258

IEEE TRANSACTIONS ON SEMICONDUCTOR MANUFACTURING. VOL 6. NO 3. AUGUST 1993

Particle Removal from Silicon Wafer Surface in Wet Cleaning Process

Mitsushi Itano, Frederick W . Kern, Jr., Masayuki Miyashita, and Tadahiro Ohmi, Member, IEEE Abstract-Particle removal from silicon wafer surfaces was studied using acid and alkaline solutions employed in wet cleaning processes found in semiconductor manufacturing. It has been demonstrated that the alkaline solutions are superior to the acid solutions in terms of particle removal efficiency. The particle removal mechanism in the alkaline solutions has been confirmed as follows: the solutions etch the wafer surfaces to lift off particles, and the particles are then electrically repelled from the wafer surfaces. It has been determined experimentally that an etch rate of 0.25 nm/min or more is required to lift off the particles adsorbed on the wafer surfaces. The etch rate of NH40H-HzOZ-Hz0 solution is 0.3 nm/min when the mixing ratio is set at 0.05 :1 :5 (NH40H:H2O2 :HzO). With this mixing ratio, the surface smoothness of the wafers is maintained at initial level. It is therefore possible to reduce the NH40H content in NH40H-HzOZ-Hz0 solution to 1/20 of the conventional level. In addition, it has been confirmed that when the pH value of NH40H-Hz0z-Hz0 solution becomes higher, polystyrene latex spheres and natural organic particles are oxidized with their surface turning into a gel and their shape changing. The particle removal efficiency has been demonstrated to be degraded by the oxidation of organic particles. The oxidation of organic particles becomes significant as the NH40H content in NH40HHzO2-Hz0solution rises above 0.1: 1 : s (pH is over 9.1). These experimental results suggest that the mixing ratio of NH40H-HzO2-H20solution should be set at 0.05 :1:5. This mixing ratio is effective in maintaining both particle removal efficiency and surface smoothness of the wafers. I. INTRODUCTION T cleaning technology remains essential to remove contaminants on silicon wafer surfaces as semiconductor devices are pursuing higher levels of integration and higher resolution patterns [l]. The RCA cleaning process presented in 1970 by Kern and Poutinen [2] is still employed across the world as a wet cleaning technology to remove contaminants on the wafer surfaces. Although N H 4 0 H - H 2 0 2 - H 2 0 solution in the RCA cleaning process is considered very effective in removing particles [3], its particle removal mechanism has not yet been fully understood. Meanwhile the mechanism of particle adsorption from solution to the wafer surfaces is now being actively studied from the perspective of colloid chemistry [4]-[ 111. It is reported that ionic strength in solutions and zeta poten-

tial of particles are very important factors of particle adsorption onto the wafer surfaces. In the study reported in this article, the mechanism to remove particles adsorbed on the wafer surfaces has been investigated based on the mechanism of particle adsorption onto the wafer surfaces. In particle removal experiments, artificially contaminated C Z (Czochralski method) and FZ (Float zone purification method) wafers have been cleaned with both acid and alkaline solutions employed in the wet cleaning processes. It has been demonstrated that the alkaline solutions are superior to the acid solutions in terms of particle removal efficiency. It has been considered that particles are electrically repelled from the wafer surfaces due to the negative zeta potential of most particles in the alkaline solutions [ 1 11. In addition to the studies from the colloid chemistry view-point, we considered it important to examine etching of the wafer surfaces caused by the alkaline solutions. For this purpose, the relation between etch rate and particle removal efficiency has been studied, using C Z and FZ wafers. Moveover, by employing polystyrene latex spheres to represent organic materials, the oxidation prosolution was investigated to cess in N H 4 0 H - H 2 0 2 - H 2 0 find out the relation between degree of oxidation and particle removal efficiency. As a result, it has been shown that the NH40H content in N H 4 0 H - H 2 0 2 - H 2 0 solution should be reduced to 1/20 of the conventional level to optimize particle removal efficiency. The remainder of this article will describe the details of the experimental process and the results. 11. EXPERIMENT Five-in C Z (1, 0, 0) wafers are used in the particle adsorption experiments. Native oxide is first removed from the wafer surfaces in a 0.5% HF solution. The wafers are then immersed for ten min in various solutions seeded with particles, and they are then rinsed and dried. After native oxide has formed on the wafer surfaces, it is again removed in a 0.5% H F solution, and they are then rinsed and dried. Five inch C Z (1, 0, 0) and four-in FZ (1, 0, 0) wafers are used in the cleaning experiments to study particle removal. The wafers are artificially contaminated employing the following contaminant sources: particles naturally slurring in city water for typical particles in liquid phase, particles naturally slurring in ambient atmosphere outside of the cleanroom for typical particles in gaseous phase,

Manuscript received June 10, 1991; revised Nov. 10. i992. The authors are with Tohoku University, Department of Electronics, Faculty of Engineering, Aoba, Sendai 980, Japan. IEEE Log Number 9209573.

0894-6507/93$03.00 0 1993 IEEE

Authorized licensed use limited to: TOHOKU UNIVERSITY. Downloaded on March 02,2010 at 02:26:28 EST from IEEE Xplore. Restrictions apply.

