Plasma-enhanced chemical vapor deposition of molybdenum

Thin Solid Films, 147 (1987) 193-202 PREPARATION

PLASMA-ENHANCED MOLYBDENUM N. J. IANNO

Department (Received

193

AND CHARACTERIZATION

CHEMICAL

VAPOR

DEPOSITION

OF

AND J. A. PLASTER

of Electrical Engineering, September

University of Nebraska-Lincoln,

16, 1985; revised May 29, 1986; accepted

August

Lincoln, NE 68588-0511 (U.S.A.) 1,1986)

The plasma-enhanced chemical vapor deposition of molybdenum films from Ar-Mo(CO), gas mixtures and H,-Mo(CO), gas mixtures was studied. Films deposited from the former mixture contain large quantities of carbon and oxygen before and after annealing, while their sheet resistance remained beyond the range of the four-point probe used here. As-deposited films from the latter mixture contain carbon, but oxygen is present apparently in the form of a hydroxide. On annealing, the resistivity reaches a minimum of 7.5 ~$2 cm and oxygen is no longer present in the film.

1.

INTRODUCTION

Plasma-enhanced chemical vapor deposition (PECVD) is being explored as an alternative to traditional methods of depositing thin films of refractory metals and their silicides for use in very-large-scale integrated (VLSI) circuitslp4. Recently, a great deal of attention has been focused on the halogen-based gases such as WF,, TiCl, and MoF, as source gases for PECVD. However, in many instances the relatively large amounts of energy required to remove the halogen from the metal and the chemically reactive nature of the halogens themselves requires minimum substrate temperatures in the 400-500 “C range to avoid etching of the substrate and halogen contamination of the deposited materia13s4. However, in the case of molybdenum deposited from MoF,-H, gas mixtures, films heavily contaminated with fluorine were produced under all deposition conditions, where the fluorine remained in the film even after annealing4. In view of the potential applications of molybdenum and PECVD, we feel it is appropriate to explore alternative source gases that may yield better quality molybdenum films. We have selected Mo(CO),, a high vapor pressure solid which is characterized by a relatively low dissociation energy and fragmentation by successive loss of CO groups5. These characteristics, combined with the inability of carbon, oxygen or CO to etch common substrates such as silicon and SiO,, led us to investigate the PECVD of molybdenum films using Mo(CO), as a source gas. Our study was carried out as a function of the type and amount of additive gas, the substrate temperature and post-deposition annealing temperature, the results of which are reported below. 0040-6090/87/$3.50

0 Elsevier Sequoia/Printed

in The Netherlands

194 2.

N.J.

I A N N O , J. A. P L A S T E R

EXPERIMENTAL APPARATUS AND PROCEDURES

The depositions were carried out in a parallel plate reactor where the grounded electrode is a stainless steel plate 25.4 cm in diameter and is 6.5 cm above a stainless steel lower electrode 10cm in diameter. The lower electrode is temperature controlled and is driven by a 13.56 M H z capacitively coupled r.f. generator in parallel with a d.c. power supply that allows independent control of the negative d.c. bias on the electrode. The Mo(CO)6 is loaded into a stainless steel nipple 15.2 cm long through a bellows valve of inner diameter 2.31 cm. The other end of the nipple is blanked off and the loading takes place in a fume hood. On completion of the loading, the valve is closed and the locked sealed nipple is removed from the hood and connected to the deposition chamber. Substrates of area 2 cm x 2 cm, consisting of 3000/~ of thermally grown SiO2 over silicon, were positioned in the center of the lower electrode, and a base pressure of at most 5 x 10 - 6 Torr was established with an oil diffusion p u m p before a continuous flow of Mo(CO)6 and the desired additive gas was initiated. The total pressure, as measured with a capacitance manometer, was held constant at 80 mTorr, while the input r.f. power was maintained at 100 W. All annealing was performed in a vacuum of 10-7 Torr, while the electrical resistivity was determined by measuring the material's sheet resistance using a fourpoint probe and multiplying by the film thickness as determined by spectroscopic ellipsometry 6. The X-ray diffraction data were acquired in a Phillips model XRG3100 X-ray diffractometer. In addition, Auger depth profiling was performed in a P H I model 540 thin film analyzer, where these data were all normalized to the same arbitrary standard. 3. RESULTS

