Variation of Electrical Transport Parameters with Large Grain Fraction in Highly Crystalline Undoped...
-
Upload
sanjay-ram -
Category
Technology
-
view
405 -
download
1
Transcript of Variation of Electrical Transport Parameters with Large Grain Fraction in Highly Crystalline Undoped...
Variation of Electrical Transport Parameters with Large Grain Fraction in Highly Crystalline
Undoped Microcrystalline Silicon
Sanjay K. RamSanjay K. Ram
Dept. of Physics, I.I.T. Kanpur, India&
LPICM (UMR 7647 du CNRS ), Ecole Polytechnique, France
Motivation: study of μc-Si:H thin films
• Promising material for large area electronics
– Good carrier mobility
– Greater stability
– Low temperature deposition
• Electrical transport properties : important for
device applications
How is film microstructure related to transport properties in µc-Si:H???
Optimization of μc-Si:H device applications: Issues
Complex microstructure of μc-Si:H
Film growth
voids
substrate
grains grain boundaries
columnar boundaries
conglomerate crystallitessurface
roughness
• To study the optoelectronic properties of well characterized μc-Si:H films
• Identify the role of microstructure in determining the electrical transport behavior
Objectives
Sample preparation
Parallel-plate glow discharge plasma deposition system
R=1/1 R=1/5 R=1/10
Substrate: Corning 1773
High purity feed gases:SiF4 , Ar & H2
Rf frequency 13.56 MHz
Flow ratio (R)= SiF4/H2
Thickness seriesTs=200 oC
μc-Si:Hfilm
R F
HSi SiNSi N
HSiH
HHN
N
H H
HHH
P E C V DR F
HSi SiNSi N
HSiH
HHN
N
H H
HHH
P E C V D
Film characterization
Structural Properties Electrical Properties
Xray Diffraction
Raman Scattering
Spectroscopy Ellipsometry
Atomic Force Microscopy
σd(T) measurement15K≤T ≤ 450K
σPh(T,∅) measurement15K≤T ≤ 325K
CPM measurement
Hall effect
TRMC
Microstructural Properties
2 3 4 5-5
0
5
10
15
20
25
30
d=390 nm
d=55 nm
d=170 nmd=590 nm
d=950 nm
E2 (4.2 eV)E1 (3.4 eV)
Energy (eV)
< ε 2 >
2 3 4 5-10
01020304050
Spectroscopic Ellipsometry : measured imaginary part of the pseudo-dielectric function <ε2> spectra
c-Sipc-Si-l
μ c-Si:H(d = 950 nm)
a-Sipc-Si-f
E2 (4.2 eV)E1 (3.4 eV)
Energy (eV)<
ε 2 >(a)
* Reference c-Si in BEMA model : LPCVD polysilicon with large (pc-Si-l) and fine (pc-Si-f) grains
thickness series of R=1/10•Ram et al, Thin Solid Films 515 (2007) 7619.
•Ram et al, Thin Solid Films (2008) in print.
Analyses of SE data: schematic view for two films
(initial and final growth stages)
TSL (7.9 nm)Fcf = 32.3 %, Fcl = 0.6 %,
Fv = 67.1%, Fa =0 %
BL (48.2 nm)Fcf = 88.4 %, Fcl = 0 %, Fv = 10.1 %, Fa = 1.5 %
d =
950
nm
TSL (8.3 nm)Fcf = 73.6 %, Fcl = 0 %,
Fv = 26.4 %, Fa =0 %
MBL (918.9 nm)Fcf = 50.4 %, Fcl = 40.8 %,
Fv=8.8 %, Fa=0%
BIL (27.7 nm)Fcf = 0 %, Fcl = 0 %,
Fv = 35.6 %, Fa =64.4 %
d =
55 n
m
Fcf = small grains
Fcl = large grains
Fv = voids
Fa = amorphous phase
