Full Paper
rative support that the orientation of parallel stacks in double-
channel photosystems is irrelevant.
Compared with the dipolar photosystems in-1 and out-1,
mixed-1 from the API series generated similar photocurrents
with similar charge separation efficiency (IPCE, i.e., similar
action spectrum, Figure 4c), clearly less charge recombination
loss (h , Figure 5b), and clearly higher activation energy (Fig-
BR
ure 5c). In the ANI series, mixed-1N generated significantly
less photocurrent than the dipolar systems in-1N and out-1N
(
Figure 5d). Moreover, mixed-1N generated photocurrent with
more charge recombination loss (Figure 5e) but clearly lower
activation energy (Figure 5 f) than the dipolar in-1N and out-
1
N. These complementary trends are best understood by con-
sidering that the mixed ANI photosystems mixed-1N were pre-
pared under thermodynamic control, whereas the mixed API
photosystems mixed-1 had to be prepared under kinetic con-
trol. The only weakly reduced photocurrent generation of
mixed-1, compared with dipolar in-1 and out-1, with increased
E and reduced h , was thus consistent with less organized ar-
Figure 5. Characteristics of API (a–c) and ANI (d–f) photosystems out-1(N)
o), mixed-1(N) (m), and in-1(N) (i). a,d) Photocurrent generation upon irradi-
a
BR
(
chitectures in more randomly mixed stacks of APIs. The more
significantly reduced activity of mixed-1N with low Ea and
high hBR was consistent with tight stacking interactions be-
tween better self-sorted antiparallel stacks of ANIs.
ation with solar simulator. b,e) Bimolecular charge-recombination efficiencies
upon irradiation with white light (unfilled bars) or above 420 nm (gray bars).
c,f) Activation energies estimated from the temperature dependence of pho-
tocurrent generation. Results are of the SOSIP photosystems with similar ab-
sorbance (a,d) or the average Æerror of data obtained with ꢂ2 independ-
ently prepared photosystems (b,c,e,f). Note that plots are scaled to highlight
overall rather small differences.
Taken together, these findings suggest that the activity of
mixed photosystems depends significantly on their method of
preparation. Moreover, they support that the activity of
double-channel photosystems with well-equilibrated and self-
sorted antiparallel push–pull stacks is significantly lower than
that with parallel stacks.
were determined from the dependence of photocurrents on ir-
[
25]
radiation power. Although the found hBR values were the
same (42%) for both systems upon excitation of APIs at
ꢂ420 nm, full white light irradiation gave rise to a slightly
Dipolar triple-channel photosystems
lower h value of 20% for in-1 compared with 22% for out-
BR
1
(Figure 5b). Activation energies E were determined from the
In double-channel photosystems, parallel stacks of push–pull
components were identified as more powerful than antiparallel
stacks, whereas the orientation of these parallel stacks turned
out to be almost irrelevant. To elaborate on the orientation of
dipolar stacks in triple-channel architectures, photosystems in-
23 and out-23 were designed (Figure 6). Reminiscent of
a
[
15,26]
temperature dependence of photocurrent generation.
A
lower E was found with in-1 than with out-1 (Figure 5c).
a
Taken together, the slightly higher photocurrent generation
of in-1 coincided with slightly lower TSE yield, blueshifted ab-
sorption maxima (Figure 4a), reduced charge recombination,
and shallower charge traps. These overall small changes in
favor of in-1 could indeed originate from the inward oriented
dipolar fields but also from higher charge mobility in better p
stacks, or from other effects. Most importantly, the found dif-
ferences between double-channel photosystems with inward
[
4d,13b,27]
donor–bridge–acceptor type triads,
triple-channel pho-
tosystems could undergo photoinduced charge separation
with electrons and holes in two differently substituted NDIs.
The parallel push–pull stacks placed in between have their di-
polar fields oriented toward the central n channel in in-23 and
toward the peripheral p channel in out-23.
(
in-1) and outward (out-1) oriented parallel API stacks were
very small. Orientation independence has also been reported
Dipolar dyads 24 and 25 were prepared by amide bond for-
mation between ANIs and NDIs with amine donors in the core
(NN-NDIs). Interestingly, the absorption maxima of NN-NDIs in
dyads were slightly (2 nm) shifted to the blue (25) and red (24)
compared with the parent NN-NDI (Figure S5 in the Supporting
Information). An appealing explanation for this finding is the
stabilization and destabilization of the LUMO of NN-NDIs by
the nearby positive and negative ends of the ANI dipoles, re-
spectively. Similar Stark shifts have been observed in other di-
[3g]
previously for APIs in DSSCs.
Compared with API photosystems, ANI photosystems gener-
ated overall more photocurrent (Figure 5a, d). More efficient
electron transfer from ANI to NDI was evident from the compa-
rable action and absorption spectra (Figure 4d, b). These re-
sults could be explained by a larger LUMO energy difference
between ANI and NDI compared with API and NDI (DELUMO
ꢀ
0.77 versus 0.34 eV, Figure 3b). The difference in activity be-
tween ANI photosystems with inward (in-1N) and outward
out-1N) oriented parallel stacks was overall even smaller than
[
5b]
polar systems.
(
These small shifts of the NN-NDI absorption maxima were
conserved in triple-channel photosystems in-23 and out-23
(Figure 7a). They were obtained from SOSIP architecture 22 by
with API photosystems (Figures 4d, 5d–f). These consistent
trends with different push–pull components provided corrobo-
Chem. Eur. J. 2016, 22, 9006 – 9014
9011
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim