Phase Diagrams for Si1xGex:H Thin Films

1
Phase Diagrams for Si1-xGexH Thin Films

• Contributors
• N. J. Podraza (1), C. R. Wronski (2), M. W. Horn
  (2),
• and R. W. Collins (1)
• (1) Department of Physics and Astronomy,
• The University of Toledo, Toledo, OH, USA
• (2) Materials Research Institute,
• The Pennsylvania State University, University
  Park, PA, USA

2
Outline

• Motivation
• Experimental details
• Real time spectroscopic ellipsometry (RTSE)
• III. Experimental details
• Plasma-enhanced chemical vapor deposition
  (PECVD)
• IV. Comparison of Si1-xGexH phase diagrams
• vs. H2-dilution flow ratio for PECVD films
  prepared on the anode and cathode at 200?C
• V. Comparison of Si1-xGexH phase diagrams vs.
• H2-dilution flow ratio at different GeH4 flow
  ratios
• VI. Comparison of SiH and Si1-xGexH phase
  diagrams vs. Helium dilution
• VII. Summary and future directions

3
I. Motivation

• We seek to optimize the components of
  a-SiH-based solar cells that apply triple
  junction n-i-p design based on an understanding
  of growth process.

• The deposition principle for the a-SiH i-layer
  is to employ the highest possible H2-dilution
  flow ratio RH2/SiH4 without crossing the
  (amorphous)-to-(mixed-phase) transition a ?
  (amc).
• A second principle is to ensure the
  largest possible thickness for the onset
  of the amorphous roughening transition a ? a
  as assessed in studies
  on c-Si substrates.
• Similar optimization principles are expected for
  Si1-xGexH alloys.
• Si1-xGexH alloys are being studied using RTSE to
  locate optimum parameter space regions which are
  then verified by other analysis methods.

4
Experimental Details PECVD of Si1-xGexH for
Phase Diagram Development

• Same single chamber rf PECVD system as that used
  previously for extensive studies of SiH phase
  diagrams
• (Native-oxide)/c-Si substrates for smoothness and
  highest sensitivity to amorphous roughening
  transitions
• Minimum rf plasma power for a stable plasma (0.08
  mW/cm2)
• Low partial pressure of SiH4GeH4 (0.06
  Torr) with a total pressure lt 1.0 Torr for all
  depositions versus H2-dilution flow ratio
• H2 flow ratio, RH2/SiH4GeH4 as the
  abscissa of the phase diagram which controls the
  phase of the film (a, amc, mc)
• Variable GeH4 flow ratio, GGeH4/SiH4GeH4
  
• ? yields room temperature optical (Tauc) gap of
  
• Eg1.8 ? 1.3 eV for G 0 ? 0.167
• ? also controls phase of the film (a, amc, mc)
• Anode (grounded) and cathode (-20 V self-bias)
  electrode config's
• Variable He flow ratio, HeHe/SiH4GeH4
  controls the phase of the film (a, amc, mc)

5
IV. Explanation of a ? a Roughening Transition
6
IV. Explanation of a ? a Roughening Transition
PECVD parameters for SiH series Substrate
c-Si T 200?C P 0.72 W/cm2 R
H2/SiH4 15 - 60 ptot 3 Torr
a ? a transitions
Fit to experimental data for a-SiH with R40
using the 1-D continuum model. In this model,
initial nuclei with a mean radius of 24 Å are
assumed, along with a diffusion length of l045 Å.
Amorphous roughening transition thickness
obtained from real time SE and the critical
diffusion length obtained from the 1-D model
plotted vs. R for the series of a-SiH samples.
R and a?a are linearly related
7
IV. Explanation of the a ? (amc) Roughening
and the (amc) ? mc Smoothening Transitions
Geometrical model for (amc)-SiH phase evolution
Motivation Conversion of RTSE data to simple
geometric parameters Assumptions (i) Abrupt
onset of mc nucleation versus thickness (ii) Nucl
ei on a square grid (iii) Constant cone angle for
the preferential growth of mcs (iv) Linearly
increasing radius of cone caps Approach From
geometric considerations Ddb db,coal -
db,trans and Dds ds,coal - ds,trans
together yield (Nd, q)
8
IV. Explanation of G Width of the Dielectric
Function
Fit using
expression from A.S. Ferlauto et al. J. Appl.
Phys. 92, 2424 (2002).
Monotonic increase in width G indicates a
reduction in excited state lifetime t h/G due
to decreased order with alloying. Using a
simple expression for mean free path Lh?/G (?
electron velocity) G 2 ? 3 eV ? L 5 ? 4 Å
, and a lack of long-range order.
9
IV. Comparison of Si1-xGexH Anode/Cathode
Diagrams
Higher deposition rates and higher a?(amc)
thicknesses (at db 4000 Å) for cathode
Si1-xGexH as compared to anode Si1-xGexH under
otherwise similar conditions here G
GeH4/SiH4GeH4 0.167

