Strength, Stiffness, and Microstructure of Cu(In,Ga)Se2 Thin Films
Deposited via Sputtering and Co-Evaporation
Supplemental Online Materials
Experimental Details
Multilayered stacks of Cu(In,Ga)Se2 (CIGS) samples were provided by National Nano
Device Laboratories (NDL) of Taiwan. Two separate batches were created, one using
selenization and the other co-evaporation processes. In both processes CIGS was
deposited on top of a 700 nm molybdenum back contact layer sputtered on Soda Lime
Glass (SLG) substrate in a two-steps process. During the first step, a 70 nm-thick Mo
layer was sputtered at a pressure of 10 mTorr; with additional 630nm sputter-deposited
during the second step at 3.6 mTorr. Cu and In metal precursors were sputtered onto the
Mo contact at 2.8 mTorr and 150 °C, followed by selenization with Se vapor with N2/H2
carrier gas at 550 °C. The total thickness of the CIGS absorber layer was measured via
Scanning Electron Microscopy (SEM) to be 1.6 ± 0.2 μm. For the co-evaporated CIGS,
the absorber layer was deposited onto 0.46 cm2 Mo coated SLG substrates using a 3-step
homogeneous thermal evaporation of elemental Cu, In, Ga and Se at 550 °C. The overall
thickness of the co-evaporated absorber layer was 1.7 ± 0.1 μm. SEM images were taken
using FEI Nova Nanolab 200 Focused Ion Beam (FIB)/SEM, and Electron Dispersive
Spectroscopy (EDS) within the same instrument was used to ascertain the chemical
composition of the sample. Samples were cut into 1 cm × 1 cm pieces for SEM imaging.
Cross-sectional SEM images were acquired at a tilt angle of 52 °. Grain sizes were
estimated based on the SEM and Transmission Electron Microscopy (TEM) images in the
following manner: (1) multiple lines of the same length were drawn across the image in
both horizontal and vertical directions emanating from different, randomly chosen points;
(2) number of intersections of the lines with the grain boundaries were recorded; (3) the
average grain size is taken to be the total length of the lines divided by the total number
of intersections. A micromanipulator (Omniprobe) within the Focused Ion Beam (FEI,
Nova 200) was used to create nano-pillars for uniaxial micro-compression experiments
and to extract site-specific locales within the films. Bright-field images and Electron
Energy Loss Spectroscopy (EELS) maps were generated from both samples using Tecnai
TF-30 TEM (FEI). The curvature induced in the substrate after the deposition of films
was measured using KLA Tencor Laser Interferometer for the neat SLG substrates,
SLG/Mo and SLG/Mo/CIGS samples. Nano-indentation into selenized films was
conducted using a standard Berkovich tip in Hysitron TI-950 Nanoindenter with the tip
effective radius of curvature of ~150 nm. Indentation depth was kept at 100 nm to avoid
the effects of the substrate. All indentations were performed under displacement-rate
control using a feedback algorithm, to an indentation depth of 100 nm at a prescribed
displacement rate of 1 nm/s, and then fully unloaded at a rate of 4 nm/s, where a slower
loading rate was used compared to unloading since the initial loading part is of major
interest in this analysis. A 300 second-long hold was set up before each indentation to
ensure that the thermal drift rate was less than 0.1 nm/s
Cylindrical nano-pillars were milled out from the co-evaporated films using Focused GaIon Beam (FIB) and compressed with a custom-fabricated 8 μm-diameter diamond flat
punch in a different nanoindenter (Agilent G200). A total of five tests were conducted,
with the top pillar diameters of 431 ± 15 nm, bottom diameters of 639 ± 32 nm, and
heights of 1.7 ± 0.2 nm. A maximum strain of 0.3 and a strain rate of 5 × 10−3 /𝑠 were
prescribed for all tests. Load-displacement data was collected in the Continuous Stiffness
Measurement (CSM) mode of the instrument. For each test, the pillar of interest was
located using a top-view 150X optical microscope; the distance between the indenter
head and the microscope objective was then calibrated on the surface of the sample away
from the pillar following a standard methodology.1 The indenter head would
subsequently locate the sample surface near the pillar and descend to 5 μm above it. After
a 300-second hold, the indenter tip would approach the pillar at a rate of 5 nm/s until
establishing contact, after which the compression experiment commenced at the set strain
rate. During each test, real-time load on the pillar, displacement from the point of contact,
and the harmonic contact stiffness were measured and recorded. The load-displacement
data was then converted to stresses and strains by following Greer2 and Lee3.