ITANO

er a l . :

PARTICLE REMOVAL FROM SILICON WAFER SURFACE

259

Si02 spheres of 0 . 3 pm and 1.0 pm in diameter as typical inorganic particles, and polystyrene latex spheres of 0.506 pm in diameter as typical organic particles. The wafers are contaminated by immersing into highly acidic solutions containing one of the above particles. Other wafers are contaminated by exposure to the ambient atmosphere outside of the cleanroom. By these procedures, wafers having a total of 5000-10 000 defects, classifies as either haze (<0.5 pm) or particles (>0.5 pm) on their surfaces are prepared for cleaning experiments. These wafers are cleaned in the following chemicals. a) NH40H-H202-H20 solution: 2 9 % N H 4 0 H: 30% H202: H 2 0 = x : 1 : 5 at 80C; hydroxide)-H202b) TMAH(tetramethy1ammonium H 2 0 solution: 2.38% or 20% TMAH: 30% H 2 0 2 : H 2 0= x : y : z at 80C; c) H2S04-H202 solution: 97 %H2S04: 3 0 % H 2 0 2 = 4 : 1 at around 120C; :H 2 0 d) HC1-H202-H20solution: 36%HC1: 3 0 % H 2 o 2 = 1 : 1 : 6 at 80-90C; e) HF-H202-H20 solution [12], [13]: 5 0 % H F : 30% H 2 0 2: H 2 0 = 1 : 33 : 66 at 25C; f ) H202-H20 solution: 3 0 % H 2 o 2 : H 2 0 = 1 : 5 at 80C. In the H2S04-H202 solution, the wafers are immersed for five min. In all other solutions, the wafers are immersed for ten min. Following the chemical cleaning, the wafers are rinsed with ultrapure water for ten minutes. In order to avoid particle contamination during the drying process of the wafers, the native oxide formed during the chemical cleaning is removed in a 0.5% H F solution. After removing the native oxide, the wafers are again rinsed with ultrapure water for ten min. Cleaning effect is evaluated by particle removal efficiency. The particle removal efficiency is evaluated by calculating the ratio of the particle counts before and after cleaning. Particle counts on all wafers are evaluated using an Aeronca WIS-100 wafer inspection system. The WIS-100 classifies detect size as particles ( > 0 . 5 pm) and haze (<0.5 pm). Before the experiments, the WIS-100 is calibrated with a relative standard wafer. Initial particle counts, before the experiments, are less than 50 per wafer. The above experimental procedure is manually performed at a wet station. As for the material of construction of the bath, the H F cleaning and the ultrapure water rinsing use PFA baths while the other chemical cleanings use quartz baths. PFA camers are used for immersing the wafers into the bath and transporting them from one bath to another. Chemicals used in these tests are all of ULSIgrade and low particle concentration.
AND DISCUSSION 111. RESULTS A . Solution p H Fig. 1 shows the relationship between the solution pH and the number of particles adsorbed on the wafer surfaces. 10% vol. of city water is added to each test solu-

%-

4 6 8 1 0 1 2 1 4 Solution P H Fig. 1. Adsorption of particles in city water onto wafer surfaces as a function of solution pH. 10-percent vol. of city water was mixed with ultrapure water. Acidityibasicity is controlled by the addition of HC1-H202or H,S04 and NH40H-H20, or TMAH-H,O,.

0 " 0

"

"

"

"

"

'

tion. The solution pH is adjusted by HC1 with H202 or H2SO4 in acidic solutions, and it is adjusted by NH40H with H202 or TMAH with H202 in alkaline solutions. Maximum particle adsorption is observed in the low pH solutions, and the number of adsorbed particles is reduced with the increase of solution pH. Fig. 2 also shows the well known effect of pH on the zeta potential of a Fe203 particle [6]. It can be seen that the Figs. 1 and 2 curves match almost exactly. It has been reported that the number of adsorbed Si02 spheres or polystyrene latex spheres on the wafer surfaces is reduced with the increase of solution pH as shown in Fig. 1 [9]-[ll]. Generally, the zeta potential of most particles is considered to be negative in alkaline solutions. In particular, the zeta potential of metal oxide particles is positive in acidic solutions and negative in alkaline solutions in accordance with the following equations from colloid chemistry [14], [15]: LOWpH: M

High pH:

+ H + + OHM - OH: + OH M - OH + H + + OHM - OH- + H f + OHM 0- + HlO + H S

-

OH

+ +

++

+ +

In addition, it is reported that both hydrophilic (covered with oxide film) and hydrophobic (bare silicon surface) wafers exhibit a negative surface charge in water [4], [5]. Relating these observations to the results shown in Fig. 1, it is believed that both the wafer surfaces and the zeta potential of particles exhibit a negative charge in alkaline solutions. As a result, the wafer surface and particles are presumed to be electrically repulsive in alkaline solutions. This indicates that alkaline solutions are beneficial for removing particles from the wafer surfaces.

B. Wafer Cleaning by Alkaline Solutions The RCA cleaning method [2], which is one of the major wet silicon wafer cleaning methods, employs a

Authorized licensed use limited to: TOHOKU UNIVERSITY. Downloaded on March 02,2010 at 02:26:28 EST from IEEE Xplore. Restrictions apply.

260

IEEE TRANSACTIONS ON SEMICONDUCTOR MANUFACTURING. VOL. 6. NO. 3 . AUGUST 1993

40
0
0

A
A

HCI HNO~ Hz SO4 NaOH KOH

H202 Mixed a t

A
0

25 "C a t 80 "C Non H202 a t 25 "C

-40'

FROM NllDA E T At

7 8 P H

10

11 12 13 14

Fig 2 Influence of pH on zeta potential of ion oxide particle (FezO,) (from Niida et al [ 6 ] )

NH40H-H202-H20 solution as the alkaline solution. The NH40H-H202-H20solution usually employs the mixing ratio of 1 : 1 : 5 (NH40H : H 2 0 2: H 2 0 ) . Fig. 3 shows the relationship between the various NH40H contents in NH40H-H202-H20 solution and the solution pH. As shown in Fig. 3 , the pH value of the 1 : 1 : 5 solution is 9.9 at 80 C. However, the results shown in Figs. 1 and 2 indicate that the NH40H content in N H 4 0 H - H 2 0 2 - H 2 0 solution can be reduced if it is to be used exclusively for particle removal. Fig. 4 shows the relationship between the pH value of the cleaning solutions and their particle removal efficiency when the NH40H content in NH40H-H202-H20 solution is varied as follows: 0.001 : 1 : 5, 0.01 : 1 : 5,

NH40H:H202:H20=

X:1:5 X:0:5

Fig. 3 . The pH value of NH4OH-H@-H2O solution as a function of NH40H content

0.05:1:5,0.1:1:5,0.25:1:5,0.5:1:5,andl:1:5.In this experiment, five in. CZ (1, 0, 0) wafers are used.