G0.167

• a g (amc) transition is shifted to much higher R
  for cathode
• Si1-xGexH
• 2) This opens up a narrow window 70 ? R ? 90
  that yields an a g a transition at a much
  higher value of db 2000 Å for cathode
  Si1-xGexH
• ? high surface stability

10
V. Optimization of Si1-xGexH Cathode PECVD
at Different GeH4 Flow Ratios
Higher deposition rates with increasing GeH4 flow
ratio, G GeH4/SiH4GeH4 at R10 For
fixed G, G decreases up to maximal R prior to
a?(amc) a ? a transition at maximum bulk layer
thickness shifts to higher R with increasing G
11
VI. Comparison of SiH with Helium Dilution

• Comparison of the growth of He0 R0 and R10
  a-SiH films with a He10 R0 film
• a ? a transition shifts to higher bulk layer
  thickness between He0 and He10 R0 films, but
  fails to reach improvement gained by R10
• H2 dilution

• Comparison of the growth of R40
• SiH films deposited with He0 and He10
• overall lower roughness amplitude
• a ? (amc) transition occurs at
• db lt 10 Å for He0, and
• db 1400 Å for He10
• ? He suppresses microcrystallinity

12
VI. Extended SiH Phase Diagram with He dilution
Higher deposition rates with He10 SiH as
compared to He0 SiH under otherwise similar
conditions here He He/SiH4

4000 Å

• a g (amc) transition is shifted to higher bulk
  layer thickness for He10 SiH allowing for an
  extended protocrystalline regime
• a ? a transition is shifted to higher bulk layer
  thickness, indicating minor improvements in film
  quality

13
VII. Comparison of Si1-xGexH with He dilution

• Comparison of the growth of He0 and He10
  a-Si1-xGexH films with R10
• a ? a transition shifts to higher bulk layer
  thickness between He0 and He10 R10 films

G0.167
G0.167

• Comparison of the growth of He0 and He20
  Si1-xGexH films with R80
• a ? a transition shifts to higher bulk layer
  thickness
• a ? (amc) transition shifts to gt 3500 Å ? He
  suppresses microcrystallinity

14
VII. Extended Si1-xGexH Phase Diagrams with He
dilution
Higher deposition rates for R10 Si1-xGexH
films up to He10, but decreasing with total
dilution S Slightly increasing depostion rate for
R80 Si1-xGexH with He20

4000 Å
G0.0167

• R10 series a g a transition is shifted to higher
  bulk layer thickness initially but saturates at
  db200 Å
• R80 series a ? a transition is shifted to higher
  bulk layer thickness while the a ? (amc)
  transition is strongly suppressed