The chemical composition of each sample measured by EDS is provided in Table. S1.
The current density-voltage (J-V) characteristics were measured on functional cells, as
shown in Fig. S1; the buffer/top contact and coating layers were not deposited on samples
for all other tests. A schematic of the sample stack is shown in Fig. S2.
Estimation of Residual Stress in the Film
Stoney’s Formula (Eqn. S1)4 was used to estimate the residual stress in the co-evaporated
films under the assumption that the stress state and the film radius of curvature is
equibiaxial (two equal components in any two mutually orthogonal in-plane directions).
𝑡2
𝐸
𝜎𝑓 = (1−𝜈𝑠 ) 6𝑡𝑠 𝐾
𝑠
𝑓
(S1)
Here, 𝜎𝑓 is the stress in the deposited thin film, 𝐸𝑠 , 𝜈𝑠 and 𝑡𝑠 are Young’s modulus,
Poisson’s ratio and thickness of the substrate, respectively; 𝑡𝑓 is the thickness of the thin
film, and 𝐾 is the radius of curvature of the substrate after thin film deposition. A
schematic showing the bent substrate caused by the presence of thin films is shown in Fig.
S3(a). Under further assumption that film thickness is small compared to substrate
thickness: 𝑡𝑓 ≪ 𝑡𝑠 , the stress introduced by the deposition of additional layers on the first
layer becomes negligible, thus stress in each film layer is only related to the additional
curvature introduced by said layer, as shown in Fig. S3(b).
Three separate curvature measurements were made on the same substrate with iteratively
added thin film layers to account for the multilayer structure of CIGS: (1) the blank SLG
substrate, (2) the same substrate after deposition of Mo layer, and (3) the same Mo/SLG
stack with the co-evaporated CIGS layer. The thicknesses of the Mo and CIGS layers,
( 𝑡𝑓 ) were measured from the SEM images of the cross-sections of each film. The
Young’s modulus of the SLG substrate (𝐸𝑠 ) was obtained from nanoindentation into
substrate to a depth of 200 nm and using Oliver-Pharr nonlinear curve fit method.5 It is
reasonable to assume that the substrate is stress-free before deposition, so the biaxial
stress in each thin after deposition can be estimated by applying Equation S1 after each
curvature measurement. These results are shown in Table SII.
Nanopillar Analysis
The analysis of compressive response of the CIGS nanopillars was conducted following
the methodology of Greer2 and Lee3. We assumed perfectly cylindrical pillars and
uniaxial compression to estimate global stresses and strains from the load-displacement
data. Conservation of volume during plastic deformation allows us to calculate the
instantaneous compression area Ap and pillar length Lp and initial pillar height L0 and area
A0, with the latter being an average of top and bottom area of an individual pillar. The
total measured displacement, dtot, can be expressed in terms of the elastic component del
and plastic component dp, while del can be expressed in terms of measured stiffness
𝑘𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 and load on pillar F, which were continuously recorded by CSM; dp would
then give us instantaneous pillar length Lp, and the stress σ.