3.1. Additive gases 3.1.1. Argon Pure M o ( C O ) 6 did not yield a stable discharge under the conditions used here. Therefore argon was initially added to the system in order to stabilize the discharge. Films were deposited from gas mixtures whose argon-to-Mo(CO)6 pressure ratios were 3:1, 6:1 and 10:1 at a substrate temperature of 100°C as a function of the externally applied d.c. bias. Auger depth profiling of the as-deposited films revealed large quantities of carbon and oxygen present in the films while an X-ray diffraction scan showed M o O 3 and M o v O , a peaks. Films deposited under any of the previously described conditions yielded sheet resistances beyond the limit of our four-point probe. On annealing for 1 h at 1000 °C, the sheet resistance was still beyond the range of the probe, the oxygen and carbon content of the films remained unchanged and the X-ray diffraction scan showed a single MoO3 peak. 3.1.2. Hydrogen Since hydrogen is known to reduce Mo(MO)6 to pure molybdenum of low resistivity under specific conditions during standard high pressure chemical vapor deposition 7, it was used as the additive gas instead of argon and deposits were made under the previously stated conditions. For each of the three gas mixtures, three

PECVD OF M O

195

different values of externally applied d.c. bias were used, namely 0, - 2 0 0 and - 400 V. The 0 V case actually corresponds to - 50 V of self-bias as measured with the r.f. generator controller. The as-deposited resistivities of these films are seen in Fig. l(a) where the resistivity in microohm centimeters is plotted v s . the gas mixture as a function of the externally applied d.c. bias. As seen, the resistivity decreases with increasing hydrogen concentration and/or d.c. bias until it reaches a relative minimum for the conditions used here. The decrease in resistivity as the hydrogen concentration increases is probably due to the fact that the hydrogen in the discharge prevents carbon from being incorporated into the film through gas phase and/or surface reactions with carbon and carbon-containing species, forming volatile compounds that are pumped out of the system. This is illustrated in Fig. 2 3

3 2

2

io 4

103 ~u

5

2 i 0 ::t. ~03

•1o

2

~

7

~ _>

5

)-

~

2

~

102

~

7 5

~ u

5

~

~L 102

_> t-tO ~

2 i0 I

7 5 2

2

i0 0 0

i01

0

2

(a)

4

6

8

I0

12

GAS RATIO ( H 2 : Mo(CO) 6 )

2

4

6

8

I0

GAS RATIO (H2=Mo(CO) 6 )

(b)

Fig. 1. Resistivityof m o l y b d e n u m films v s . the H 2 : M o ( C O ) 6 pressure ratio for three different d.c. bias values ( O , 0 V; [~, - 200 V; V, - 400 V): (a) as-deposited films; (b) films annealed at 1000 °C for 1 h in a vacuum of 10 7 Torr.

0¢ n~ I-

v

_o

== 0.

<[ 0

2

4

6

8

I0

12

14

16

18

20

2

4

6

8

I0

12

14

16

18

20

SPUTTERING TIME (MINUTES) SPUTTERING TIME (MINUTES) (a) (b) Fig. 2. Auger depth profiles ( O , oxygen; V, carbon; I-q, molybdenum) of films deposited at - 4 0 0 V d.c. bias and H2:Mo(CO)6 pressure ratios of(a) 3:1 and (b) 10:1.

196

N . J . IANNO, J. A. PLASTER

where the Auger depth profiles of samples deposited from H2:Mo(CO)6 at pressure ratios of 3:1 and 10:1 and at - 4 0 0 V d.c. bias are shown. The large difference in carbon signals between the 3:1 film and the 10:1 film should be noted. Also shown in this figure is the oxygen signal in each of the films. While it is smaller in the 10:1 film than in the 3:1 film, it is still quite large. The resistivity of the as-deposited films also decreases as the d.c. bias increases for a given gas mixture, as also seen in Fig. l(a). Again this can be related to a decrease in the carbon content of the film as seen in Fig. 3, where the normalized average carbon and oxygen Auger signals for films deposited from the 10:1 H2:Mo(CO)6 mixture as a function of d.c. bias are shown. The carbon content of the film decreases as the bias increases, while the oxygen content of the films exhibits only a slight change as the bias is increased. On annealing at 1000 °C for 1 h, the resistivity of these films drops dramatically (as seen in Fig. l(b)) with a minimum value of 7.5 ~tf2cm being achieved. This value compares favorably with the value for bulk molybdenum of 5.57 p.l)cm a and the values achieved for sputtered and evaporated films which lie between 6 and 10 p.f2 cm respectively 9'1°. Accompanying this drop in resistivity is a change in crystal structure and film composition where the as-deposited and post-annealed X-ray diffraction scans and the post-annealed Auger depth profile of a film deposited from a 10:1 H2:Mo(CO)6 mixture at - 400 V d.c. bias are shown in Fig. 4. As seen in the figure, the large hydroxide peak in the as-deposited film is replaced by several peaks corresponding to molybdenum and by one molybdenum carbide peak. The Auger depth profile also shows a significant decrease in the amount of carbon in the film and a complete removal of oxygen from the annealed film compared with the asdeposited film (see Fig. 2(b) for comparison). Also of practical concern is the adherence of the film to the substrate. In order to evaluate the film's adherence, the Scotch tape test was performed, where a piece of tape is placed on the film and then removed. If the film remains on the substrate the adherance is considered to be good. For the as-deposited films, the adherence was

