Self-assembled nc-Si/a-SiNx: H quantum dots thin films: An alternativesolid-state light emitting material

Basudeb Sain, Debajyoti Das n

Nano-Science Group, Energy Research Unit, Indian Association for the Cultivation of Science, Jadavpur, Kolkata 700032, India

a r t i c l e i n f o

Article history:Received 22 July 2014Received in revised form8 September 2014Accepted 9 September 2014Available online 22 September 2014

Keywords:Photoluminescencenc-Si/a-SiNx:H QDs thin filmsHetero-structured dielectric materialQuantum confinement effectNonradiative dangling bondsICP-CVD

a b s t r a c t

Nanocrystalline silicon quantum dots (QDs) of varying size from �5.4 to 2.2 nm embedded in amorphoussilicon-nitride matrix (nc-Si-QDs/a-SiNx:H) were prepared via ICP-CVD (13.56 MHz), using a mixture of silane(SiH4), ammonia (NH3) gases in H2-diluted plasma, by changing the low deposition pressure in a narrow range,10–50mTorr at low substrate temperature (250 1C). Strong visible photoluminescence (PL) tunable over asignificantly wide range (1.67–3.02 eV) has been observed. The origin of the PL is mostly attributed to band-to-band recombination due to the quantum confinement effect (QCE) in the nanocrystalline silicon QDs. A largeamount of atomic hydrogen flux that originates due to the high degree dissociation of the gas molecules inhigh-density inductively coupled plasma (ICP) passivates well the nonradiative dangling bonds and helps ingrowing plenty of ultra-nanocrystallites that demonstrate intense visible photoluminescence. The red-shift ofRaman peak and the corresponding line broadening have been associated to the confinement of opticalphonons within the nc-Si QDs. The widening of band gap and tune-ability of visible photoluminescence over anotably wide range along with significantly high electrical conductivity of the material demonstratesenormous promise for its utilization in the fabrication of effective solid-state light emitting devices.

& 2014 Elsevier B.V. All rights reserved.

1. Introduction

Light emission from silicon has been identified as a topic of generalscientific interest. Owing to its indirect band gap of 1.12 eV, silicon inthe bulk is considered as a material of very poor optical radiativeefficiency and can only produce light below the visible range. How-ever, when silicon is in the form of low dimensional structures likenanocrystal (nc-Si) or quantum dot (Si-QD), nanowire (Si-NW), super-lattices, etc., it reveals improved light emission at room temperature[1–17]. Such reports have claimed efficient photonic applications ofsilicon in future, where silicon-based light-emitting diodes representpromising candidates for the next generation of full-color flat paneldisplays, optical interconnections, telecommunications, and lasers. Theadvantages of silicon-based light emitting devices include comple-mentary metal-oxide-semiconductor compatibility, system feasibility,and their low cost of fabrication. Although a variety of emission colorsfrom porous silicon [11] and nanocrystalline silicon [17–19] show asufficiently high efficiency for applications, the tuning of emissioncolor, particularly in the short-wavelength region, continues to be achallenge because the silicon/dielectric interface plays an importantrole in the formation of radiative states [20]. In the recent past, a greatdeal of research on silicon nanocrystals embedded in a silicon oxide

matrix has been conducted because of their potential for applicationsin silicon-based optoelectronic devices. However, silicon oxide as anefficient matrix material that hosts silicon nanostructures has limita-tions in device fabrication because of several reasons reported in theliterature [20–22]. Nanocrystalline silicon embedded in hydrogenatedamorphous silicon-nitride matrix (nc-Si/a-SiNx:H) is currently attract-ing considerable attention as a candidate for efficient light emittingdevices [2,9,10,22,23] wherein the matrix silicon-nitride (SiNx) playsan important role because of its lower tunneling barrier and higherdielectric constant.

In this work, a tunable wide range (1.67–3.02 eV) visible photo-luminescence is being demonstrated from inductively coupledplasma synthesized nc-Si/a-SiNx:H QDs thin films by varying thedeposition pressure in a narrow range at low substrate temperature�250 1C. A comprehensive correlation between photoluminescence(PL) and structural properties has been made from which quantumconfinement effect (QCE) has been assigned to be mostly respon-sible for the light emission.

2. Experimental section

2.1. Material synthesis

Inductively Coupled Plasma-CVD technique in radio frequ-ency (13.56 MHz) low-pressure plasma was employed to prepare

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/jlumin

Journal of Luminescence

http://dx.doi.org/10.1016/j.jlumin.2014.09.0230022-2313/& 2014 Elsevier B.V. All rights reserved.

n Corresponding author. Fax: þ91 3324732805.E-mail address: erdd@iacs.res.in (D. Das).