20 30 40 50 60 70
Cu Kα 2θ (degrees)
(400)
(311)(220)
(111)
Inte
nsity
(arb
.uni
t)
X-ray diffraction
thickness ~ 1 µm
68.0 68.5 69.0 69.5 70.0
Exp. XRD peak (400) Total Fit Peak 1 (22.4 nm) Peak 2 (9 nm)
2θ (degree)
Inte
nsity
(arb
. uni
t)
55 56 57 582θ (degree)
Inte
nsity
(arb
. uni
t)
Exp. XRD peak (311) Total Fit Peak 1 (48 nm) Peak 2 (11.4 nm)
26 27 28 29 30 31 32 33
Exp. XRD peak (111) Total Fit Peak 1 (14.8 nm) Peak 2 (4.8 nm)
2θ (degree)
Inte
nsity
(arb
. uni
t)
45 46 47 48 49 502θ (degree)
Inte
nsity
(arb
. uni
t) Exp. XRD peak (220) Total Fit (11.4 nm)
20 30 40 50 60 70
Cu Kα 2θ (degrees)
(400)
(311)(220)
(111)
Inte
nsity
(arb
.uni
t)
68.0 68.5 69.0 69.5 70.0
Exp. XRD peak (400) Total Fit Peak 1 (22.4 nm) Peak 2 (9 nm)
2θ (degree)
Inte
nsity
(arb
. uni
t)
26 27 28 29 30 31 32 33
Exp. XRD peak (111) Total Fit Peak 1 (14.8 nm) Peak 2 (4.8 nm)
2θ (degree)
Inte
nsity
(arb
. uni
t)
45 46 47 48 49 502θ (degree)
Inte
nsity
(arb
. uni
t)
Exp. XRD peak (220) Total Fit (11.4 nm)
55 56 57 582θ (degree)
Inte
nsity
(arb
. uni
t)
Exp. XRD peak (311) Total Fit Peak 1 (48 nm) Peak 2 (11.4 nm)
thickness ~ 1 µm
X-ray diffraction analysis
0 100 200 300 400
Freq
uenc
y (a
rb. u
nit)
Conglomerate surface grain size (nm)
d = 55 nm
d = 180 nm
d = 390 nm
d = 590 nm
d = 950 nm
σrms= 2.1 nm + 0.2 nm
σrms= 7 nm + 0.1 nm
σrms= 4.3 nm + 0.4 nm
σrms= 3.3 nm + 0.1 nm
σrms= 4 nm + 0.3 nm
thickness series of R=1/10•Ram et al, Thin Solid Films 515 (2007) 7619.
•Ram et al, Thin Solid Films (2008) in print.
Surface morphology by AFM
R=1/1, thickness = 1200 nm
400 425 450 475 500 525 550
(a) Expt. data (glass side)
Inte
nsity
(arb
. uni
t)
Raman Shift (cm-1)
(b) Expt. data (film side)
Bifacial Raman Spectroscopy
collection
excitation
film
glass
glassfilm
excitation
collection
Deconvolution of Raman Spectroscopy Data
• Microstructure of our samples: – No a-Si:H phase– Presence of two (mean) sizes of crystallites
• Conventional deconvolution:– Single mean crystallite size – A peak assigned to grain boundary material– An amorphous phase : asymmetric tail
• Previous efforts – Deconvolution based on two asymmetric Lorentzian peaks
[Touir et al, J. Non-Cryst. Solids 227-230 (1998) 906]
– Method of subtracting the amorphous contribution and fitting the resulting crystalline part of spectrum with three or five Gaussian peaks [Smit et al, J. Appl. Phys. 94 (2003) 3582]
In absence of amorphous phase
• Asymmetry in the Raman lineshape of RS profiles (low energy tail) distribution of smaller sized crystallites
• Incorporation of a bimodal CSD in the deconvolution of RS profiles avoids:– Overestimation of amorphous content
– Inaccuracy in the estimation of the total crystalline volume fraction •Islam & Kumar, Appl. Phys. Lett. 78 (2001) 715.
•Islam et al, J. Appl. Phys. 98 (2005) 024309.
•Ram et al, Thin Solid Films 515 (2007) 7619.
