DOUBLE-RESONANCE-ENHANCED RAMAN-SCATTERING IN LASER-RECRYSTALLIZED AMORPHOUS-SILICON FILM

M.

C.

Lee,

C.

R.

Huang,

Y.

S.

Chang, and

Y.

F.

Chao

Department ofElectrophysics and Institute ofElectro Op-tical Engineering, National Chiao Tung University, Hsin Chu, Taiwan 30049,TheRepublic ofChina

(Received 6 April 1989;revised manuscript received 21 July 1989)

Amorphous silicon films after picosecond laser excitation have been investigated by Raman mi-croprobe analysis. The recrystallized Raman peak intensity can be more than four times that of crystalline Si. By carefully examining the annealed microstructure, the enhancement isinterpreted asa double-resonance effectdue tomicrocrystals.

I.

INTRODUCTION

The laser-induced phase transition in amorphous

sil-icon (a-Si) has been extensively studied in recent years for

fundamental interests and potential applications.

'

Us-ing laser annealing techniques on a-Si thin films, we

ob-served that the Raman intensity

of

the recrystallized peak can be higher than four times that

of

crystalline silicon

(c-Si). Bytaking the Raman intensity ratio

of

annealed Si

to c-Si

(I,

„„„„d/I,

s;) and tuning the probe wavelength from the visible to the ultraviolet (uv), the Raman enhancement is revealed. This enhancement in the visi-ble (blue-green) region does not appear to be due to the resonance Raman scattering

of

c-Sinear the Eo and

_E,

critical points. After examining the scanning-electron microscopy (SEM) pictures

of

picosecond laser excited a

Si films, we found that the annealed microstructure is within the range

of

Mie scattering resonance. However,

if

wejust consider the Mie absorption cross section, ' _the

calculated enhancement is not large enough. Because the Raman scattering involves absorption

of

incident light and scattering

of

Stokes and anti-Stokes photons, when both the incident and the scattered wavelengths are close

to the size

of

the annealed microcrystals, a double reso-nance could occur and give rise to the enhancement. Such an interpretation is made since the product

of

the Mie absorption and scattering cross sections indeed fit our measurements will except in the long-wavelength

re-gion which is due to significant interference e6'ects.

II.

EXPERIMENT

The details

of

the film preparation were described in

Ref.

3.

The film thicknesses are inferred from the

spec-trophotometric measurements to be

0.

8,

1.

2, and

1.

8pm, for the 20-,30-,and 45-min deposition time, respectively.

The experimental setup for pulsed-laser annealing and Raman measurements is essentially the same as that in

Ref.

8,except a dye laser (PRA LN-107 at 580nm)

of

600

ps was employed. After exciting the a-Sifilms with vari-ous Auences from

0.

1to

1.

4

J/cm,

the microstructure

of

them was examined under microscope. The Raman in-tensity was measured by microprobe

(10

pm at

1/e

)

(Ref.8)at

=2

mW which also serves as a monitor for the

local structure variation. All major laser lines from

ar-gon, He-Ne to He-Cd lasers, were employed. The

scat-tered signals were imaged onto a double monochromator

(Jobin Yvon U-1000). A multichannel detector

(Prince-ton Instruments

(IRY-1024G)

was used to increase the signal-to-noise ratio by detecting a wide spectral range

for long exposure.

For

the enhancement analyses, we take the Raman intensity ratio

of

annealed Si to c-Si at

each probe wavelength instead

of

the intensity itself. By

doing so, not only the wavelength dependence

of

the

Ra-man susceptibility and the detection system eKciency can be removed, but the probe power Auctuation isalso elim-inated. As the result

of

this normalization, the Raman enhancement is revealed.

III.

RESULTSAND DISCUSSION

In

Fig.

1, the Raman spectrum

of

c-Siexhibits a strong and sharp peak at 520 cm ' with a linewidth

of

-3

cm

',

while that

of

a-Si shows a very broad feature

(-60

cm ') around 475 cm

'.

As the ffuence on the a-Siis in-creased to

0.