Journal of Luminescence 158 (2015) 11–18

silicon-nitride films on p-type (100) silicon substrate, Cornings

Eagle2000™ glass and carbon coated copper microscope grids. Amixture of silane (SiH4), ammonia (NH3), and hydrogen (H2) wasused as reactant gas source, maintaining a flow rate ratio SiH4:NH3:H2¼2 sccm:0.6 sccm:38 sccm. The substrate temperatureduring deposition was fixed at 250 1C, while the chamber pressurewas varied from 10 to 50 mTorr. Other process parameter includesRF power 500 W. No post-annealing process was employed aftergrowing the silicon nitride films.

2.2. Characterizations of the material

Thickness of the films was determined by a Dektak 6M stylusprofiler. For uniform comparison, thickness of all the films wasmaintained at �4000 Å. The room temperature PL spectra wereobtained using a He–Cd laser source, with a uniform excitationof 325 nm. The luminescence spectrum was monitored using aTRIAX 320 monochromator fitted with a cooled Hamamatsu R928photomultiplier detector. The Raman spectra were obtainedby a Renishaw inVia Micro-Raman spectrophotometer (Serial no.12W143) at room temperature in a backscattering geometry, using514 nm Arþ Laser as the excitation source, at a power density of2 mW/cm2. High-resolution transmission electron micrographs(HR-TEM) as well as selected-area electron diffraction (SAED)patterns were obtained on �30 nm thick samples deposited oncarbon coated copper microscope grids supplied by Pacific Grid-Tech, USA, using a JEOL-JSM2010 transmission electron micro-scope operating at 200 kV. To observe the surface morphology, aVeeco (di CP-II) atomic force microscope (AFM; Bruker AXS,Mannheim, Germany) has been used. The optical absorption andreflection measurements in the UV–visible region at room tem-perature were performed using a double beam spectrophotometer(Hitachi 330, Japan). The silicon–hydrogen–nitrogen bondingstructure of the films was investigated from the infrared vibra-tional spectra obtained by a Nicolet Magna-IR 750 FTIR spectro-meter. Electrical conductivity of the samples was measured withcoplanar Al electrodes deposited by thermal evaporation at roomtemperature, and the measurement was carried out in vacuum of�10�6 Torr, using a Keithley 6517A electrometer, after annealingthe samples at 450 K for 1 h. The evacuation and the thermalannealing processes were adopted to eliminate the effects ofadsorbed gases, moisture and light induced degradation, if any.

3. Results

Fig. 1 shows the photoluminescence (PL) spectra of nc-Si/a-SiNx:H QDs thin films deposited by inductively coupled plasma atdifferent pressures, p, varying from 10 to 50 mTorr. All thesesamples showed intense photoluminescence (PL) of individualcolor in the visible range. With the increase of pressure the PLpeak energy has been found to be blue-shifted. The PL spectra forsamples p10 and p20 contain two prominent peaks in the nearvisible region, whereas, the remaining samples (p30, p40 and p50)show single peak photoluminescence at higher energies. Therelative intensity ratio of the lower energy peak of the samplesp10 and p20 decreases with the increase of pressure and finallydiminishes from the PL spectrum of sample p30. The PL peakenergy shifts from �1.67 to 3.02 eV and the PL peak intensity(height) increases systematically with increasing chamber pres-sure from 10 to 50 mTorr, as shown in Fig. 1. On varying thesubstrate temperature between 400 and 150 1C we have pre-viously reported photoluminescence from nc-Si/a-SiNx:H thinfilms over the energy range 1.66–2.47 eV [9]. The wider variationof PL peak energy (�1.67 to 3.02 eV) with chamber pressure in thepresent work is very much promising for its effective application

into light emitting devices. All the individual PL spectra are notbeing found as symmetric; however, the shape of the two mosthigh energy spectra (samples p40 and p50) can be considered asroughly symmetric Gaussian. Deconvolution of the remainingthree, corresponding to the samples p10, p20 and p30, might behelpful in order to find the different origins of photoluminescence.

The optical density data of the samples prepared at differentpressures was obtained from the absorbance and reflectancespectra of the films in the UV–vis region. The optical gap (Eg)was estimated from Tauc's plot [24] and its variation withpressure, p, is shown in Fig. 2. The gradual widening of opticalgap of the hetero-structure in the range 2.02–3.05 eV is beingconsidered as a result of band gap engineering in the material byplasma processing and the large variation of the absorption gaphas been observed by varying only the pressure within a narrowrange, 10–50 mTorr.