RS Data Deconvolution : Our Model
inclusion of crystallite size distribution (CSD)
Bifacial Raman Study
400 425 450 475 500 525 5500.0
0.3
0.6
0.9
1.2 glass side exp. data of F0E31 cd1 cd2 a fit with - cd1cd2a
Inte
nsity
(arb
. uni
t)
Raman Shift (cm-1)450 475 500 525 550
0.0
0.3
0.6
0.9
1.2 film side exp. data of F0E31 cd1 cd2 fit with - cd1cd2
Raman Shift (cm-1)
Inte
nsity
(arb
. uni
t)
Small grain (cd1) Large grain (cd2) a-Si:H
Size (nm) [σ (nm)] XC1 (%) Size (nm) [σ (nm)] XC2 (%) Xa (%)
Film side cd1+cd2 6.1, [1.68] 20 72.7, [0] 80 0
Glass side cd1+cd2+a 6.6, [1.13] 8.4 97.7, [4.7] 52.4 39.2
Sample #E31 (1200 nm, R=1/1) Fitting Model
RS(F) data bimodal CSD RS(G) data bimodal CSD +amorphous phase
400 450 500 550
fit model "cd+a"
fit model "cd+a"
a
a
a
fit model "cd1+cd2+a"
cd
cd
cd2
cd1
fit model "cd1+cd2"
cd2
cd1
d = 55 nm, RS(G)
d = 55 nm, RS(F)
d = 950 nm, RS(G)
d = 950 nm, RS(F)
Inte
nsity
(arb
. uni
t)
Raman shift (cm-1)
RS analysis
3 models:
cd1 + cd2
cd1 +cd2 +a
cd +a
200 400 600 800 1000 12000
20
40
60
80
100 (a)
Film Thickness (nm)
F cf ,
F cl ,
F v (%)
by S
E
Fcf Fcl Fv
200 400 600 800 10000
20
40
60
80
100(b)
Xa, X
c1, X
c2 (%
) by
RS
Film Thickness (nm)
Xc1 (%) Xc2 (%) Xa (%)
200 400 600 800 1000 12000
20
40
60
80
100 (a)
Film Thickness (nm)
F cf ,
F cl ,
F v (%)
by S
E
Fcf Fcl Fv
200 400 600 800 10000
20
40
60
80
100(b)
Xa, X
c1, X
c2 (%
) by
RS
Film Thickness (nm)
Xc1 (%) Xc2 (%) Xa (%)
Fractional composition of films: Qualitative agreement between RS and SE studies
Samples belong to thickness series of R=1/10
Summary of variation in fractional compositions and roughness with film growth
300 600 900 12000
20
40
60
80
100(a)
R = 1/1
FcFcf
Fcl
F c , F cf
and
Fcl (%
) by
SE
σSE
d (nm)300 600 900 1200
(b)
R = 1/5
FcFcf
Fcl
σSE
d (nm)300 600 900 1200
2
4
6
8
(c)
Fcl
FcfFc
R = 1/10
d (nm)
σSE
Rou
ghne
ss b
y SE
, σSE
(nm
)
d = 390 nm
σrms= 3.3 nmσrms= 7 nm
d = 170 nm
σrms= 2.1 nm
d = 55 nm d = 590 nm
σrms= 4.3 nm
d = 950 nm
σrms= 5 nm
Surface morphologies of the samples belonging to thickness series of R=1/10
Dark Electrical Transport Properties
–Above Room Temperature
• Microstructure:
– >90% crystallinity, no amorphous phase• Electrical transport:
– Crystallinity ?– Crystallite size ? – Interfacial regions between crystallites or columns ?
• Carrier transport is influenced by:– Film morphology – Compositional variation in constituent crystallites
• large grain fraction?• Microstructure ↔ Electrical transport:
– Need for investigation of correlation with large grain fraction
Electrical transport in single phase µc-Si:H
2.1 2.4 2.7 3.0 3.310-10
10-8
10-6
10-4
10-2
(a)
1200; 0.2 920; 0.15 450; 0.55 180; 0.58 62; 0.58 Fit
d (nm); Ea (eV)
R = 1/1
σ d (Ω
.cm
)-1
1000/T (K -1)2.1 2.4 2.7 3.0 3.3
10-7
10-6
10-5
10-4
10-3
σ d (Ω
.cm
)-1
(b)
d (nm); Ea (eV)
R = 1/10
1000/T (K -1)
950; 0.33 590; 0.44 390; 0.44 170; 0.54 150; 0.54 55; 0.54 Fit
Above room temperature dark electrical conductivity (σd) shows
Arrhenius type thermally activated behavior: σd(T)=σo e –Ea / kT
•Ram et al, Thin Solid Films 515 (2007) 7469.
200 400 600 800 1000 120010-10
10-8
10-6
10-4
10-2
(a)
P =1.5 TorrTS =250 0C
TS =150 0C
TS =100 0C
σ d (Ω c
m)-1
Film thickness (nm)
R, TS
1/10, 200 0C 1/5, 200 0C 1/5, variable T
S
1/1, 200 0C 1/20, 200 0C
200 400 600 800 1000 12000.1
0.2
0.3
0.4
0.5
0.6
0.7 (b)
TS =250 0C
P=1.5 Torr
TS =150 0C
TS =100 0C
Film thickness (nm)
E a (eV
)
200 400 600 800 1000 1200 10-3
10-1
101
103
105
(c)
P=1.5 Torr
TS =250 0C
TS =150 0C
TS =100 0C
Film thickness (nm)
σ 0 (Ω
cm
)-1
0 20 40 60 800
200
400
600
800
1000
1200
Fi
lm T
hick
ness
(nm
)
Fcl (%)0
200
400
600
800
1000
1200
10-7 10-6 10-5 10-4 10-3 10-2
Film
Thi
ckne
ss (n
m)
σd (Ω cm)-1
0.1 0.2 0.3 0.4 0.5 0.6Ea (eV)
0 100 200 300 400
Freq
uenc
y (a
rb. u
nit)
Conglomerate surface grain size (nm)
d = 55 nm
d = 180 nm
d = 390 nm
d = 590 nm
d = 950 nm
Samples belong to thickness series of R=1/10
0 20 40 60 80 100
10-2
10-1
1
101
102
103
104
type-Ctype-Btype-A
σ0Ea
0.1
0.2
0.3
0.4
0.5
Fcl (%)
E a (eV
)
σ 0 (Ω c
m)-1
Classification of films: electrical transport behavior and Fcl
•Ram et al, J. Non-Cryst. Solids 354 (2008) 2263.
Classification of films
Type-A material• Small grains (SG)
• Low amount of conglomeration (without column formation)
• High density of intergrain boundary regions containing disordered phase.
Type-C material• Highest fraction of LG.
• Well formed large columns
• Least amount of disordered phase in the columnar boundaries.
Type-B material• Rising fraction of LG.
• Marked morphological variation: column formation
• Moderate amount of disordered phase in the columnar boundaries.
0 20 40 60 80 100
10-2
10-1
1
101
102
103
104
type-Ctype-Btype-A
σ0Ea
0.1
0.2
0.3
0.4
0.5
Fcl (%)
Ea (e
V)
σ 0 (Ω c
m)-1
Meyer Neldel Rule (MNR) & anti-MNR behaviors in dark electrical transport
Meyer Neldel Rule (MNR)Observed in:
Materials:
Ionic Materials Chalcogenide glassesOrganic thin filmsAmorphous Silicondoped μc-Si:H
Processes:
Annealing Phenomena Trapping in crystalline SemiconductorsAging of insulating polymersBiological death ratesChemical reactionsElectrical conductionmicroscopic origin of MNR
& physical meaning of G ??
electrical transport in a-Si:H/disordered semiconductor: MNR σ0=σ00 eGEa ,
where G or EMN (=1/G)
and σ00 are MNR parameters
σd=σ0.exp(-Ea/kT)
Statistical shift of Fermi level
Activated process: Y=A.exp (-B/X)
MNR A=A’.exp(GB)where G and A’ are MNR parameters
Statistical Shift Model
According to Mott: σd(T) =σM exp{-(EC - EF)/kT}
EC(T ) = EC0 - γCT ; EF(T ) = EF
0 - γFT
Ea= EC0 - EF
0, at T=0 K
σd=σo exp (–Ea / kT )
σo=σM exp {(γC - γF) / k}
σ0=σ00 exp (GEa) --- MNR
Anti Meyer Neldel Rule
Correlation between σ0 and Ea appears to change sign– a negative value of MN energy (EMN) is seen
Experimentally observed in:– Heavily doped μc-Si:H
– Heterogeneous Si (het-Si) thin film transistor
– Organic semiconductors
Lucovsky & Overhof (LO) model
in a degenerate case Efmoves above Ec in the crystalline phase
consequently Ef can move deeply into the tail states in the disordered region, giving rise to anti MNR behavior.
