You are on page 1of 7

Thin Solid Films 518 (2010) 6811–6817

Contents lists available at ScienceDirect

Thin Solid Films


j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / t s f

Effect of hydrogen plasma treatment on the surface morphology, microstructure and


electronic transport properties of nc-Si:H
P. Dutta ⁎, S. Paul, D. Galipeau, V. Bommisetty
Department of Electrical Engineering & Computer Science, EECS Building, South Dakota State University, Brookings, SD 57007, USA

a r t i c l e i n f o a b s t r a c t

Article history: Hydrogenated nanocrystalline silicon (nc-Si:H) films, deposited by reactive radio-frequency sputtering with
Received 17 January 2010 33% hydrogen dilution in argon at 200 °C, were treated with low-power hydrogen plasma at room
Received in revised form 28 May 2010 temperature at various power densities (0.1–0.5 W/cm2) and durations (10 s–10 min). Plasma treatment
Accepted 9 June 2010 reduced the surface root mean square roughness and increased the average grain size. This was attributed to
Available online 17 June 2010 the mass transport of Si atoms on the surface by surface and grain boundary diffusion. Plasma treatment
under low power density (0.1 W/cm2) for short duration (10 s) caused a significant enhancement of
Keywords:
crystalline volume fraction and electrical conductivity, compared to as-deposited film. While higher power
Nanocrystalline silicon
(0.5 W/cm2) hydrogen plasma treatment for longer durations (up to 10 min) caused moderate improvement
Hydrogen plasma treatment
Atomic force microscopy in crystalline fraction and electrical properties; however, the magnitude of improvement is not significant
Microstructure compared to low-power (0.1 W/cm2)/short-duration (10 s) plasma exposure. The results indicate that low-
power hydrogen plasma treatment at room temperature can be an effective tool to improve the structural
and electrical properties of nc-Si:H.
Published by Elsevier B.V.

1. Introduction exposure to hydrogen plasma at high temperature were carried out


sequentially [17,18]. The structural rearrangements in a-Si:H due to
Hydrogenated nanocrystalline silicon (nc-Si:H) is a promising hydrogen plasma treatment led to the formation of hydrogenated
material for photovoltaics [1–3]. It is a biphasic material composed of microcrystalline silicon [19,20]. Recently, it has been shown that
silicon nanocrystals embedded in amorphous silicon matrix and electron cyclotron resonance (ECR) hydrogen plasma treatment of
microvoids. Thin films of nc-Si:H are typically deposited using plasma ECR-CVD deposited a-Si:H for 60 min at 325 °C resulted in an increase
enhanced chemical vapor deposition (PECVD) [4,5], hot-wire chemical in the crystalline fraction by 10% to 15% [19]. Conversely, a similar
vapor deposition [6,7], low-energy plasma enhanced chemical vapor hydrogen plasma treatment on mc-Si:H caused a pronounced
deposition [8], very-high-frequency chemical vapor deposition [9] and decrease in the crystalline fraction by 20%, which was attributed to
radio-frequency (RF) sputtering [10,11]. Among these techniques, RF an increase in disorder in the material [21]. In general, high-power
sputtering provides a better control of hydrogen incorporation in the hydrogen plasma exposure for long durations at high temperatures
film during deposition [12]. causes etching of surface Si atoms resulting in deterioration in the
Efforts have been made in recent years to modify and improve the electrical quality of the films.
structural and optoelectronic properties of a-Si:H through post- However, the effect of post-deposition low-power hydrogen
deposition high-power argon and hydrogen plasma treatments at plasma treatment at room temperature (RT) on nc-Si:H has not
elevated temperatures [13–18]. Post deposition hydrogen plasma been reported. Low-power hydrogen plasma treatment at RT may
treatment induced crystallinity in PECVD-deposited hydrogenated provide a fast and cost-effective technique to control the electronic
amorphous silicon (a-Si:H) and was explained by hydrogen in- and structural properties of nc-Si:H. In addition, it may be an effective
diffusion and insertion of hydrogen atoms in strained Si–Si bonds tool to control the properties of the growth-front (and subsequent
through the formation of intermediate bond-centered Si–H–Si growth processes) through brief growth interruptions. Moreover, nc-
configurations [13–15]. It has also been reported that room temper- Si:H deposited on flexible substrates requires low temperature
ature hydrogen plasma exposure can reduce the thermal crystalliza- processing due to the low melting point of the substrate. Hydrogen
tion time of a-Si:H by a factor of five [16]. In a layer-by-layer PECVD plasma treatment at RT can be an effective tool for modifying the
deposition technique, deposition of nm-thin layer of a-Si:H and microstructure of these films. This paper reports the influence of low-
power density (0.1–0.5 W/cm2) and short-duration (10 s–10 min)
post-deposition RF hydrogen plasma treatment on the microstruc-
⁎ Corresponding author. tural and nanoscale electronic properties of sputter deposited nc-Si:H
E-mail address: pavel.dutta@sdstate.edu (P. Dutta). thin films. The effect of hydrogen plasma treatment on grain size,

