Characterisation of room temperature blue emmiting Si/SiO2 multilayers

www.elsevier.com/locate/optmat

Optical Materials 27 (2005) 1020–1025

Characterisation of room temperature blue emmitingSi/SiO2 multilayers

M. Modreanu a,*, E. Aperathitis b, M. Androulidaki b, M. Audier c, O. Chaix-Pluchery c

a University College Cork, National University of Ireland, Lee Maltings, Prospect Row, Cork, Irelandb IESL-FORTH, Heraklion, Crete, Greece

c LMPG-CNRS/INPG, BP 46, Saint-Martin d’Heres Cedex, F-38402, France

Available online 12 October 2004

Abstract

In this study we report results on a series of Si/SiO2 multilayers periods deposited by standard CMOS processes on thermally

oxidised Si substrates. Each period was formed by alternating a low-pressure chemical vapour deposition (LPCVD) nanocrystalline

silicon thin film and a SiO2 thin film obtained by atmospheric pressure chemical vapour deposition (APCVD). High-resolution TEM

(HRTEM) and Raman spectroscopy have been used for the physical characterisation of the nc-Si/SiO2 multilayers. The HRTEM

analysis revealed a columnar structure, with an average grain size of 15nm for the silicon layers. The Si–SiO2 interfaces were smooth

with a surface roughness for silicon layers less than 1nm, as estimated from HRTEM analysis. Raman measurements demonstrate

that the nanocrystalline silicon (nc-Si) layers are free-of-stress. Room temperature photoluminescence (PL) spectra, obtained by

using a 325nm continuous wave (CW) laser excitation, showed a broad blue peak centred on 440nm. The intensity of the blue

PL band increased with the number of periods in the nc-Si/SiO2 multilayers. The origin of this intense room temperature blue

PL band is discussed.

� 2004 Elsevier B.V. All rights reserved.

1. Introduction

Since the first report [1] on efficient visible light emis-

sion from porous silicon in 1990, much effort has been

devoted to investigations on the luminescence properties

of silicon-based structures, including porous silicon (PS)

[1,2], silicon nanoclusters (SNC) in amorphous SiO2

[2–4] and Si/SiO2 multilayers [5–7]. Although there is asignificant amount of experimental data available in

the recent years, the mechanism underlying the light

emission in silicon-based structures is still unclear. Sev-

eral models have been proposed to explain the lumines-

cence of PS and SNC/SiO2 systems, involving either the

defects at the Si/SiO2 interface or in SiO2 itself [8,9] or

0925-3467/$ - see front matter � 2004 Elsevier B.V. All rights reserved.

doi:10.1016/j.optmat.2004.08.056

* Corresponding author.

E-mail address: [email protected] (M. Modreanu).

the confinement of electrons and holes in the silicon

dots/nanoclusters [1].

Several deposition techniques have been used to ob-

tain Si/SiO2 multilayers: molecular beam epitaxy [10],

electron beam deposition [11] or alternating sequences

of low-pressure chemical vapour deposition (LPCVD)

of thin silicon layers and high temperature thermal oxi-

dation [12].A new approach for obtaining the efficient light

emitting nanocrystalline-Si (nc-Si)/SiO2 multilayers by

employing a technology compatible with the one used

for Si ICs is investigated. In this study we report results

on a series of Si/SiO2 multilayers periods deposited by

standard CMOS processes on thermally oxidised Si sub-

strates. High-resolution TEM (HRTEM) and Raman

spectroscopy have been used for the physical character-isation of the nc-Si/SiO2 multilayers. The low tempera-

ture and room temperature PL emission from these

M. Modreanu et al. / Optical Materials 27 (2005) 1020–1025 1021

nc-Si/SiO2 multilayers is reported here and the origin of

light emission is discussed.

2. Experimental

The new preparation technique for obtaining effi-

cient light emitting multilayer involves the use of two

low temperature standard complementary-metal-oxide-

silicon (CMOS) processes, LPCVD and APCVD.