Fig. 3a represents the typical low magnification HR-TEM image ofthe sample prepared at p¼40 mTorr that identifies a homogeneousdistribution of nanocrystalline silicon quantum dots (nc-Si QDs)within the amorphous silicon nitride matrix. Fig. 3b represents themagnified view of single QD, which is spherical in shape and thecrystal planes are noticeable within it. In general, quantum dots aredenser material and relatively dark in appearance, randomly dis-tributed within a less dense matrix, here a-SiNx:H. The bright diffuse

Fig. 1. Room temperature photoluminescence (PL) spectra, under 325 nm excita-tion by He–Cd laser of nc-Si/a-SiNx:H QDs thin films deposited by ICP-CVD atdifferent pressures, p, varying from 10 to 50 mTorr.

Fig. 2. Variation of optical (Tauc) gap, Eg, for nc-Si/a-SiNx:H QDs thin films as afunction of deposition pressure, p.

B. Sain, D. Das / Journal of Luminescence 158 (2015) 11–1812

rings with absence of spots in the selected area electron diffraction(SAED) pattern (Fig. 3c) indicate the existence of tiny crystallinegrains with two prominent orientations, o1114 and o2204 ,having interplaner spacing 0.313 nm, and 0.198 nm, respectively,while the less prominent ring represents o3114 orientation withinterplaner spacing 0.165 nm. The average diameter of the wellseparated nanocrystalline silicon quantum dots (nc-Si QDs) wasestimated to be �2.7 nm for the sample p40 with a number densityof �1.2�1012 cm�2.

The surface morphology and roughness of the deposited filmswere studied using an atomic force microscope (AFM) and thecorresponding 2D topographic images for the films prepared atp¼20 and 40 mTorr are shown in Fig. 4. Both the surfaces containclusters, very few isolated grains, and pits; however, the voids arenot clearly visible in the surfaces, providing an overall goodpacking of the networks. The average cluster size seems to reduceon increasing pressure providing smaller surface roughness. Theroot mean square surface roughness (δ), as estimated from thestatistical analysis of AFM images, has been found to reduce from3.5 to 2.2 nm when pressure increases from 20 to 40 mTorr.

First order Raman spectra of the samples deposited at differentp have been displayed in Fig. 5. With increase of p from 10 to50 mTorr, a continuous red-shift of peak frequency associated withasymmetric broadening in the line shape was identified, attribut-ing modification in the nature of the network structure withsimultaneous reduction in size of the crystalline grains. Thequantitative estimation of the crystalline volume fraction in eachsample was performed from Raman spectra by deconvolution ofeach spectrum into three Gaussian components corresponding tonanocrystalline (nc-Si), amorphous (a-Si) and the intermediateultra-nanocrystalline (unc-Si) components and the typical decon-volution corresponding to the sample p10 has been shown in theinset of Fig. 5. The volume fractions of the individual componentswere estimated and the total crystalline volume fraction (XC) wasobtained as: XC¼XncþXunc [24]. The variations of XC and Xunc with

pressure are shown in Fig. 6a. The reduction in overall crystallinity(corresponding to XC varying from 75% to 46%) along withsimultaneous increase in the ultra-nanocrystalline fraction, Xunc,from 18% to 27% identifies an exciting structural modification ofthe network with increasing deposition pressure, which in turnwas controlled by the variation of nitrogen incorporation in thenetwork. The enhanced magnitude of ultra-nanocrystalline com-ponent indicates the favored growth of tiny crystallites withincreasing pressure. In addition, the consecutive blue-shift of thePL band can be correlated with the red-shift of TO crystallineRaman peak identifying continuous size reduction of the nano-crystallites on increasing pressure. The nanocrystalline peak posi-tion was observed at �514.63 cm�1 for the sample deposited atp¼10 mTorr, whereas, it shifted to �511.35 for the sample atp¼50 mTorr.

The average grain size has been estimated from Raman spec-trum using the following relation [24]:

½ωL�ω0�2þΓ0

2

� �2

ffi 13L

expð�π2Þ ð1Þ

where ωL is the frequency of the crystalline like mode for ananocrystal of size L. The values of ω0 and Γ0 are 520 and3.5 cm�1, respectively, for crystalline silicon. In our previousworks, a comparison between the estimated size from XRD, HR-TEM and Raman using Eq. (1) was reported and in all cases a closeproximity in grain sizes estimated from above mentioned threedifferent processes was observed that established the generalacceptance of the formula [in Eq. (1)] for finding out the grainsize from Raman spectra [24,25]. The size of nc-Si QDs decreasesfrom �5.4 nm at p¼10 mTorr to �2.2 nm at p¼50 mTorr and itsnature of variation has been shown in Fig. 6b.

Fig. 7a shows the typical FTIR spectrum of the film prepared atp¼30 mTorr, where the characteristic features of silicon nitride,notably the Si–N symmetric and asymmetric stretching modes

Fig. 3. (a) Low magnification HR-TEM micrograph of the nc-Si/a-SiNx:H QDs films prepared at p¼40 mTorr, (b) high magnification image of a single Si quantum dot (QD)and (c) the corresponding SAED pattern, demonstrating three different rings of increasing diameter, concerning (111), (220) and (311) c-Si planes, respectively.

B. Sain, D. Das / Journal of Luminescence 158 (2015) 11–18 13

at �480 and �850 cm�1, respectively, Si–H stretching modesat �2150 cm�1, the Si–H wagging mode at 640 cm�1 and the N–Hwagging/rocking mode at �1170 cm�1, have been exhibited[26,27]. With the increase of the deposition pressure, the inten-sities of the Si–N absorption peaks increase, whereas, that of theSi–H absorption peaks decrease. The decreased intensity of Si–Hwagging peak with increasing deposition pressure has been shownin the inset of Fig. 7a. The evolution of the Si–N asymmetricstretching absorption spectra at �850 cm�1 with pressure hasbeen shown separately in Fig. 7b. The absorption intensity of theSi–N asymmetric stretching vibration increases monotonicallywith increasing p. Besides the increased amplitude, the peakposition of the Si–N asymmetric stretching peak shows a slightblue-shift when p is decreased (marked by a arrow in Fig. 7b).Cheng et al. [28] considered it as an outcome of the inclusion ofhigher electro-negative nitrogen atoms in the network. Theabsorption coefficient of Si–H stretching mode vibration around�1925 to 2300 cm�1 has been plotted separately for the filmsdeposited at different pressures between 10 and 50 mTorr, asshown in Fig. 7d which demonstrates a continuous reduction ofabsorption intensity with increasing p.

The bonded nitrogen content x in SiNx is estimated from the Si–N asymmetric stretching mode vibration (�850 cm�1), as

x¼ AωNSi

ZαðωÞdω

ωð2Þ

where the oscillator strength, Aω¼6.3�1018 cm�2 and NSi¼5�1022 cm�3, is the atomic density of crystalline silicon [29]. Thenitrogen content increases monotonically with the increase ofdeposition pressure, as shown in Fig. 7c. At p¼10 mTorr, theestimated x value was �0.18, which increased with increasing pand attained �0.69 at p¼50 mTorr. The bonded hydrogen contentCH (in at%) was estimated from the absorption peaks of the Si–H

Fig. 4. 2D AFM images of the films p20 and p40 prepared at deposition pressure, p¼20 and 40 mTorr, respectively. The length scale of both the images is in the unit of μm(i.e. area is 4 μm�4 μm).

Fig. 5. 3D view of the first order Raman spectra for different p. Red-shifting of peakfrequency and widening of Raman band have been directed by arrow. Insetpresents deconvolution of the first order Raman spectrum corresponding to samplep10, into three satellite components.

Fig. 6. (a) Variations of the crystalline (XC) and ultra-nanocrystalline volumefraction (Xunc) with deposition pressure, p. (b) Variations of the size of nc-Si QDsestimated from first order Raman spectra with p.

B. Sain, D. Das / Journal of Luminescence 158 (2015) 11–1814

wagging mode at �650 cm�1 and N–H stretching mode at�3300 cm�1 using the following formula:

C H ¼ AωNSi

ZαðωÞdω

ω

� �� 100% ð3Þ

where Aω¼1.6�1019 cm�2 for Si–H wagging and 2.8�1020 cm�2

for N–H stretching, NSi¼5�1022 cm�3 [30,31]. The hydrogencontent decreased monotonically, opposite to the variation ofnitrogen content with the increase of pressure, as demonstratedin Fig. 7c. At p¼10 mTorr, the estimated value of CH was �9.2 at%that gradually decreased to �3.7 at% at p¼50 mTorr.