0040-6090/$ – see front matter. Published by Elsevier B.V.


doi:10.1016/j.tsf.2010.06.037
6812 P. Dutta et al. / Thin Solid Films 518 (2010) 6811–6817

surface roughness, crystalline fraction, surface and bulk electrical 3. Results


conductivity is discussed in detail.
3.1. Atomic force microscopy studies
2. Experimental details
Fig. 1(a), (b) and (c) shows topographic AFM images of untreated and
The thin films of nc-Si:H were deposited on glass substrates hydrogen plasma treated (Samples A and B) nc-Si:H surface. The surface
(Corning 1737) using reactive RF magnetron sputtering in H2/Ar of the untreated sample was composed of isolated grains, clusters of
atmosphere in a custom built deposition system. The base pressure of
the deposition system was ≈10− 5 Pa. Samples were deposited under
33% hydrogen dilution (rest Ar) at 200 °C. The process pressure and
plasma power for all the depositions were 2.15 Pa (measured using
capacitance manometer) and 12.5 W/cm2 respectively. Substrate to
target distance was kept constant at 15 cm. Thickness of the films
were measured and found to be around 100–150 nm.
Freshly deposited nc-Si:H samples of the same batch were then
exposed to RF generated hydrogen plasma at 2.67 Pa in the deposition
chamber at various plasma powers (0.1–0.5 W/cm2) and durations
(10 s–10 min). No substrate heating was used during plasma
treatment. All the plasma treatments were carried out within 1 day
of nc-Si:H deposition. Samples were stored in vacuum desiccators to
minimize exposure to air between deposition and plasma treatment.
Only two plasma treatments (shortest duration/smallest plasma
power density to longest duration/largest plasma power density)
are detailed in this report for reasons of brevity. Throughout the
paper, the 0.1 W/cm2 plasma treatment for 10 s is called low-
power/short-duration process or Process A (Sample A); 0.5 W/cm2
plasma treatment for 10 min is called moderate-power/long-duration
process or Process B (Sample B). In all the cases as-deposited nc-Si:H
was used as the starting material.
An Agilent Nanoscope 5500 scanning probe microscope was used
for Atomic Force Microscopy (AFM) and Conductive Atomic Force
Microscopy (C-AFM) studies. Intermittent contact AFM images were
measured in a hermitically sealed chamber under dry nitrogen
atmosphere. Silicon cantilever/tips (Budget sensors, resonance fre-
quencies ≈300 kHz and nominal force constant of ≈40 N/m) of tip
diameter of b10 nm were used for AFM imaging. Local electrical
conductivity was measured using C-AFM [21] in contact mode using
gold coated Si tips (Nanosensors). A positive fixed bias of 3 V was
applied to the tip and the sample was grounded. The Al electrodes
were deposited on sample surface. Placing an electrode on the top of
the surface allows charge carriers to flow on the surface and not
through the thickness of the film as usually obtained from bottom
electrode geometry. This allows us to measure the surface electrical
conductivity of the material at nanoscale. X-ray diffraction (XRD)
patterns were measured at beam line 33-BM-C of Advanced Photon
Source (Argonne National Laboratories, Argonne, IL) using 8 keV X-
rays (λ ∼ 1.54 Å) to determine the crystallinity, crystallite size and
preferential growth orientation of the films. Raman spectra were
measured with a Jobin-Yvon Confocal Raman spectrometer at RT,
using an excitation wavelength of 532 nm from a diode pumped solid-
state laser. The power density of the laser at the sample surface was
10 mW/cm2 and the spot diameter was 2 μm. The excitation
wavelength was chosen so as to reduce light penetration depth and
prevent formation of the substrate spectrum. Sample probing depth
for 532 nm was estimated to be ∼120 nm. The excitation power was
kept low in order to avoid laser induced crystallization. The laser spot
was focused on the sample through an optical microscope (Olympus
MPLAN 20X/0.4) and the scattered radiation was detected by a CCD
camera having 1800/nm grating. The experiments were done in
backscattering configuration and Raman spectra of the films were
recorded with a resolution of 0.6 cm− 1. The bulk electrical measure-
ments were carried out using Agilent 4155C Semiconductor Param-
eter Analyzer. Al contacts were thermally evaporated and the Fig. 1. Atomic force microscopy (topography) images of nc-Si:H: (a) untreated,
(b) Sample A, and (c) Sample B. Change in surface morphology with plasma treatment
current–voltage (I–V) measurements were performed using two- is noticeable. White circles in Fig. 1(a) show clusters of grains on the surface. The inset
point probe method. Photoconductivity measurements were done in Fig. 1(a) shows an enlarged view of one such cluster of 80 nm diameter that consists
using an A.M 1.5 (100 mW/cm2) light source. of several grains.
P. Dutta et al. / Thin Solid Films 518 (2010) 6811–6817 6813

