|Home | About | Journals | Submit | Contact Us | Français|
A potentiometric method for measuring redox potentials of colloidal semiconductor nanocrystals (NCs) is described. Fermi levels of colloidal ZnO NCs are measured in situ during photodoping, allowing correlation of NC redox potentials and reduction levels. Excellent agreement is found between electrochemical and optical redox-indicator methods. Potentiometry is also reported for colloidal CdSe NCs, which show more negative conduction-band-edge potentials than in ZnO. This difference is highlighted by spontaneous electron transfer from reduced CdSe NCs to ZnO NCs in solution, with potentiometry providing a measure of the inter-NC electron-transfer driving force. Future applications of NC potentiometry are briefly discussed.
The redox potentials of colloidal semiconductor nanocrystals (NCs) play central roles in many current and envisioned technologies. For example, electron-transfer (ET) kinetics and reaction spontaneity for NC-sensitized solar photocatalysis are governed by the redox potentials of the NC photoabsorbers.1–6 Likewise, relative potentials of band-like and surface-trapped electronic configurations dictate NC electronic doping,7 which governs the utility of NCs for electronic and optoelectronic technologies such as photovoltaics. 8,9 Although critical for many target applications, in situ measurements of colloidal NC redox potentials have proven challenging.
Cyclic voltammetry (CV) is the most commonly employed electrochemical technique for measuring colloidal NC redox potentials.10–12 Irreversibility of NC CV waves, low current-to-NC ratios, redox-active surface states, and surface-composition inhomogeneities have all been found to complicate solution-phase NC electrochemistry. CV measurements of NCs immobilized on electrode surfaces have been successful,12–15 but NC redox potentials are very sensitive to their surface chemistry,16,17 and the redox potentials of the same NCs as free-standing colloids may therefore differ substantially. As a consequence of these complications, it is common for driving forces of ET reactions involving colloidal semiconductor NCs to be discussed in terms of band-edge potentials estimated from vacuum ionization and electron-affinity measurements, often of the corresponding bulk material. Although this approach has powerful intuitive value, observations16,18 that altering surface ligation alone can shift NC band edges by as much as 1 eV highlight the need for in situ redox measurements of colloidal NCs in their native form. Here, we report a potentiometric method for measuring colloidal NC redox potentials. Potentiometry has been a valuable tool in metal nanoparticle research.19 By coupling potentiometry with optical detection of conduction-band (CB) electrons in colloidal semiconductor NCs generated via photodoping,20,21 redox potentials associated with these electrons can be deduced. As a simple proof of concept, we show that our colloidal CdSe NCs have CB-edge potentials more negative than our ZnO NCs, leading to spontaneous inter-NC ET from photoreduced CdSe NCs to ZnO NCs in solution. Additional mechanistic details are revealed by the transient open-circuit potentials.
Figure 1 illustrates the apparatus used to measure Fermi levels (EF) during NC photodoping. In an airtight optical cuvette, solutions of NCs under N2 atmosphere are photoreduced using hole quenchers.7 The average number of CB electrons per NC (n) is quantified during photodoping using absorption spectroscopy. Simultaneously, electrodes in the NC solution track changes in EF under galvanostatic (I = 0 A) control, i.e., the potentiostat biases the working electrode in response to the photoinduced increase in EF (Figure 1, right). The electrode and solution EF remain equivalent at all times. Consequently, no depletion region at the electrode/electrolyte interface develops, and the recorded half-cell potential represents EF of the NC suspension.19 From these combined data, NC redox potentials at various electron densities can be determined. Accurate transient potentiometry requires a stable reference electrode. We use a leakless Ag/AgCl reference electrode, which avoids instabilities due to solution contamination, ionic activity, or electrode/electrolyte junction potentials (see Supporting Information (SI)). To account for possible electrochemical drift, CVs of an internal standard (cobaltocenium hexafluorophosphate, [Cp2Co][PF6]) were collected before and after most experiments. Drift was generally very small (<~10 mV). All data are referenced experimentally to the ferrocenium/ferrocene couple (Fc+/Fc, see SI).