For efficient light emitting application of a material, significantelectrical conductivity is always needed. Hence, it was wise tomeasure the electrical conductivity simultaneously. Fig. 8 presentsthe variation of room temperature electrical conductivity (σ) of nc-Si/a-SiNx:H QDs thin films with change in deposition pressure. Atp¼10 mTorr, σE3.77�10�3 S cm�1 which is significantly highcompared to the similar undoped silicon films reported in theliterature [25]. The electrical conductivity of the undoped nc-Si/a-SiNx:H QDs thin films strongly depends on the film crystallinity.The large value of σ of the undoped nc-Si/a-SiNx:H film atp¼10 mTorr can be correlated to the very high crystallinity(�75%) of the network. With the increase of p, σ decreasesmonotonically when the network is identified to be graduallyamorphous dominated. The lowest dark conductivity of the nc-Si/a-SiNx:H QDs thin film is found to be �8�10�6 S cm�1 atp¼50 mTorr, corresponding to XC�46%.

4. Discussion

The overall shift of PL spectra towards higher energy withincreasing chamber pressure could be considered as a conse-quence of quantum confinement effect (QCE) occurring in thenanostructured system, as the size of the Si QDs had beenestimated to reduce systematically with chamber pressure(Fig. 6b). However, the asymmetric Gaussian nature of the PLspectra and the appearance of corresponding shoulders and

Fig. 7. (a) Typical FTIR spectrum of nc-Si/a-SiNx:H QDs films deposited at p¼30 mTorr. Inset presents the magnified view of the variation of Si–H wagging mode. (b) Theintensity variation of Si–N asymmetric stretching absorption band. (c) Variation of bonded nitrogen content, x in SiNx and hydrogen content CH (in at%), with p.(d) Absorption co-efficient spectrum of Si–H stretching band for films with different p.

Fig. 8. Variation of electrical conductivity, σ, of nc-Si/a-SiNx:H QDs thin films as afunction of deposition pressure, p.

B. Sain, D. Das / Journal of Luminescence 158 (2015) 11–18 15

additional peaks for the samples p10, p20 and p30 indicate thepresence of many other origins of photoluminescence [32] to co-exist with band-to-band recombination due to QCE. Quantumconfinement effect occurs when dot dimension is of the order ofthe Bohr radius, rB (for silicon rB�5 nm) of a material. In addition,radiative defects in the films could be the other origin of PL forindirect band gap material like, silicon. The major radiative defectenergy levels at 1.8, 2.4 and 3.0 eV were reported earlier forsilicon-nitride films [33,34]. Besides this, oxygen related defects,inherently residing within Si-based micro- and nano-structuredthin films, also play vital role in the appearance of visibleluminescence from silicon nanocrystal [35]. For defect related PL,the peaks should appear only near the corresponding defectenergy levels. Therefore, the PL spectra with variable peak posi-tions should mostly be attributed to the quantum size effect.

For a detailed understanding regarding the origin, the PLspectra corresponding to the samples p10, p20, and p30 havebeen deconvoluted into possible satellite components and theseare shown in Fig. 9. The component (Q1) at �1.67 eV for p10 mayresult from band-to-band recombination in nc-Si QDs and thatappears as (Q2) at �1.95 eV for p20 and as (Q3) at �2.24 eV forp30 sample. The origin of the satellite band (a1) at �1.5 eV forsample p10 is not clear to us. In one way, it may be attributed tothe recombination at defect energy level; however, this band isbeing found to appear as (a2) at �1.53 eV for sample p20 and thisslight shift toward higher energy may rise question regarding itsorigin as defect states. Alternatively, the components (a1) and (a2)may be considered to originate from similar quasi direct band-to-band recombination in a separate group of Si nanocrystals of largersize, additionally present in the network. This narrow PL bandbecomes further narrower, less intense and blue-shifted in (a2)due to the average size reduction of this group of larger size Sinanocrystals and their presence in lower density within thenetwork for sample p20, grown at higher pressure. The (b1),(b2) and (b3) components for samples p10, p20 and p30, respec-tively, at �1.8 eV can be attributed to the following recombinationprocesses: between conduction edge of the intrinsic band and Nþ

4level, and between the Nþ

4 and N02 states. It has been noted that

the intensity of �1.8 eV PL component decreases with the increaseof pressure from 10 to 30 mTorr and diminishes from the PLspectrum at 40 mTorr sample that basically indicates the presenceof reduced number of defects at higher pressure.

According to effective mass theory, the energy gap, E, for three-dimensionally confined silicon nanocrystals can be expressed as

E¼ EbulkþQ=L2 ð4Þwhere Ebulk is the bulk crystal silicon band gap, L is the dot size,and Q is the confinement parameter. The variation in the band gapenergy (E, in eV) of the nc-Si/a-SiNx:H QDs estimated from theassociated PL peak position has been plotted in Fig. 10, as afunction of average size (L, in nm) of the QDs. Considering thequantum confinement phenomena occurring in the silicon–nitro-gen–hydrogen complex system, the least square fit plot to the datain the form of above equation leads to the semi-empirical relation

E¼ 1:10þ12=L2 ð5Þwhere Ebulk¼1.10 eV matches well with the literature and areasonably high quantum confinement parameter, Q¼12 eV nm2,has been demonstrated [4,36,37]. The widening of optical gap(Tauc's gap), Eg, on reduction in the size of the quantum dots hasalso been plotted in Fig. 10, showing close proximity with the PLdata and demonstrating similar quantum confinement effectsinfluencing the optical gaps [38].

At higher deposition pressure the dissociation of Si–H and N–Hbonds increases, resulting in enhanced silicon and nitrogen atomshaving dangling bonds. The increase in dangling bond populated

silicon atoms facilitates the creation of nucleation sites and theformation of silicon clusters in the silicon nitride film during thegrowth process. Increasing nucleation sites induces size-reductionof the silicon clusters because of the distribution of growth-precursors. Simultaneously, enhanced nitrogenation to the net-work reduces the overall crystallinity of the material. High surfacecoverage by atomic hydrogen, contributed from high densityinductively coupled plasma produced by high density of electrons(n0�1012 cm�3 at 10 mTorr) at very low pressure (in the mTorrrange), helps in forming plenty of ultra-nanocrystallites within theamorphous nitrogenated network grown at 250 1C [39–42].

In case of p¼10 mTorr, at very high crystalline volume fraction(XC) with larger grain size the barrier width happens to reducewhen the tunneling between two adjacent crystallites dominatesthe conduction mechanism. In addition, low defects and disordersin an overall better crystalline network reduce the density of

Fig. 9. Deconvolution of PL spectra corresponding to the samples prepared atp¼10, 20 and 30 mTorr into possible satellite components arising out of defectcontributions and band-to-band recombination due to quantum confinementeffect.

B. Sain, D. Das / Journal of Luminescence 158 (2015) 11–1816

trapping centers, in general, virtually reducing the height of thepotential barrier. As the overall crystallinity decreases in thenetwork with increase in pressure, the barrier width in theelectrical transport increases in the amorphous dominated net-work and the tunneling current gradually becomes insignificant,when the extended state conduction prevails [25].

In the present deposition system the plasma is generated by abuilt in four-antenna low inductance flat spiral coil and the plasmais confined at the vicinity of the quartz plate of ICP source. ICP-discharges are produced by the RF power applied across adielectric window via electromagnetic coupling advantageous forhigher plasma density, lower plasma sheath potentials at thegrowth surface, etc., compared to other low-pressure plasmadischarges. The growth holder is located at such a distance thatthe direct contact of the growth from the plasma can be avoided sothat high quality film can be produced.

In general the Si-QDs have been reported to be producedthrough multi-step routes, the most fundamental of which is thehigh temperature processing as either pre-deposition or post-deposition annealing at �1000 1C or above. As regards thefabrication and the subsequent integration of high-performancedevices, a high annealing temperature is unsuitable from atechnical point of view since the thermal budget is extremelyhigh and the diffusion of dopants would be intolerably fast.Compared to various usual methods of Si-QD formation, thesupremacy of plasma processing has been well established, beingthe most favored route compatible with even miniaturized inte-grated circuit design and fabrication. Using low temperature one-step and spontaneous plasma processing by CVD, we have pre-viously reported the growth of nc-Si QDs of controlled size,density and distribution within a-Si/SiOx matrix [4]. Regardingthe growth of nc-Si QDs in a-SiNx matrix, only few reports areavailable in the literature [36,37]. Present work deals with thecontrolled growth of nc-Si QDs in a-SiNx:H dielectric matrixthrough one-step low temperature spontaneous plasma proces-sing, utilizing the high density low pressure inductively coupledplasma enriched in high atomic H density and operating in avirtual remote plasma configuration.