z O3

t~

,9, N .-I

oz

i

0

-100

i

-200

I

-300

I

-400

Fig. 3. Normalized average carbon (V) and oxygen (O) Auger signal for films deposited from a pressure ratio of 10:1 as a function of d.c. bias.

H2:Mo(CO)6 at

PECVD

OF Mo

197

7 deposited

hnr

>I--

Mo annealed

Z W I-Z

20

MoC

Mo

i

i

i

i

i

i

i

40

60

80

I00

120

140

160

(a)

2(9 ( D E G R E E S )

I--

I1: al

I....z bJ I W i

IM O.

0

2

4

6

8

I0

12

14

16

18

20

(b) SPUTTERING TIME { M I N U T E S ) Fig. 4. (a) X-ray diffraction scans of an as-deposited and an annealed film deposited from a 10:1 Hz:Mo(CO)6 mixture at -400 V d.c. bias. (b) Auger depth profile of the annealed film seen in (a): O, oxygen; V, carbon; m, molybdenum. good, except for the with the tape.

- 4 0 0 V d.c.

bias case, when the films had a tendency to peel off

3.2. Annealing temperature In order to evaluate the effect of annealing temperature on the film properties we selected films deposited from H2:Mo(CO)6 at a 10:1 pressure ratio and at - 2 0 0 V d.c. bias where the substrate temperature during deposition was held at 100 °C. These films exhibited g o o d adhesion and the lowest as-deposited resistivity. As-deposited samples were annealed for 1 h at a specific temperature and removed from the furnace for analysis. The resistivity as a function of annealing temperature is shown in Fig. 5, where the as-deposited value is shown for comparison. This result is typical of that observed for the P E C V D of tungsten, where high temperature anneals are also necessary to reduce the film's resistivity 3"4. X-ray diffraction scans of the as-deposited film and films annealed at 500, 700 and 1000 °C are shown in Fig. 6(a), while the normalized average carbon and oxygen signals of the films as determined from their respective Auger depth profiles are shown in Fig. 6(b). As seen in the figure, at an annealing temperature of 700 °C the oxygen and most of the carbon are removed from the film and the hydroxide is absent from the X-ray scan.

198

N.

J. IANNO,

J. A.

PLASTER

2,50

200

150

{J :L

I00

50

0

I

I

I

I

I

I

,500

600

7'00

800

900

Iooo

Anneoling

Temperoture

(o C)

Fig. 5. Resistivityvs. annealing temperature for films deposited from a 10:1 H2:Mo(CO)6 mixture at - 200 V d.c. bias: ©, as deposited; O, annealed. However, at 1000 °C the MoC peak is present along with the pure molybdenum peaks, which are also intensified compared with those for the 700 °C case. 3.3. S u b s t r a t e temperature

While deposition at a substrate temperature of 100°C has been achieved, slightly higher temperatures may produce a lower as-deposited resistivity and reduce the maximum required annealing temperature to achieve the minimum resistivity. In view of this, we deposited films from 10:1 Hz:Mo(CO)6 gas mixtures at - 2 0 0 V d.c. bias and at substrate temperatures of 100, 200 and 300°C. The resistivities of the as-deposited films are given in Fig. 7(a) while X-ray diffraction scans are shown in Fig. 7(b). The increased substrate temperature has clearly lowered the as-deposited resistivity, while also eliminating the hydroxide peak in the X-ray diffraction pattern, leaving an apparently amorphous film. Figure 8 indicates that, while the oxygen concentration is reduced to almost zero by the increased substrate temperature, the carbon content drops by only a factor of 2. The films deposited at 300 °C were annealed in a manner similar to that already described. As seen in Fig. 9(a) the resistivity behaves in a manner similar to that of films deposited at 100 °C, except that the as-deposited value is lower. The X-ray diffraction scans for films annealed at 500, 700 and 1000 °C are presented in Fig. 9(b), where the 500 °C film still exhibits no peaks and the films annealed at higher

PECVD OF MO

.