grains, and pits. A magnified image of one such cluster of about 80 nm RMS roughness of the surface also decreased from 5.6 to 3.9 nm with
diameter containing 9 smaller grains is shown in the inset of Fig. 1(a). A plasma treatment Process B.
typical cluster contained about 5–10 smaller grains and has an average Fig. 4(a) shows the schematic of the C-AFM measurements. The flow
cluster diameter in the range of 60–100 nm. The average diameter of the of carriers on the film surface from the AFM tip to the Al electrode is
individual grains was ∼18 nm. The space between the clusters was filled
with smaller isolated grains of 8–10 nm in diameter, deep pits and voids
of about 15 nm in average depth.
Process A resulted in a surface with more individual grains than
clusters (Fig. 1(b)). The average size of the clusters, density of clusters and
the number of grains per cluster decreased. The average cluster diameter
was about 40–60 nm and 3–4 grains were typically present in each
cluster. Clusters were scattered across the surface and the regions
between clusters were occupied by isolated individual grains. The
average diameter of the grains was about 20 nm. The surface had a
large number of pits and voids of about 10 nm in average depth.
Therefore hydrogen plasma, even at low power, modified the surface
structure by breaking larger clusters into smaller clusters and individual
grains. An increase in the size of grains from 18 nm to 20 nm was also
observed. Process B caused further breakdown of larger clusters and
increase in number of individual grains on the nc-Si:H surface (Fig. 1(c)).
About 2–3 grains were in each cluster in most cases and the size of the
clusters was about 40–50 nm. The majority of the clusters as well as the
individual grains inside the clusters had an oval shape. The surface was
mostly occupied by oval shaped individual grains of about 22 nm in
diameter. Pits and voids of ∼5 nm average depth were observed between
some clusters.
A detailed grain-size distribution analysis was carried out to
determine the extent of the surface modification. The diameter (both
short axis and long axis, where applicable) and heights of several
thousand grains of each sample were carefully measured from
topography images. Histograms of grain-size (grain area) distribution
of untreated and plasma treated nc-Si:H are shown in Fig. 2. The solid line
on each histogram shows a polynomial fit to serve as a guide to the eye.
Both isolated grains and grains within clusters were taken into account
when plotting the histograms. Fig. 2(a) shows the grain-size distribution
of untreated nc-Si:H. The area of a vast majority of the grains was
between 200 and 2000 nm2 with predominantly smaller sized grains in
the range of 200–400 nm2. The surface root mean square (RMS)
roughness was 5.6 nm. Fig. 2(b) shows the grain-size distribution of a
sample after process A which resulted in a smoother surface with broader
and uniform distribution of grain sizes, mostly in the range of
200–3000 nm2. The surface RMS roughness decreased to 4.7 nm. After
Process B (see Fig. 2(c)), the peak in the grain-size distribution plot
shifted to the right, indicating presence of larger grains. The distribution
became broader suggesting that Process B had also smoothed the surface.
The RMS surface roughness after Process B was 3.9 nm. Fig. 3 shows the
height distributions derived from the topographic images of nc-Si:H
before and after plasma treatment [20]. These plots show the number of
pixels at a given height in the topographic images and are important for
statistical height variation studies. All images were taken using the same
AFM tip under similar imaging conditions, which was important for
statistical accuracy of the data [21]. The untreated sample had a broad
height distribution indicating the presence of particles (“particles” are
statistical quantities representing both grains and granular clusters) with
non-uniform heights ranging from 1 to 43 nm. This was attributed to the
presence of large clusters on the surface. For sample A the distribution
became narrower with the peak shifting to the left indicating that the
heights of the particles were in a smaller range of 1 to 37 nm, with most of
them confined in the 10–20 nm range. This also indicated that the surface
had become smoother with plasma treatment causing a decrease in
surface RMS roughness to 4.7 nm. The average height of the particles
decreased from 21.5 to 19.3 nm with plasma treatment Process A. With
Process B, the peak of the distribution shifted further to the left indicating
a decrease in the average height of the particles and that the particles Fig. 2. Grain size (grain area) distribution of nc-Si:H: (a) untreated, (b) Sample A, and
were in a much smaller range of 1 to 30 nm. The average height decreased (c) Sample B. Polynomial fit (black line) enveloping grain size distributions is shown to
from 21.5 to 14.6 nm with plasma treatment Process B. Consequently, the guide the eye.
6814 P. Dutta et al. / Thin Solid Films 518 (2010) 6811–6817

