pure HBr plasma can be taken as the proof that the physical sputtering of the native oxide surfaces does not contribute the total etch rate in HBr-rich plasmas. Also, this allows one to infer that for the given set of input parameters, the etch processes for both materials are not limited by the ion-surface interaction kinetics and appear, probably, in the reaction rate limited or in the transitional etch regime with the Br atoms as the main chemically active species. The last conclusion might be expected when the following reasons are taken into account. From Ref. 21, it can be understood that both In–Br and Sn–Br bonds are stronger than the In–O and Sn–O ones 共414⫾ 21, ⬎552, ⬍320.1, and 531.8⫾ 12.6 kJ/ mol, respectively兲. Oppositely, the In–H 共243.1 kJ/mol兲 and Sn–H 共264⫾ 17 kJ/ mol兲 bonds are weaker than the oxide ones. That is why the bromination of both etched materials is more favorable than the hydration because the last needs the additional energy to break the oxide bonds.

Neutral flux, 10

230

c)

6

8

10

2.0

Pressure [mTorr]

FIG. 4. Model-predicted fluxes of neutral species 共dashed lines兲 and the parameter mi␧1/2Jis characterizing the ion energy flux 共solid line+ symbols兲 as functions of HBr/Ar mixing ratio 共a兲, input power 共b兲, and pressure 共c兲. The conditions are q = 40 SCCM, p = 6 mTorr, W = 700 W, and Wdc = 200 W 共a兲; 25% Ar, q = 40 SCCM, p = 6 mTorr, and Wdc = 200 W 共b兲; 25% Ar, q = 40 SCCM, W = 700 W, and Wdc = 200 W 共c兲.

Form the comparison of Figs. 1共b兲 and 4共b兲, it can be understood that an increase in both SnO2 and In2O3 etch rates is slower than that for ⌫Br but faster compared to the change in mi␧1/2Jis. Also, an increase in gas pressure, which is accompanied by the increasing ⌫Br, results in the opposite effect in both SnO2 and In2O3 etch rates that formally correspond to the behavior of mi␧1/2Jis. All these are in disagreement with the pure reaction-rate-limited etch regime and point out on the transitional regime where the etch rate is controlled by neutral and ion fluxes simultaneously. In this regime, since the etched surface is partially covered by the low-volatile reaction products, a decrease in mi␧1/2Jis with increasing gas pressure lowers the fraction of free surface 共1 − ␪兲 and thus, lowers the overall etch rate in spite of increasing ⌫Br. IV. CONCLUSION In this work, we investigated the etch characteristics and mechanisms for In2O3 and SnO2 thin films in the HBr-based

231


inductively coupled plasma. The etch rates were measured in the range of 0%–100% Ar 400–700 W of input power and 4–10 mTorr of gas pressure fixed bias power 共200 W兲 and gas flow rate 共40 SCCM兲. Plasma diagnostics by double Langmuir probe and a global 共zero-dimensional兲 plasma model provided the data on plasma parameters as well as on densities and fluxes of plasma active species on the etched surface. The model-based analysis of etch kinetics allows one to infer that in the given range of experimental conditions, both SnO2 and In2O3 etch mechanisms correspond to the transitional regime of ion-assisted chemical reaction with Br atoms as the main chemically active species. The low reaction probability for H atom is, probably, connected with unfavorable energy balances for the hydration reactions. ACKNOWLEDGMENTS “This work was supported by the National Research Foundation of Korea 共NRF兲 grant funded by the Korea government 共MEST兲共No. 2009-0085863兲.” 1

A. B. Djurišić, C. Y. Kwong, P. C. Chui, and W. K. Chan, J. Appl. Phys. 93, 5472 共2003兲. 2 H. Kupfer, R. Kleinhempel, B. Graffel, Th. Welzel, Th. Dunger, F. Richter, W.-M. Gnehr, and T. Kopte, Surf. Coat. Technol. 201, 3964 共2006兲. 3 C. H. Jeong, J. H. Lee, J. T. Lim, M. S. Kim, and G. Y. Yeom, Surf. Coat. Technol. 201, 5012 共2007兲. 4 G. Bradshaw and A. J. Hughes, Thin Solid Films 33, L5 共1976兲. 5 T. Ratcheva and M. Nanova, Thin Solid Films 141, L87 共1986兲. 6 L. Francioso, M. Russo, A. M. Taurino, and P. Siciliano, Sens. Actuators B 119, 159 共2006兲.

JVST A - Vacuum, Surfaces, and Films

7

231

E. S. Braga, A. P. Mammana, and C. I. Z. Mammana, Thin Solid Films 73, L5 共1980兲. 8 D. Vaufrey, M. Ben Khalifa, M. P. Besland, C. Sandu, M. G. Blanchin, V. Teodorescu, J. A. Roger, and J. Tardy, Synth. Met. 127, 207 共2002兲. 9 I. H. Yang, Y. Lee, J. N. Jang, and M. Hong, Thin Solid Films 517, 4165 共2009兲. 10 M. Kim, N.-K. Min, S. J. Yun, H. W. Lee, A. Efremov, and K.-H. Kwon, Microelectron. Eng. 85, 348 共2008兲. 11 A. Efremov, B. G. Choi, S. Nahm, H. W. Lee, N.-K. Min, and K.-H. Kwon, J. Korean Phys. Soc. 52, 48 共2008兲. 12 E. O. Johnson and L. Malter, Phys. Rev. 80, 58 共1950兲. 13 M. Sugavara, Plasma Etching. Fundamentals and Applications 共Oxford University Press, New York, 1998兲. 14 P. J. Chantry, J. Appl. Phys. 62, 1141 共1987兲. 15 M. A. Lieberman and A. J. Lichtenberg, Principles of Plasma Discharges and Materials Processing 共Wiley, New York, 1994兲. 16 C. Lee and M. A. Lieberman, J. Vac. Sci. Technol. A 13, 368 共1995兲. 17 G. P. Kota, J. W. Coburn, and D. B. Graves, J. Vac. Sci. Technol. A 17, 282 共1999兲. 18 B. J. Wood and H. Wise, J. Phys. Chem. 65, 1976 共1961兲. 19 A. M. Efremov, G. H. Kim, D. I. Balashov, and C.-I. Kim, Vacuum 81, 244 共2006兲. 20 A. M. Efremov and V. I. Svetsov, High Temp. 44, 189 共2006兲. 21 D. R. Lide, Handbook of Chemistry and Physics 共CRC, New York, 1998兲. 22 B. Chapman, Glow Discharge Processes: Sputtering and Plasma Etching 共Wiley, New York, 1980兲. 23 D. C. Gray, I. Tepermeister, and H. H. Sawin, J. Vac. Sci. Technol. B 11, 1243 共1993兲. 24 H. W. Winters and J. W. Coburn, Surf. Sci. Rep. 14, 162 共1992兲. 25 C. Lee, D. B. Graves, and M. A. Lieberman, Plasma Chem. Plasma Process. 16, 99 共1996兲. 26 W. Jin, S. A. Vitale, and H. H. Sawin, J. Vac. Sci. Technol. A 20, 2106 共2002兲. 27 A. M. Efremov, D. P. Kim, and C.-I. Kim, IEEE Trans. Plasma Sci. 32, 1344 共2004兲.