Energy band diagram as proposed by Lucovsky et al, J.N.C.S. 164-166, 973 (1993)
The reason for observed anti MNR in doped µc-Si:H
0.0 0.2 0.4 0.6 0.8
10-2
100
102
104
anti MNR parametersG = -44.6 eV-1
or EMN=-22.5 meVσ00= 87 (Ωcm)-1
MNR parametersG=25.3 eV-1 (EMN=39.5 meV)σ00=7.2x10-4 (Ωcm)-1
γf ~ 0
γf ~ γc
σ 0 (Ω c
m)-1
Ea (eV)
type-A type-B type-C
σ0 vs. Eaσo and Ea follow linear
relationship Type-A and Type-B samples.
Type-A samples are having high values of Eaand σ0
This shows γF is extremely small in Type-A samples due to its pinning
The values of MNR parameters nearly the same as found in a-Si:H.
Correlation between σoand Ea appears to change sign for type-C samples:anti-MNR
Findings
MNR & anti MNR in single phase μc-Si:H•Ram et al, Phys. Rev. B 77 (2008) 045212.
•Ram et al, J. Non-Cryst. Solids 354 (2008) 2263.
MNR: type-A μc-Si:H
• Consists mainly of SG with an increased number of SG boundaries. – No question of formation of
potential barrier (i.e., transport through crystallites)
– transport will be governed by the band tail transport.
0.0 0.2 0.4 0.6 0.8
10-2
100
102
104
anti MNR parametersG = -44.6 eV-1
or EMN=-22.5 meVσ
00= 87 (Ωcm)-1
MNR parametersG=25.3 eV-1 (EMN=39.5 meV)σ00=7.2x10-4 (Ωcm)-1
γf ~ 0
γf ~ γc
σ 0 (Ω c
m)-1
Ea (eV)
type-A type-B type-C
• Ea saturates (≈ 0.55 eV) and σo ≈ 103 (Ωcm)-1.– EF is lying in the gap where the DOS does not vary much and there is a
minimal movement of EF, or γF ≈ 0 • The initial data points for type-A have higher σo [≈ 104 (Ωcm)-1] and Ea (≈
0.66 eV)– because of a shift in EC and/or a negative value of γF, as happens in
a-Si:H for Ea towards the higher side.
MNR: type-B μc-Si:H
Improvement in film microstructuredelocalization of the tail states– EF moves towards the band edges,
closer to the current path at EC. – γF depends on T and initial position
of EF, and when EF is closer to any of the tail states and the tail states are steep, γF is rapid and marked.
Transition between Type-A and Type-B materials– Nearly constant σo [70-90 (Ωcm)-1] with the fall in Ea (0.54-0.40 eV),– γF ≈ γC, canceling each other out in σo=σM exp [(γC - γF) / k]– EF pinned near the minimum of the DOS between the exponential CBT
and the tail of the defect states (DB–)– With increasing crystallinity and/or improvement in microstructure,
minimum shifts towards EC leading to a decrease of Ea.
0.0 0.2 0.4 0.6 0.8
10-2
100
102
104
anti MNR parametersG = -44.6 eV-1
or EMN=-22.5 meVσ
00= 87 (Ωcm)-1
MNR parametersG=25.3 eV-1 (EMN=39.5 meV)σ
00=7.2x10-4 (Ωcm)-1
γf ~ 0
γf ~ γc
σ 0 (Ω c
m)-1
Ea (eV)
type-A type-B type-C
Is the model by Lucovsky et al applicable for
explaining Anti MNR in type-C μc-Si:H ?
• The value of EMN = -22.5 meV is close to the value reported in heavily doped µc-Si:H (-20meV)
• EB diagram as suggested by LO model seems inapplicable to
our undoped µc-Si:H case
– Calculated free electron concentrations do not suggest
degenerate condition.
– Consideration of equal band edge discontinuities at both ends of
c-Si and a-Si:H interface Doubtful
– Also, in a degenerate case, the conductivity behavior of
polycrystalline material is found to exhibit a T 2 dependence of σd
• Considering transport through the encapsulating disordered tissue, a band tail transport is mandatory.
• The large columnar microstructure in a long range ordering delocalizes an appreciable range of states in the tail state distribution.
• In addition, higher density of available free carriers and low value of defect density can cause a large increase in DB–
density together with a decrease in DB+ states in the gap a lower DOS near the CB edge possibility of a steeper CB tail.