Fig. 3. Height distributions derived from the topographic images of nc-Si:H before and
after plasma treatment. These plots show number of pixels at a given height in the
topographic images and are important for statistical height variation studies.

marked with a dotted line. Fig. 4(b), (c) and (d) shows the C-AFM images
(0.5 μm× 0.5 μm) of the untreated and hydrogen plasma treated samples
(Samples A and B) respectively. Brighter regions in the images
correspond to higher surface electrical conductivity and darker regions
correspond to lower surface electrical conductivity. A strong correlation
was observed between conductivity and the grains and grain boundaries.
It was determined that grain interiors were brighter regions which
provided paths for electrical conduction. Grain boundaries,
corresponding to lower current (darker regions), were highly defective
regions that hindered the flow of electrons. Fig. 4(b) shows a C-AFM
image of an untreated sample. The current flow on the surface of the
untreated sample was nearly uniform with an average value of 0.5 pA.
The average current and local electrical conductivity was estimated from
multiple images from different regions on the sample surface with a
standard deviation of 5%. The average surface electrical conductivity of
untreated sample was 3.7 ×10− 6 S/cm. With plasma treatment Process A
on the untreated sample (Fig. 4(c)) there was a significant rise in current
flow through the grains, and distinctive high-and low-conductivity
regions became clearly visible. The average current on the surface and
local electrical conductivity of Sample A increased 4 times to 2 pA and
1.47 ×10− 5 S/cm respectively. A few regions were completely dark
suggesting highly defective regions [21]. Process B plasma treatment on
the untreated sample (Fig. 4(d)) also resulted in an increase of average
current on the surface from 0.5 to 1 pA and local electrical conductivity
from 3.7 ×10− 6 to 7.35× 10− 6 S/cm (2 times increase). However, the
enhancement of average current and local electrical conductivity on the
surface was much more pronounced in case of short duration and low-
power plasma treatment process (Process A) than in the case of long-
duration/moderate-power plasma treatment (Process B).

