Ultrafast exciton decay in PbS quantum dots through simultaneous electron and hole recombination with a surface-localized ion pair

Article abstract of DOI:10.1016/j.chemphys.2015.09.012

This paper describes the ultrafast decay of the band-edge exciton in PbS quantum dots (QDs) through simultaneous recombination of the excitonic hole and electron with the surface localized ion pair formed upon adsorption of tetracyanoquinodimethane (TCNQ). Each PbS QD (R = 1.8 nm) spontaneously reduces up to 17 TCNQ molecules upon adsorption of the TCNQ molecule to a sulfur on the QD surface. The photoluminescence of the PbS QDs is quenched in the presence of the reduced TCNQ species through ultrafast (≤15-ps) non-radiative decay of the exciton; the rate constant for the decay process increases approximately linearly with the number of adsorbed, reduced TCNQ molecules. Near-infrared and mid-infrared transient absorption show that this decay occurs through simultaneous transfer of the excitonic electron and hole, and is assigned to a four-carrier, concerted charge recombination mechanism based on the observations that (i) the PL of the QDs recovers when spontaneously reduced TCNQ^1- desorbs from the QD surface upon addition of salt, and (ii) the PL of the QDs is preserved when another spontaneous oxidant, ferrocinium, which cannot participate in charge transfer in its reduced state, is substituted for TCNQ.

Full text of DOI:10.1016/j.chemphys.2015.09.012

Accepted Manuscript
Ultrafast Exciton Decay in PbS Quantum Dots through Simultaneous Electron
and Hole Recombination with a Surface-Localized Ion Pair
Kedy Edme, Stephanie Bettis Homan, Alexander B. Nepomnyashchii, Emily A.
Weiss
PII:
S0301-0104(15)00300-6
DOI:
http://dx.doi.org/10.1016/j.chemphys.2015.09.012
CHEMPH 9422
Reference:
Chemical Physics
To appear in:
Please cite this article as: K. Edme, S.B. Homan, A.B. Nepomnyashchii, E.A. Weiss, Ultrafast Exciton Decay in
PbS Quantum Dots through Simultaneous Electron and Hole Recombination with a Surface-Localized Ion Pair,
Chemical Physics (2015), doi: http://dx.doi.org/10.1016/j.chemphys.2015.09.012
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Ultrafast Exciton Decay in PbS Quantum Dots through Simultaneous Electron and
Hole Recombination with a Surface-Localized Ion Pair
Kedy Edme, Stephanie Bettis Homan, Alexander B. Nepomnyashchii, and Emily A. Weiss *
Department of Chemistry, Northwestern University, 2145 Sheridan Rd., Evanston, IL 60208-
3113
RECEIVED DATE (automatically inserted by publisher);
*Email: e-weiss@northwestern.edu
ABSTRACT
This paper describes the ultrafast decay of the band-edge exciton in PbS quantum dots (QDs)
through simultaneous recombination of the excitonic hole and electron with the surface localized
ion pair formed upon adsorption of tetracyanoquinodimethane (TCNQ). Each PbS QD (R= 1.8
nm) spontaneously reduces up to 17 TCNQ molecules upon adsorption of the TCNQ molecule to
a sulfur on the QD surface. The photoluminescence of the PbS QDs is quenched in the presence
of the reduced TCNQ species through ultrafast (≤15 ps) non-radiative decay of the exciton; the
rate constant for the decay process increases approximately linearly with the number of
adsorbed, reduced TCNQ molecules. Near-infrared and mid-infrared transient absorption show
that this decay occurs through simultaneous transfer of the excitonic electron and hole, and is
assigned to a four-carrier, concerted charge recombination mechanism based on the observations
that (i) the PL of the QDs recovers when spontaneously reduced TCNQ1- desorbs from the QD
surface upon addition of salt, and (ii) the PL of the QDs is preserved when another spontaneous
oxidant, ferrocinium, which cannot participate in charge transfer in its reduced state, is
substituted for TCNQ.
1

INTRODUCTION
This paper describes the ultrafast (≤15-ps) decay of the lowest-energy exciton in PbS
quantum dots (QDs) through simultaneous decay of the electron and hole in the presence of
surface-adsorbed tetracyanoquinodimethane (TCNQ), Scheme 1. Upon adsorption to the QD
surface, TCNQ spontaneously accepts electron density from a surface sulfur to form either a
localized ion pair (oxidized sulfur-TCNQ1-) or charge transfer (CT) complex (oxidized sulfur-
TCNQx-, 1<x<2).[1] A single PbS QD forms ground state ion pairs and CT complexes with
multiple (up to 17 for these QDs) adsorbed TCNQ molecules simultaneously. Upon
photoexcitation of the QD within these complexes, we observe non-radiative, picosecond-
timescale recombination of the exciton; the rate constant for this recombination increases
approximately linearly with an increasing number of oxidized sulfur-TCNQx- pairs per QD.
Desorption of TCNQ1- from the surface of the QD (by adding a counterion), or substitution of
TCNQ with another spontaneous oxidant whose reduced
form is not redox active with the QD, recovers the PL.
These data strongly suggest that excitonic decay occurs
via simultaneous charge recombination of the excitonic
electron and hole with the electron and hole of the
surface-localized ion pair or CT complex. This proposed
charge transfer mechanism is a Dexter-type energy
transfer,[2] the typical mechanism for triplet-triplet
Scheme 1. The structure of TCNQ.
energy transfer,[3,4] where the excitonic electron and
hole are recombining with a charge-separated state
spread over two distinct sites, rather than to a single
redox or chromophoric center. A similar concerted mechanism was observed for the ultrafast
transfer of both an electron and hole from the charge-transfer-type excited state of a molecular
chromophore, 4-aminonaphthalene-1,8-imide, to separate covalently linked redox partners.[5]
The excited state of this chromophore, which has ~70% charge transfer character, is an organic
analog of the exciton in a colloidal QD.
Identifying and eliminating the recombination centers for QD excited states is important for
the application of QDs as photosensitizers in heterogeneous or solution-phase photocatalytic
systems. QDs have size-tunable absorption spectra that span the ultraviolet, visible, and near-
2