• In this situation, if Ef is lying in the plateau region of the DOS, it may create an anti MNR situation.
Applying the statistical shift modelin explaining Anti MNR in type-C μc-Si:H
Evidence of Anti MNR in μc-Si:H in
Literature
# 1 undoped µc-Si:H# 2 lightly p-doped µc-Si:H
0.0 0.2 0.4 0.6 0.810-3
10-1
101
103
105
Ea(eV)
anti-MNR line of type-C μc-Si:H
MNR line of types: A & B μc-Si:HMNR line of a-Si:H
#1 (rH=21) #1 (rH=32) #2 #3 (a-Si:H) this work
σ 0 (Ω
.cm
)-1
•Ram et al, Phys. Rev. B 77 (2008) 045212.
Undoped µc-Si:H
0.0 0.2 0.4 0.6 0.810-3
10-1
101
103
MNR line (#7) [a-Si,C:H+μc-Si,C alloy]
anti MNR line (#7) [heavily doped μc-Si:H]
#4 (thickness series) #4 (doped series) #5 dope series, p-nc-Si-SiC:H alloy#5 dilution series, p-nc-Si-SiC:H alloy #6 (Boron doped μc-Si:H) #7
σ 0 (Ω.c
m)-1
Ea (eV) •Ram et al, Phys. Rev. B 77 (2008) 045212.
Doped µc-Si:H
MNR parameters Anti MNR parameters
Samples σ00
(Ω.cm)-1 G
(eV-1) EMN
(meV)σ00
(Ω.cm)-1 G
(eV-1) EMN
(meV) This workType-A&B
Type-C Published
Data Case#1 (rH=21) Case#1 (rH=32) Case#2 Case#3 Case#4 Case#5 Case#6 Case#7 Case#8 Case#9
7.2×10-4
--
4×10-3
3.2×10-6
1.7×10-4 7.7×10-3
0.32 4.2×10-3 3.2×10-6
2.3 0.5
7.2×10-3
25.3
--
20.7
36.6
23.4 24
15.4 15.3 31.3 8.5
11.8 20
39.5
--
48.4
27.3
42.7 41.6 65.1 65.4 31.9
118.384.5 50
-- 87
1.26×1010
--
6 -- 59 21 2.4 309 -- --
--
-44.6
-97.7
--
-32.5 --
-66.1 -64.9 -39.9 -49.5
-- --
--
-22.5
-10.2
--
-30.8 --
-15.1 -15.4 -25.1 -20.2
-- --
5 10 15 20 25 30 35 4010-6
10-4
10-2
100a-Si,C:H alloy (#7)
a-Si:H (#3)
p-nc-Si-SiC:H alloy (#5)
Porous Si (#9)
#1 (rH=21) #1 (rH=32) #2 #3 #4 #5 #6 #7 #8 #9 this work FitσM=100 (Ωcm)-1 (at γf=γc)
Emin=0.61 eVσ0=1.2x103 (Ωcm)-1 (at γf=0)
σ 00 (Ω
.cm
)-1
G (eV-1)
If one has a collection of G and σ00 then:
σ00=σM exp [(γC- γF)/k –GEa]
σ00=σM exp [(γC- γF)/k –G(EC0 –EF
0)]
At a position of EF in DOS where
γF(EC0-Emin)=0
σ00=σM exp [(γC/k) –GEmin]
The quantity Emin is a measure for the
position of the DOS minimum within
the mobility gap.
If γC is known then for such a value of
σ00 where G=0, one can obtain σM•Ram et al, Phys. Rev. B 77 (2008) 045212.
Summary• In single phase µc-Si:H films, film morphology shows
correspondence with large grain fraction independent of film thickness and deposition conditions
• Percentage fraction of constituent large crystallite grains can be used as an empirical parameter to correlate a wide range of microstructures to the electrical transport properties
• Both MNR and anti MNR can be seen in the dark conductivity behavior of this material, depending on the microstructure and the correlative DOS features.
• The statistical shift model can successfully explain both the MNR and anti MNR behavior in our material.
• Corroborative evidence of similar electrical transport behavior of µc-Si:H in literature is present
Acknowledgements
• Dr. Satyendra Kumar (I.I.T. Kanpur, India)• Dr. Pere Roca i Cabarrocas (LPICM, France)
Thank you