3.2. Structural studies by X-ray diffraction

Fig. 5 shows the XRD patterns of the untreated and plasma treated
nc-Si:H films. The three peaks at 2θ = 28.5°, 47° and 56° corresponded
to (111), (220) and (311) crystallographic orientations respectively
for all the samples. Average crystallite size was calculated from the
FWHM of the (111) peak using Scherrer's formula (d = 0.9λ/B cos θ,
where d is the mean grain size of the silicon crystallites, λ is the
incident wavelength of X-ray radiation, B is the full width at half-

Fig. 4. (a) Schematic of conducting atomic force microscopy (C-AFM) set up. Carriers
flow between the conducting AFM tip and the Al bonding pad on the film surface when
a dc bias is applied between them. The tip scans over a certain area and provides a map
of the local electrical conductivity of the surface. C-AFM images in Fig. 4(a), (b) and (c)
show the local current distribution of untreated, Sample A and Sample B respectively. A
significant change in the surface electrical conductivity with plasma treatment is
observed.
P. Dutta et al. / Thin Solid Films 518 (2010) 6811–6817 6815

Fig. 5. X-ray diffraction patterns of nc-Si:H: untreated, sample A and Sample B. Fig. 6. Raman spectra of untreated, Sample A and Sample B. Inset shows the
Preferred crystallographic orientation was (111). The vertical axis was shifted deconvoluted Raman spectrum of Sample A. Peaks corresponding to amorphous,
appropriately for clarity. grain boundary and crystalline phases are marked as (a), (b) and (c) respectively.

maximum (FWHM) of the diffraction peak, and θ is the Bragg 14.4% and 12.8% for the untreated sample and samples A and B
diffraction angle). With plasma treatment, the intensity of all the respectively. It was observed that grain boundary fraction decreased
three peaks increased indicating an increase in crystallinity. The large monotonously with plasma treatment.
(111) peak suggested that the (111) crystallographic direction was
the preferred growth orientation for all samples. The preferential 3.4. Bulk current–voltage measurements
growth along the (111) direction was attributed to the low surface
energy of the (111) crystal facets of Si compared with other Dark (σD) and illuminated (σI) electrical conductivity measure-
orientations [22]. The change in the crystalline fraction could be ments were performed using a two probe method under dark and AM
correlated to the intensity of the (111) peak. The plasma treatments 1.5 illumination. The results are summarized in Table 2. Photocon-
(both Processes A and B) resulted in an increase of the intensity of the ductivity (σp) is the difference between the illuminated and dark
(111) peak, suggesting increased crystalline fraction. The increase in conductivity for a given sample. σD, σI and σp for the untreated sample
the intensity was more pronounced for Process A than Process B. were 0.76 × 10− 6, 2.13 × 10− 6 and 1.37 × 10− 6 S/cm respectively.
Similar hydrogen plasma treatments on unhydrogenated a-Si did not Process A caused an increase in σD, σI and σp to 2.75 × 10− 6,
induce crystallinity, suggesting that initial hydrogen content in the 7.7 × 10− 6 and 4.95 × 10− 6 S/cm respectively. This increase was
film was critical for improving film properties. about 3.6 times with respect to that of the untreated sample.
However, process B resulted in an increase in the conductivity by a
3.3. Raman spectroscopy studies factor of 1.5, which was much smaller than Process A.

The degree of crystallinity of the films was investigated by Raman 4. Analysis


spectroscopic measurements. Fig. 6 shows the Raman spectra of the
untreated and plasma treated (Process A and Process B) nc-Si:H. The Heavy ion bombardment (such as Ar ions, mass number = 40) is
inset shows the Gaussian deconvolution of a typical Raman spectra known to modify the surface structure (such as formation of mounds,
identifying three components corresponding to a broad distribution large pits, rough topography) without significantly affecting the
centered at 480 cm− 1, an intermediate peak at about 515 cm− 1 and a crystalline quality of the film [25–27]. The surface roughness is usually
sharp narrow peak at 520 cm− 1. The 480 cm− 1 component was removed by heating the surface to sufficiently high temperatures,
assigned to the transverse optical (TO) vibrational mode of a-Si and whereas in the present case, bombardment of low energy (∼0.1–0.5 eV)
the 520 cm− 1 component was assigned to the TO vibrational mode of and light mass H ions (mass number 1) on Si surface created by local RF
c-Si. The intermediate 515 cm− 1 component corresponded to the plasma at room temperature reduced the surface roughness and
bond dilation at the grain boundaries [8,23,24]. The crystalline volume increased the crystallinity of the material. The reduction of surface
fraction Xcr (crystallinity) and the grain boundary fraction XGB was roughness and increase in grain size by hydrogen plasma treatment may
determined using: be explained by diffusion assisted mass transport and surface migration
of Si atoms [28]. Two main diffusion pathways generally attributed to
Xcr = ðIc + IGB Þ = ðIc + IGB + βIA Þ ð1Þ surface modification are surface diffusion and grain boundary diffusion
[29]. The energy of the hydrogen plasma provided the external driving
XGB = IGB = ðIc + IGB + βIA Þ ð2Þ

Where Ic, IGB, and IA are the peak values of the crystalline, grain Table 1
Surface roughness, average height, crystallite size and crystalline volume fraction
boundary and amorphous bands and β is a constant related to the variation of untreated and plasma treated nc-Si:H.
crystalline phase to amorphous phase scattering cross section ratio. A
value of 0.8 was chosen for β [15]. Untreated Sample A Sample B

The crystalline volume fractions (Xcr) of the untreated sample and Average height (nm) 21.5 19.2 14.5
samples A and B were 72%, 77% and 75% respectively (see Table 1). The RMS roughness (nm) 5.6 4.7 3.9
Crystallite size from XRD (nm) 24 25 24
increase in crystallinity was more pronounced for sample A as
Crystalline Volume Fraction (Xc) (%) 72 77 75
compared to sample B. Grain boundary fractions (XGB) were 16.4%,
6816 P. Dutta et al. / Thin Solid Films 518 (2010) 6811–6817

Table 2 of disorder due to etching of surface Si atoms and breaking of Si–H bonds
Dark, illuminated and photoconductivity of untreated and plasma treated nc-Si:H. inside the bulk of the material due to increased power and longer
Sample Dark conductivity Illuminated conductivity Photoconductivity exposure of the growth surface to plasma (Process B) [18].
The increase in the surface electrical conductivity was four-fold after
σD (S/cm) σI (S/cm) σp = (σI–σD) (S/cm)
Process A, whereas Process B resulted in a two-fold increase. This
Untreated 0.76 × 10− 6 2.13 × 10− 6 1.37 × 10− 6
behavior is similar to that of bulk conductivity, although the magnitude
Sample A 2.75 × 10− 6 7.7 × 10− 6 4.95 × 10− 6
Sample B 1.14 × 10− 6 3.2 × 10− 6 2.1 × 10− 6 of increase is larger for surface conductivity. A larger increase in surface
conductivity can be explained by improved passivation of Si near the
surface, as expected. It is also possible that more of the near surface
force for surface Si atoms to acquire sufficient energy to move the amorphous tissue has been crystallized compared to that in the bulk.
surrounding atoms apart or even drag them along and land at a site that Thus, Processes A and B produced similar variation to surface and bulk
is energetically more favorable. Significant mass transport through the electrical conductivity as observed from C-AFM and I–V measurements
diffusion of surface atoms from grain interiors into the surface valleys (Fig. 7), implying that hydrogen plasma has affected the surface and
and voids between adjacent grains reduced surface roughness [29,30]. bulk states in a similar manner. These results can be correlated to the
Moreover, for a granular structure with high surface-to-bulk ratio, grain variation in crystallinity of the material as confirmed by XRD and Raman
boundary (GB) diffusion provided another important pathway through experiments.
which significant mass transport occurred. Since the diffusivity of atoms
is much higher at GBs because of the presence of defects and
imperfections, which reduces the activation energy for diffusion, 5. Conclusion
atoms at GB sites are less tightly bound and more mobile than atoms
at grain interiors and undergo mass transport under an external force Low-power hydrogen plasma treatment can have a pronounced
gradient [29]. Therefore, the larger grains lose atoms by surface and GB effect on crystalline fraction, surface structure, and macroscopic and
diffusion, thereby reducing the height of the grains and surface nanoscale electrical properties of RF sputter deposited nc-Si:H thin
roughness. Also, as a result of mass transport on the surface, the films. Both the crystalline fraction and nanoscale electrical conductivity
aggregates or clusters of grains separated out into individual grains. increased with plasma treatment but the increase was more pro-
With longer exposure time, the relatively higher energy of the plasma nounced during low-power/short-duration plasma treatment (Process
caused a further decrease in the height of the grains and surface A). Hydrogen plasma treatment caused a significant change in the