Strong visible photoluminescence (PL) tunable over wide range(1.67–3.02 eV), from nc-Si/a-SiNx:H QDs thin films having signifi-cant electrical conductivity, deposited in single step synthesis byinductively coupled plasma CVD, described in the present work

would be an appropriate candidate for fabrication of light emittingdevices (LEDs). The tune-ability of PL response over the signifi-cantly wide range was achieved by varying the size of the nc-SiQDs from �5.4 to 2.2 nm, obtained by controlling a singleparameter, the low pressure of the plasma within a narrow range(10–50 mTorr), maintaing a notably low growth temperature(250 1C), and here lies the novelty of this work.

5. Conclusions

Intense visible photoluminescence tunable over a wide range(1.67–3.02 eV) has been observed from nc-Si/a-SiNx:H QDs thinfilms. Planer ICP-CVD of 13.56 MHz RF plasma was employed toprepare the films using a mixture of silane (SiH4), ammonia (NH3),and hydrogen (H2) gases, by varying the deposition pressurewithin a narrow range (10–50 mTorr) at low substrate tempera-ture (250 1C). The origin of the PL spectra with variable peakposition was attributed mostly to the band-to-band recombinationdue to the quantum confinement effect (QCE) in the nanocrystal-line silicon QDs of size�Bohr radius. A large amount of atomichydrogen flux that originates due to the high degree dissociationof the gas molecules in high-density inductively coupled plasma(ICPs) provides very high surface coverage during growth, passi-vates well the nonradiative dangling bonds in the Si-network andproduces tiny nc-Si quantum dots demonstrating visible photo-luminescence. The red-shift of Raman peak and the correspondingbroadening of the Raman line due to the confinement of opticalphonons indicate the size reduction of Si QDs within amorphousdominated nature of the network, during the increase of plasmapressure. Size estimated from the Raman spectra varies from5.4 nm to 2.2 nm as pressure increases from 10 to 50 mTorr. HR-TEM image of the sample gives the physical evidence of thepresence of Si quantum dots (Si QDs) within the material.

The tunable photoluminescence and the widening of band gapas a consequence of quantum size effect along with significantlyhigh electrical conductivity from nc-Si/a-SiNx:H QDs thin filmsprepared at a very low pressure, described in the present work,demonstrate enormous promise for light emitting applicationssuch as the fabrication full-color flat panel displays, etc.

Acknowledgments

The work has been done under nano-silicon projects funded bythe Department of Science and Technology (Nano-Mission Pro-gram) and the Council of Scientific and Industrial Research,Government of India. The HR-TEM studies have been performedusing facilities of Unit on Nano-Science at IACS. One of the authors(B.S.) acknowledges the Council of Scientific and IndustrialResearch, Government of India, for providing him with a researchfellowship (via Grant no. 09/080(0706)/2010-EMR-I) for the work.

References

[1] P.M. Fauchet, Mater Today 8 (2005) 26.[2] B.H. Kim, C.H. Cho, T.W. Kim, N.M. Park, G.Y. Sung, S.J. Park, Appl. Phys. Lett. 86

(2005) 091908.[3] A.G. Cullis, L.T. Canham, Nature 353 (1991) 335.[4] D. Das, A. Samanta, Nanotechnology 22 (2011) 055601.[5] A.R.V. Sickle, J.B. Miller, C. Moore, R.J. Anthony, U.R. Kortshagen, E.K. Hobbie,

ACS Appl. Mater. Interfaces 5 (2013) 4233.[6] A. Samanta, D. Das, J. Mater. Chem. C 1 (2013) 6623.[7] D.C. Hannah, J. Yang, P. Podsiadlo, M.K.Y. Chan, A. Demortière, D.J. Gosztola,

V.B. Prakapenka, G.C. Schatz, U. Kortshagen, R.D. Schaller, Nano Lett. 12 (2012)4200.

[8] A. Rodriguez, J. Arenas, J.C. Alonso, J. Lumin. 132 (2012) 2385.[9] B. Sain, D. Das, Phys. Chem. Chem. Phys. 15 (2013) 3881.[10] K. Ma, J.Y. Feng, Z.J. Zhang, Nanotechnology 17 (2006) 4650.

Fig. 10. Band gap widening on reduction in the size of nc-Si QDs. Data: obtainedfrom PL peak position (red circle), obtained from Tauc's plot (blue square) and fittedPL data using effective mass theory (navy line). Reasonably good match betweendata and the plot demonstrates quantum confinement phenomena to occur in thesilicon–nitrogen–hydrogen heterostructure. (For interpretation of the references tocolor in this figure legend, the reader is referred to the web version of this article.)