Fig. 5 shows the results of the same experiment as Fig. 4, except that it uses four inch FZ (1, 0, 0) wafers. Fig. 6 shows the relationship between the pH value of the cleaning solutions and their particle removal efficiency when the TMAH-H202-H20 solution is used instead of the NH40H-H202-H20solution. The TMAH content in TMAH-H202-H20 solution is varied as follows: 0.1 : 1 : 5, 1 : 1 : 5 , 4 : 1 : 0 , 9 : 1 : O (using 2.38% TMAH), and 2: 1 : 4 , 4 : 1 : 2 (using 20% TMAH). Five inch CZ (1, 0, 0) wafers are used in this experiment. Before the cleaning experiments, each wafer is contaminated with city water particles, Si02 spheres, polystyrene latex spheres, and ambient atmosphere particles. As for the cleaning, the following sequence is employed: a) Immersion into NH40H-H202-H20 solution or TMAH-H202-H20solution for 10 min at 80C; b) Ultrapure water rinsing for 5 min at 25C; c) Immersion into hot ultrapure water for 10 min at 80-90 O C ; d) Ultrapure water rinsing for 10 min at 25C; e) Immersion into 0.5% H F solution for 1 min at 25C; f) Ultrapure water rinsing for 10 min at 25C. As can be seen in Figs. 4-6, the particle removal efficiency goes up in most cases as the pH value of the cleaning solution increases. However, the particle removal efficiency of polystyrene latex spheres -and ambient at\

I I

City Water aSi02 .--a PSL e--* Enviro. Air

10

Solution P H at 80 'C ( NH4 OH- H202 )

Fig. 4. Relationship between the pH value in the cleaning solution and the particle removal efficiency on the CZ wafers using NH,OH-H2O2-H,O solution. The wafers are contaminated with city water particles, 0.3 pm and 1.0 pm S i 0 2 spheres, 0.506 pm polystyrene latex spheres, and ambient atmosphere particles (Enviro. Air).

mosphere particles is not improved even when the pH value increases. These two cases will be discussed later. The experimental results shown in Figs. 4-6 indicate that the electrical repulsion generated between the wafer sur-

Authorized licensed use limited to: TOHOKU UNIVERSITY. Downloaded on March 02,2010 at 02:26:28 EST from IEEE Xplore. Restrictions apply.

ITANO et a l . : PARTICLE REMOVAL FROM SILICON WAFER SURFACE

261

100

5 a

80

aJN
( 1 :

6 ,0

.v 2 r:
U

L+!m

2 '-40 U
a 0
a ,

I/
0 0

c c

20

j
c
I
I

City Water Si02 m--m PSL e---* Enviro. Air

A

10

Solution P H at 80 'C ( NH40H - H2 02 1

Fig. 5 . Relationship between the pH value in the cleaning solution and the particle removal efficiency on the FZ wafers using NH,0H-H202-H20solution. The wafers are contaminated with city water particles, 0.3 pm and 1.0 pm S i 0 2 spheres, 0.506 pm polystyrene latex spheres. and ambient atmosphere particles (Enviro. Air).

100

, ,
\

faces and particles in alkaline solutions is by itself not sufficiently effective to remove the particles adsorbed on the wafer surfaces. Fig. 7 illustrates the etch rate on the wafer surfaces using the NH40H solution and the TMAH solution. Fig. 7 also illustrates the etch rate on the wafer surfaces using the NH,0H-H20,-H20 solution and the TMAH-H,O,H 2 0 solution. To determine the etch rate, half of the wafer surface is masked with an oxide film and then the wafers are immersed in the alkaline solutions at 80C. After removing the oxide film, the step on the wafer surfaces generated by etching is measured with a Dectak 3030ST Surface Analyzing System to determine the etch rate. The etch rate on the wafer surfaces is considerably lowered when H 2 0 2 is added. However, it is true that the wafer surfaces are etched even in the solutions with HzOzadded. The etch rate goes up whether or not H z 0 2is added as the pH value increases. Figs. 8-1 1 show the particle removal efficiency as a function of the etch rate when each wafers are contaminated with city water particles, SiO, spheres, polystyrene latex spheres, and ambient atmosphere particles. These figures are obtained from Figs. 4-6 in which the pH value is converted to the etch rate by the data shown in Fig. 7. The particle removal efficiency shows a rapid increase up to an etch rate of 0.25 nm/min. When the etch rate reaches 0.25 nm/min, the particle removal efficiency levels off. In order to remove particles adsorbed on the wafer surfaces by means of the alkaline solutions, the etching of the wafer surfaces is necessary in addition to the electrical repulsion effect described before. An etch rate of 0.25 nm/min is required to efficiently remove particles when the wafers are cleaned by the alkaline solutions for ten minutes. It is therefore possible to reduce the content of NH,OH or TMAH until the etch rate on the wafer surface reaches 0.25 nm/min. When the mixing ratio of NH,OH-H,O,-H,O solution is 0.05 : 1 : 5 , the etch rate of both CZ and FZ wafers reaches 0.3 nm/min. On the other hand, when the mixing ratio of TMAH-H202-H20solution is 9 : 1 : 0 (2.38 wt. o/o TMAH : HzOz: H 2 0 ) , the etch rate of the C Z wafer reaches 0.28 nm/min.