roughness. The depth of the voids decreased from 15 to 5 nm with surface structure of the films by increasing the average grain size and
plasma treatment which indicates that atoms from the top of the grain reducing the average height of grains, causing a significant decrease in
interiors moved to the low-lying voids and GB regions. The average surface roughness. The decrease in surface roughness with hydrogen
grain size (diameter) increased from 18 to 22 nm, almost by 20%. This plasma treatment was attributed to GB and surface diffusion of atoms
increase in grain size may be attributed to the enhanced mass transport leading to massive mass transport resulting in the decrease in average
of Si atoms through GB diffusion [29]. With the increase in grain size, GB height of the grains and smoothening of the surface. Improvement in
fractions decreased (see Table 1) because of a reduction in GB regions. surface and bulk electrical conductivity during Process A of plasma
The increase in crystallinity with Process A, observed from the treatment was attributed to improved passivation of surface and bulk
increase in Xcr and (111) peak intensity in Raman and XRD measure- dangling bonds. A four-fold increase in photoconductivity was observed
ments respectively, can be explained by the insertion of H atoms into in Sample A. Prolonged plasma treatment at moderate power (Process
strained Si–Si bonds through the formation of intermediate bond- B) did not increase the surface and bulk electrical conductivity as much
centered Si–H–Si configurations as the H atoms diffuse through the nc- as Process A. This was attributed to the degradation of the surface due to
Si:H film, as reported in case of a-Si:H [13,14]. The amorphous tissue etching of Si atoms and creation of defects inside the bulk by high-
surrounding the nanocrystals is preferentially transformed into crys- energy plasma. Results show that low-power (0.1 W/cm2) RF hydrogen
talline form during this process. Xcr values were obtained from Raman plasma treatment for short durations (10 s) and at room temperature
measurement by using 532 nm light whose probing depth in nc-Si:H can be an effective tool to improve the microstructural and nanoscale
was ∼120 nm, almost equal to the film thickness. Thus it can be inferred electrical properties of nc-Si:H and can provide a fast and efficient tool to
that 10 s of exposure to low-power plasma (0.1 W/cm2) was enough to fabricate high-quality solar grade nc-Si:H films.
cause an increase in crystallinity in the bulk of the material by in-
diffusion of hydrogen. The 5% increase in crystallinity (from 72% to 77%)
due to diffusion of hydrogen in the amorphous silicon tissue during the
initial few seconds of plasma treatment (Process A) is followed by the
degradation of the material due to etching of Si atoms by the hydrogen
plasma during Process B. Etching of Si by hydrogen ions is non-
preferential and it etches away Si atoms from both amorphous and
crystalline regions on the surface. Thus Process B led to only a 3%
increase in crystallinity (from 72% to 75%) as opposed to 5% obtained
from Process A. Interestingly, similar hydrogen plasma treatment on
unhydrogenated a-Si films did not induce crystallinity. This implies that
the presence of atomic hydrogen is crucial for low-power hydrogen
plasma mediated crystallization.
From the results of I–V measurements it was observed that Process A
caused ∼3.6 times increase in both dark and illuminated electrical
conductivity. Enhanced electrical conductivity can be attributed to
improved passivation of dangling bonds by hydrogen diffusion into the
bulk of the material [13,14] or due to an increase in the crystalline
fraction. Process B also led to an increase in dark and illuminated
electrical conductivity but the increment was only 1.5 times, much Fig. 7. Histograms showing the variation of crystalline volume fraction, surface and bulk
smaller as compared to Process A. This can be attributed to the creation electrical conductivity of untreated, Sample A and Sample B.
P. Dutta et al. / Thin Solid Films 518 (2010) 6811–6817 6817