~

~

199

Oa (OH)a

Oa (OH] 8

as-

deposited

500* C

Q~ bc~

MO ,/M._._

}-

Mo

700 * C

~

I(300 * C

Z W }Z

MO

20

Mo

i

,

40

60

(a)

MoC

i

80

i

I00

i

120

i

I

140

160

20 (DEGREES)

_1 z (.9 I/) i..iJ uJ w N -J ri-

O Z _

500

_~

600 700 800 900 ANNEALING TEMPERATURE (*C)

I000

(b) Fig. 6. (a) X-ray diffraction scans of films deposited from a 10:1 H 2 : M o ( C O ) 6 mixture at - 200 V d.c. bias after annealing at 500, 700 a n d 1000 °C. (b) Normalized average carbon (V) and oxygen (O) Auger signals of the films of (a).

temperatures are similar to those deposited at 100°C and annealed at 700 and 1000 °C. 4. DISCUSSION As previously mentioned, we consider that the reduction in resistivity of the asdeposited films for a given d.c. bias when hydrogen is used as an additive gas as opposed to argon is due to hydrogen's ability to prevent carbon from being incorporated into the film, by either gas phase or surface reactions. This conclusion is based on Auger depth profiling of films deposited from an H2:Mo(CO)6 mixture and from Ar: Mo(CO)6 mixtures and the knowledge that carbon can greatly increase the resistivity of molybdenum films TM. Also, the hydrogen apparently limits the formation of oxides in the as-deposited films, causing the formation of hydroxides instead. The hydroxides apparently volatilize during annealing, removing most of the oxygen from the film and allowing a polycrystalline molybdenum film with a low resistivity to form. It should be noted that the molybdenum oxides are volatile a at

200

N . J . IANNO, J. A. PLASTER

300

200

~

I00

i

i

2OO

I00

(a)

SUBSTRATE

300 TEMPERATURE ('C)


200"C

iv" >l.-

300* C

Z IJJ IZ I

20

40

(b)

I

i

i

60

80

I00

2e

i

120

i

140

(DEGREES)

Fig, 7. (a) Resistivity vs. substrate temperature for films deposited from a 10:1 H2:Mo(CO)6 mixture at - 200 V d.c. bias. (b) X-ray diffraction scans of the films of(a).

I

.J
c[

z I00

200 SUBSTRATE

300 T E M P E R A T U R E (°C)

Fig. 8. Normalized average carbon [XT)and oxygen (O) Auger signals v s . substrate temperature for films

deposited from a 10:1 H2:Mo(CO)6 mixture at - 2 0 0 V d.c. bias. the annealing temperature used here and therefore any oxides that form could also be volatilized during the anneal. However, the results obtained by annealing the films deposited from Ar:Mo(CO) 6 mixtures indicate that oxides can still remain after the annealing cycle used here. Therefore the amount of oxides formed during

PECVD OF M o

201

100

:::L. 50 Q.

500

600

(a)

700

800

900

tO00

ANNEALING TEMPERATURE (*C) 500 *C

I.-

Mo

Mo

700 * C

Mo MO

MoC

IO00*C

~

I'-

_z

20

(b)

40

60

80

tOO

t

I

120

140

2e (DEGREES)

Fig. 9. (a) Resistivity v s . annealed temperature for films deposited at 300°C from a 10:1 H2:Mo(CO)6 mixture at - 200 V bias. (b) X-ray diffraction scans of the films in (a) after annealing at 500, 700 and 1000 °C.