Figure 2A plots EF and the absorbance at λ = 1000 nm (A1000) measured simultaneously during ZnO NC photodoping using ethanol as the hole quencher.22,23 A1000 increases with ZnO photoexcitation, reflecting photodoping.21,24–26 Concomitantly, EF becomes more negative. From the per-electron extinction at λ = 1000 nm (ε1000 = 1097n0.7 M−1 cm−1, see SI), n ≈ 20 e− CB/NC at its maximum (nmax), corresponding to an average electron density of Nmax ≈ 1.21 × 1020 cm−3, in agreement with previous reports.7,21,23,26 Because EF and A1000 were measured simultaneously, it is valuable to plot EF against n as shown in Figure 2B. EF rises steeply at ~−70 mV/n between n = 0 and 2, after which its rise decreases to ~ −4 mV/n until photodoping is complete.
It is instructive to compare these potentiometric data with those obtained using a solvated optical redox indicator (ORI),1–3 an approach we applied recently to monitor ZnO NC photodoping.27 Here, EF is measured during photodoping using the optically detected equilibrium constant of a solvated redox couple that is also in equilibrium with the NCs. For the present comparison, ORI data were collected while photodoping the same ZnO NCs as probed electrochemically, under the same experimental conditions, and the ratio [Cp2Co+]/[Cp2Co] measured spectrophotometrically to determine EF. These results are included in Figure 2B. The two methods yield nearly indistinguishable results.
Despite yielding the same results, potentiometry offers an important advantage over the ORI method: Potentiometry circumvents the need for a transparent spectroscopic window in which to monitor the ORI (e.g., for Cp2Co, λprobe ≈ 500 nm). Because of this advantage, the redox potentials of narrower-gap NCs can be readily monitored potentiometrically, making this the more general approach. As proof of concept, potentiometry and absorption were measured simultaneously during photodoping of CdSe NCs (with absorption overlapping that of Cp2Co). Figure 3A plots electronic absorption spectra of undoped and maximally photodoped d = 4.1 nm CdSe NCs, photoexcited at 405 nm in the presence of Na[Et3BH] (hole quencher),20,28 [Bu4N][PF6] (electrolyte), and TOPO (NC stabilizer). Photodoping causes the first NC excitonic transition to bleach to A/A0 ≈ 0.3 (A0 = absorbance before photodoping) and redshift slightly, consistent with prior results.20 From the established linear relationship between n and A/A0,20,29 these data imply nmax = 1.4, again consistent with previous results.20,30 Note that the CdSe NC photodoping experiment is considerably quicker than the ZnO NC photodoping experiment (Figure 2) because of ~5 times greater photoexcitation rates, greater conversion yields using [Et3BH]− hole quenchers,21 and smaller nmax in the CdSe NCs.
Figure 3B plots EF and A/A0 data collected transiently during CdSe NC photodoping for two experiments: one performed with Cp2Co+ as an electron shuttle and internal redox standard and the other without Cp2Co+. Prior to irradiation, EF and A/A0 are both constant, but EF is ~100 mV more positive in the sample containing Cp2Co+. This difference reflects a small amount of Cp2Co+ reduction prior to deliberate CdSe irradiation.31 Upon irradiation of the sample without Cp2Co+, EF immediately shifts more negative, reaching a value near −1.52 V vs Fc+/Fc after 4 min. Similarly, A/A0 decreases immediately, reaching a new value of ~0.3. Upon irradiation of the sample with Cp2Co+, EF again immediately shifts more negative, reaching a similar value near −1.52 V vs Fc+/Fc after 4 min. Interestingly, the onset of CdSe photodoping (as indicated by the inflection in A/A0) is clearly delayed by ~40 s in the presence of Cp2Co+, even though EF starts shifting more negative immediately upon photoexcitation. This delayed photodoping reflects EF equilibration between the CdSe NCs and Cp2Co+/Cp2Co redox couples, which initially strongly favors Cp2Co+ reduction. Reduction of Cp2Co+ by photodoped CdSe NCs continues until the CdSe CB-edge potential is reached, at which point both Cp2Co+ reduction and CdSe NC electron accumulation proceed simultaneously with further photoexcitation. This observation is an example of the new insights that can be gained from potentiometry in the time domain.