infrared regions of the solar spectrum,[6] and highly degenerate band-edge orbitals, which allows
them to accommodate and transfer multiple redox equivalents to molecular catalysts (or directly
to a catalytic substrate) upon photoexcitation.[7,8] During a sequential multi-electron transfer
process, the QD may be photoexcited in the presence of a pre-formed ion pair, where one or both
radical ions are located on a proximate redox center, in the core of the QD, or on the surface of
the QD. Here we show that charge recombination of the excitonic carriers with these ions is
potentially competitive with subsequent electron transfer steps, even if those electron transfers
are on the picosecond timescale.
EXPERIMENTAL METHODS
Synthesis and Purification of PbS Quantum Dots. We synthesized PbS QDs (R = 1.8 nm)
using a procedure adapted from that reported by Hines and Scholes.[9] We added 2 mL of oleic
acid (OA), 18 mL of 1-octadecene (ODE), and 0.36 g of PbO in a three-neck 50-mL round-
bottom flask, and heated the mixture to 150 °C with stirring under nitrogen flow for 30 minutes
until the solution became clear and colorless. After cooling the solution to 110 °C, we injected 8
mL of a hexamethyldisilathiane (TMS)2S solution, initially prepared by adding 0.17 mL of
(TMS)2S in 8 mL of ODE, into the round-bottom flask. The PbO reaction mixture went from
colorless to dark brown ~3 seconds after injection of (TMS)2S. We allowed the QDs to grow for
10 minutes, and then submerged the flask in an ice bath until the reaction mixture reached room
temperature.
We purified the QDs by equally dividing the room-temperature PbS QDs dispersions into
two falcon tubes to which we added acetone in a 1:1 ratio by volume. The tubes were centrifuged
at 3500 rpm for 5 minutes, which produced a brown pellet and a supernatant containing excess
ODE. After removing the acetone, we redispersed the pellets in a minimal amount of hexanes,
added methanol in 2:1 excess (by volume), and centrifuged the tubes at 3500 rpm for 5 minutes.
The methanol wash also resulted in a brown pellet and a supernatant containing excess ligands
and small nanocrystalline clusters. We repeated the methanol wash a second time, centrifuged
the mixture, dried the resulting pellets over nitrogen, and redispersed them in chloroform to
make stock solutions of PbS QDs.
Bulk Electrolysis. We carried out bulk electrolysis of TCNQ in anhydrous dichloromethane.
The samples were prepared in a nitrogen-purged glovebox. We used a three-electrode
3

configuration, with a platinum mesh as the working electrode, a platinum counter electrode, and
a silver wire as the reference electrode. A porous frit was used to separate the electrodes. We
used a CH660 potentiostat (CH Instruments) to apply a constant potential of -0.3 V vs. ferrocene
for 40 minutes to a solution of 0.54 mM TCNQ with 0.1 M TBAPF6 as the supporting
electrolyte.
Fourier-transform Infrared (FTIR) Spectroscopy. We performed FTIR spectroscopy on a
Thermo-Electron Nicolet NEXUS 870 FTIR in a 0.52-mm path length KCl cell. We performed
measurements on samples in CHCl3 containing 17.8 µM PbS QDs in the presence of 0, 0.30,
0.62, 1.2, 3.7, 6.2, 8.6, 12.4, 30.9, 37.1, 43.3 and 49.4 equivalents of TCNQ.
Near-Infrared Transient Absorption (TA) Measurements. Samples for visible-to-NIR TA
were prepared at a QD concentration of 17.8 µM and placed in a 2-mm path-length cuvette with
quartz windows. The TA system is described in detail elsewhere.[10] For measurements of
picosecond-timescale dynamics, we split the output of a commercial amplified Ti-sapphire
(Soltice, 1 kHz, 350 fs) with output energy of 2.3 mJ into an optical parametric amplifier (OPA)
(TOPAS-C, Light Conversion), to produce a pump wavelength of 900 nm, and into a TA
spectrometer (Helios, Ultrafast systems) to produce a white light continuum, which serves as the
broadband probe. The pump and the probe were spatially and temporally overlapped on the
sample in a 2-mm quartz cuvette.
We measured the microsecond dynamics of the samples in a 2-mm quartz cuvette using the
same pump and a commercial EOS spectrometer, which is also described in detail elsewhere.[10]
Mid-Infrared TA. Samples for mid-IR TA were prepared at a concentration of 17.8 µM and
placed in a 390 µm path length IR cell with CaF2 windows. Our ultrafast mid-IR TA set-up splits
the output of a commercial amplified Ti:Sapphire laser (Spitfire, 1 kHz, 100 fs, Spectra Physics),
and guides 50% to an optical parametric amplifier (OPA) (TOPAS-C, Light Conversion), which
produces the pump wavelength (900 nm) for sample excitation, and 50% to a second OPA
(TOPAS, Light Conversion) to generate the probe pulse. The probe pulses at 4 µm and 5 µm are
the difference frequencies of the signal and idler outputs of the second OPA. The signal is set to
1333 nm and the idler is set to 2000 nm for the 4 µm output. The signal is set to 1379 nm and the
idler is set to 1905 nm for the 5 µm output. The differential absorption spectrum (∆A) was
obtained through active background subtraction of the ground state spectrum by chopping the
pump pulse at 500 Hz. The pump pulse is focused to 390 µm spot size, and overlapped with the
4