15
Phase Diagrams for Si1-xGexH Thin Films
VII. Summary A. SiH and Si1-xGexH phase
diagrams have been developed at T200oC,
comparing the anode and cathode electrode
configurations and the role of low-energy ion
bombardment. These comparisons show that 1)
the ag(amc) transitions shift to higher R for
films deposited on the cathode for both SiH
and Si1-xGexH a. for SiH, this shift creates
an extended protocrystalline regime at higher
dilution levels, while maintaining very high
a ? a transition thicknesses at lower
dilutions b. for Si1-xGexH, this shift opens
a narrow window leading to a ? a transitions
at much higher bulk thicknesses, suggesting
significant increases in precursor surface
diffusion for a-Si1-xGexH deposited at the
cathode 2) The higher surface stability for the
cathodic films is accompanied by smoother
surfaces
16
Phase Diagrams for Si1-xGexH Thin Films

• VII. Summary (continued)
• B. Si1-xGexH phase diagrams versus R have been
  developed at T200oC under
• cathodic deposition conditions for different
  values of the GeH4 flow ratio G.
• This comparison shows that
• 1) the a?a transitions shift to higher
  thicknesses with increasing R right up
• to the a?(amc) transitions, indicating the
  beneficial effect
• of H on precursor surface diffusion under all
  alloying conditions
• 2) the dielectric function broadening parameter
  G decreases with increasing H2
• dilution up to the a?(amc) transition,
  indicating improved film quality
• SiH and Si1-xGexH (G0.167) have been studied
  with He dilution.
• 1) The presence of He in both a-SiH and
  a-Si1-xGexH has been shown to
• shift the a?a roughening transition to higher
  bulk layer thicknesses,
• indicating enhanced surface diffusion, but the
  effect is weaker than for H2
• 2) Similarly, He has been shown to suppress
  microcrystalline formation,
• shifting the a?(amc) transition in thin film
  SiH and Si1-xGexH to
• higher bulk layer thickness, creating an
  extended protocrystalline regime

17
Phase Diagrams for Si1-xGexH Thin Films

• VII. Future Directions
• Explore additional non-standard deposition
  techniques to
• improve film quality and/or increase rate
• Controlled low-energy ion bombardment
• Total pressure variation
• Vhf plasma excitation frequencies
• Preparation and optimization of graded layer
  a-SiH,
• a-Si1-xGexH, and mc-SiH structures through
• simultaneous variation in G, R, and He

18
(No Transcript)
19
Outline

• I. Statement of the Problem
• II. Review of Deposition Phase Diagrams
• III. Instrumentation Unique to Univ. Toledo
• IV. Approaches for Real Time Analysis
• 1. Rigorous Virtual Interface Analysis
• 2. Empirical E2 Amplitude Analysis
• V. Comparison of Approaches
• VI. Results for One-Step Microcrystalline SiH
• VII. Results for Two-Step Microcrystalline SiH
• VIII. Summary and Future

20
Statement of Problem Optimization of i-layer
for mc-SiH Solar Cells
A. Shah et al. Sol. Energy Mater. Sol. Cells
78, 469 (2003) "If we add hydrogen to the plasma,
... the layer remains amorphous until we reach a
threshold concentration. So far, the best
microcrystalline solar cells are deposited near
the threshold concentration for very high values
of hydrogen dilution, the material does not
appear to be usable for solar cells, because of
the cracks ... that appear also to act as
channels by which contamination ... can enter
into the layer." J. Bailat, Ph.D. Thesis (2004)
Neuchâtel "Amorphous tissue is responsible for
the passivation of the nano-crystals surface ..."
21
Review of Phase Diagrams and Optimization
Procedures
Question The substrates are R0 a-SiH films
however, once the overlying film begins to grow
in the first step, a new "substrate" is
established, either R40 a-SiH or R200
mc-SiH. Then why should the same diagram be
valid for multistep deposition? Answer The same
phase diagram is NOT valid and in both cases the
phase boundary shifts to lower R in subsequent
deposition steps. This effect must be taken into
account in the optimization prescription. It
is important to analyze in greater detail the
substrate dependence of the phase evolution
including the effect of crystallite size.
22
Unique Instrumentation for RTSE and Graded Solar
Cell Fabrication
? H2 dilution grading has a great effect on
mc-SiH cell performance. ? The film growth
processes of sample GD1955 were measured using
RTSE.
23
Approaches for Real Time Analysis