Nanocrystalline-Si/SiO2 multilayers, up to five peri-

ods, have been prepared using alternative layers of

150A nominal thickness LPCVD (from pure SiH4 at

T = 625 �C) silicon thin film and 1000A nominal thick-ness APCVD SiO2 (from SiH4 and O2 at T = 400 �C)on 3’’ silicon wafer initially with 2000A thermal oxide

buffer layer. The samples were labelled A1, A2, A3

and A5 where the number indicates the number of Si/

SiO2 periods for each sample.

High resolution TEM observations (on a JEOL

200CX) were carried out on thin fragments obtained

by scratching the surface of samples with a diamondtip, which were afterwards deposited on copper grids

coated with a carbon film. Among these fragments some

of them corresponded to a thin cross-section of the

multilayer structure. Then, HREM images were re-

corded after orienting the Si substrate along a [110]

zone axis parallel to the (001)Si plane interface and

allowing to observe two sets of 111 Si fringes. Such

111 Si fringes were used as a scale for measuring accu-rately the different thicknesses of the alternating Si and

SiO2 deposited layers.

Raman spectra were collected using a Dilor XY

multichannel spectrometer and a CCD detector. Exper-

iments were conducted in micro-Raman mode at room

temperature in a backscattering geometry. The

514.5nm line of an Ar+ ion laser was focused to a spot

size smaller than 1lm2. The incident laser power wasadjusted to 5mW in order to avoid thermal effects.

The PL measurements were made at temperatures

ranging from 22K to 280K, using a 325nm (3.81eV)

CW He–Cd laser. The photoluminescence spectra were

dispersed by a 150g/mm grating monochromator and

detected by a calibrated cooled CCD detector.

Fig. 1. TEM bright (a) and (111)Si dark (b) field images and

corresponding electron diffraction pattern (c) of the SiO2–Si–SiO2

multilayers sample (i.e. sample A1). The multilayer is observed along a

direction close to the normal of the multilayer plane. The dark field

image was obtained by selecting part of the 111 reflection ring of Si.

3. Results and discussion

3.1. High resolution TEM results

Both the microstructure and structure of four SiO2–Si

multilayers deposited on (001) Si wafers, were investi-

gated by transmission electron microscopy (TEM). As

an advantage of the sample preparation method usedit is that observations either perpendicular or parallel

to the (001) Si interface plane can directly be compared.

For instance, both the Figs. 1 and 2, related to sample

A1, correspond to these two types of results. From

bright (see Figs. 1a and 2a respectively) and dark (see

Figs. 1b and 2b respectively) field images shown in these

figures, the crystalline Si layer appears to be constituted

of grains of less than 150A in diameter while the SiO2

layers exhibit homogeneous grey contrasts typical of

an amorphous state. For the plane view close to an ori-

entation of the incident electron beam normal to the

multilayer plane (see Fig. 1c), the corresponding selected

area electron diffraction pattern exhibits reflection rings111, 220 and 311 of crystalline Si and a broad diffuse

scattering ring due to amorphous SiO2 layers (centred

at about 1.5 A�1). From the cross-section view corre-

sponding to a [110] zone axis of orientation of the Si

substrate (see Fig. 2c), the successive SiO2, Si and SiO2

layer thicknesses from the Si substrate interface were

found to be 1433 ± 15A, 250 ± 2.5A and 1037 ± 11A.

The thickness uncertainties (<±1%) are related here tothose of TEM image magnifications. We have verified

that such thickness values are in agreement with thick-

ness values deduced from high resolution imaging where

the d1 1 1 spacing of the Si substrate fringes can be used

as a scale (i.e. spacing of 111 fringes as shown in insert

in the left side of Fig. 2c); for instance, we have found

that the thickness of the first SiO2 layer varies in a range

of 1425–1435A on account of the SiO2–Si layer interfaceroughness. The roughness of SiO2–Si layer interfaces

was estimated from high resolution imaging on thin

Fig. 3. TEM high resolution images of two Si layers in between SiO2

layers corresponding successively to the third and second Si layers

from the SiO2–Si substrate interface of the sample A3. Interfaces are

indicated by full and dotted lines. A Fourier Transform pattern of the

first Si layer is shown in insert.