probe in the sample cell. The transmitted probe pulse is focused into a monochromator that
disperses it onto a multichannel mercury cadmium telluride (MCT) detector array (Infrared
Systems).
1
1H Nuclear Magnetic Resonance Spectroscopy (NMR). We performed H NMR
experiments on an Agilent DD2 600 MHz spectrometer. The samples consisted of 17.8 µM PbS
-
QDs in the absence and presence of 50 µM ferrocinium tetrafluoroborate (Fc+/BF4 ) in CDCl3.
RESULTS AND DISCUSSION
Spontaneous reduction of TCNQ by PbS QDs. Figure 1A (inset) shows the ground state
absorption spectrum of the 17.8 µM PbS QDs used in this study.[11] The spectrum includes a
first excitonic peak centered at 985 nm and a continuum that spans the entire visible region.
Analysis of TEM images of the QDs yields an average radius of 1.8 0.3 nm (see the Supporting
Information, Figure S1B,C).
Figure 1A also shows absorption spectra of reduced TCNQ produced by bulk electrolysis in
the presence of TBAPF6 (“TBA+/TCNQ1-”, red), as described in the Experimental Methods
section, and of a dispersion of PbS QDs with 2.5 added equivalents of TCNQ (black) after
subtraction of the spectrum of QDs alone. The shape of the spectrum of TCNQ in the presence of
QDs is very similar to the spectrum of electrochemically reduced TCNQ, which contains
characteristic TCNQ1- absorption features at 686, 749 and 850 nm;[1,12,13] these spectra
indicate that spontaneous electron transfer from PbS QDs to TCNQ forms a stable QD1+/TCNQ1-
complex. From the intensity of the signal corresponding to neutral TCNQ (centered at 400 nm,
not shown), we calculate that the mixture of QDs with TCNQ contains 2.4 µM of neutral TCNQ
in solution, and therefore (4.2 – 2.4 =) 1.8 µM of reduced TCNQ in the sample (there is no signal
from TCNQ2-). Using the observed absorbance of reduced TCNQ at 842 nm (0.026 a.u), we
calculate the extinction coefficient of the surface-adsorbed, reduced TCNQ to be 14,000 cm-1 M-
1, which is a factor of 3 smaller than the published value of 43,300 cm-1 M-1 in CH2Cl2.[12] For
comparison, we have scaled the spectrum of pre-electrolyzed TCNQ1- such that the two spectra
correspond to the same concentrations of reduced TCNQ; we see, again, that the signal from free
TCNQ1- in solution is a factor of 3 larger than that of the TCNQ reduced in situ.
The ground state absorption of the QD does not bleach upon addition of TCNQ (see the
Supporting Information, Figure S1A). The lack of a ground state bleach and the absence of the
5

characteristic intraband absorptions[14] of an excitonic hole in the mid-IR spectrum of QD-
TCNQ complexes (see the Supporting Information, Figure S2) are consistent with our previous
conclusion[1] that TCNQ is reduced by the surface sulfurs to which they adsorb, and not by the
QD core. In that previous study[1], we supported our assertion that the oxidized species is at the
surface of the QD (as opposed to a freely diffusing molecular species) by showing that the spin-
lattice (T1) relaxation times of protons on the oleate ligands bound to the QDs are accelerated by
the reduction of TCNQ in mixtures with the QD, due to the paramagnetism of TCNQ1-, of the
surface hole, or both. The degree to which T1 is accelerated is precisely proportional to the
amount of reduced TCNQ in the system (as measured by FTIR). Regardless of the origin of the
paramagnetism (i.e., whether it is all due to TCNQ1-, or whether some originates from the hole of
the reductant species), the paramagnetic species must be localized on the surface of the QD in
order to see this acceleration. If the hole were localized on a free molecular species that could
diffuse away from the QD, then TCNQ1- would also diffuse away because these ions would not
exist unpaired in organic solvent, and we would see no acceleration of T1.
6

Figure 1. A) Absorption spectra of reduced TCNQ produced by bulk electrolysis in the presence
of TBAPF6 (“TBA+/TCNQ1-”, red) and of a mixture of PbS QDs (R = 1.8 nm) with 2.5
equivalents of TCNQ (black) after subtraction of the QD-only spectrum, which peaks at 985 nm
(inset). B) The nitrile stretching region within FTIR spectra of TCNQ with no added QDs (red),
and of 17.8 µM PbS QDs with increasing concentrations of added TCNQ (black to blue). The
peaks correspond to C-N stretching modes for neutral TCNQ (1), for TCNQ1- (2 and 4) and for
TCNQx- where 1<x<2 (3, 5 and 6). C) Plot of the number of TCNQx- formed in situ per PbS QD
(R = 1.8 nm, [QDs] =17.8 µM) vs. the number of equivalents of added TCNQ per QD. We
determined the number of TCNQx-/QD using the combined signal intensity of peaks 2 and 3 in
panel B. Inset: The same plot as C, but plotted on a linear scale to show more clearly that the
number of reducing sites on the QD surfaces for TCNQ saturates at approximately 17 per QD.
7