A
M

C. Particle Removal Mechanism

City Water A Si02 m--m PSL e---* Enviro. Air

Solution P H at 80 'C ( TMAH- H202 )

Fig. 6. Relationship between the pH value in the cleaning solution and the particle removal efficiency on the CZ wafers using TMAH-H202-H20solution. The wafers are contaminated with city water particles, 0.3 pm and 1.0 pm SiOz spheres, 0.506 pm polystyrene latex spheres, and ambient atmosphere particles (Enviro. Air).

Table I shows the particle removal efficiency when acid solutions are used. Table I1 shows the particle removal efficiency when alkaline solutions of the NH,0H-H20,H 2 0 solution and the TMAH-HzO2-H20 solution are used. Among the acid solutions, the H2S04-H202 solution exhibits very high particle removal efficiency. It is expected that the H2SO4-H2O2 solution oxidizes and decomposes the particles absorbed on the wafer surfaces due to its strong oxidizing force. In addition, the number of particles adsorbed on the wafer surfaces is limited in the H,S04 solution as shown in Figs. 1 and 2. The H2S04-H,0, solution is believed to have a high particle removal efficiency because of the strong oxidizing force as well as the electrical state between the wafer surfaces and particles.

Authorized licensed use limited to: TOHOKU UNIVERSITY. Downloaded on March 02,2010 at 02:26:28 EST from IEEE Xplore. Restrictions apply.

262

IEEE TRANSACTIONS ON SEMlCONDUCl O R LIANLIFAC Tl.'RIUG.

V O L 6. KO 3. AUGUST 1993

103

10-2

9
Solution

11 12 P H at 80 "C

10

Fig. 7 . Etch rate of the CZ and FZ wafer surfaces as a function of solution pH. The wafers are immersed into the NH,OH solution or TMAH solution or NH,OH with H,O, solution or TMAH with H,O, solution at 80 C .

On the other hand, the particle removal efficiency of HFH202-HZ0 solution is low. even though the etch rate is very high at 0.42 nm/min as shown in Table I. This indicates that etching the wafer surface is by itself not an effective particle removal technique. As shown in Table 11. the alkaline solutions exhibit high particle removal efficiency for all of the examined methods. This is due to the electrical repulsion effect which is generated between the wafer surfaces and particles in the alkaline solutions as shown in Figs. 1 and 2, as well as the etching effect of the wafer surfaces in the alkaline solutions as shown in Figs. 8-1 l . Fig. 12 illustrates the mechanisms to remove the particles adsorbed on the wafer surfaces. In the solutions with an extremely strong oxidizing force (such as the H2S04H 7 0 2 solution). the particles adsorbed on the wafer surfaces are oxidized. decomposed, and dissolved into the solutions. On the other hand. in the alkaline solutions (such as the NH40H-H2O2-H,O) solution, the particles adsorbed on the wafer surfaces are lifted off when the surfaces are etched and then dislodged from the wafer surface by electrical repulsion. Alkaline solutions require both the etching of the wafer surfaces and the electrical repulsion to remove the particles. It is observed from Tables I and I1 that the alkaline solutions are superior to the acid solutions in terms of particle removal efficiency.

TABLE I COMPARlSOh OF VARIOUS ACID SOLLTIOLS I \ TERMS OF PAKIICLt REMOVAL EFFICIEWY ON CZ W A F E R S

D. Sur-ice Micro-Roughness
Fig. 13 shows the change of the surface micro-roughness on CZ and F Z wafers when the NH40H content in NH4OH-H2O2-H,O solution is varied. The surface microroughness is evaluated with a scanning tunneling microscope (STM). As shown in Fig. 13, the surface microroughness on the C Z wafers is degraded when the NH40H content in NH,0H-H202-H20 solution is set above 0.1 : 1 : 5 . This correlates well with the change of the etch rate of the wafer surfaces when the wafers are immersed into the NH,OH-H,O,-H:O solution. On the other hand, the surface micro-roughness on the F Z wafers is not degraded. The reason for this lack of degradation is not precisely known. It has been demonstrated that the micro-roughness on the wafer surfaces greatly affects the electrical characteristics of the thin oxide film which is formed on it [ 11, [ 161, [ 171. Therefore, the NH4OH-H2O2-H20 solution to be used for particle removal should adopt a mixing ratio which does not degrade the micro-roughness of the wafer surfaces. As shown in Figs. 8-1 1 , the particle removal efficiency levels off when the etch rate exceeds 0.25 nm /min. the mixing ratio of NH,OH-H7O2-H20solution which exhibits the etch rate close to 0.25 nm/min is NH,OH : H,02 : H 2 0 = 0.05 : 1 : 5 . As shown in Fig. 13, the water surfaces is not degraded with this mixing ratio. In order to obtain both sufficient surface smoothness and sufficient particle removal efficiency, the NH40H content in NH40H-H202-H20 solution should be reduced from the conventional level of 1 : 1 : 5-0.05 : 1 : 5 .

4 7 2 Rat io Temp P H Etchme(,%) ('C) 0

3.0
0.42

4.8

N . D.

N.D.

N.D.

TABLE I1 COMPARISON OF VARIOLS ALKALI\E sOLLTIO\S I\ TFKMS REMOVAL E F F I C I E I Co\ ~ CZ W \FERS

Of

PAKTICLF

Enviro Air Ratio

4 5 4
-

8 9 0 0051 5 80 8 8 0 30

88 9
1 1 5

9 1 80 1 0 5 0 28

Temp P H

("C)

80-90
7 6

80 9 9

Etchme(,%)

N D

0 82

Authorized licensed use limited to: TOHOKU UNIVERSITY. Downloaded on March 02,2010 at 02:26:28 EST from IEEE Xplore. Restrictions apply.