Fig. 2. TEM bright (a) and dark (b) field images and corresponding

electron diffraction pattern (c) of the same multilayers sample as the

one shown in Fig. 1. The observed cross-section correspond to [110]

zone axis of orientation of the Si substrate. In this case only reflections

of the Si substrate are visible because diffracting features from the

multilayer are of much too weak intensity. The insert in the above right

part of the figure is a high resolution image of [110] zone axis for the Si

substrate observed on the thinner right part of the SiO2–Si substrate

interface.

1022 M. Modreanu et al. / Optical Materials 27 (2005) 1020–1025

sample parts. Fig. 3 shows two Si layers in between SiO2

layers corresponding successively to the third and sec-

ond Si layers from the SiO2–Si substrate interface of

the sample A3. Both types of crystalline Si and amor-

phous SiO2 layers are distinguished from differences in

their contrast variation, mainly those of Si interference

fringes and an orange skin-like contrast, respectively.

In the case of amorphous SiO2 layers, the contrast var-

iation results from the effect of the microscope transferfunction depending on the image defocus and on several

parameters such as spherical and chromatic aberrations

of the objective lens. From Fourier transform analyses,

all the fringes observed in the Si layers are 111 fringes.

As complete 111 rings were observed on Fourier trans-

form patterns (e.g. insert of Fig. 3) and on account of

complete reflection rings observed in Fig. 1, an isotropy

is implied for the crystallographic orientations of Sigrains. Besides, as some 111 fringe domains (arrows in

Fig. 3) appear to extend through all the Si layer thick-

ness, it might be concluded that the microstructure of

Si layers is (or tends to be) columnar. The roughness

of Si–SiO2 interfaces was estimated to be in the range

of 8–10A from the variations of the Si layer thicknesses

observed in Fig. 3 (i.e. 125–141A and 128–149A).

Therefore, as through thick sample parts of a cross-sec-

tion, all the thickness variations of a Si layer superim-

pose, the measured thickness is in this case equal to

the sum of the average layer thickness and the average

roughness of a Si–SiO2 interface. Therefore, for Si layersshown in Fig. 3, such values correspond to 141 and

149A and are in agreement with values measured on

thick sample parts, i.e. 141 and 151A, respectively (see

third and second Si layers from the substrate interface

in Fig. 4c). The Fig. 4 yields an overview of all the Si

and SiO2 thickness layers of the four samples which

have been used in the present study. Let us note, how-

ever, that such thickness measurements are only relatedto the observed sample fragments and do not indicate

whether smooth thickness variations occur on large

area.

3.2. Raman spectroscopy results

The first-order Raman spectra for monocrystalline

silicon consist of a one-phonon peak located at520.5cm�1 [13]. The transverse optical (TO) and the

longitudinal optical (LO) phonons in silicon are degen-

erated at the zone-centre. Therefore, this phonon

line will be called LO–TO phonon in this work. The

LO–TO shift provides information about the possible

presence of stresses in polysilicon thin films [14]. Fig. 5

shows the Raman spectra for the A1, A2, A3 and A5

samples. No noticeable LO–TO shift has been foundfor these samples, which indicates the absence of strains

in the nc-Si layers even for the sample with five Si–SiO2

periods.

Fig. 4. TEM bright field images of [110] zone axis cross-sections of the four nc-Si/SiO2 samples (a-A1, b-A2, c-A3 and d-A5) used in the present

study.

480 500 520 540 5600

100

200

300

400

500

600

I [a.

u.]