Figure 1B shows the nitrile stretching region of the FTIR spectra of neutral TCNQ (red, peak
1), and of reduced TCNQ generated within samples containing a mixture of 17.8 µM PbS QDs
with varying concentrations of TCNQ. The spectrum of TCNQx- in this region, where 1≤x<2, is
known to contain a pair of peaks separated by ~25 cm-1 with energies that decrease as x increases
if the charge is spread symmetrically over the four nitrile groups.[15] In our samples, when the
molar ratio TCNQ:QD is < 3.7:1, the dominant pair of peaks in the spectra in Figure 1B are
those at 2179 cm-1 (“2”) and 2154 cm-1 (“4”). Based on the FTIR spectrum of mixtures of
decamethylferrocene (DFe) and TCNQ, where DFe spontaneously reduces TCNQ to TCNQ1-,
we determined that the peaks (“2”) and (“4”) correspond to TCNQ1- (see Supporting
Information, Figure S3). When the molar ratio TCNQ:QD ≥ 3.7:1, the pair of peaks at 2170 cm-1
(“3”) and 2145 cm-1 (“5”) become the dominant pair. The growth of peaks “3” and “5” is
correlated with the growth of a peak at 2100 cm-1 (“6”).[15] These peaks correspond to TCNQx-,
where 1<x<2. The different charge states of TCNQ correspond to different adsorption
geometries. TCNQ1- exists within an electrostatically bound ion pair of TCNQ with surface
sulfur, while TCNQx-, where 1<x<2, exists within strongly coupled CT complexes of TCNQ
with surface sulfur, where charge is shared between the donor and acceptor. We discuss these
geometries in detail elsewhere.[15] At molar ratios TCNQ:QD > 30.9:1, we observe
contributions from neutral TCNQ (peak 1) in the FTIR spectra of the mixtures, due to saturation
of available adsorption sites on the QD surfaces and the presence of a steady-state population of
freely diffusing TCNQ.
Using the intensities of peaks 2 and 3 (representative of the two charge states of TCNQ), in
these FTIR spectra, we quantify the number of spontaneously reduced TCNQ molecules per QD
for each concentration of added TCNQ. We first calculate an effective extinction coefficient for
the reduced TCNQ species by summing the areas of peaks 2 and 3 within spectra of samples
containing no neutral TCNQ – that is, samples where all added TCNQ is reduced by the QDs –
and plotting this total area vs. the concentration of added TCNQ. In these samples, we assume
that the concentration of added TCNQ equals the concentration of reduced TCNQ. This
procedure yields εeff = 2.4×104 M-1cm-2 (see Supporting Information, Figure S4). We then use
this extinction coefficient to calculate the concentration of all species of reduced TCNQ in every
sample. Figure 1C shows that the number of TCNQx- per QD increases with increasing
8

concentration of added TCNQ, with a maximum of approximately 17 TCNQx- molecules per QD
in the sample.
TCNQ quenches the photoluminescence of PbS QDs. Figure 2A shows a decrease in the
photoluminescence (PL) intensity of the QDs as the concentration of added TCNQ increases.
The absorption cross-section of the QDs at the band-edge wavelength does not change upon
addition of TCNQ, and we do not observe any defect state emission at wavelengths longer than
the band-edge wavelength, so the decrease in the PL intensity of the samples is due to an
additional non-radiative relaxation pathway for the photogenerated exciton in the QDs due to the
adsorption of TCNQ. The Supporting Information details how the PL measurements were
performed.
Figure 2B shows a plot of “PL/PL0”, the fraction of originally emissive QDs in the sample
that still undergo radiative decay after the addition of TCNQ, vs. the concentration of bound
TCNQx-. At a molar ratio TCNQx-:QD of 8.6:1, the PL of the QD sample is quenched by more
than 99%.
Surface-adsorbed TCNQx- induces an ultrafast decay of the QD excited state. In order to
determine the mechanism by which adsorbed TCNQx- quenches the PL of QDs, we studied the
decay dynamics of the excitonic state of the QDs in the presence of TCNQ with transient
absorption (TA) spectroscopy in the near-IR and mid-IR. Figure 3A shows a kinetic trace
extracted from the TA spectrum of 17.8 µM PbS QDs in the absence of TCNQ at 1000 nm
(corresponding to the ground state bleach, orange); as expected, it does not recover on the
picosecond timescale. We show in the Supporting Information, Figure S5, that the exciton
lifetime of the QDs in the absence of TCNQ is 2.3 µs. Figure 3A also shows that increasing the
number of bound TCNQx- per QD (for 3.7, 6.2, 8.6, and 12.4 bound TCNQx- per QD) induces an
ultrafast recovery of the ground state bleach. We fit the kinetic traces to a sum of exponential
functions convoluted with an instrument response function (IRF).
9

Figure 2. A) The PL spectra of 17.8 µM PbS QDs (R= 1.8 nm) with increasing concentrations of
added TCNQ in CHCl3, with a maximum ratio TCNQ:QD of 12.4:1. The negative feature at
1140 nm is due to absorption of the solvent. B) Fraction of originally emissive QDs that remain
emissive (PL/PL0) upon adsorption of various numbers of TCNQ molecules per QD.
10

Figure 3. A) Kinetic traces extracted at 1000 nm from the TA spectra of 17.8 µM PbS QDs (R =
1.8_nm) in the presence various equivalents of bound TCNQx-, in CHCl3, fit to a sum of simple
exponential functions convoluted with an IRF (green dashed lines). Table 1 contains the fitting
parameters for these traces. B) Normalized kinetic traces extracted at 4 and 5 µm from the mid-
IR TA spectra of samples of PbS QDs (R = 1.8 nm) in the absence of TCNQ (orange squares and
blue open triangles), and in the presence of 3.3 bound TCNQx- per QD (black triangles and green
circles). The 4-µm kinetic trace overlays the 5-µm trace, showing that the electron and hole are
annihilated in one concerted step. C) Plot of the measured rate constant for simultaneous
electron/hole decay in the presence of adsorbed TCNQx- (1/τ1) in ps-1, vs. the number of
11

