02 plasma

Reactive ion etching deposited aluminum Jae-Whan mechanism of plasma enhanced oxide film in CF4/02 plasma Kim, Yong-Chun Kim,al and Won-Jong chemi...
6 downloads 0 Views 576KB Size
Reactive ion etching deposited aluminum Jae-Whan

mechanism of plasma enhanced oxide film in CF4/02 plasma

Kim, Yong-Chun

Kim,al and Won-Jong

chemically

vapor

Lee

Department of Materials Science and Engineering, Korea Advanced Institute of Science and Technology, Taejon 305701, Korea

(Received 6 September 1994; accepted for publication 6 April 1995) Aluminum oxide film prepared by plasma enhanced chemical vapor deposition (PECVD) is one of the promising candidates for an etch mask or an etch barrier material in very large scale integrated fabrication. We have investigated the reactive ion etching mechanism of the PECVD aluminum oxide films in the CFJO, plasma. The dependences of the aluminum oxide etch rate on the atomic fluorine concentration and the incident particle bombardment energy are studied at various etching conditions. The etch products and their depth distribution are also determined by analyzing the surface layer of the etched aluminum oxide films with Auger electron spectroscopy. The CFdO, plasma fluorinates the aluminum oxide surface layer through the particle bombardment activated reaction, producing etch products in the form of AlFs or AlO,F,, which is nonvolatile but has a higher sputtering yield than aluminum oxide. The reactive ion etching of aluminum oxide proceeds by the formation and the sputter removal of the etch product. The etch rate strongly depends on the particle bombardment energy because it determines not only the formation rate but also the removal rate of the etch products. However, the atomic fluorine concentration has little effect on the etch rate because the consumption rate of fluorine radicals is so low that they are always superfluous under any experimental conditions. 0 1995 American Institute of Physics.

I. INTRODUCTION

II. EXPERIMENT

Aluminum oxide offers high chemical stability, high radiation resistance, good thermal conductivity and very low permeability to alkali ions and other impurities.‘~2 These properties make aluminum oxide attractive in microeiectronits as well as in various other applications.3-” For multilevel interconnection in very large scale integrated (VLSI) fabrication, an etch barrier having a high etching selectivity in the dry etch plasma is required. Aluminum oxide filmy3 prepared by the plasma enhanced chemical vapor deposition (PECVD) method, has been reported to be one of the promising candidates for an etch mask or an etch barrier material in VLSI fabrication. In order to utilize this material as an etch mask or barrier, thorough studies on the dry etching properties of the PECVD aluminum oxide film are needed. To the best of our knowledge, no systematic research has been performed on the dry etching mechanism of the PECVD aluminum oxide film. In this research, ahuninum oxide films were prepared at a low temperature of 220 “C by the PECVD method and their reactive ion etching (RIE) mechanisms were studied in the CFd/OZ plasma. In order to analyze the physical and the chemical aspects inthe etching process, not only the etch rate of the aluminum oxide film, but also the atomic fluorine concentration [F] and the plasma sheath voltage Vsh were monitored at various etching conditions. The etch product and its depth distribution were also determined by analyzing the surface layer of the etched films with surface spectroscopy.

Aluminum oxide films were prepared by the PECVD method on 50 s1 cm p-type (100) Si wafer. The deposition conditions were as follows: pressure, 130 Pa; deposition temperature, 220 “C; radio frequency (rf) power, 0.05 W/cm’; reaction gases, 1.5 seem trimethylaluminum (TMA), 4.5 seem NZO and 84 seem He; deposition time, 10 min. The thickness and the refractive index of the deposited film were measured by an ellipsometer of X=632.8 run. The deposited film has a thickness of 200 rmr and a refractive index of 1.50. The film was microcrystalline y-A120s with an O/Al concentration ratio of 1.63. The compositional and the structural analysis of the PECVD aluminum oxide film was reported in detail elsewhere.’ Figure 1 shows the schematic diagram of the planar-type RIE system. The specimen, coated with an aluminum oxide film, was put on the rf powered electrode, where a constant temperature of 20 “C was kept by means of water cooling. The opposite electrode and the chamber wall were grounded. The electrodes were made of stainless steel. The reaction gases used were CF, and O2 of which the flow rates were adjusted by mass flow controllers. The RIE experimental parameters were as follows: CF,+O, flow rate,. 15 seem; O2 percent, O%-40%; system pressure, 8-40 Pa; 13.56 MHz rf power, 50-400 W (0.33-2.6 W cm-‘); electrode spacing, 45 mm. The etch rate of the aluminum oxide film was determined by measuring the thickness change with an ellipsometer. Atomic fluorine (F), CF, radical and CF;’ ion are the possible etching species generated in CF, plasma. If the CF, radical or ion is an important etching species, carbon must remain at the aluminum oxide surface in a form of fluorocarbon after etching in CF, plasma. By using an x-ray pho-