annealing of films deposited from H2:Mo(CO)6 mixtures must be small enough that they are volatilized during the anneal. The d.c. bias also has an effect on the as-deposited resistivity, as seen in Fig. l(a), where an increasing bias, in general, lowers the resistivity of the film for a given gas mixture. Again Auger depth profiling of the as-deposited films reveals that the carbon content of the film decreases as the bias increases for a given gas mixture. However, no significant or consistent changes in the film microstructure as indicated by X-ray diffraction analysis could be found as the bias increases. Also, it should be noted that the effect of the bias is smaller on films deposited from gas mixtures containing the most hydrogen and that after annealing the resistivities depend almost exclusively on the gas mixture for the ratios 6:1 and above. These results tend to indicate that at lower hydrogen concentrations the bias may enhance the removal of high resistivity compounds at the surface of the depositing film through sputtering 12, while at higher hydrogen concentrations these species are removed chemically during both deposition and annealing, accounting for the limited effect of the bias on the resistivity at these concentrations before and after annealing. The effect of the substrate temperature on the as-deposited film properties is similar to that of the d.c. bias, where the resistivity decreases to a relative minimum as the substrate temperature is increased. In this case, however, Figs. 7 and 8 clearly show that the increased substrate temperature prevents oxygen incorporation and the formation of hydroxides. Finally, while post-deposition annealing dramatically reduces the film's

202

N. J. IANNO, J. A. PLASTER

resistivity, it appears that a m i n i m u m temperature of 900 °C is necessary to reduce the resistivity to a point where it is comparable with those of sputtered or evaporated films. This result is independent of the substrate temperature during deposition even though films deposited at lower substrate temperatures contain a hydroxide while those deposited at higher temperatures do not. The annealing process drives off the hydroxide as seen in Fig. 6. However, once it is removed the film behaves as though the hydroxide was never present as seen by a comparison of the resistivity and diffraction patterns of films deposited at 100 and 300 °C (see Figs. 5 and 9(a)) which show identical behavior above an annealing temperature of 700 °C. 5. SUMMARIZING REMARKS

The P E C V D of m o l y b d e n u m films from A r : M o ( C O ) 6 and H 2 : M o ( C O ) 6 gas mixtures was performed. The Xay diffraction scans of films deposited from A r : M o ( C O ) 6 mixtures show oxides before and after annealing, while Auger depth profiling shows carbon and oxygen to be present in the films. These films have electrical resistivities beyond the range of the four-point probe used here. In direct contrast with these films were those deposited from H 2 : M o ( C O ) 6 mixtures, where the electrical resistivity and elemental composition of these films as a function of the H2: Mo(CO)6 pressure ratio, the external d.c. bias, the substrate temperature and the annealing temperature were studied. It was found that asdeposited resistivities as low as 100p,Ocm could be produced at a substrate temperature of 300 °C and a - 2 0 0 V d.c. bias. While post-deposition annealing above 700 °C lowers the resistivity, a m i n i m u m annealing temperature of 900 °C is required to minimize the resistivity to 7.5 t~O cm. ACKNOWLEDGMENTS The authors would like to thank Mr. Peter Zaemes for performing the X-ray diffraction scans and careful sample preparation. This work was supported in part by the National Science F o u n d a t i o n under G r a n t ECS-83-07025. REFERENCES 1 K. Akimoto and K. Watanabe, Appl. Phys. Lett., 39 (1981). 2 F. Okyama, Appl. Phys. A, 28 (1982). 3 J.K. Chu, C.C. TangandD. W. Hess, Appl. Phys. Lett.,41(1982)75.

4 C.C. Tang, J. K. Chu and D. W. Hess, SolidState Technol., 26 (1983) 125. 5 I. Wender and P. Pino, Organic Synthesis via Metal Carbonyls, Vol. I, Interscience, New York, 1968. 6 R.M.A. Azzam and N. M. Bashara, Ellipsometry and Polarized Light, North-Holland, New York, 1977. 7 L.H. Kaplan and F. M. d'Heurle, J. Electrochem. Soc., 117 (1970) 693. 8 L. Northcott, Molybdenum, Academic Press, New York, 1956. 9 M. Suzuki and K. Asai, J. Electrochem. Soc., 131 (1984) 185. 10 R.A. Holmwood and R. Glang, J. Electrochem. Soc., 114 (1965) 857. 11 T. Sugano, H. Chou, M. Yoshida and T. Nishi, Jpn. J. Appl. Phys., 7 (1968) 1028. 12 J.C. Angus and J. Segal, Diamondlike carbon films, NASA Final Rep. NASA CR 165588, February 1982 (National Aeronautics and Space Administration).