adsorbed TCNQx- per QD, determined from FTIR spectra. The slope of the linear fit to this data
yields an “intrinsic” (per-adsorbed-TCNQx-) time constant for excitonic decay of 15 ps.
The kinetic trace for the sample with 3.7 bound TCNQx- per QD fits to a sum of three
exponential components: τ1 = 3.3 ps (74 % of the recovery of the bleach signal), τ2 = 16 ps (5%
of the recovery of the bleach signal), and τ3, which corresponds to the offset of the bleach signal
from zero amplitude at long times. We fix τ3 to 2.3 µs to account for recombination of excitons
within QDs that do not participate in the ultrafast exciton decay. The Supporting Information,
Figure S6, shows that inclusion of the smallest-amplitude component, τ2, is necessary to achieve
a symmetric scattering of residuals around the zero line, and therefore an adequate fit. Mid-IR
TA experiments, described below, which differentiate electron dynamics from hole dynamics,
suggest that the τ2 component is due to a small fraction of QDs participating in an electron
trapping mechanism (see the Supporting Information, Figure S7); we will not discuss this
component further. The value of τ1 shortens to 2.4 ps for the sample with 6.2 TCNQx-, to 1.7 ps
for 8.6 TCNQx-/QD, and to 1.3 ps for 12.4 TCNQx-/QD. Traces for the samples with 8.6 and 12.4
TCNQx-/QD do not require the τ3 component to achieve an adequate fit since the PL of these
samples is quenched by more than 99% and the bleach signal recovers to zero on the picosecond
timescale. Table 1 summarizes the values of the fitting parameters for the bleach dynamics of
each sample.
It appears from these experiments that, for QDs with adsorbed TCNQ, both excitonic carriers
(electron and hole) decay with the same ultrafast lifetime (τ1), and that this lifetime shortens as
the number of adsorbed TCNQ per QD increases. One possibility is that one of the charge
carriers is decaying on a timescale that is faster than the instrument response. In this scenario,
however, the amplitude of the bleach within the TA spectrum of a sample of PbS QDs in the
presence of TCNQ would be less than that of a sample without TCNQ at the earliest delay times
that we can monitor, because the bleach would have already partially recovered due to removal
of one the excitonic carriers from the band-edge. In the Supporting Information, Figure S8, we
show that the initial amplitudes of the ground state bleach in the TA spectra of samples
consisting of 17.8 µM PbS with 3.7 bound TCNQx- and without added TCNQ are nearly
identical at a delay time of 600 fs. We can therefore conclude that neither of the charge carriers
decays on a timescale faster than the instrument response.
12

Table 1. Fitting Parameters for the near-IR and Mid-IR TA Decay Traces for PbS QDs and
Mixtures of PbS QDs and TCNQ.
Sample
Probe
Wavelength
τ1 (A1)
τ2 (A2)
τ3 (A3)
PbS QDs
1000 nm
not present
not present
2.3 µs
PbS QDs/3.7 bound TCNQx- 1000 nm
3.3 ps (0.74) 16 ps (0.05)
3.1 ps (0.55) 8.9 ps (0.09)
3.6 ps (0.60) not present
2.4 ps (0.93) 16 ps (0.04)
1.7 ps (0.95) 16 ps (0.05)
1.3 ps (0.92) 16 ps (0.08)
2.3 µs (0.21) a
2.3 µs (0.36) a
2.3 µs (0.40) a
PbS QDs/3.3 bound TCNQx-
PbS QDs/3.3 bound TCNQx-
4 µm
5 µm
PbS QDs/6.2 bound TCNQx- 1000 nm
PbS QDs/8.6 bound TCNQx- 1000 nm
a
2.3 µs (0.03)
not present
not present
PbS
QDs/12.4
bound 1000 nm
TCNQx-
a
Time constant is fixed at the value measured for the PbS QD-only sample.
In order to confirm that the electron and hole are annihilated in one concerted step, we
monitored the photoinduced absorption signals of these carriers in the mid-IR. Lian and
coworkers have shown that kinetic traces extracted from the mid-IR TA spectrum of PbS QDs
between 3 and 4 µm have contributions from intraband transitions of both the excitonic electron
and hole, whereas traces extracted at 5 µm only have contributions from intraband transitions of
the hole.[14] We have independently confirmed this finding (see the Supporting Information,
Figures S10 and S11). Figure 3B shows kinetic traces extracted from the mid-IR TA spectra, at
4 µm and 5 µm, of a sample of PbS QDs in the absence of TCNQ (orange squares and blue open
triangles) and in the presence of 3.3 bound TCNQx- per QD (black triangles and green circles).
As expected, the intraband signal does not decay on the picosecond timescale for the sample
containing only PbS QDs at either wavelength. For the sample of PbS QDs with 3.3 bound
TCNQx-, we observe a fast decay component with a lifetime of 3.1 ps (similar to the 3.3-ps
lifetime measured at 1000 nm), and a long-lived shelf (τ = 2.3 µs) at both 4 µm and 5 µm. The
fact that the kinetic traces at 4 µm (where both electron and hole contribute to the signal) and 5
µm (where only the hole contributes to the signal) overlap perfectly further prove that the
electron and the hole decay simultaneously in the presence of TCNQx-.
The concerted, ultrafast, non-radiative annihilation of the electron and hole of the band-edge
exciton of the QD could be due to (i) Auger recombination, (ii) Forster-type energy transfer, or
13

(iii) a concerted electron-hole charge transfer (a Dexter-type energy transfer). The single-
picosecond time constants we observe for this process, and the lack of a highly coupled,
delocalized charge carrier to accept the energy from the exciton, probably rule out the Auger
process. Time constants for Auger recombination are typically 10-400 ps for biexcitons in PbS
QDs, [16-18], and 50 ps – 1.5 ns for trions in CdSe and CdSe/CdX core/shell QDs.[19-21].
Forster energy transfer is not possible in this system, as there is no overlap between the emission
spectrum of the QD and the absorption spectrum of reduced TCNQ, or the complex it makes
with sulfur on the surface of the QD (see the Supporting Information, Figure S13). The
simultaneous electron-hole transfer we propose as the quenching process is similar to Dexter
energy transfer, where instead of the electron and hole being transferred to one localized
chromophore, they are transferred to the oxidized sulfur-TCNQ1- ion pair (or charge-transfer
complex, oxidized sulfur-TCNQx-, in the case of some adsorbed TCNQs). The plot in Figure 3C
supports the mechanism of a concerted charge transfer as the PL quenching mechanism. In this
figure, we show that the observed rate constant for ultrafast exciton decay (1/τ1) is approximately
a linear function of the number of bound TCNQx- per QD, a dependence characteristic of charge
transfer in single donor-multi-acceptor systems composed of nanoparticles, [22,23] and
molecular redox centers.[24]
In order to support the concerted charge transfer mechanism for PL quenching, and to clarify
the roles of both TCNQx- and the surface hole formed upon the spontaneous reduction of TCNQ
in this mechanism, we added pre-electrolyzed TCNQ1- (with tetrabutylammonium, TBA+,
counterion) to the QDs rather than allowing the TCNQ to be reduced in situ. When we add 2.5
equivalents of TBA+/TCNQ1- to 1.7 µM PbS QDs in CHCl3, we observe that 3.4% of the
sample’s PL is quenched, whereas 40% of the sample’s PL is quenched when we add this same
number of equivalents of TCNQ to the QDs in solution, (see the Supporting Information, Figure
S14). It appears, however, that the pre-electrolyzed TCNQ1- does not quench the PL of the QDs
in this experiment because the TBA+ stabilizes the TCNQ1- in solution such that it does not
adsorb in high yield to the surfaces of the QDs. The presence of TCNQ1- in solution, rather than
on the QDs, is shown by the higher extinction of TCNQ1- in the spectrum where it is pre-
electrolyzed (Figure S14A) vs. when it is formed in situ (Figure S14C).
Furthermore, when we add 3.5 equivalents of neutral TCNQ to 17.8 µM PbS QDs in CHCl3,
all of the TCNQ is reduced in situ, Figure 4A (red), and the PL of the sample is quenched by
14

85% (similar to the 80% quenching we reported for a separate sample), Figure 4B (red). If we
then add salt (TBAPF6) to the sample, some of the TCNQ1- that was reduced on the surfaces of
the QDs desorbs, increasing its extinction coefficient, Figure 4A (green) and the PL of the QDs
partially recovers, Figure 4B (green). FTIR spectra of 17.8 µM PbS QD/TCNQ mixtures, with
12.4 TCNQx- bound per QD, show that, upon addition of TBAPF6, peaks (“2”, 2179 cm-1) and
(“4”, 2154 cm-1) from the spectrum in Figure 1B, corresponding to TCNQ1- electrostatically
bound to the QD surface, are selectively enhanced as this species of reduced TCNQ is released
into solution, Figure 4C. This result is reasonable given that TCNQx- (1<x<2), which
corresponds to peaks “3” and “5” in the spectrum, shares charge with the surface sulfur and
therefore cannot desorb as a distinct anion.
The experiments we just described show that TCNQ1- – specifically TCNQ1- adsorbed to the
surface of the QD – is at least partially responsible for the PL quenching of the QDs, but do not
reveal the role of the surface hole in the PL quenching, and do not reveal the mechanism by
which the quenching occurs. To answer these questions, we added to the QDs ferrocinium (Fc+),
which is spontaneously reduced to ferrocene upon adsorption to the surfaces of PbS QDs.
Alivisatos and co-workers have previously observed a similar reaction, in which
decamethylferrocinium is spontaneously reduced by PbSe films.[25] We chose this oxidant
because, unlike TCNQ1-, the reduced form, ferrocene, cannot participate in either electron or
hole transfer with the QDs – both processes are energetically uphill. With this system, we
therefore eliminate the participation of the reduced species in charge-transfer-type quenching of
the QDs’ PL.
15

Figure 4. A) Absorption spectra of 17.8 µM PbS QDs (black), a mixture of 17.8 µM PbS QDs
and 63 µM TCNQ (red), and a mixture of 17.8 µM PbS QDs, 63 µM TCNQ, to which we
subsequently added 0.1 M TBAPF6 (green). B) PL spectra of the same set of samples. When we
add 0.1 M TBAPF6 to the QD-TCNQ mixture, the QD PL recovers by 25%. The negative feature
in the PL spectra is due to reabsorption by a chloroform overtone. C) FTIR spectra of 17.8 µM
PbS QDs and 12.4 equivalents of bound TCNQx- before (blue) and after (black) adding 0.1 M
TBAPF6. The selective enhancement of the peaks at 2179 cm-1 and 2154 cm-1 (peaks “2” and “4”
from Figure 1B), shows that adding salt desorbs the TCNQ reduced to the 1- state, as opposed to
TCNQx-, where 1<x<2.
16

-
Figure 5A (black) shows the proton NMR spectra of 17.8 µM PbS QDs, of 2 mM Fc+/BF4
-
(green), and of a mixture of 17.8 µM PbS QDs and 50 µM Fc+/BF4 (red), in CDCl3. Fc+ is
paramagnetic, and therefore does not display any proton resonance in its NMR spectrum. When
-
50 µM Fc+/BF4 is added to PbS QDs, a peak corresponding to 48 µM ferrocene appears as a
resonance at 4.14 ppm (Figure 5A, inset). Figure 5B shows that the PL of the QDs is quenched
by only 2% upon the addition of Fc+ and its spontaneous reduction to Fc (the small amount of
quenching is probably due to residual Fc+ in solution), despite the formation of a surface hole in
this reaction. This result indicates that i) the surface hole does not quench the PL of the QDs on
its own, and ii) PL quenching does not occur when the charge transfer pathway is eliminated.
In summary, our data indicate that reduced TCNQ must be surface-adsorbed in order to
quench the PL of the QDs, and strongly suggest that this quenching is through charge transfer.
The simultaneous electron and hole relaxation that we observe in the near-infrared and mid-
infrared TA spectra of the samples indicates that both TCNQx- and the surface hole probably
participate in a double-charge transfer, Scheme 2.
It is difficult to conclusively say which excitonic carrier recombines with which member of
the oxidized sulfur-TCNQx- ion pair, because the surface sulfur state, the valence band-edge of
the QD, and the TCNQ LUMO are closely spaced in energy. We believe that we do not detect
the recombined state because, after the proposed four-carrier charge recombination occurs, the
re-reduction of TCNQ to TCNQx- (the thermodynamically favored state) occurs spontaneously
and fast enough such that we never observe neutral TCNQ in our transient experiments. The
ultrafast PL quenching and re-reduction of TCNQ to TCNQx- observed in this study suggest the
presence of strong coupling between TCNQ and the surface sulfur. We believe that the covalent-
type interactions between TCNQ and the QD observed in the mid-IR (nitrile stretching peaks
“3”, “5”, and “6” in Figure 1B) are evidence for the presence of strong coupling. Spectroscopic
signatures of this covalent-type interaction are not clearly discernable in the NIR or visible
region of the ground state or TA spectra, probably because they are obscured by the QD spectra,
but we have observed them previously for complexes of TCNQ and CdSe QDs.[15]
17

Figure 5. A) Proton NMR spectra of 17.8 µM PbS QDs (black), a mixture of 17.8 µM PbS QDs
-
-
and 50 µM Fc+/BF4 (red), and a 2mM solution of Fc+/BF4 (green), in CDCl3. The peak centered
at 5.3 ppm is the vinyl proton resonance of the native oleate ligand on the surface of the QDs.
Fc+ does not display any proton resonance in its NMR spectrum because it is a paramagnetic
species. Inset: The Fc proton resonance at 4.14 ppm only appears upon mixing Fc+ with the QDs.
B) PL spectra of the same QD samples. The “dip” observed in the PL spectra is due to
reabsorption by a CDCl3 overtone.
Scheme 2. A sulfur state on the surface of a PbS QD spontaneously reduces a neutral TCNQ in
solution to surface-adsorbed TCNQx- (where 1≤x<2), and becomes oxidized. Selective
photoexcitation of the PbS QD in the QD-Sδ+-TCNQδ- complex generates the excited state of the
QD. The excited QD decays with an intrinsic lifetime of 15 ps, probably through a double-charge
transfer process whereby the QD returns to its ground state, and the sulfur and TCNQ are
18

restored to their original (pre-adsorbed) oxidation states. This excited state decay process is
followed by the fast, spontaneous re-reduction of TCNQ by the surface sulfur.
CONCLUSIONS
We have shown using ground state absorption spectroscopy that PbS QDs (R= 1.8 nm)
spontaneously reduce up to 17 TCNQ molecules per QD in solution to form multi-plexed charge
transfer complexes. Within these complexes, some adsorbed TCNQ molecules undergo a one-
electron reduction to form TCNQ1-, and some share electron density with their chalcogenide
surface site to form TCNQx-, 1<x<2. The presence of surface-adsorbed, reduced TCNQ species
quenches the PL of the QDs. Using a combination of time-resolved near-IR and mid-IR transient
absorption spectroscopy, we determined that the PL quenching occurs through an ultrafast, non-
radiative, simultaneous decay of the excitonic electron and hole. The rate constant for this decay
increases linearly with the number of adsorbed TCNQx- per QD, and has an intrinsic time
constant of 15 ps. Removal of adsorbed TCNQ1- from the surface of the QD through addition of
a counterion (tetrabutylammonium) causes the PL to recover. Addition of a different
spontaneous oxidant, ferrocinium, to the QDs also produces a localized hole on the QD surface,
but does not quench the QDs’ PL, probably because the reduced product, ferrocene, is not redox
active with respect to the QD exciton. These data allow us to propose that the PL quenching and
ultrafast exciton decay upon adsorption of TCNQ and spontaneous reduction to TCNQx- occurs
through a four-carrier concerted charge recombination, where the excitonic electron and hole
simultaneously recombine with the oxidized sulfur-TCNQx- ion pair on the surface of the QD.
We do not directly detect the product of this recombination because, we believe, the system
spontaneously reverts to the oxidized sulfur-TCNQx- state too fast for us to measure the neutral
intermediate.
The observation of ultrafast exciton recombination in the presence of reduced TCNQ
indicates that, if one is to use a QD as a sensitizer and initial electron or hole donor within a
sequential photoinduced multi-charge transfer process (such as those necessary for many
catalytic reactions), the presence of a surface-localized pre-formed ion pair potentially serves as
a recombination site for the QD exciton, with charge recombination rates that can outcompete
subsequent charge transfer steps. It may therefore be difficult to accumulate electrons or holes on
19