16

J/cm,

the broad feature diminishes and a distinct peak energies at 513 cm ' with a large width. This peak gradually shifts to 518cm ' with the increas-ing fluence.

It

looks similar to that

of

c-Si after

0.

38

J/cm . This indicates that the recrystallization volume enlarges with the excitation. However, after excitation

~ 0.

4J/cm the Raman-enhancement eAect begins to

ap-pear.

For

extracting the enhancement

of

the recrystal-lized Raman line, the normalized intensity data obtained by diff'erent probe photon energies (they are much larger than the Raman shift) are plotted in

Fig.

2 to show whether it is due to the band-gap resonance about

3.

4 eV. When the excitation Auence isincreased from

0.

3to

1.

2

J/cm,

the enhancement peak only slightly shifts from

2.7 to 2.6 eV (457.9 to 476.5 nm). Its profile is sharp around

0.

4

J/cm,

and becomes broadened above

it.

This enhancement is obviously not due to the band-gap reso-nance but to the annealed microstructure

of

the film.

The detailed structure

of

three annealed films obtained by SEM are shown in Figs. 3(a)—3(c). The fine crystals formed in thicker films

(1.

2 and

1.

8 pm) are in similar sizes and shapes but the thinnest film (0.8 pm) contains lots

of

larger crystals. Because the substrate is glass, there is no epitaxial growth. The ball-like particles are

mixed in crystalline-amorphous structure. Thus, the

40 DOUBLE-RESONANCE-ENHANCED RAMAN SCATTERING

IN.

.

.

10421

c-Si

1.

1

J/c

_x0.

5

0.

38J

km

0.

3

J/cm

2

x1

0.

16

Jlcm

~RLl a

~~

JJg~

~.

~

~a~~'

~pmR~Sf ~

W~

~r

-

W~

.

~ 2QQ

a-si

320

370

I

420

I

4-7

0

I

520

I 5 I

570

620

RAMAN

SHIFT

[

cm

)

FIG.

1. The Raman spectra ofc-Si and 1.2-pm a-Sifilm obtained by 488-nm probe. The recrystallized Raman intensity increases

with the Auence that can show the annealing status directly. (X no.)indicate magnifications.

cA I 13 Q

Q3—

a 1.1 4

083

0.6 Jlcrn2 v

042

x _0.38 1.2 ~m film

standard Lorenz-Mie scattering formulas for spherical particles are used

to

fit the enhancement profile. The

normalized efficiency factors

of

absorption

(Q,

b,) and

scattering

(Q„,

«)cross sections, are given by

oo

Q,

b,

=,

g

(2p+1)[2

—

~2a„(x,

m)

—

1~' 2x

—

_I2b„(x,

_m)

—

_1~

_],

UJ QJ 1.

5

2.

0

2.5

3.

0

PROBE

PHOTON ENERGY

(eV)

3.

5

FIG.

2. The Raman enhancement

_(I,

„„„],

_d/I, $;) of 1.2 pm film after diA'erent Auences vsprobe photon energy. Solid lines

are the calculated _Q,b,

Q„,

«and the dashed line is the calculat-ed _Q,

„,

for 1.1

J/cm,

under asymmetric Gaussian distribution.

oo

Q„„,

=

_g

_{(2@+1)[~a„(x,}

_m)~

₊

_~b„(x,

_m)~2],

x

v=1

(2)

where

x

=~d/A,

isthe size parameter

of

particle with di-ameter d, and incident wavelength A,

.

Due to the optical

dispersion

of

c-Si, the corresponding complex refractive

indices, m

=n+ik,

are used to calculate the complex

electric and magnetic multipole expansion coefficients

of

the pth order

(a„and

b„)

in the Mie theory. However, the calculated enhancement for single-particle size is much narrower than what we observed.

Because

of

rapid nucleation by picosecond laser pulse, the annealed microcrystals are in difFerent sizes. In addi-tion they are not isolated but connected in a porous net-work as shown 111Flg. 3(d), tile X 10klllaglllflcatloll

SEM

micrograph. The real cause

of

the network formation is unclear, which may be ascribable to explosive crystalliza-tion and its generated shock wave. '

'"

In order to fit the

small size deviations o._{,(cr,) are introduced} besides the mean particle diameter. This assumption

of

the

asym-metric distribution isbased on the SEM annealed

micros-tructure and supported by the good fitting to the enhancement. Since recrystallization is more eA'ective from proper crystal size, pulsed annealing should pro-duce unequal size distribution.

Considering the submicrometer crystals as spheres

line in Fig.2)isnot as large as the measured results.

Be-cause the Raman process involves absorption

of

incident light and scattering

of

Stokes and anti-Stokes photons, a double resonance' issuspected. Since the first-order

Ra-man shift

of

c-Siissmall compared with the incident

pho-ton energy

(-2%),

when the incident wavelength isclose

to the particle size for Mie resonance, the scattered wave-length is also near the resonance. Thus, both the

incom-FIG.

3. SEMmicrographs of{a)0.8-, {b)1.2-,and {c) 1.8-pm a-SiAlms excited by a 580-nm dye laser

{1.

1J/cm~). There are

DOUBLE-RESONANCE-ENHANCED RAMAN SCATTERING

IN.

. .

10423 4 632.8 ~ 5]4.5 nm g3 a 363.8 LU X hl2-1.2em film

/

~ ~ Ul z1 X ~ a-~ I II

0,

0.4 0.8 1.2 FlUENCE(J/cm )

FIG.

4. The Raman enhancement vs annealing fluence. The sharp transition isresulted from a recrystallization layer which is about the probe penetration depth (300 A at 0.3 J/cm'). Above the transition, the large Auctuations are due to interfer-ence.

ing and the outgoing resonances could conceivably occur, giving rise to the enhancement. Taking it in proportion tothe product

of

the Mie absorption and scattering cross

sections, several annealed conditions (0.38, to

1.

1 J/cm )

are analyzed by choosing (d, o

„and

o,

)from (0.

09,

0.

03,

and

0.

01)

to (O.

l,

O.

l,

and

0.01)

pm. The calculated

Q,

b,

Q„,

« indeed give rise to the resonance in the wave-length domain which are in good agreement with our measurements (solid lines in

Fig.

2). Therefore, the

mi-crostructure effect is believed to be the cause

of

the enhanced Raman scattering.

From

Fig.

1, the higher Raman intensity results from the stronger excitation which produces the thicker

recry-stallization layer.

For

understanding its thickness effect, we only plot the Raman enhancement versus annealing Iluence for three difFerent probes in

Fig.

4 (all visible probes have similar behavior). As the annealed thickness approaches the probe penetration depth, the enhance-ment sharply rises about the transition fluence

(-0.

4

J/cm ). Above the transition, large Auctuations are ob-served in all visible probes but no such variations ob-served in the uv

(363.

8 nm) probe. Because the uv pho-ton energy (3.4 eV) is close to the silicon direct gap, due

to the strong optical absorption, it is reasonable to esti-mate the recrystallization thickness to be 300 A at

0.

3

J/cm;

If

the Raman cross sections

of

the annealed film and c-Si are not significantly different, then the effective scattering is inversely proportional to the probe penetra-tion depth.

For

the thin recrystallized layer (

(

0.

4

J/cm ), the scattering eKciency

of

the uv probe is larger than that

of

the visible probes. This isconsistent with the early enhancement rise

of

the

363.

8-nm probe and the low enhancement

of

the

632.

8-nm probe about

0.

3J/cm

.

However, for the thick recrystallized layer

()

0.

4J/cm ), the large absorption at

363.

8nm eventually suppresses all possible enhancement, so the normalized intensity tails

ofT' _to 1 (i.e., no enhancement variations). But weak ab-sorption in the long wavelength allows multipole reAection and/or scattering to occur, so that large

Auc-5.

0

I

~

₄₀

a

3.

0-UU

0

UJ

F20

x-UJ 0 0.8 x 1.2 ~m film 1.8 after 1J/cm

~

10-

_gl

lX I I

2.3

2.

7

3.