ITANO et al.: PARTICLE REMOVAL FROM SILICON WAFER SURFACE

263

100

100

a
0
V .c

s 80

n
v

a
0

s 80
E

E $60

$60
a,
V .c

aJ

2 40
-0

Ii
1 . 1
I

City Water

2 40
U
rd

a
a , 20

a20

a r
0
I
I

0 ND

0.25 Etching

0.5

0.75

1.0

1 . 1

OND

0.25 Etching

0.5

0.75

1.0

Rate ( nmlmin

Rate ( nmlmin )

Fig. 8. The particle removal efficiency as a function of the etch rate when the wafers are contaminated with city water particles. NH,0H-H202-H20 solution and TMAH-H,O,-H,O solution are used for cleaning.

Fig. IO. The particle removal efficiency as a function of the etch rate when the wafers are contaminated with the polystyrene latex spheres. NH,OHH20,-H20 solution and TMAH-H20Z-H20 solution are used for cleaning.

1OC
/
n

?-r
100
C I

a
0

80

a
0

s 80
>

$60

E $60
aJ .U
c L

.-

2
Q,

40

2 40
U

Enviro. Air

a
aJ20

a aJ 20 a
0 OND
0.25 Etching 0.5
1 .

a r

0.75

1.0

0 ND

0.25 Etching

0.5

0.75

1.0

Rate ( nmlmin

Rate ( nmlmin

Fig. 9. The particle removal efficiency as a function of the etch rate when the wafers are contaminated with the SiO, spheres. NH,OH-H,O,-H,O solution and TMAH-H,0,-H20 solution are used for cleaning.

Fig. 1 I . The particle removal efficiency as a function of the etch rate when the wafers were contaminated with ambient atmosphere particles (Enviro. Air). NH,OH-H,O2-H2O solution and TMAH-H2O2-HZO solution are used for cleaning.

E. Haze Increase in Hot Ultrapure Water Bath

Fig. 14 shows the particle removal efficiency as a function of cleaning solution pH. The wafers are contaminated with city water particles, S O 2 spheres, polystyrene latex

spheres, and ambient atmosphere particles. In this figure, the following cleaning sequence is employed; a) Immersion into NH,OH-H,O,-H,O min at 80C; solution for 10

Authorized licensed use limited to: TOHOKU UNIVERSITY. Downloaded on March 02,2010 at 02:26:28 EST from IEEE Xplore. Restrictions apply.

264

IEEE TRANSACTIONS ON SEMICONDUCTOR MANUFACTURING, VOL. 6, NO. 3, AUGUST 1993

100

a
0

>

E
9,

~ 6 0
U

Particle on Water

uv) .+.a LJ=

In Alkaline Solution Fig. 12. The mechanism of particle removal from the wafer surfaces

n .G40
I n

a *
c

a c

a
l d

20

8
Solution

9
P H at

10

80 "C

NH4OH- H202
Fig. 14. Relationship between the pH value in the cleaning solution and the particle removal efficiency using NH40H-H20,-H,0 solution. The CZ wafers are contaminated with city water particles, 0.3 pm and 1.0 pm S i 0 2 spheres, 0.506 pm polystyrene latex spheres and ambient atmosphere particles (Enviro. Air). The cold ultrapure water rinsing at 25C between NH,0H-H202-H20solution cleaning and the hot ultrapure water cleaning is excluded.

0.0

BlaA 0.05 0.10 0.25 1.0

Ratio of
NH,OH

in NH,OH-H,O,-H,O

Content solution

( a : l 5) Fig. 13. Relationship between the surface micro-roughness (Ra) on CZ wafers and the ratio of NH4OH-H2O2-H2O solution. Surface micro-roughness (Ra/lO min) is measured after I O minutes immersion in the solution.

b) Immersion into hot ultrapure water for i0 min at 80-90C; c) Ultrapure water rinsing for 10 min at 25C; d) Immersion into 0.5% HF solution for 1 min at 25C; e) Ultrapure water rinsing for 10 min at 25C. Comparing with the cleaning sequence shown in the Fig. 4, the ultrapure water rinsing at 25C between the NH40H-H202-H20cleaning and the hot ultrapure water cleaning is neglected. The particle removal efficiency is reduced as the pH value in the cleaning solution increases. Mishima et al. have reported that the particle removal efficiency of the NH,OH-H202-H20 solution is reduced as the NH40H content in this cleaning solution increases [3]. This is because the wafer surfaces are roughened by

the high pH NH40H-H202-H20solution and this roughness is reported as a haze. Table I11 shows the pH value and the etch rate of the wafer surfaces in the NH40H-H202-H20 solution bath and in the following hot ultrapure water bath when the above cleaning sequence is employed. In this experiment, PFA carriers are used when the wafers are immersed to the bath and transported from a bath to another. The reason the hot ultrapure water in the bath turned alkaline is that the NH40H solution is carried over to the hot ultrapure water bath from the NH40H-H202-H20bath by means of the wafers and the PFA carriers. As shown in Table 111, the etch rate on the wafer surfaces in the hot ultrapure water bath is higher than that in the NH,0H-H202-H20 bath. This means the wafer surfaces become rougher more seriously in the hot ultrapure bath. The WIS-100 detects this roughness as a haze, so that the particle removal efficiency is degraded as the pH value increases when the above cleaning sequence is adopted. These results indicate that the particle removal efficiency is improved by the introduction of cold ultrapure water rinsing at 25C between the NH40H-H202-H20 cleaning and the hot ultrapure water cleaning. The particle removal efficiency of polystyrene latex spheres and ambient atmosphere particles shows the greater drop than that of city water and SiOz spheres when the above cleaning sequence is adopted.

Authorized licensed use limited to: TOHOKU UNIVERSITY. Downloaded on March 02,2010 at 02:26:28 EST from IEEE Xplore. Restrictions apply.