𝐴0 𝐿0 = 𝐴𝑝 𝐿𝑝
(S2)
𝐿𝑝 = 𝐿0 − 𝑑𝑝 = 𝐿0 − (𝑑𝑡𝑜𝑡 − 𝑑𝑒𝑙 ) = 𝐿0 − (𝑑𝑡𝑜𝑡 − 𝑘
𝑃
𝐹
𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑
)
𝑃𝐿𝑝
σ=𝐴 =𝐴
𝑝
(S3)
(S4)
0 𝐿0
The true strain ε consists of the elastic part ε𝑒𝑙 , given by Hooke’s Law, and plastic part ε𝑝 ,
related to Lp:2
ε = ε𝑒𝑙 + ε𝑝 = 𝐸
𝐹
𝐶𝐼𝐺𝑆 𝐴𝑝
𝐿
+ 𝑙𝑛 (𝐿0 )
𝑝
(S5)
Where 𝐸𝐶𝐼𝐺𝑆 is Young’s Modulus for CIGS, obtained from the stiffness 𝑘𝐶𝐼𝐺𝑆 :3
𝐿𝑝
𝐸𝐶𝐼𝐺𝑆 = 𝑘𝐶𝐼𝐺𝑆 𝐴
𝑝
(S6)
Which was in turn obtained from 𝑘𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 and taking into account the substrate effect
using the formula proposed by Sneddon:6
𝑘𝐶𝐼𝐺𝑆 = (𝑘
1
𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑
𝑘𝑆𝑛𝑒𝑑𝑑𝑜𝑛 =
2𝐸𝑀𝑜
 𝑀𝑜
2
1−
−𝑘
1
𝑆𝑛𝑒𝑑𝑑𝑜𝑛
)
−1
𝐴𝑀𝑜
√
𝜋
Where Young’s Modulus and Poisson’s Ratio for Molybdenum, 𝐸𝑀𝑜 and 𝜈𝑀𝑜 were used
instead of those for the SLG substrate because the displacement at the bottom of the pillar
is likely on the order of 10 nm,3 well within the thickness of the Mo film. Area at the
bottom of the pillars was used as the contact area between CIGS pillars and Mo layer
𝐴𝑀𝑜 . EMo was measured by nanoindentation into the as-deposited Mo film as 329 GPa,
and 𝜈𝑀𝑜 was taken to be 0.3.
1
Agilent, Agilent Nano Indenter G200 User Manual (2013).
2
J.R. Greer, W.C. Oliver, and W.D. Nix, Acta Mater. 53, 1821 (2005).
3
S.-W. Lee, S.M. Han, and W.D. Nix, Acta Mater. 57, 4404 (2009).
4
G.G. Stoney, Proc. R. Soc. Lond. A 82, 172 (1909).
5
W.C. Oliver, G.M. Pharr J. Mater. Res. 7 (1992)
6
I.N. Sneddon, Int. J. Eng. Sci. 3, 47 (1965).
(S7)
(S8)
FIG. S1. Device performance of co-evaporated (-c) and selenized (-s) CIS and CIGS
cells. X-axis shows the voltage (V) measured across the cell under illumination and yaxis shows the corresponding current density (J). Shown in red is the J-V curve for
selenized CIS cell; blue is selenized CIGS cell, and yellow is co-evaporated CIGS cell.
Note that the cells tested here are different than the films that were tested later in this
work.
FIG. S2 Schematic of (a) final product CIGS solar cell, and (b) actual sample tested in
this work, shown with thickness measurements of each layer.
FIG. S3 (a): Schematic of the bending effect caused by deposition of thin films, 𝐸𝑠 , 𝜈𝑠
and 𝑡𝑠 are Young’s modulus, Poisson’s ratio and thickness of the substrate, and 𝑡𝑓 is the
thickness of the thin film, and (b): Stress and bending effect caused by depositing
additional layer is independent of the previous layers under the thin film approximation
𝑡𝑓 ≪ 𝑡𝑠 . Subscript 1 and 2 refer sequentially deposited thin film layers.
FIG. S4 Nine sets of displacement (d, y-axis) and load (F, x-axis) data from indentation
of selenized film overlaid with curve-fitting results from Hertzian contact theory. For
each data set, up to 1% of total displacement was considered to be in the linear regime
and showed in the graph. The smeared curves are the actual data while the thin curves are
the polynomial fit according to Eqn. 2. The standard Berkovich tip used and the
individual grains on the film were approximated as spheres. R square values of above
9.87 were reached for all data sets.
TABLE SI. Atomic composition of selenized and co-evaporated films measured by EDS
Element
Selenized Co-evaporated
[AT%]
[AT%]
Cu
23.68
25.06
In
25.54
20.93
Se
50.78
47.01
Ga
0
6.69
TABLE SII. Sample deposition conditions, measured radius of curvature and residual
stress calculated using Stoney’s formula
Layer
Glass
substrate
Thickness Process
temperature
[nm]
[oC]
Radius
curvature
1× 106
228
N/A
Molybdenum 700
<150
CIS
(selenized)
Max.
(three
process)
1300
CIGS
(co1700
evaporated)
550
[m]
237.3
550
N/A
step
7.43
of
Residual Stress
[MPa]
N/A
14.7 (tensile)
N/A
1230
(Compressive)