• PECVD conditions
• Substrate temperature T 200?C
• Plasma frequency f 60 MHz (VHF)
• Plasma power density P 0.08 mW/cm2
• Hydrogen dilution R H2/SiH4 20
• Total gas pressure ptot 0.54 Torr
• Four growth regimes
• I d lt 100Å a-SiH nucleation/coalescence
• II 100Å lt d lt 500Å microcrystal nucleation from
  
• amorphous phase
• (amc)-SiH growth
• III 500Å lt d lt 1500Å microcrystal coalescence
• IV d gt 1500Å single-phase mc-SiH
  growth

Two different approaches were applied to deduce
the volume fraction evolution throughout the
growth 1) Rigorous virtual interface analysis.
Advantages high accuracy depth
resolution Disadvantages difficult to implement
in real time 2) Empirical E2 amplitude
analysis Advantages easy to implement in real
time after C. Ross et al. MRS (2003) Disadvantage
s indirect -- requires calibration low depth
resolution
24
Comparison of Approaches

• Depth profiles are from
• Rigorous VI analysis (red circles)
• (i) a-SiH dielectric function obtained
  from
• y(E, t), D(E, t) in first 120 Å.
• (ii) mc-SiH dielectric function obtained
  from
• y(E), D(E)at a thickness of
• (1500 Å).
• (iii) ds(t), fmc(t), tt ? t ? tc
  determined from VI analysis r(t) can be
  fixed at value
• obtained in uniform growth analysis.
• Empirical E2 amplitude analysis of D(E) (blue
  circles)
• performed as described on previous
  viewgraph

Differences in profiles are attributed to
differences in depth resolution dd Rigorous
VI analysis (solid circles) dd r Dt ? 20 Å
Empirical analysis (open circles) dd ? 2a
(4.05 eV)-1 ? 2/(1.55x106) cm ? 130 Å
25

• Cone model of ds(t) vs. db(t)
• Model is used to extract nucleation density Nd
  and cone angle q.
• Assumptions of the model
• All nuclei originate at d db,t
• Nuclei grow with cone angle q
• Tops of cones are spherical caps with radius
  d ? db,t.

Results for One-Step mc-SiH

• Correlation of Nd and q with db,t
• Cone angle nearly constant at
• q ? 15-20
• Nucleation density decreases rapidly
• with increasing db,t
• Good agreement with AFM and XTEM.

Increasing R
26
Results for Two-Step mc-SiH
Deposition Procedure Substrate R0 a-SiH (to
simulate a-SiH n-layer) Deposit SiH with R40
(no seeding) this leads to a 200 Å amorphous
(protocrystalline) layer before the a ? (amc)
transition Reduce R to R20 when fmc reaches 0.5
to slow the rate of crystallite growth
R20 a-SiH
Modeling the Crystallite Evolution in Two
Steps Crystallite density can be estimated by
fitting fmc versus db in the initial stages of
growth or the final stages of coalescence using
the cone growth model. The crystallite density
is found to decrease with decreasing R.
R40 a-SiH
R0 a-SiH
27
Summary
Amorphous phase passivation at the grain
boundaries of mc-SiH i-layers appears necessary
for the fabrication of optimum solar cells. This
requires depositing the film near the boundary
between amorphous and mixed-phase (amc)-SiH
film growth. Because the boundary between
amorphous and mixed-phase (amc)-SiH film
growth shifts to decreasing R with increasing
thickness, it becomes necessary to grade the
H2-dilution during the growth of mc-SiH for
optimum i-layer material. However, because of
its high substrate sensitivity, the deposition
phase diagram provides only qualitative guidance
for identifying the optimum graded layer
process. As a result, real time
characterization and control of graded layer
growth is important. We have compared two
procedures for determining the depth profile in
the volume fraction of crystallinity. Both yield
reasonable results and are promising for real
time control.