ITANO

er al.:

PARTICLE REMOVAL FROM SILICON WAFER SURFACE

265

THEpH VALUE AND

TABLE I11 ETCHRATEAT 80C ON C z WAFERS I N NH.,OH HOT ULTRAPURE WATER BATH H20,-H20 BATHA N D THE FOLLOWING
THE

I\
.075:1:5

NH40H H&

Bath

Hot U P W Bath

PH 9.3

*I

0.25:l : 5

8.9

0.35

8.0
( C Z Wafer

0.9

QJm Fig. 15. Influence of solution pH on the shape of 1.0 pm SiO, spheres by oxidation.

F. Oxidation of Polystyrene Latex Spheres

Fig. 15 shows the shapes of the SiOz spheres in each solution. In this experiment, the SiO, spheres of 1.0 pm in diameter are added to each solution at 80C and then the solution is dropped onto the wafer surfaces after 10 minutes has passed. The pictures are taken after the droplet has dried. The inorganic particles of SiO, spheres maintain their original shape in each solution as shown in Fig. 15. Further the particle removal efficiency of Fig. 14 is similar to that of Fig. 4. Fig. 16 shows the shapes of polystyrene latex spheres in each solution. The polystyrene latex spheres with a diameter of 1.866 pm are used in this experiment. When the pH value in the NH40H-H202-H20solution exceeds 9.3 (NH40H : H 2 0 2: H 2 0 = 0.25 : 1 : 5 ) , the organic particles of polystyrene latex spheres are oxidized and their surfaces turn into a gel, as shown in Fig. 16, resulting in the change of their shape. It is believed that the polystyrene latex spheres adsorbed on the wafer surfaces are also oxidized during the NH40H-H202-H20 cleaning. The decline of particle removal efficiency shown in Fig. 14 takes place when the pH value in the cleaning solution exceeds the specific level at which the shape of polystyrene latex spheres starts to be distorted. It is therefore possible that the shape and the surface condition of the particles have some effect on the capability of the solutions to remove those particles. On the other hand, the particle removal efficiency of polystyrene latex spheres shown in Fig. 4, which includes the cold ultrapure water rinsing at 25C between the NH40H-H202-H20 cleaning and the hot ultrapure water cleaning, shows no degradation-even when the pH value exceeds 9.3 - . The reason for this lack of degradation is not clear. We suspect that the cold ultrapure water rinsing at 25 "C following the NH40H-H202-H20cleaning has some effect on the capability to remove those oxidized particles. However, whenever the pH value exceeds 11.1 using TMAH-H202-H20solution which includes the cold ultrapure water rinsing at 25C following the TMAH-H202H 2 0 solution shown in Fig. 6, the polystyrene latex spheres can not be completely removed. Since the oxidizing power of H 2 0 2is enhanced as the pH value increases
In U.P.W
-1 . SL

PH 9.9 at BO'C

In N H q C H - W 1 :1:5
PH 9.3
"

0.25:l : 5

50 pm

Fig. 16. Influence of solution pH on the shape of 1.866 ptn polystyrene latex spheres by oxidation.

[ 181, we suspect that a great degree of oxidation of polystyrene latex spheres is caused by the higher pH of the TMAH-H202-H20solution.

G. Ambient Atmosphere Particles Outside of the Cleanroom Fig. 17 shows the pictures of particles adsorbed on the wafer surfaces from ambient atmosphere particles outside of the cleanroom. These particles feature various shapes since they are existing in the natural environment. In addition, they are composed of various constituents including inorganic and organic materials. Fig. 4 presents the experimental result when the NH40H-H202-H20cleaning is applied to the particles shown in Fig. 17. As shown in Fig. 4 , the particle removal efficiency varies to a great degree when the pH value of the cleaning solution exceeds 9 . 1 . Based on the findings concerning the polystyrene latex spheres, it is postulated that the surface of the organic particles turns to a gel and their shape becomes distorted. Fig. 17 also shows pictures of the particles which remain on the wafer surfaces after the cleaning process. Most of the residual particles feature the distorted shapes. Some residual particles have experienced surface oxidation and have changed to a gel. Therefore, the mixing ratio of 0.05 : 1 : 5 (NH40H : H 2 0 2: H,O), which keeps the pH level below

Authorized licensed use limited to: TOHOKU UNIVERSITY. Downloaded on March 02,2010 at 02:26:28 EST from IEEE Xplore. Restrictions apply.

266

IEEF TRANSACTIONS ON SEMICONDUCTOR MANCFACTURlhG. VOL 6. NO. 3. AUGUST 1993

Particles

frwn Enviro. Air

After Cleaning

WCm

Fig. 17. The adsorbed particles from ambient atmosphere outside of the cleanroom and their shape after high pH alkaline solution cleaning (NH,OH : HzOz: H,O = 1 : 1 : 5).

9 , is preferable for removing the particles in actual manufacturing conditions. IV. CONCLUSIONS 1) The mechanism for removing particles adsorbed on the wafer surfaces with the alkaline chemical plus H,02 solution has been explained. The particles adsorbed on the wafer surfaces are first lifted off when the wafer surfaces are etched. These particles are then electrically repelled from the surfaces by the alkaline solution. 2) It has been demonstrated in the experiments that the etch rate required for the alkaline chemical plus H 2 0 z solution to remove the particles adsorbed on the wafer surfaces is more than 0.25 nm/min for both the CZ and FZ wafers. The etch rate of both the CZ and FZ wafers becomes 0.3 nm/min when the mixing ratio of NH40HH202-H20solution is set at 0.05: 1 : 5 . This means the NH40H content in NH40H-H20,-H20solution can be reduced to 1 /20 of the conventional level of 1 : 1 : 5 . 3 ) The experiments have revealed that the surface micro-roughness of CZ wafer is degraded when the NH40H content in NH40H-H202-H20 solution is set above 0.1 : 1 : 5 . On the other hand, the surface micro-roughness of FZ wafer is not degraded. The reason for the absence of degradation of the FZ wafer is unknown. However, when the mixing ratio of N H 4 0 H - H 2 0 2 - H 2 0 solution is set at 0.05 : 1 : 5 , this degradation of the CZ wafer is not detected. It has therefore been determined that the mixing