B. Sain, D. Das / Journal of Luminescence 158 (2015) 11–18 17

[11] L.T. Canham, Appl. Phys. Lett. 57 (1990) 1046.[12] X. Zeng, W. Liao, G. Wen, X. Wen, W. Zheng, J. Appl. Phys. 115 (2014) 154314.[13] G.R. Lin, C.J. Lin, H.C. Kuo, Appl. Phys. Lett. 91 (2007) 093122.[14] R. Ghosh, P.K. Giri, K. Imakita, M. Fujii, Nanotechnology 25 (2014) 045703.[15] G.R. Lin, C.J. Lin, C.K. Lin, L.J. Chou, Y.L. Chueh, J. Appl. Phys. 97 (2005) 094306.[16] B. Rezgui, A. Sibai, T. Nychyporuk, M. Lemiti, G. Brémond, J. Lumin. 129 (2009)

1744.[17] C.H. Cheng, Y.C. Lien, C.L. Wu, G.R. Lin, Opt. Express 21 (2013) 391.[18] W.L. Wilson, P.F. Szajowski, L.E. Brus, Science 262 (1993) 1242.[19] B.H. Lai, C.H. Cheng, G.R. Lin, IEEE J. Quantum Electron. 47 (2011) 698.[20] M.V. Wolkin, J. Jorne, P.M. Fauchet, G. Allan, C. Delerue, Phys. Rev. Lett. 82

(1999) 197.[21] A. Puzder, A.J. Williamson, J.C. Grossman, G. Galli, Phys. Rev. Lett. 88 (2002)

097401.[22] G.R. Lin, Y.H. Pai, C.T. Lin, C.C. Chen, Appl. Phys. Lett. 96 (2010) 263514.[23] C.D. Lin, C.H. Cheng, Y.H. Lin, C.L. Wu, Y.H. Pai, G.R. Lin, Appl. Phys. Lett. 99

(2011) 243501.[24] B. Sain, D. Das, Sci. Adv. Mater. 5 (2013) 188.[25] D. Das, B. Sain, J. Appl. Phys. 114 (2013) 073708.[26] G. Lucovsky, J. Yang, S.S. Chao, J.E. Tyler, W. Czubatyj, Phys. Rev. B 28 (1983)

3234.

[27] D.V. Tsu, G. Lucovsky, M.J. Mantini, Phys. Rev. B 33 (1986) 7069.[28] Q.J. Cheng, S. Xu, K. Ostrikov, J. Mater. Chem. 20 (2010) 5853.[29] S. Hasegawa, H. Anbutsu, Y. Kurat, Philos. Mag. B 59 (1989) 365.[30] W.A. Lanford, M.J. Rand, J. Appl. Phys. 49 (1978) 2473.[31] C.J. Fang, K.J. Gruntz, L. Ley, M. Cardona, F.J. Demond, G. Muller, S. Kalbitzer,

J. Non-Cryst. Solids. 35–36 (1980) 255.[32] D. Das, Thin Solid Films 476 (2005) 237.[33] S.V. Deshpande, E. Gulari, S.W. Brown, S.C. Rand, J. Appl. Phys. 77 (1995) 6534.[34] C. Mo, L. Zhang, C. Xie, T. Wang, J. Appl. Phys. 73 (1993) 5185.[35] S. Veprek, Z. Iqbal, R.O. Kuhne, P. Capezzuo, F.A. Sarott, J.K. Gimzevski, J. Phys.

C: Solid State Phys. 16 (1983) 6241.[36] T.W. Kim, C.H. Cho, B.H. Kim, S.J. Park, Appl. Phys. Lett. 88 (2006) 123102.[37] T.Y. Kim, N.M. Park, K.H. Kim, G.Y. Sung, Y.W. Ok, T.Y. Seong, C.J. Choi, Appl.

Phys. Lett. 85 (2004) 5355.[38] D. Das, Solid State Commun. 108 (1998) 983.[39] Q.J. Cheng, S. Xu, S.Y. Huang, K. Ostrikov, Cryst. Growth Des. 9 (2009) 2863.[40] Q.J. Cheng, E. Tam, S. Xu, K. Ostrikov, Nanoscale 2 (2010) 594.[41] D. Das, Solid State Phenom. 44 (1995) 227.[42] D. Das, M. Jana, Sol. Energy Mater. Sol. Cells 81 (2004) 169.

B. Sain, D. Das / Journal of Luminescence 158 (2015) 11–1818