Figure 3C plots EF vs n for both experiments of Figure 3B. Although EF is very different for the two samples prior to photodoping, the onset of CdSe NC reduction occurs at ~−1.47 V (±0.01 V) in both experiments. Once CB electrons begin to accumulate, the change in EF between n = 0 and n = 1 is small, with a slope of ~−10 mV/n. The slope of EF vs n increases as nmax is approached, and photodoping maximizes at nmax ~ 1.4 and ~−1.52 V (±0.01 V) for both experiments. Plotted in this manner, the electrochemical data from these two experiments, which initially appeared markedly different (Figure 3B), are now essentially super-imposable. From this result we conclude that the CdSe CB-edge potential is independent of the presence of Cp2Co+ under these conditions.
Comparing EF data (Figures 2B and and3C),3C), we note that the CdSe NCs at n = 1 are ~260 mV more reducing than the ZnO NCs at n = 1 (−1.48 vs −1.22 eV, respectively). This difference is notably smaller than would be estimated from bulk data (~1.1 eV, see SI), but it still indicates a driving force for inter-NC ET. To illustrate, a mixture of similar CdSe and ZnO NCs was prepared containing Li[Et3BH] as the hole quencher,28 with all conditions similar to those of Figures 2 and and3.3. Figure 4 shows absorption spectra of this solution collected after selective CdSe photoexcitation for various durations. The broad NIR (<2 eV) absorption characteristic of n-ZnO (Figure 2A) grows with photoexcitation time. The CdSe excitonic absorption maximum redshifts by ~20 meV over the same time window, but there is no evident bleach, allowing tentative attribution of this shift to a Stark effect associated with surface charge redistribution. Control experiments performed in the absence of CdSe NCs (see SI) show no spectroscopic changes, ruling out direct ZnO photodoping under these conditions. The absence of CdSe excitonic absorption bleach and growth of ZnO NIR absorbance indicate that photodoped CdSe NCs indeed transfer their electrons to ZnO NCs under these conditions, as anticipated from the favorable ET driving force measured electrochemically.
During the course of these experiments, several interesting complexities were noted. First, as anticipated from prior observations,16,18 CdSe NC redox potentials are found to be extraordinarily sensitive to sample preparation and measurement conditions, varying reproducibly by hundreds of mV depending on the specific details. Consequently, the redox potentials reported here reflect the particular reaction conditions employed, just as standard reduction potentials (E°) of molecular reagents correspond to a standard set of conditions. These observations will be described in detail in a subsequent report, but this preliminary observation already highlights the utility of this technique for identifying sample-specific redox potentials through in situ measurements. Additionally, we found it possible to measure the potentials of sub-CB electron traps in CdSe NCs by combining potentiometry with photoluminescence spectroscopy (PL, see SI). Here, we observe PL brightening as EF is raised, starting at least 120 mV below the CB edge, before the characteristic darkening that coincides with CB filling and the resulting Auger recombination.20 NC PL brightening at sub-CB potentials is consistent with several recent observations32–35 and indicates reductive passivation of surface electron traps. We note that in some cases CdSe NC surface-trap reduction appears to have exactly the opposite effect of quenching PL,18,20,36 reflecting the complexity of these surface chemistries and highlighting the need for in situ electrochemical measurements.
Overall, the results presented here demonstrate potentiometry as a powerful and broadly applicable approach to semiconductor NC electrochemistry. With this approach, it is possible to quantify band-edge potentials in situ, without special apparatus or modification of NC surface chemistries. The impact of NC composition (isovalent or aliovalent impurities, etc.),37–39 charge-compensating cations (H+, Li+, [CoCp2]+, etc.),1,2,7,40 or NC surface ligands (with dipoles, conjugation, etc.)16,18,41 should be readily quantified, and extension to other redox-active NC heterostructures3,23 or nonphotochemical reductants appears equally promising. The transient potentiometry described by Figures 2 and and33 further suggests interesting possibilities for probing dynamical processes. NC potentiometry thus opens new opportunities for future fundamental and applied research involving redox-active colloidal semiconductor NCs.
This research was supported by the NSF (CHE-1506014 to DRG, Graduate Research Fellowship DGE-1256082 to KHH), NIH (Postdoctoral Fellowship F32GM110876 to EYT), and the State of Washington through the Clean Energy Institute via funding from the Washington Research Foundation (to CKB).