Acknowledgements [10] A. Achiq, R. Rizk, F. Gourbilleau, P. Voivenel, Thin Solid Films 348 (1999) 74.
[11] C. Goncalves, S. Charvet, A. Zeinert, M. Clin, K. Zellama, Thin Solid Films 403
(2002) 91.
This research was supported by National Science Foundation and [12] B. Garrido, A. Perez-Rodriguez, J.R. Mornte, A. Achiq, F. Gourbilleau, R. Madelon, R.
South Dakota EPSCoR (Grant No. 0091948). Also the authors would like Rizk, J. Vac. Sci. Technol. B. 16 (1998) 1851.
[13] S. Sriraman, S. Agarwal, E.S. Aydil, D. Maroudas, Nat. London 418 (2002) 62.
to acknowledge help from Dr. P. Zschack and Dr. G. Karapetrova, [14] S. Sriraman, M.S. Valipa, E.S. Aydil, D. Maroudas, J. Appl. Phys. 100 (2006) 053514.
Advanced Photon Source, Argonne National Laboratory, Argonne, IL, for [15] C. Godet, N. Layadi, P. Roca Cabarrocas, Appl. Phys. Lett. 66 (1995) 3146.
XRD measurement; Dr. C. Jiang, Department of Chemistry, University of [16] K. Pangal, J.C. Sturm, S. Wagner, T.H. Büyüklimanli, J. Appl. Phys. 85
(1999) 1900.
South Dakota for Raman measurements; and Dr. M. Kumar, SDSU, for [17] T. Akasaka, I. Shimizu, Appl. Phys. Lett. 66 (1995) 3441.
helpful discussions. [18] K. Saitoh, M. Kondo, M. Fukawa, T. Nishimiya, A. Matsuda, W. Fukato, I. Shimizu,
Appl. Phys. Lett. 71 (1997) 3403.
[19] I. Kaiser, N.H. Nickel, W. Fuhs, Phys. Rev. B 58 (1998) 1718.
References
[20] Y. Chen, W. Huang, Meas. Sci. Technol. 15 (2004) 2005.
[21] D. Cavalcoli, Solid State Phenom. 547 (2008) 131.
[1] A.V. Shah, J. Meier, E. Vallat-Sauvain, U. Kroll, N. Wyrsch, J. Guillet, U. Graf, Thin Solid
[22] H. Kakinuma, M. Mohri, M. Sakamato, T. Tsuruoka, J. Appl. Phys. 70 (1991) 7374.
Films 403 (2002) 179.
[23] W. Wei, Vacuum 81 (2007) 857.
[2] J. Meier, R. Fluckiger, H. Keppner, A. Shah, Appl. Phys. Lett. 65 (1994) 860.
[24] C. Gonçalves, A. Zeinert, S. Charvet, M. Lejeune, A. Grosman, H.J. von Bardeleben, K.
[3] S. Veprek, V. Marecek, Solid State Electron. 11 (1968) 683.
Zellama, Thin Solid Films 451 (2004) 370.
[4] A.M. Ali, S. Hasegawa, Thin Solid Films 68 (2003) 437.
[25] M.S. Hoogeman, M.A.J. Klik, R. van Gastel, J.W.M. Frenken, J. Phys. Condens. Matter
[5] P. Roca Cabarrocas, J. Non-Cryst. Solids 31 (2000) 266.
11 (1999) 349.
[6] D. Das, K. Bhattacharya, J. Appl. Phys. 100 (2006) 103701.
[26] J. Naumann, J. Osing, A.J. Quinn, I.V. Shvets, Surf. Sci. 388 (1997) 212.
[7] A.R. Middya, A. Lloret, J. Perrin, J. Huc, J.L. Moncel, J.Y. Parey, G. Rose, Mater. Res. Soc.
[27] M. Ritter, M. Stindtmann, M. Farle, K. Baberschke, Surf. Sci. 348 (1996) 243.
Symp. Proc. 119 (1995) 377.
[28] M. Ohring, Materials Science of Thin Films, Elsevier, Singapore, 2006.
[8] C. Droz, E. Vallat-Sauvain, J. Bailat, L. Feitknecht, J. Meier, A. Shah, Sol. Energy Mater.
[29] M.A. Wank, R.A.C.M.M. van Swaaij, M.C.M. van de Sanden, Appl. Phys. Lett. 95
Sol. Cells 87 (2005) 11.
(2009) 021503.
[9] G. Yue, B. Yan, G. Ganguly, J. Yang, S. Guha, C.W. Teplin, Appl. Phys. Lett. 88 (2006)
[30] T. Karabacak, Y.P. Zhao, G.C. Wang, T.M. Lu, Phys. Rev. B 64 (2001) 085323.
263507.

You might also like