UCLA
UCLA Previously Published Works
Title
Comparing matched polymer:Fullerene solar cells made by solution-sequential processing and
traditional blend casting: Nanoscale structure and device performance
Permalink
https://escholarship.org/uc/item/6fq3t34t
Journal
Journal of Physical Chemistry C, 118(31)
ISSN
1932-7447
Authors
Hawks, SA
Aguirre, JC
Schelhas, LT
et al.
Publication Date
2014-08-07
DOI
10.1021/jp504560r
Peer reviewed
eScholarship.org Powered by the California Digital Library
University of California
Comparing Matched Polymer:Fullerene Solar Cells Made by Solution-
Sequential Processing and Traditional Blend Casting: Nanoscale
Structure and Device Performance
Steven A. Hawks,
†
Jordan C. Aguirre,
‡
Laura T. Schelhas,
‡
Robert J. Thompson,
‡
Rachel C. Huber,
‡
Amy S. Ferreira,
‡
Guangye Zhang,
‡
Andrew A. Herzing,
∥
Sarah H. Tolbert,*
,†,‡,§
and Benjamin J. Schwartz*
,‡,§
†
Department of Materials Science and Engineering and
‡
Department of Chemistry and Biochemistry, University of California, Los
Angeles, Los Angeles, California 90095-1569, United States
§
California NanoSystems Institute, University of California, Los Angeles, Los Angeles, California 90095, United States
∥
Materials Measurement Science Division, Material Measurement Laboratory, National Institute of Standards and Technology,
Gaithersburg, Maryland 20899, United States
*
S
Supporting Information
ABSTRACT: Polymer:fullerene bulk heterojunction (BHJ) solar cell active layers
can be created by traditional blend casting (BC), where the components are mixed
together in solution before deposition, or by sequential processing (SqP), where
the pure polymer and fullerene materials are cast sequentially from different
solutions. Presently, however, the relative merits of SqP as compared to BC are
not fully understood because there has yet to be an equivalent (composition- and
thickness-matched layer) comparison between the two processing techniques. The
main reason why matched SqP and BC devices have not been compared is because
the composition of SqP active layers has not been accurately known. In this paper,
we present a novel technique for accurately measuring the polymer:fullerene film
composition in SqP active layers, which allows us to make the first comparisons
between rigorously composition- and thickness-matched BHJ organic solar cells
made by SqP and traditional BC. We discover that, in optimal photovoltaic
devices, SqP active layers have a very similar composition as their optimized BC
counterparts (≈44−50 mass % PCBM). We then present a thorough investigation of the morphological and device properties of
thickness- and composition-matched P3HT:PCBM SqP and BC active layers in order to better understand the advantages and
drawbacks of both processing approaches. For our matched devices, we find that small-area SqP cells perform better than BC
cells due to both superior film quality and enhanced optical absorption from more crystalline P3HT. The enhanced film quality
of SqP active layers also results in higher performance and significantly better reproducibility in larger-area devices, indicating that
SqP is more amenable to scaling than the traditional BC approach. X-ray diffraction, UV−vis absorption, and energy-fi ltered
transmission electron tomography collectively show that annealed SqP active layers have a finer-scale blend morphology and
more crystalline polymer and fullerene domains when compared to equivalently processed BC active layers. Charge extraction by
linearly increasing voltage (CELIV) measurements, combined with X-ray photoelectron spectroscopy, also show that the top
(nonsubstrate) interface for SqP films is slightly richer in PCBM compared to matched BC active layers. Despite these clear
differences in bulk and vertical morphology, transient photovoltage, transient photocurrent, and subgap external quantum
efficiency measurements all indicate that the interfacial electronic processes occurring at P3HT:PCBM heterojunctions are
essentially identical in matched-annealed SqP and BC active layers, suggesting that device physics are surprisingly robust with
respect to the details of the BHJ morphology.
1. INTRODUCTION
Photovoltaics based on mixtures of semiconducting polymers
and functionalized fullerenes have attracted significant interest
as low-cost solar energy harvesters.
1
Improvements in device
architecture and polymer design have yielded single-junction
power conversion efficiencies (PCEs) above 9%
2
and tandem-
cell efficiencies over 10%.