1

3.5

PROBE

PHOTON ENERGY(eV)

FIG.

5. The Raman enhancement vs probe photon energy. The thinnest film has the largest enhancement. No apparent enhancement beyond 2.8eV. The large difference below 2eV is due to interference. Curves are drawn through data points.

tuations are observed beyond

0.

4 J/cm for most

of

the visible probes.

As shown in the SEM pictures, the annealed micro-structure is different from one film thickness to another

under the same Auence. Since the Raman enhancement is closely related to the annealed microstructure, the results

of

0.

8-,

1.

2-, and

1.

8-pm samples are collected in

Fig.

5 for comparison. Between

2.

4 and 2.8 eV

(514.

5to

441.

6-nm blue-green region), the largest enhancement appears in the thinnest film and less prominent in the thicker

ones. But they all approach unity beyond

2.

8 eV, in-dependent

of

the film thickness. This confirms that

strong absorption indeed prevents any enhancing effect. However, the situation below 2 eV is complicated, which can vary from

0.

5 to more than 2 as compared with c-Si

depending on both the excitation fluence and the film thickness. An interference-enhanced Raman scattering

on a relatively transparent multilayer Ti02/SiO2 from

570 to 630 nm was observed by Craig et

al.

Because

of

weak absorption at

1.

96 eV (632.8 nm), the interference effect has to be considered in our experiment to account for the large enhancement variation. Nevertheless, for

probe wavelengths between

2.

4 and

2.

8 eV, the sharp enhancement feature can we11 be explained by the

Due to the annealed microstructure eA'ect, the Raman enhancement

of

a-Si films is well interpreted as a double resonance by using Mie scattering formulas between 2.4

and

2.

8 eV. In long-wavelength probes, the multiple in-terference eA'ect becomes important and causes intensity variations. However, strong absorption in short wave-lengths prevents any enhancement from happening.

We are grateful to Professor

Y.

C.

Lee for his valuable suggestions. %'e also appreciate Mr. S.

Y.

Wu for his technical help in taking the SEM pictures. The financial support from the National Science Council

of

the

Repub-lic

of

China isalso acknowledged.

~Amorphous Silicon Semiconductors —Pure' and Hydrogenated,

Vol. 95 ofthe Materials Research Society Proceedings, edited

by A. Madan, M. Thompson, D. Adler, and Y.Hamakawa {MRS,Pittsburgh, PA, 1987).

2M. Bohm, Solid State Technol.,September (1988).

M. C.Lee, C.

J.

Tseng, C.R.Huang, and

T.

H.Huang, Jpn.

J.

Appl. Phys. 26, 193(1987).

~A. Compaan and H.

J.

Trodahl, Phys. Rev. B 29,793(1984). 5D. V.Murphy and S.

R.

J.

Brueck, Opt. Lett. 8,494(1983).

S.

R.

J.

Brueck, in Laser Processing and Diagnostics, Vol. 39of

Springer Series in Chemica/ Physics, edited by D. Bauerle (Springer-Verlag, Berlin, 1984), p.446.

7R. A. Craig, G.

J.

Exarhos,

%.

T.Pawlewicz, and

R.

E.

Willi-ford, Appl. Opt. 26, 4193 (1987).

M. C.Lee, C.

R.

Huang, and C.C.Lin, Jpn.

J.

Appl. Phys. Suppl. 26,67(1987).

D.

F.

Edwards, in Handbook

_of

Optical Constants ofSolids,

edited by

E.

D. Palik (Academic, New York, 1985), pp. 547-569.

M. O.Thompson, G.

J.

Galvin,

J.

W.Mayer,

P.

S.Peercy,

J.

M. Poate, D. C. Jacobson, A. G. Cullis, and N. G. Chew, Phys. Rev.Lett.52,2360 (1984).

J. J.

P. Bruines,

R.

P. M. van Hal, B.H. Koek, M. P.A. Viegers, and H. M.

J.

Boots,Appl. Phys. Lett. 50, 507(1987). H.Chew and D.-S.Wang, Phys. Rev.Lett. 49, 490 (1982).