Letter pubs.acs.org/JPCL
Photocharging Artifacts in Measurements of Electron Transfer in Quantum-Dot-Sensitized Mesoporous Titania Films Nikolay S. Makarov,†,‡ Hunter McDaniel,†,‡ Nobuhiro Fuke,§ Istvan Robel,† and Victor I. Klimov*,† †
Center for Advanced Solar Photophysics, Chemistry Division, Los Alamos National Laboratory, Los Alamos, New Mexico 87545, United States § Materials & Energy Technology Laboratories, Corporate Research and Development Group, Sharp Corporation, 282-1 Hajikami, Katsuragi, Nara 639-2198, Japan S Supporting Information *
ABSTRACT: Transient absorption and time-resolved photoluminescence measurements of high-performance mesoporous TiO2 photoanodes sensitized with CuInSexS2−x quantum dots reveal the importance of hole scavenging in the characterization of photoinduced electron transfer. The apparent characteristic time of this process strongly depends on the local environment of the quantum dot/TiO2 junction due to accumulation of long-lived positive charges in the quantum dots. The presence of longlived photoexcited holes introduces artifacts due to fast positive-trion Auger decay (60 ps time constant), which can dominate electron dynamics and thus mask true electron transfer. We show that the presence of a redox electrolyte is critical to the accurate characterization of charge transfer, since it enables fast extraction of holes and helps maintain charge neutrality of the quantum dots. Although electron transfer is observed to be relatively slow (19 ns time constant), a high electron extraction efficiency (>95%) can be achieved because in well-passivated CuInSexS2−x quantum dots neutral excitons have significantly longer lifetimes of hundreds of nanoseconds. SECTION: Energy Conversion and Storage; Energy and Charge Transport
T
bleach recovery.11 In the TA experiment, excitation of the QD promotes an electron into the conduction band, which then rapidly relaxes to the lowest-energy states at the band edge. Next, the photoexcited electron can either recombine radiatively (i.e., by emission of a photon), become trapped in a midgap state (e.g., due to dangling bonds at the surface), or transfer to the mp-TiO2. Interestingly, reports on electron transfer from CuInS2 QDs to mp-TiO2 vary significantly (time constants, τeT, from ∼2 ps12 to ∼20 ns13) despite nominally similar conditions used in the measurements. In the case of the CdSe QD/TiO2 system, the size-dependent τeT has been reported in the range of 83 ps to 100 ns.11 For PbS QDs, the literature values vary between 6 fs14 and 100 ns.15 While some variations in the reported electron transfer rates can be understood in the context of differences in QD size, composition-dependent energetic driving force,11 or the strength of the QD-TiO2 coupling, it is also likely that some of the inconsistencies relate to measurement artifacts. Although multiple factors can affect electron transfer, it is important to understand the role of other potentially unexpected phenomena that might obscure the transfer process in time-resolved spectroscopic measurements. To shed light on the interplay between charge transfer from a QD to mp-TiO2,
he ever-changing landscape of global energy supply and demand makes it difficult to foresee the next breakthrough technology or the markets it may serve. Clearly, lowercost and higher-efficiency renewable energy supplies are desired, but efficient means of energy storage and usage are also needed. Heterojunctions of colloidal semiconductor nanocrystals, also known as quantum dots (QDs), and widegap metal oxides are employed in a number of energy applications from light-emitting diodes1−3 and solar cells4,5 to solar fuels6 and CO2 reduction.7 The critical function of the metal oxide/QD junction is the efficient separation of electrons and holes, which ideally should outcompete various loss mechanisms such as surface trapping and/or other deleterious recombination channels. Mesoporous titania (mp-TiO2) is perhaps the most promising and frequently investigated widegap (3.2 eV) metal oxide semiconductor because of its low fabrication cost, high stability, large surface-area-to-volume ratio, and attractive optoelectronic properties. In this report, we investigate interfacial electron transfer in photoanodes consisting of CuInSexS2−x (CISeS) QD-sensitized mp-TiO2 films, which have recently been utilized in solar cells achieving power conversion efficiencies above 5%.8,9 Transient absorption (TA) spectroscopy is an effective tool for investigating carrier dynamics at nanostructured heterojunctions.10 It has been widely used to characterize electron transfer from photoexcited QDs to TiO2 in sensitized configurations by probing dynamics of the QD band-edge © 2013 American Chemical Society
Received: October 29, 2013 Accepted: December 6, 2013 Published: December 6, 2013 111
dx.doi.org/10.1021/jz402338b | J. Phys. Chem. Lett. 2014, 5, 111−118
The Journal of Physical Chemistry Letters
Letter
Scheme 1. Two Drastically Different Time Constants Can Be Measured for Population Decay in QDs Attached to an mp-TiO2 Film Depending on Experimental Conditionsa