ratio of 0.05 : 1 : 5 is effective in maintaining both surface smoothness of the wafer and particle removal efficiency. 4) It has been confirmed that the CZ wafer surfaces become rough more seriously in the hot ultrapure water bath just after the NH,0H-H,02-H,0 cleaning, resulting in the increase of haze and the degradation of the particle removal efficiency. However, this degradation of the C Z wafer surfaces has been reduced by the introduction of the cold ultra-pure water rinsing at 25C between the NH40H-H202-H20 cleaning and the hot ultrapure water cleaning. 5 ) It has been revealed that the particle removal efficiency of polystyrene latex spheres adsorbed on the wafer surfaces becomes degraded because the spheres are oxidized in the alkaline chemical plus H2O2 solution. When the mixing ratio of the NH10H-H,02-H,0 solution is set above 0.25 : 1 : 5 (pH : higher than 9.3), the surface of polystyrene latex spheres turns to a gel and their shape becomes distorted. The particle removal efficiency is reduced or levels off as this phenomenon proceeds. However, the artificial polystyrene latex spheres can be completely removed when the cold ultra-pure water rinsing at 25C is adopted between the NH,0H-H202-H20cleaning and the hot ultra-pure water cleaning. 6) The particle removal efficiency of ambient atmosphere particles adsorbed on the wafer surfaces becomes variable when the pH level of the alkaline chemical plus H,O, increases. This phenomenon takes place when the mixing ratio of the NH,0H-H202-H20 solution is set above 0.1 : 1 : 5 (pH : higher than 9.1). Based on the findings concerning the polystyrene latex spheres, it is postulated that the surface of the organic materials turns to a gel and their shape becomes distorted. However, the particles in the natural environment can not be fully removed by the simple modification of the rinsing process. 7) The above experimental results suggest that the mixing ratio of the NH40H-H20,-H,0 solution employed in RCA cleaning process to remove particles should be set at 0.05 : 1 : 5 (NH40H : HzOl : H 2 0 ) . Furthermore, it is essential to introduce the cold ultrapure rinsing between the NH,0H-H202-H20 cleaning and the hot ultrapure water cleaning in order to obtain both sufficient particle removal efficiency and sufficient surface smoothness. ACKNOWLEDGMENT This research has been carried out in the Mini Super Cleanroom of the Faculty of Engineering, and in the Super Cleanroom of the Laboratory for Microelectronics, Research Institute of Electrical Communication at Tohoku University. REFERENCES
[ I J T . Ohmi, Foresightedness in RCA cleaning concept and their present problem. Proc. 8th Workshop Ultru Clean Technology, Tokyo, Japan, Dec. 1990. pp. 5-15. 121 W . Kern and D. A . Poutinen, Cleaning solution based on hydrogen peroxide for use in silicon semiconductor technology. RCA Rev., vol. 3 1 . pp. 187-205. June 1970.

Authorized licensed use limited to: TOHOKU UNIVERSITY. Downloaded on March 02,2010 at 02:26:28 EST from IEEE Xplore. Restrictions apply.

ITANO et al.: PARTICLE REMOVAL FROM SILlCON WAFER SURFACE

261

[3] H. Mishima, T . Yasui, T. Mizuniwa, M. Yabe and T . Ohmi. Particle free wafer cleaning and drying technology, IEEE Trans. Semicond. Manufact., vol. 3 , pp. 98-108, Aug. 1989. [4] D. J . Riley and R. G . Carbonel, The deposition of liquid-based contaminants onto silicon surfaces, Proc. 36th Ann. Technical Meering of the Institute of Environmental Science. New Orleans, Apr. 1990. pp. 224-228. [5] D . J . Riley and R. G . Carbonel, The deposition of contaminants from deionized water onto hydrophobic silicon wafers, in Proc. 37th Annual Technical Meeting of the Institute of En\Yronmental Science, San Diego, pp. 886-891, May 1991. 161 T . Niida, Controlling for particle\ adhesion in liquids: a basic way of thinking, Chemical Engineering Instirute of Japan. Kansai Branch, Osaka, Japan, pp. 14-25, Sept. 1989. 171 A. Saito, K . Ohta and H. Oka, Particle deposition mechanism onto silicon wafer, Extended Abstracts of the 21st Conf on Solid State Devices and Materials, Tokyo, Aug. 1989, pp. 409-412. [8] A. Saito, K. Ohta and H. Oka, Particle deposition mechanism onto silicon wafer, Proc. 8th Workshop Ultra Clean Technologj, Tokyo. Japan, Dec. 1990, pp. 19-31. [9] R. W . Rosenberg, M. Itano, I. Kawanabe, F. W . Kern, Jr., M. Miyashita and T . Ohmi, Particles in ULSI grade chemicals and their adhesion on silicon surface. Proc. Microconramination 90, Santa Clara, CA, Oct. 1990, pp. 273-291. [IO] -, Particle adhesion and separation in RCA cleaning process, Proc. 8th Workshop Ultra Clean Technologj. Tokyo, Japan. Dec. 1990, pp. 35-55. [ I I ] -, Particle deposition and removal in wet cleaning processes for ULSI manufacturing, IEEE Trans. Semicond. Manufacr., vol. 5 , pp. 114-120, May 1992. [I21 T. Shimono, M. Morita, Y. Muramatsu and M. Tsuji, Device degradation by metallic contamination and evaluation of metallic contaminants, Proc. 8th Workshop Ultra Clean Technology, Tokyo, Japan, pp. 59-68, Dec. 1990. 1131 T . Shimono and T . Tsuji, A new cleaning solution for metallic impurities on the silicon wafer surface, Extended Abstracts of Spring Meeting of the Electrochemical Society. Washington, DC, pp. 278279, May 1991. [14] A. Kitahara and K. Furusawa, Chemistn of Dispersion arid Emulsion, 7th ed. Tokyo: Kougakutosyo Ltd., 1979. [I51 A. W . Adamson, Physical Chemistry of Surface. 4th ed. New York: Wiley, 1973. 1161 M. Miyashita, M. Itano, T. Imaoka. I . Kawanabe and T . Ohmi, Optimizing NH,0H/H,02 cleaning process for ultra-clean wafer surface preparation, Extended Abstracts of Spring Meeting of the Electrochemical Society, Washington DC, pp. 709-710, May 1991. [17] T. Ohmi, M. Miyashita, M. Itano, T . Imaokaand I. Kawanabe, Dependence of thin oxide films quality on surface microroughness, IEEE Trans. Electron Devices, vol. 39, no. 3, pp. 537-545. Mar. 1992. [I81 J . E. A. M. yan den Meerakker and M. H. yan der Staaten, A mechanistic study of silicon etching in NH,/HZOZ cleaning solution, J . Electrochem. Soc.. vol. 137, no. 4. pp. 1239-1243. Apr. 1990.