3−5
The ability to achieve high PCEs
with these materials is predicated on forming a nanoscale
polymer−fullerene network, known as a bulk heterojunction
(BHJ), that must simultaneously dissociate excitons, transport
separated mobile charges, and suppress recombination of excess
photogenerated carriers.
6,7
An extensively studied approach for
Received: May 8, 2014
Revised: June 27, 2014
Published: July 8, 2014
Article
pubs.acs.org/JPCC
© 2014 American Chemical Society 17413 dx.doi.org/10.1021/jp504560r | J. Phys. Chem. C 2014, 118, 17413−17425
creating such networks is the blend-casting (BC) method,
wherein the polymer and fullerene are dissolved together in
solution and then cast into a thin film. This approach is highly
sensitive to processing conditions and to material properties
because it relies on poorly understood spontaneous nanoscale
phase separation to create the desired donor −acceptor network
morphology.
8
Thus, even though the BC approach is simple
and amenable to extensive optimization, it introduces
irreversible interdependencies between material properties,
processing, and morphology that limit control over BHJ
network formation and thus also device performance.
9,10
Additionally, with the BC approach, it is difficult to determine
if an optimized morphology is kinetically trapped and unstable
or near a reasonable thermodynamic minimum and thus is
more suitable for long-term solar energy harvesting.
Recently, we presented an alternative to the BC method that
involves sequential deposition of the polymer and fullerene
layers from semiorthogonal solvents.
11,12
This solution
sequential-processing (SqP) route involves interdiffusing the
acceptor molecule into a precast donor underlayer. Although it
involves two processing steps for the active layer instead of one,
the SqP method is advantageous for making a BHJ compared to
the BC method because it affords more control over the
polymer:fullerene network formation while still preserving
device efficiency and the ease of solution-based fabrication.
13−16
A schematic illustrating the methodology behind these two
processing routes is presented in Figure 1. Though initially
misunderstood,
11
it is now accepted that extensive mixing of
the donor and acceptor components must occur for optimal
SqP device performance.
17−25
These interdiffusion processes in
SqP result from selective swelling of the amorphous regions of
the (donor) underlayer with the solvent used to deposit the
(acceptor) overlayer material. Additionally, thermal diffusion
from annealing can be used to intermix the two components.
The added versatility of this interdiffusion-based SqP approach
has allowed researchers to better understand the underlying
factors that give rise to functional polymer:fullerene morphol-
ogies and to use techniques that are inapplicable or detrimental
to the BC method.
26−37
For instance, the polymer layers in the
SqP approach are amenable to cross-linking,
33,34
chemical
doping,
26
nanopatterning,
36,38,39
polarization by chain align-
ment,
37
and controlled solution-based deposition using a gas-
permeable cover layer,
35
whereas these treatments are typically
harmful or nonbeneficial to BC device performance.
40
Recently,
the SqP approach also has surpassed the BC method in
comparisons of globally optimized device performance in
multiple semiconducting polymer systems.
41,42
Despite all of these advantages, there still has not been a
stringent comparison of the nanoscale networks formed via SqP
and BC to determine what differences, if any, exist between
them. The main reason for this lack of comparison is that the
polymer:fullerene film composition in the SqP processing route
is not accurately known because the components are deposited
separately instead of from a premixed solution. To the best of
our knowledge, only approximate, indirect estimates of the SqP
film composition have been made using a variety of methods,
including solid-film UV−vis absorption spectroscopy,
43
photo-
luminescence (PL) quenching,
18
neutron reflectivity,
17,19
and
time-resolved microwave conductivity (TRMC).
20
Since all of
this previous work gives only approximate compositions, the
overall morphology/processing/performance relationships are
not well-known for SqP active layers, so equivalent head-to-
head comparisons of the two approaches summarized in Figure
1 have not been carried out to date.
In this article, we present a rigorous comparison of the
nanoscale m orphological and photovolt aic properties of
composition- and thickness-matched SqP and BC bulk
heterojunction solar cells made from active layers composed
of poly(3-hexylthiophene-2,5-diyl) (P3HT) and [6,6]-phenyl
C
61
butyric acid methyl ester (PCBM). To do this, we first
developed a new method for accurately determining the overall
SqP film composition and then used this method to produce
composition- and thickness-matched BC and SqP layers. When
examining these thickness- and composition-matched BC and
SqP P3HT:PCBM solar cells, we find that the optimal SqP
active-layer composition is between 44 and 50 mass % PCBM.