(a) Excitation of the QD results in electron transfer to the mp-TiO2 film with the characteristic time constant τeT. (b) If the recombination of the resulting charge-separated state is slow compared to the excitation rate kexc = jσν (j is the per-pulse excitation fluence, σ is the absorption cross section, and ν is the pump pulse repetition rate), then the hole still remains in the QD when the next photon is absorbed, which results in generation of a positively charged exciton (a positive trion). This trion decays via the Auger process with time constant τX+, which can be confused with that of true electron transfer. (c) If the hole is rapidly removed from the QD (e.g., with the help of an electrolyte), then the next absorbed photon excites a neutral exciton. In this case, the measurement yields the true electron transfer time τeT. a
next photon is absorbed (Scheme 1b). The result is a positively charged exciton (i.e., a positive trion) that has a very short lifetime (tens to hundreds of picoseconds) determined by the nonradiative Auger recombination rate.16−19 This fast trion relaxation thus dominates the time-resolved dynamics and can be confused with electron transfer. A different situation is realized if the extraction of a hole from the QD is relatively fast compared to the rate of photon absorption (Scheme 1c), and one way this can be achieved is by immersing the QD/mp-TiO2 film into a hole-scavenging electrolyte (e.g., aqueous polysulfide solution) that facilitates QD discharging (see below). In this case, the QD regains a charge-neutral state before the next photon-absorption event, and thus the measured TA dynamics will be dominated by true electron transfer. In the discussion that follows, we will refer to the first case described above (absence of the redox couple) as “Pathway I” and to the second case (having a reducing electrolyte present) as “Pathway II”, according to Scheme 1. In our studies of electron transfer at the QD/mp-TiO2 interface, we use CISeS QDs that were recently employed to demonstrate high-performance QDSSCs9 with certified power efficiencies greater than 5%. These QDs have been synthesized based on a method introduced in ref 20, the details of which are reported in refs 8 and 9. Transmission electron microscopy (TEM) measurements (Figure 1a,b) indicate that these QDs have a tetrahedral shape and are fairly monodisperse, with a typical size standard deviation of ∼15%. In the example shown in Figure 1, the height of a QD face is 5.3 ± 0.8 nm, and the average volume is ∼27 nm3. The
recombination of charge separated states, re-excitation of a charged QD, and nonradiative Auger decay of charged species, we propose a simple model that explains variations in the measured rate constants in the context of the differences in QD’s surroundings. We show that indeed, depending on the local environment, different scenarios can be observed experimentally. We study lifetimes of neutral and charged excitons in isolated CISeS solution-based QDs using TA and time-resolved photoluminescence (TR PL) spectroscopies to show that the observed relaxation time constants agree with our model. We then explore the dynamics of charge carriers at the QD/TiO2 interface upon photoexcitation and find the transfer to be surprisingly slow, thus allowing competing recombination processes (both radiative and nonradaitive) to affect the efficiency of electron transfer and ultimately the performance of QD-sensitized solar cells (QDSSCs). These results help rationalize our previous observation that the modification of QD surface properties and corresponding changes in recombination dynamics have a significant effect on the power conversion efficiency of QDSSCs.8,9 Scheme 1 shows two possible pathways leading to drastically different results in measurements of electron transfer depending on relative values of recombination and photon absorption rates. In both scenarios, photon absorption by a neutral QD results in the generation of a single exciton, which then undergoes charge separation as the electron is transferred to the mp-TiO2 film (Scheme 1a). If recombination of separated charges (electron in the mp-TiO2 film and hole in the QD) is relatively slow, then the hole still remains in the QD when the 112
dx.doi.org/10.1021/jz402338b | J. Phys. Chem. Lett. 2014, 5, 111−118
The Journal of Physical Chemistry Letters
Letter
Figure 2. Relaxation of the band-edge bleach measured at 800 nm with excitation at 400 nm for the CISeS QDs/TiO2 film for four different environments: air (brown), argon (blue), water (orange), and a polysulfide electrolyte (green); ⟨N⟩ = 0.25. Symbols show experimental data points, while lines are double-exponential fits.
Surprisingly, the TA dynamics change dramatically if the sample is immersed into a polysulfide electrolyte (1 M of Na2S and 1 M of S in a methanol−water 1:1 mixture), which is similar to that previously utilized for hole extraction and transport in high-performance QDSSCs.8,9 In this case, the fast 60 ps component vanishes, and instead the dynamics become dominated by slow relaxation (not measurable with the TA experiment, which is limited to quantifying time constants