Wavenumber [cm-1]

A1

A2A3

A5

Fig. 5. Raman spectra for the samples A1, A2, A3 and A5. The 514nm

laser power was kept at 5mW in order to avoid heating of the samples.

350 375 400 425 450 475 500 5250

300

600

900

1200

PL (a

.u.)

Wavelength (nm)

280K

22K

100K

Fig. 6. Temperature dependence of the PL spectra for sample A1 (one

period nc-Si/SiO2). The dash and the dash-dot lines correspond to two

of the three-Gaussians deconvolution of the 22 K PL spectrum.

M. Modreanu et al. / Optical Materials 27 (2005) 1020–1025 1023

3.3. PL results

Fig. 6 shows the temperature dependence of PL spec-

tra for sample A1 (one period nc-Si/SiO2) obtained

using the excitation of the CW 325nm He–Cd laser.

The 22 K PL spectrum can be deconvoluted with three

Gaussians: one violet PL band centred at 409nm (see

the dash line in Fig. 6), and two blue PL bands, one cen-

tred at 436nm (see dash-dot line in Fig. 6) and the sec-

ond blue band centred at 485nm (not shown in Fig. 6).When the temperature increases from 22K to 280K the

PL spectra are red-shifted and the violet PL band cen-

tred at 405nm is quenched. There is only a slight de-

crease in the PL intensity of the blue band centred at

435nm when the room temperature is reached inside

the cryostat.

Figs. 7 and 8 show the temperature dependence of the

PL spectra for the sample A2 (two periods nc-Si/SiO2)

and A5 (five periods nc-Si/SiO2). It can be observed that

the PL intensity for sample A2 and A5 increases with the

number of nc-Si/SiO2 periods and red-shifts as the tem-

perature increases. When compared to sample A1, the

violet band is now centred at 411nm and the first blue

band becomes much larger and centred at 442nm. Thesecond blue band (485nm for sample A1 see Fig. 6) is

not present, probably being obscured by the first one.

A new ultraviolet band appears for sample A2 and A5

centred at 390nm. This band becomes very week at

room temperature (see Figs. 7 and 8).

350 375 400 425 450 475 500 525

2000

4000

6000

8000

10000

12000

14000

16000

18000PLE=325nm

PL (a

.u.)

Wavelength (nm)

280K

200K

100K

22K

Fig. 7. Temperature dependence of the PL spectra for sample A2 (two

periods nc-Si/SiO2).

350 375 400 425 450 475 500 525

5000

10000

15000

20000

25000 PLE=325nm

PL (a

.u.)

Wavelength (nm)

22K

100K

200K

280K

Fig. 8. Temperature dependence of the PL spectra for sample A5 (five

periods nc-Si/SiO2).

1024 M. Modreanu et al. / Optical Materials 27 (2005) 1020–1025

The origin of light emission in nc-Si is still under de-bate. Among the different mechanisms proposed for the

radiative recombination processes the most likely ones

are related to the direct band-to-band recombination

(usually named ‘‘quantum confinement models’’-QCM,

because the emission features strongly depend on the

size of the nanostructures [15]) and the exciton recombi-

nation via interface luminescent levels (usually called

‘‘surface recombination models’’-SRM [16]). The originof violet–blue light in Si/SiO2 multilayers is unclear. As

the TEM results showed that the nc-Si layer appeared to

be constituted of grains of greater than 100A, the QCM

cannot be used to explain the violet–blue light emission

from the samples investigated in this work. The rough-

ness of Si–SiO2 interfaces is estimated from TEM meas-

urements in the range of 8–10A so is not expected to

have a major contribution to the light emission mecha-nism. The Raman results have demonstrated the absence

of strains in the nc-Si layer so strain effects can also be

ruled out. Recently Wu [17] have reported a blue dou-

ble-peak PL centred around 416nm and 437nm in a

number of Si nanostructures and they concluded that

Si vacancies in silicon nanocrystals are responsible for

the blue light emission. Their conclusion is in agree-

ments with our findings which ruled out quantum con-

finements and surface roughness contributions to the

blue light emission.The origin of violet–blue light in Si/SiO2 multilayer remains unclear and further investiga-

tions are required in order to elucidate the emission

mechanism in this part of the spectrum.