This is more PCBM-rich than expected based on our previous
work
20,43
and lies in the same optimal regime as BC
P3HT:PCBM films.
44
Furthermore, structural characterization
shows that SqP films have a higher degree of both polymer and
fullerene molecular ordering than equivalent BC films and that
the SqP P3HT:PCBM films are blended on a slightly finer scale
than the matched BHJs produced by BC. Despite these
differences in molecular crystallinity a nd nanoscal e B HJ
morphology, matched SqP and BC films have remarkably
similar electronic and photovoltaic properties in small-scale
devices. However, when we compare matched films in larger
active area devices, the SqP route yields higher device
performance and significantly better reproducibility due to
enhanced film quality. Overall, even when matched as closely as
possible, SqP and BC produce different nanoscale BHJ
architectures; however, these different architectures lead to
similar device performance (in small active areas), showing that
polymer-based BHJ photovoltaics surprisingly can tolerate a fair
range of nanometer-scale BHJ structures.
2. METHOD FOR DETERMINING THE COMPOSITION
OF POLYMER:FULLERENE FILMS
Experimental details for the standard, well-established,
techniques used in this work can be found in the Supporting
Figure 1. Active-layer BHJ formation approaches for the SqP and
traditional BC methods. The SqP method creates a BHJ network by
interdiffusion of the acceptor into a host donor matrix, whereas the
traditional BC approach relies upon spontaneous nanoscale phase
separation. The questions we aim to address here are: is the final BHJ
structure from the two methods the same, or not, and what
implications does the respective processing route have for device
performance?
The Journal of Physical Chemistry C Article
dx.doi.org/10.1021/jp504560r | J. Phys. Chem. C 2014, 118, 17413−1742517414
Information. Here we detail only our original method for
obtaining the composition of SqP active layers.
SqP films were made by casting a PCBM overlayer dissolved
in dichloromethane (DCM) on top of a predeposited 130 ± 5
nm thick P3HT underlayer cast from o-dichlorobenzene
(ODCB). Since the amount of PCBM deposited onto the
P3HT film is unknown, the total mass ratio of P3HT to PCBM
in the final film is also not known. Although it may seem safe to
assume that UV−vis absorption measurements on thin- film
samples can be used to calculate solid-state film compositions,
in fact the polymer extinction coefficients can vary significantly
with the crystallinity of the polymer, and a myriad of other
effects also can affect solid-film absorbance measurements (e.g.,
scattering, interference, reflectivity, etc.). As a result, the
composition cannot be accurately determined from solid-state
optical absorption, as will be revealed in detail below.
Instead of using solid-film absorbance measurements, our
approach for determining an active layer’s stoichiometry
involves redissolving the film after casting/processing and
fitting the resulting dilute-solution absorption spectrum to a
linear combination of the individual-component spectra (Figure
2a and inset to Figure 2b). Procedurally, we first remove the
outer edges of the film with a razor, leaving only the area where
solar cells are fabricated. We scratch away the outer edge of the
films as a precautionary measure because by eye this region
looks different, and it has no relevance to the questions at hand.
Once the edges are removed, the active layer is then redissolved
in ODCB and transferred to a 1 mm thick cuvette. Even after
thermal annealing, we found that P3HT:PCBM films readily
dissolve in ODCB. We carried out the redissolving/washing
step at least 3−4 times for each film in order to fully remove all
material from the surface of the substrate. If the substrate was
insufficiently cleaned, the compositions determined for BC
films appeared anomalously rich in PCBM by roughly 10 mass
%. We suspect that these anomalous compositions arise from
insufficient cleaning are due to the higher propensity for P3HT
to remain on the substrate rather than enter solution upon
redissolving. Fortunately, this issue can be easily avoided by
simply washing away the entirety of the film. The final solutions
typically had peak optical densities in the range of 0.1−0.2 and
concentrations on the order of 0.05 mg/mL for each
component.
Figure 2a shows the solution-phase absorption spectrum of a
redissolved 1:0.8 P3HT:PBCM mass ratio BC film (black
circles) along with a
fit (blue curve) of the absorption to the
sum of the individual solution-phase components:
λλλ=+ABOD ( ) OD ( ) OD ( )
Soln PCBM PCBM P3HT P3HT
(1)
where A
PCBM
and B
P3HT
are fitting coefficients representing the
amount of each material, OD
Soln
(λ) is the measured optical
density of the composite (dissolved fi lm) solution, OD
PCBM
(λ)
is the normalized optical density of a dilute pure PCBM
solution in ODCB (re d curve ), and OD
P3HT
(λ)isthe
normalized optical density of a dilute pure P3HT solution in
ODCB (black curve). We fit the entire solution spectrum to
take advantage of the full spectral information and also to
minimize/recognize any effects of impurities or aggregation.
We found that the fits to eq 1 were excellent (Figure 2a is
typical) and unaffected by the use of different P3HT or PCBM
material batches or extensive thermal annealing. Clearly this
approach is general and can be extended to a wide range of
soluble organic molecule combinations.
Since the dilute redissolved solutions faithfully follow Beer’s
law with invariant extinction coefficients, the ratio of the fitted
coefficients A
PCBM
/B
P3HT
is equal to the PCBM/P3HT mass
ratio of the solution/redissolved film. To confirm this, Figure
2b plots the ratio of eq 1 fit coefficients for as-prepared dilute
BC solutions (red circles) as a function of their actual PCBM/
P3HT mass ratio. As expected, all points fall on a line of slope
one and intercept zero. Figure 2b also plots the ratio of eq 1 fit
coefficients obtained from a series of redissolved BC films (blue
squares) as a function of their actual mass ratio, which falls on
the same line of slope one and intercept zero, proving that a
film’s composition can be accurately determined by our
method. We note that the dissolved BC film data in Figure 2
are averages over three separate substrates with standard
deviations that are smaller than the symbol size, demonstrating
that the method is highly reproducible.
Figure 2. (a) Solution-phase absorption spectrum (blue curve) of a redissolved 1:0.8 P3HT:PCBM weight ratio BC film (obtained from the
procedure shown in the inset of panel b), along with its fit to a linear combination of the pure solution-phase P3HT (black curve) and PCBM (red
curve) components. (b) Test of this procedure on BC films and solutions with known composition. The fitted P3HT:PCBM mass ratio (eq 1) of BC
solutions (red spheres) and redissolved BC films (blue squares) as a function of their actual mass ratio; the black line is a reference with slope 1 and
intercept zero. Clearly, the solution UV−vis of a redissolved blend film can accurately recover the film’s composition. Each point is the average of
three substrates/solutions, and the error bars (one standard deviation) are smaller than the plotted symbols.
The Journal of Physical Chemistry C Article
dx.doi.org/10.1021/jp504560r | J. Phys. Chem. C 2014, 118, 17413−1742517415
3. COMPARING THE ACTIVE-LAYER COMPOSITION
AND MORPHOLOGY OF MATCHED SqP AND BC
P3HT:PCBM FILMS
3.1. Revealing the SqP Film Composition. After
establishing that our composition measurement technique
was accurate and reproducible, we applied it to P3HT:PCBM
SqP films processed over a range of conditions representative of
what is employed in the literature.
11,23,24,31
Figure 3 shows SqP
film compositions resulting from a series of active layers made
by casting a PCBM overlayer from DCM with a variety of
different concentrations and spin speeds on top of a
predepo sited 130 ± 5 nm thick P3HT underlayer (see
Supporting Information for more detailed experimental
procedures). The results in Figure 3 show that SqP films are
richer in PCBM than would be expected based upon our
previous estimates.
20,43
When we fabricated photovoltaic
devices out of these active layers, we found that the optimal
device performance was achieved when the overlayer was spun
from a 10 mg/mL PCBM solution in DCM at a spin speed of
419 rad/s (4000 rpm), which corresponded to film
compositions with mass ratios between 1:0.8 an d 1:1
P3HT:PCBM. We found that the run-to-run compositions
we obtained were highly reproducible when using solutions of
PCBM in DCM with concentrations less than 8 mg/mL but
that the commonly used 10 mg/mL concentration was too near
the PCBM solubility limit and gave run-to-run results that
varied between 1:0.8 and 1:1 P3HT:PCBM by mass for the
same spin conditions. In a particular run, though, we found that
a 10 mg/mL PCBM solution in DCM can give reproducible
compositions (Figure 3 error bars). In comparing the results of
Figure 3 to previously reported processing conditions for
P3HT:PCBM SqP devices, we conclude that optimal
P3HT:PCBM SqP active layers have a composition in the
same range as their optimal BC counterparts (1:0.8 to 1:1
P3HT:PCBM mass ratio; yellow bar in Figure 3), which is
surprising because it is not necessarily obvious that SqP should
have the same optimal composition as BC.
