reSeArCH Letter
exchange using F4TCNQ together with either Li-TFSI or EMIM-TFSI.
Surprisingly, the conductivity of the film anion-exchanged with EMIM-
TFSI was increased by a factor of 1.7 compared with doping solely with
F4TCNQ, while that of the Li-TFSI specimen was 2.4 times higher
(Fig. 3c), which suggests that the doping level was improved as a result
of the anion-exchange phenomenon when using an optimized com-
bination of X+ and Y−. It could be argued that the concentration of
F4TCNQ (F4TCNQ•−) and TFSI− affected the conductivity. We foundthat the doping efficiency (conductivity) becomes less when the supply
of reactant, here TFSI−, is lower (Extended Data Fig. 6). Indeed, at the
concentrations of F4TCNQ•− and TFSI− used here, the entropy gain is
estimated to be approximately 200 meV, which is comparably large with
respect to the ionic interactions. In considering the role of this entropy
gain in our hybrid anion system, our focus is how the ionic interac-
tions (binding preference) can have an effect on the doping efficiency
(conductivity), and to what extent the doping level can be increased by
tuning ionic compounds in anion-exchange doping. Specifically, the
most efficient anion exchange, resulting in the highest conductivity of
620 S cm−^1 , is realized when using Li-TFSI; this is one of the highest
values yet reported for doped PBTTT thin films.
We also monitored the degree of doping with photoelectron yield
spectroscopy. Figure 3d plots the photoemission yield, γ1/3, as a func-
tion of the incident photon energy. From the threshold, we estimate the
ionization potential, Ip, to be 4.83 eV for the pristine PBTTT thin film,
which is identical to the energy level at the edge of the highest occupied
molecular orbital (HOMO) band for this material, and close to the
literature value (−4.7 eV; ref.^14 ). Compared with F4TCNQ doping,
a larger shift in the threshold photon energy is clearly obtained fol-
lowing anion-exchange doping with Li-TFSI. In this case, we estimate
Ip to be 5.38 eV, exceeding the lowest unoccupied molecular orbital
(LUMO) level of F4TCNQ and suggesting that anion-exchange dop-
ing energetically stabilized the final state by a factor of Δex (Extended
Data Fig. 7), which has a value of several hundred millielectronvolts.
Thus, anion exchange is expected to be a driving force to overcome
the redox potential limitations in molecular doping. We further
verified this hypothesis through the successful anion-exchange doping
of the donor–acceptor copolymer poly[2,5-(2-octyldodecyl)-3,6-diketo-
pyrrolopyrrole-alt-5,5-(2,5-di(thien-2-yl)thieno[3, 2-b]thiophene)]
(PDPP-2T-TT-OD). We found anion-exchange doping to allow rea-
sonably high doping levels even in donor–acceptor polymers that typ-
ically have a deep HOMO level (Extended Data Fig. 8). Another control
experiment is shown in Extended Data Fig. 8, where a weak acceptor,
tetracyanoquinodimethane (TCNQ), was used as an initiator instead
of F4TCNQ. Even though the LUMO level of TCNQ does not exceed
the Ip value of PBTTT, the introduction of Li-TFSI promotes doping (as
shown by the bleaching of the neutral absorbance). Therefore, we con-
clude that anion exchange is indeed a driving force behind the doping
level. Supplementary Information section 1.6 provides a summary of
the kinetic mechanism responsible for the increased doping levels. The
overall results show that introducing a salt into a conventional binary
molecular doping system can overcome the redox potential limitationsX+
Li+ Na+EMIM+ MtBPho+MPPyrr+BtMA+
BMIM+1 2 3 4 5Y–BF 4 –5 4 3 2 1PF 6 –BOB–FAP– TFSI–PFSI–00Reff (Å)Reff (Å)BPyri+Best combinationsPositive NegativeElectrostatic potentialLi+ Na+EMIM+ MtBPho+MPPyrr+BtMA+
BMIM+1 2 3 4 5BF 4 –4 3 2 1PF 6 –BOB–FAP– TFSI–PFSI–^0000000000000BPyyyyyyyyyyyyyyyyyyri+Electrostatic potentialFig. 2 | Molecular structures and electrostatic potential maps of the
cations and anions used here. These molecular structures and calculated
electrostatic potentials are based on van der Waals surfaces of the anions
and cations. Reff denotes the effective molecular radius, equal to the radius
of a sphere with the same volume as the organic ion. Ionic radius values
were used for Li+ and Na+. The spatial distributions of electrostaticpotentials were calculated using DFT with the B3LYP functional and
6-311+G(d) basis set (Spartan’16 software). Detailed molecular structures
are shown in Extended Data Fig. 1. BOB, bis(oxalato)borate; BPyri,
1-butylpyridinium; BtMA, butyltrimethylammonium; MPPyrr, 1-methyl-
1-propylpyrrolidinium; PFSI, bis(pentafluoroethanesulfonyl)imide.0.30.20.10.0Absorbance500 1,0001,5002,0002,500
Wavelength (nm)BMIM-BF 4BMIM-PF 6BMIM-FAPBMIM-TFSI
0.30.20.10.0Absorbance500 1,0001,5002,0002,500
Wavelength (nm)Li-TFSI
EMIM-TFSI
MtBPho-TFSIF4TCNQ
dopedEMIM-TFSI Li-TFSI8006004002000Conductivity (S cm–1)1/3 γ(a.u.)4.0 4.5 5.0 5.5 6.0 6.5
Photon energy (eV)Pristine
F4TCNQ
EMIM-TFSI
Li-TFSIabcdFig. 3 | Variations in anion exchange and doping concentration with
the strength of ionic interactions. a, Optical absorption spectra of anion-
exchange-doped PBTTT thin films using various Y− anions (blue, TFSI−;
green, FAP−; orange, PF 6 −; and red, BF 4 −). b, Optical absorption spectra
of anion-exchange-doped PBTTT thin films, using various spectator X+
cations (red, Li+; blue, EMIM+; and orange, MtBPho+). c, Variations in
the conductivity of anion-exchange doped PBTTT thin films. The error
bars in the conductivity show uncertainty in the thickness of PBTTT thin
films, and represent one standard deviation. d, Photoelectron yield spectra
acquired from doped PBTTT thin films.636 | NAtUre | VOL 572 | 29 AUGUSt 2019