Frederick W. Kern, Jr. received the B.S. degree in chemical engineering from the University of Rochester, Rochester, New York. He is a Staff Engineer at IBM Corporation General Technology Division, Essex Junction, VT, responsible for process equipment design for high density memory devices. Currently he is on temporary assignment, as a Visiting Scholar, at the Solid State Electronics Laboratory, Department of Electronics at Tohoku University in Sendai, Japan. From 1974 until 1978, he worked at the Joseph C. Wilson Center for Technology of Xerox Corporation where he was responsible for the process development for the manufacturing of Polymeric toner dispersions and colloidal suspensions of sub-micron crystalline selenium. Mr. Kern has authored papers on equipment aerodynamics, particle transport mechanisms in cleanrooms. process equipment design and evaluation. He is a member of the American Institute of Chemical Engineers, the Institute of Environmental Sciences, the Fine Particle Society and serves on the advisory board for the Center for Advanced Materials Processing at Clarkson University in Potsdam. NY.

Masayuki Miyashita was born in Osaka, Japan, on June 24, 1961. He received the B.S. degree in chemistry form Ritumeikan University, Kyoto, Japan in 1984. He joined Hashimoto Chemical Co. Ltd. in 1984 and since then he engaged in the research and development of high quality fluoride chemicals. From September 1988 to August 1991, he was the visiting researcher of Faculty of Engineering of Tohoku University, where he was carrying out research works on gas phase etching of oxide films and wet cleaning process technologies

Mitsushi Itano is a researcher at Daikin Industries, LTD. in Osaka, Japan where he is involved in the research and development of high purity fluoride chemicals used in semiconductor manufacturing. He holds a B.S. and a M . S . degree in Environmental Engineering from Osaka University in 1978 and 1980, respectively. He was a Visiting Researcher on the Faculty of Engineering of Tohoku University from January 1990 to March 1991, where he engaged in the research and development of wet chemical cleaning process for use in semiconductor manufacturing.

Tadahiro Ohmi (M81) was born in Tokyo, Japan, on January 10, 1939. He received the B.S., M.S., and Ph.D. degrees in electrical engineering from the Tokyo Institute of Technology, Tokyo, in 1961, 1963, and 1966, respectively. Prior to 1972, he served as a Research Associate in the Department of Electronics at the Tokyo Institute of Technology, where he worked on Gunn diodes such as velocity overshoot phenomena, multi-valley diffusion and frequency limitation of negative differential mobility due to an electron transfer in the multi-valleys. high field transport in semiconductors such as unified theory of space-charge dynamics in negative differential mobility and block oscillators, and dynamics in injection layers. He is presently a Professor in the Department of Electronics, Faculty of Engineering, Tohoku University, where he is engaged in research on high-performance ULSI such as ultra high speed ULSI; Current Overshoot Transistor LSI, HBT LSI and SO1 on metal substrate; base store image sensor (BASIS) and high speed flat panel display; and advanced semiconductor process technologies, i.e.. ultra clean technologies such as high quality oxidation. Also, high quality metallization by low kinetic energy particle bombardment; very low temperature Si epitaxy having simultaneous doping capability by low kinetic energy particle bombardment; crystallinity film growth technologies from single crystal; grain size controlled polysilicon and amorphous due to low kinetic energy particle bombardment; highly selective CVD; highly selective RIE; high quality ion implantation having low temperature annealing capability, etc.. based on the new concept supported by newly developed ultra clean gas supply system: ultra high vacuum compatible reaction chamber with a self-cleaning function; ultra clean wafer surface technology: etc. He has 260 original papers and 190 patent applications. He received the Ichimura Award in Industry-Meritorious Achievement Prize in 1979, Teshima Award in 1987, Inoue Harushige Award in 1989, the Ichimura Prize in Industry-Meritorious Achievement Prize in 1990, and the Okochi Memorial Technology Prize in 1991, He serves as the president of the Ultra Clean Society. Dr. Ohmi is a member of the Institute of Electronics, Information and Communication Engineers of Japan, the Institute of Electrical Engineers in Japan, and the Japan Society of Applied Physics, and ECS.

Authorized licensed use limited to: TOHOKU UNIVERSITY. Downloaded on March 02,2010 at 02:26:28 EST from IEEE Xplore. Restrictions apply.