‘brrent address: Research and Development Center, Samsung ElectcoMechanics Co., Suwon, Kyungki-do 441-743, Korea. J. Appl. Phys. 78 (3), 1 August 1995

0021-8979/95/78(3)/2045/5/$6.00

Q 1995 American Institute of Physics

2045

Downloaded 21 Sep 2011 to 143.248.115.181. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

E.R.

Vsh

[ F1

(v)

(lp’3cm-s)

(nm/min)

11

1

pumpingout

pumpingout I

GenEator

Oscilloscope

II computer

8

800

6

600

4

400

2

200

-10

0 FIG. 1. Schematic diagram of the reactive ion etching system.

0 0

10

Oxygen toelectron spectroscopy (XI’S), we analyzed the surface layer of the aluminum oxide film etched in CF, plasma at 100 W. No sign of fluorocarbon was observed at the surface layer, while large amounts of fluorine atoms were found to be incorporated in the film. Therefore, it was concluded that atomic fluorine F is the main etching species for the aluminum oxide film in the fluorine-based plasma. It has been known that Si and SiO, are also etched mainly by F in the fluorine-based plasma.“-t6 In the present research, the atomic fluorine concentration in the CF,/O, plasma was measured by Ar actinometry,‘3-‘6 which is one of the spectral analysis techniques. Ar gas of 0.5 seem was injected into the CF,/O, plasma system and the optical emissions from the excited states of the F and Ar gas atoms were monitored with a monochrometer. For the quantitative analysis of the atomic fluorine concentration [F], the following equation proposed by Lee and Chen17 was used. [F]=0.68(ZnIZ&[Ar], where [AI-] is the atomic concentration of Ar, IF is the intensity of 703.7 nm F emission peak and ZAr is the intensity of 751.3 nm Ar emission peak. Another factor exerting an important effect on the etch rate of the aluminum oxide film is the energy of particles bombarding the sample surface. In the absence of collisions, an ion would accelerate across the sheath region and bombard the sample with an energy equivalent to the sheath voltage. But the ion usually does collide with neutrals in the sheath region. The collisions attenuate the ion energy and increase the neutral energy so that not only ions but also neutrals bombard the sample. Therefore, the sheath field is such as to furnish energies to the bombarding particles, and the average bombarding energy increases with increasing sheath voltage. The plasma sheath voltage V,, at the rf powered electrode is given by where Vppis the rf peak-to-peak voltage and Vdcis the direct current (dc) self bias. lrslsh was obtained by measuring Vppand V dc with an oscilloscope in order to determine the influences of particle bombardment energy on the aluminum oxide etching. The surface layer of the etched aluminum oxide film was analyzed with Auger electron spectroscopy (AES) to determine the etch products and their depth distribution. Auger 2046

J. Appl. Phys., Vol. 78, No. 3, 1 August 1995

-20

20

30

-0

40

Percent

FIG. 2. Aluminum oxide etch rate (E.R.), [F] and V, as a function of oxygen percent at 13 Pa and 100 W in CFJ02 plasma.

electron spectra were obtained under the following operating conditions: incident electron beam, 5 keV, electron current, 250 nA; modulation voltage, 4 V, specimen tilt angle from the surface normal to avoid charging effects, 50”; analysis area, 100X 100 ,um2. In order to obtain the depth profiles of the elements (Al, 0, F, C) of concern, the etched aluminum oxide film was sputtered with an erosion rate of 2 nm/min. No preferred sputtering was observed in the aluminum oxide film, as reported previously.18T’g 111.RESULTS