11,44−48
Additionally,
in the Supporting Information (Figure S4) we show that a
PCBM content of ≈1:0.8 in SqP devices is necessary for good
device performance, and that SqP active layers do not operate
well at lower PCBM contents.
Figure 3 further shows that the SqP film composition can be
tuned over a wide range from 31 mass % PCBM (1:0.45 weight
ratio) to 58% PCBM by mass (1:1.37 weight ratio) by simply
changing the processing parameters for the PCBM overlayer
(i.e., solution concentration and spin speed). This allows us to
make better sense of the wide range of processing conditions in
the literature for P3HT:PCBM SqP films.
18−21,23,24,26,45,46
Figure 3 suggests that when optimizing an SqP active layer
for device performance, the PCBM solution concentration and
deposition conditions are tuned for a given P3HT underlayer to
achieve a composition that is approximately the same as the
optimal composition for an equivalent BC film. Finally, we also
note that compositions for processing-condition combinations
not indicated in Figure 3 can be estimated from linear
extrapolation from the data in Figure 3. This analysis also can
be extended to previously published SqP morphology studies
where the overall composition was unknown.
17,19,46
3.2. Morphology Differences of Matched SqP and BC
P3HT:PCBM Films. Given that the optimal device processing
conditions lead to SqP active layers with a similar composition
as that for optimal BC films, the next important question we
ask is whether or not the two di fferent processing routes
produce the same nanoscale BHJ architecture. There have been
claims that SqP simply provides a more complex route to the
same BHJ structure as BC,
45,46
so it is important to determine if
thickness- and composition-matched films produced via SqP
and BC have the same morphology. To investigate this
question, we fabricated a series of 1:0.8 P3HT:PCBM mass
ratio active layers via SqP, determining the composition as
above, and then made corresponding BC films with matching
composition and thickness (see Supporting Information for the
detailed matching recipe). In the following analysis, when we
refer to “matched” BC and SqP films, we mean films with
identical 1:0.8 composition ratios and identical total thicknesses
of 165 nm.
We begin by examining absorption spectra of matched films
produced by the two different processing routes since the
absorptive features of P3HT directly reflect its molecular
ordering.
49
Figure 4 shows the thin-fi lm absorbance of matched
annealed P3HT:PCBM SqP and BC films prepared in the
manner described in the Supporting Information. When
compared to the equivalent BC film, the P3HT:PCBM SqP
film shows stronger absorbance in the region associated with
aggregated P3HT as well as more pronounced vibronic
structure.
49
We note that when repeating this measurement,
the absorbance of SqP films was more reproducible than the
corresponding BC films because the absorption of BC films is
highly sensitive to drying history, irrespective of thermal
annealing (see Supporting Information Figure S1).
50
Although
this sensitivity to drying is especially prevalent with P3HT-
based BC films, aggregation-dependent absorption is a feature
of many molecular materials that are of interest for SqP solar
cells.
51
Thus, Figure 4 provides the first evidence that SqP films
are less sensitive to processing kinetics than BC films, showing
that SqP can yield high-quality active layers that are less affected
by the details of the drying conditions.
Figure 3. Composition of P3HT/PCBM SqP active layers as a
function of the PCBM solution concentration in DCM and spin speed
used to create the overlayer. In all cases the P3HT underlayer was 130
± 5 nm thick. In this comparison, the optimal conditions for SqP solar
cell performance (cf. Figure 7) are 10 mg/mL, 419 rad/s (4000 rpm),
and 10 s. The optimal BC solar cell composition range is from ref 44.
The Journal of Physical Chemistry C Article
dx.doi.org/10.1021/jp504560r | J. Phys. Chem. C 2014, 118, 17413−1742517416