the surface of a QD, in which case either charge transfer to the catalytic substrate or scavenging
of the second carrier will have to be efficient for the QD to perform multi-electron reactions.
ACKNOWLEDGEMENTS
This research was supported as part of the Argonne-Northwestern Solar Energy Research
(ANSER) Center, an Energy Frontier Research Center funded by the U.S. Department of Energy
(DOE), Office of Science, Basic Energy Sciences (BES), via Grant DE-SC0001059, and by the
National Science Foundation through a Graduate Research Fellowship to K.E. (DGE-1324585).
Any opinion, findings, and conclusions or recommendations expressed in this material are those
of the authors(s) and do not necessarily reflect the views of the National Science Foundation.
FTIR was performed in the Keck-II facility of NUANCE Center at Northwestern University. The
NUANCE Center is supported by the International Institute for Nanotechnology, MRSEC (NSF
DMR-1121262), the Keck Foundation, the State of Illinois, and Northwestern University. The
authors thank Michael Mattei for his help in performing bulk electrolysis experiments, and Dr.
Christopher Thompson for helpful discussions.
References
[1]
K.E. Knowles, M. Malicki, R. Parameswaran, L.C. Cass, E.A. Weiss, J. Am. Chem. Soc.
135 (2013) 7264.
D.L. Dexter, J. Chem. Phys. 21 (1953) 836.
[2]
[3]
N.J. Thompson, M.W.B. Wilson, D.N. Congreve, P.R. Brown, J.M. Scherer, Thomas S.
Bischof, M. Wu, N. Geva, M. Welborn, T.V. Voorhis, V. Bulović, M.G. Bawendi,
Marc A. Baldo, Nat Mater 13 (2014) 1039.
[4]
[5]
A. Khetubol, S. Van Snick, A. Hassinen, E. Fron, Y. Firdaus, L. Pandey, C.C. David, K.
Duerinckx, W. Dehaen, Z. Hens, M. Van der Auweraer, J. Appl. Phys. 113 (2013)
083507.
S.E. Miller, A.S. Lukas, E. Marsh, P. Bushard, M.R. Wasielewski, J. Am. Chem. Soc.
122 (2000) 7802.
[6]
[7]
[8]
[9]
M.G. Bawendi, M.L. Steigerwald, L.E. Brus, Annu. Rev. Phys. Chem. 41 (1990) 477.
J.B. Sambur, T. Novet, B.A. Parkinson, Science 330 (2010) 63.
B.L. Wehrenberg, P. Guyot-Sionnest, J. Am. Chem. Soc. 125 (2003) 7806.
M.A. Hines, G.D. Scholes, Adv. Mater. 15 (2003) 1844.
[10] E.A. McArthur, A.J. Morris-Cohen, K.E. Knowles, E.A. Weiss, J. Phys. Chem. B 114
(2010) 14514.
[11] I. Moreels, K. Lambert, D. Smeets, D. De Muynck, T. Nollet, J.C. Martins, F.
Vanhaecke, A. Vantomme, C. Delerue, G. Allan, Z. Hens, ACS Nano 3 (2009) 3023.
20

[12] S.E. Bell, J.S. Field, R.J. Haines, M. Moscherosch, W. Matheis, W. Kaim, Inorg. Chem.
31 (1992) 3269.
[13] L.R. Melby, R.J. Harder, W.R. Hertler, W. Mahler, R.E. Benson, W.E. Mochel, J. Am.
Chem. Soc. 84 (1962) 3374.
[14] Y. Yang, W. Rodríguez-Córdoba, T. Lian, J. Am. Chem. Soc. 133 (2011) 9246.
[15] L.C. Cass, N.K. Swenson, E.A. Weiss, J. Phys. Chem. C 118 (2014) 18263.
[16] Y. Yang, W. Rodríguez-Córdoba, T. Lian, Nano Lett. 12 (2012) 4235.
[17] J.T. Stewart, L.A. Padilha, M.M. Qazilbash, J.M. Pietryga, A.G. Midgett, J.M. Luther,
M.C. Beard, A.J. Nozik, V.I. Klimov, Nano Lett. 12 (2012) 622.
[18] F. Gesuele, M.Y. Sfeir, W.K. Koh, C.B. Murray, T.F. Heinz, C.W. Wong, Nano Lett. 12
(2012) 2658.
[19] A.W. Cohn, J.D. Rinehart, A.M. Schimpf, A.L. Weaver, D.R. Gamelin, Nano Lett. 14
(2014) 353.
[20] K. Gong, D.F. Kelley, J. Phys. Chem. C 119 (2015) 9637.
[21] H.M. Zhu, N.H. Song, T.Q. Lian, J. Am. Chem. Soc. 135 (2013) 11461.
[22] M. Tachiya, J. Chem. Phys 76 (1982) 340.
[23] A.J. Morris-Cohen, M.T. Frederick, L.C. Cass, E.A. Weiss, J. Am. Chem. Soc. 133
(2011) 10146.
[24] M.J. Ahrens, R.F. Kelley, Z.E.X. Dance, M.R. Wasielewski, PCCP 9 (2007) 1469.
[25] J.H. Engel, Y. Surendranath, A.P. Alivisatos, J. Am. Chem. Soc. 134 (2012) 13200.
21

Graphical abstract
22

We synthesize complexes of PbS quantum dots (QDs) and tetracyanoquinodimethane (TCNQ).
Each PbS QD spontaneously reduces up to 17 TCNQ molecules upon adsorption of the TCNQ
molecule to a sulfur on the QD surface.
The photoluminescence of the PbS QDs is quenched in the presence of the reduced TCNQ
species through ultrafast (≤15 ps) non-radiative, simultaneous decay of the electron and hole.
We assign this decay to a four-carrier, concerted charge recombination mechanism with the
surface localized sulfur-TCNQx- ion pair.
23