AND DISCUSSION

It has been known that the addition of O2 to CF, plasma increases [F], since 0 or O2 react with CF, radicals to liberate F. In order to study the effect of [F] on the etch rate of the PECVD aluminum oxide film, the films were etched in CF,/O, plasma at various O2 percent. The total flow rate, the system pressure and the rf power were fixed at 15 seem, 13 Pa and 100 W, respectively. Figure 2 shows the etch rate, [F] and V,, as a function of O2 percent. [F] for the CF4/02(20%) plasma is about 8X1013 cme3 that is 20 times greater than the value (0.38X10i3 cmw3) obtained without 0, addition. The aluminum oxide etch rate, however, does not seem to be affected by the huge increase of [F], which is quite a contrast to the case of Si etching,i3-I6 where the etch rate increases tens of times as 0, is added to the CF, plasma. On the other hand V,, changes a little with varying 0, percent. Figure 3 shows the dependences of the etch rate, [F] and vs,, on the system pressure in CF,/02(20%) plasma. The rf power was fixed at 100 W (0.65 W cm-‘). As the pressure increases from 8 to 40 Pa, [F] increases 20 times while the etch rate of the aluminum oxide film decreases from 2.6 to 0.4 nm/min. On the other hand, Vshdecreases from 470 to 170 V as the pressure increases. The real bombarding energy of the incident particles decreases even more rapidly with increasing pressure than the V,, because the probability of collision in the sheath is greater at the higher pressure. Therefore, we considered that the reduction in etch rate with increasing pressure is due to the decrease in the energy of particles bombarding the aluminum oxide film surface. Figure 4 shows the etch rate, LF] and V,, as a function of Kim, Kim, and Lee

Downloaded 21 Sep 2011 to 143.248.115.181. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

Vsh

E.R.

E.R. (nm/min)

[F]

01) (ld”cm3

(nm/min)

, 1600

16 T

Vsh 03

16

1600

14

1400

12

1200

50

50

40

lo-

800 1

30

/

:I Jd:::

cF7

(ld3cm-3

Vsh

k

a6-

20

40

30 - 800 20

-600

10 10 200 0

0-o 0

0

0 10

20

30

40

100

200

300

RF Power

400

500

(W)

PresSure(Pa) FIG. 3. Aluminum oxide etch rate (E.R.), [F] and V, as a function of system pressure at 100 W in CF,/O,(20%) plasma.

rf power at a constant pressure of 13 Pa in CF,/O,(20%) plasma. As the rf power increases from 50 to 400 W, [F] and V,, increases 2 times and 4.5 times, respectively, while the aluminum oxide etch rate increases 40 times. Since the doubling of [F] will not affect the etch rate, the large increase in etch rate is attributed to the rise of the particle bombardment energy caused by the increased sheath field. From the previous experiments, we found that the etch rate of the aluminum oxide film is not controlled by the atomic fluorine concentration, but by the particle bombardment energy. In order to examine whether or not the reactive ion etching of the aluminum oxide film in CF,/O, plasma is conducted simply by the physical sputtering involving momentum transfers between the incident particles and the surface atoms, the aluminum oxide etch rate in an inert Ar plasma was measured as a function of rf power. The result is given as a broken line in Fig. 4. The yq, values in the Ar plasma were almost the same as those in the CFJO, plasma. -4s the rf power increases, the etch rate in Ar plasma also increases linearly. But its absolute values are much smaller compared with those in CF4/0, plasma and the difference between the etch rates in CFJO, and Ar plasmas becomes enlarged as the rf power (so V,,j increases. This result implies that the existence of the fluorine radical is important in reactive ion etching of the aluminum oxide film even though its concentration has little effect on the etch rate. Therefore, it was concluded that the etching of the aluminum oxide film in CF,/O, plasma is not conducted simply by the physical sputtering, but involves the formation of etch products through the chemical reactions with fluorine radicals. J. Appl. Phys., Vol. 78, No. 3, 1 August 1995

FIG. 4. Aluminum oxide etch rate (E.R.), [F] and V,, as a function of rf power at 13 Pa in CFJO,(20%) plasma. The etch rate in Ar plasma is also plotted as a broken line for the purpose of comparison.

The etch rate of the aluminum oxide film was measured by varying the electrode area covered with silicon wafers. The etch rate of the silicon was also measured for the purpose of comparison. The results are given in Fig. 5. The general etch rate of aluminum oxide is much smaller than that of silicon by two orders of magnitude. As the silicon loaded area increases from 0% to 75% of the electrode area, the etch rate of silicon decreases rapidly while that of alumi-

16

1600

1000 & .800 cn % 600

2

$4

400

2 XI

22 "2

200

ilLi

V

W 0

0

0

20

Si Loaded

40

60

Area

80

(%)

FIG. 5. Etch rates of the aluminum oxide and the silicon as a function of the Si loading percent at 13 Pa and 100 W in CFh/O,(ZO%) plasma. Kim, Kim, and Lee

2047

Downloaded 21 Sep 2011 to 143.248.115.181. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

60 50 40

0.0

0.4

0.8

Sputtering

1.2

T ime

1.6

10

10

20

05

10

G

2.0

(min)

num oxide increases slightly. The changes in [F] and V,, were also measured with increasing silicon loading from 0 % to 75%: [F] decreased by 70% and V,, increased by 20%. Since the,etch product of silicon in fluorine-based plasma is SiF,, which is volatile at room temperature, the etch rate of silicon is very high, so is the consumption of the etching speciesF. As the area of silicon to be etched becomes larger, the supply of fluorine radicals becomes insufficient, resulting in an inverse relationship between the etch rate and the silicon-loaded area. On the other hand, the etch rate of aluminum oxide film is very small, so is the consumption of the fluorine radicals. Therefore, the fluorine radicals were always superfluous under any conditions during the present experi-’ ments, which explains why the etch rate of the aluminum oxide film was not affected by [F]. The slight increase in the aluminum oxide etch rate with increasing silicon loading is certainly due to the slight increase of V,,. The surface layer of the aluminum oxide film etched in CF,/O,(20%) was analyzed using AES in conjunction with ion sputtering to investigate the etch products and their depth distribution. Auger peak-to-peak heights of Al,, (1378 eV), OKLL (503 eV) and F KLL (647 eV) were evaluated from AES differential spectra and the corresponding atomic concentration were calculated. The sputtering rate of the bulk aluminum oxide film was 2 nm/min. However, the fluorinated surface layer is expected to have a higher sputtering rate, as will be discussed later. Figure 6 shows the atomic concentration depth profiles of Al, 0 and F for the aluminum oxide film reactively ion etched at 100 W and 13 Pa for 10 mm. As the ion-sputtering proceeds, the F concentration decreaseswith a similar rate as the 0 concentration increases, indicating that F substitutes 0 to the depth of about one minute’s sputtering. The atomic concentrations of F and 0 at the “surface” (i.e., at zero sputtering time) is about 52% and 18% respectively. The term “surface” does not mean the real surface, but the surface layer with a thickness of l-2 nm because the samJ. Appl. Phys., Vol. 78, No. 3, 1 August 1995

15

& 0

FIG. 6. AES sputtered depth profile for the aluminum oxide film reactivbly ion etched at 13 Pa and 100 W in CFd02(20%) plasma. The sputtering rate of the bulk aluminum oxide film is 2 nmImin. However, the fluorinated surface layer is expected to have a higher sputtering rate.

2048

,.cj

zs 0 *s 0 -2

30

0

0

55

5

5

0

0 0.0

0.4

0.8

Sputtering

1.2

T ime

1.6

2.0

(min)

FIG. 7. AES sputtered depth profiles of F and C concentrations for the aluminum oxide film reactively ion etched at various rf power at 13 Pa in CF,/Os(20%) plasma. The sputtering rate of the bulk aluminum oxide film is 2 nm/min. However, the fluorinated surface layer is expected to have a higher sputtering rate.

pling depth of the Auger signal is about 1-2 nm. Consequently, the F concentration at the surface monolayer is expected to be higher than 52%, while the 0 concentration is expected to be smaller than 18%. Therefore, we may consider that the surface monolayer of the aluminum oxide film is fully fluorinated in the form of AlF, and the subsurface layer is partially fluorinated to the depth of a few nm in the form of AlOxFr . The etch products, both AIF, and AIOxFy , are nonvolatile at room temperature, and thus a physical sputtering by energetic particles is needed to remove them. This explains why the etch rate of the aluminum oxide film is so small and why it is so sensitively dependent on the particle bombardment energy. The etch rate of the aluminum oxide film is higher in fluorine-based plasma than in Ar plasma because the sputtering yield of the etch products (AIF, or AlOxF,) is higher than that of the aluminum oxide. Figure 7 shows the AELSdepth profiles of F for the aluminum oxide films reactively ion etched at various rf powers ranging from 50 to 400 W . As the rf power increases, aluminum oxide becomes further fluorinated: the F concentration at the surface layer becomes higher and the fluorinated layer becomes thicker to several nm with increasing V,,. This indicates that the formation rate of the etch products increases with increasing rf power, which is probably because the higher particle bombardment energy promotes the fluorination reaction by helping to break the strong Al-O bonds. The broken line in Fig. 8 representsthe carbon concentration profile for the film etched at rf power of 250 W , showing that carbon is penetrated to the depth of a few nm. At rf powers below 100 W , carbon was hardly detected in the etched film. It appears that the high impact energy gained from the high rf power caused the CF,f ions to dissociate, depositing carbon on the aluminum oxide surface. The reactive ion etching of the aluminum oxide film in CF,/Oz plasma is conducted by the formation of etch prodKim, Kim, and Lee