4. Conclusions

In summary, a new approach was used for obtainingefficient light emission from nc-Si/SiO2 multilayer by

employing a technology compatible with the one used

for Si ICs. This approach involved the use of two stand-

ard CMOS processes, namely LPCVD and APCVD and

at temperatures as low as 625 �C.The HRTEM analysis revealed a columnar structure,

with an average grain size of 150A for the silicon layers

and smooth Si–SiO2 interfaces. Raman measurementsshowed the absence of stress in the nc-Si layers.

An intense room temperature blue light emission lo-

cated at 440nm can be clearly observed in nc-Si/SiO2

multilayers. The PL blue band is red-shifted and its

intensity increases with the increase of the number of

nc-Si/SiO2 periods. The origin of blue light in Si/SiO2

multilayers can be related to the presence of Si vacancies

within the nanocrystals grains.

Acknowledgement

This work was partial supported by Enterprise Ire-

land grant IF/2002/624.

References

[1] L.T. Canham, Appl. Phys. Lett. 57 (1990) 1046.

[2] A.G. Gullis, L.T. Canham, G.M. Williams, P.W. Smith, O.D.

Dosser, J. Appl. Phys. 75 (1994) 493.

[3] T. Shimizu-Iwayama, K. Fujita, S. Nakao, et al., J. Appl. Phys.

75 (1994) 7779.

[4] S. Hayashi, T. Nagareda, Y. Kanzawa, K. Yamamoto, Jpn. J.

Appl. Phys. 32 (1993) 3840.

[5] C. Ternon, F. Gourbilleau, C. Dufour, J.L. Doualan, B. Garrido,

J. Lumines. 99 (2002) 361.

[6] Z. Ma, L. Wang, K. Chen, W. Li, L. Zhang, Y. Bao, X. Wang, J.

Xu, X. Huang, D. Feng, J. Non-Cryst. Solids 299–302 (2002) 648.

[7] M. Modreanu, M. Gartner, E. Aperathitis, N. Tomozeiu, M.

Androulidaki, D. Cristea, P. Hurley, Physica E 16 (2003) 461.

[8] H. Nishikawa, E. Watanabe, D. Ito, M. Takiyama, A. Liki, Y.

Ohki, J. Appl. Phys. 78 (1995) 842.

[9] Y. Kanemitsu, T. Futagi, T. Matsumoto, H. Mimura, Phys. Rev.

B 49 (1994) 14723.

[10] D.J. Lockwood, Z.H. Lu, J.M. Baribeau, Phys. Rev. Lett. 76

(1996) 539.

M. Modreanu et al. / Optical Materials 27 (2005) 1020–1025 1025

[11] S. Nihonyanagi, K. Nishimoto, Y. Kanemitsu, J. Non-Cryst.

Solids 299–302 (2002) 1095.

[12] P. Photopoulos, A.G. Nassiopoulou, D.N. Kouvatsos, A. Trav-

los, Mater. Sci. Eng. B 69–70 (2000) 345.

[13] A. Zwick, R. Carles, Phys. Rev. B 48 (1993) 6024.

[14] E. Anastassakis, J. Appl. Phys. 86 (1999) 249.

[15] S. Guha, M.D. Pace, D.N. Dunn, I.L. Singer, Appl. Phys. Lett.

70 (1997) 1207.

[16] Y. Kanemitsu, S. Okamoto, Phys. Rev. B 56 (1997) R1696.

[17] X.L. Wu, S.J. Xiong, G.G. Siu, G.S. Huang, Y.F. Mei, Z.Y.

Zhang, S.S. Deng, C. Tang, Phys. Rev. Lett. 91 (2003)

157402-1.