Downloaded 21 Sep 2011 to 143.248.115.181. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

ucts and by the consecutive sputter-removal of them with energetic particle bombardment. The etch products are in the form of AlFs or AIOxFy , which are nonvolatile at room temperature, but have a higher sputtering yield than aluminum oxide. The sputter rate of the material usually increases linearly with the bombardment energy of incident particles.” The AES analysis showed that the formation of the etch products is also promoted by the bombardment of energetic particles. IV. CONCLUSIONS Reactive ion etching mechanism of PECVD aluminum oxide film was studied in a fluorine-based plasma. The reactive ion etching of the aluminum oxide film is conducted by the formation of etch products and by the consecutive removal of them. The CFJO, dry etch plasma fluorinates the aluminum oxide surface, producing etch products in the form of AlFs (at surface) or AIOxFy (in the subsurface layer). Since the etch products are nonvolatile at room temperature, the reactive ion etching of the aluminum oxide film proceeds through the sputter removal by energetic particle bombardment. The etch products have a higher sputtering yield than aluminum oxide, which led to a higher etch rate in fluorinebased plasma than in Ar plasma. However, the aluminum oxide etch rate was little affected by the atomic fluorine concentration as far as fluorine radicals are supplied. It is because the consumption rate of fluorine radicals is so low that the radicals are always superfluous under any experimental conditions. Energetic particle bombardment activates the fluorination reaction, enhancing the formation rate of the etch products. The reactive ion etch rate of the aluminum oxide film strongly depends on the particle bombardment energy because it determines not only the removal rate of the etch products but also their formation rate by activating chemical reactions.

J. Appl. Phys., Vol. 78, No. 3, 1 August 1995

This work has been supported by the Korea Science and Engineering Foundation through the Research Center for Thin Film Fabrication and Crystal Growing of Advanced Materials.

’K. H. Zaininger and A. S. Waxman, IEEE Trans. Electron Devices 16,333 (1969). “D. A. Abbott and T. I. Kamins, Solid State Electron. 13, 565 (1970). 3Y. Shinoda and T. Kobayashi, .I. Appl. Phys. 52, 6386 (1984). 4J. Tsyjide, S. Nakamnra, and Y. Ikushima, J. Electrochem. Sot. 117, 703 (1970). ‘B. E. Deal and J. M. Early, J. Electrochem. Sot. 126, 20 (1979). %Z. Nylander, M. Argarth, and 0. Svensson, J. Appl. Phys. 56, 177 (1984). 7C. J. Kang, Y. C. Kim, C. 0. Park, W. J. Lee, and J. S. Chun, High Peqbmance Ceramic Films and Coatings, edited by P. Vincenzini (Elsevier, Amsterdam, 1991), p. 391. ‘5. W. Kim and W. J. Lee, J. Korean Appl. Phys. 7,289 (1994). 9Y. C. Kim, H. H. Park, J. S. Chun, and W. J. Lee, Thin Solid Films 237.57 (1994). “C. J. Mogab, J. Electrochem. Sot. 124, 1262 (1977). “C. J. Mogab, A. C. Adams, and D. L. Flamm, J. Appl. Phys. 49, 3796 (1978). “5. W. Butterbaugh, D. C. Gray, and H. H. Sawin, J. Vat. Sci. Technol. B 9, 1461 (1991). 13J. W. Coburn and M. Chen, J. Appl. Phys. 51, 3134 (1980). 14R A. Gottscho and V. M. Donnelly, J. Appl. Phys. 56, 245 (1984). t5 V. M Donnelly, D. L. Flamm, W. C. Dautremont-Smith, and D. J. Werder, J.‘Appl. Phys. 55, 242 (1984). 16A. D. Richards, B. E. Thompson, K. D. Allen, and H. H. Sawin, J. Appl. Phys. 62, 792 (1987). r7Y. H. Lee and M. M. Chen, J. Appl. Phys. 54,5966 (1983). *sK. S. Kim, W. E. Baitinger, J. W. Amy, and N. Winograd, J. Electron Spectrosc. Relat. Phenom. 5, 351 (1974). rgS. Hofmann and J. M. Sanz, Trace Micro. Tech. 1, 213 (1982). *‘B. Chapman, Glow Discharge Processes (Wiley, New York, 1980). p. 182.

Kim, Kim, and Lee

2049

Downloaded 21 Sep 2011 to 143.248.115.181. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions