Photo-Excitation Dynamics in Electrochemically Charged CdSe Quantum Dots: from Hot Carrier Cooling to Auger Recombination of Negative Trions

Fulfilling the potential of the colloidal semiconductor quantum dots (QDs) in electrically driven applications remains a challenge largely since operation of such devices involves charged QDs with drastically different photo-physical properties compared to their well-studied neutral counterparts. In this work, the full picture of excited state dynamics in charged CdSe QDs at various time-scales has been revealed via transient absorption spectroscopy combined with electrochemistry as direct manipulation tool to control the negative charging of CdSe QDs. In trions, excited states of single charged QDs, the additional electron in the conduction band speeds up the hot electron cooling by enhanced electron-electron scattering followed by charge redistribution and polaron formation in picoseconds timescale. The trions are finally decayed by Auger process in 500 ps timescale. Double charging in QDs, on the other hand, decelerates the polaron formation process while accelerates the following Auger decay. Our work demonstrates the potential of photo-electrochemistry as a platform for ultrafast spectroscopy of charged species and paves a way for further studies to develop comprehensive knowledge of the photophysical processes in charged QDs more than the well-known Auger decay preparing their use in future optoelectronic applications.


1.INTRODUCTION
Quantum dots (QDs) 1 as semiconductor nanocrystals with size smaller than Bohr radius, have been widely investigated both as model systems for fundamental research and for numerous applications [2][3][4][5][6] .The applications typically rely on separation and extraction of photo-generated electron-hole pairs for the efficient photon to free charge carrier conversion 7 .This can be achieved through transfer of photo-generated charges to electron (hole) acceptors which highly depends on the relative position of the band edges 8 .Charge dynamics in such systems have been extensively investigated by time-resolved spectroscopies [7][8][9] .Most of these studies are implemented on halfcell systems (i.e.only photoanodes or photocathodes) or open-circuit conditions where electron/hole extraction through external circuits does not occur.In full functioning devices, extraction of charges to/from the electrodes is essential for device operation, which means that under working conditions, the photoactive layer can accumulate charges due to defects or Schottky barrier formation at the interfaces 10 .Charge accumulation at the interfaces and also in the QDs considerably influences the device performance and optoelectronic behavior [11][12][13][14] .Therefore, understanding of the effect of an external bias and the presence of extra charges in the QD-acceptor system, is of great importance for the development of high-performance QD devices.
By absorption of light in a semiconductor an electron-hole pair, exciton, is created.Two excitons in a QD interact and form a biexciton, which rapidly decays through Auger process leaving behind a single exciton 15 .Auger recombination in QDs is efficient 16 and drastically influences performance of the corresponding devices at high excitation intensities.Photoexcitation of a charged QD leads to a so called trion which is a three body state of electron-hole pair and an additional charge.Depending on the charge, a trion can be either positive or negative.Analogously to the biexciton, the trion also can decay through Auger recombination 15 .The depopulation of trions through the Auger process competes with the extraction of charges and consequently undermines the charge separation efficiency in the devices.Most of the existing studies on trion dynamics utilized time-resolved photoluminescence spectroscopy to measure radiative trion decay in core/shell heterojunction structures 15,17 where the shell structure has direct effect on the trion recombination 18 .Such core-shell structure isolates the QD core from surface defects, at the same time the shell also slows down or even prevents charge carrier transport to and from the core making integrated optoelectronics based on QDs difficult 19 .
In this work we combine electrochemistry with ultrafast transient absorption spectroscopy (TA) to monitor changes in the excited state dynamics of the QD under controlled charging 12,[20][21][22][23] .By observing the changes in the steady state absorption during spectro-electrochemistry measurement, we have identified distinct potentials which correspond to the injection of one and two electrons to the QDs.The presence of these negative charges leads to negative trions and tetrons upon excitation by laser light in the TA measurement 15 .We observed a series of changes in the relaxation processes due to the extra electrons including nonradiative Auger recombination in 500 ps timescale.We anticipate that our findings will pave the way to comprehensive knowledge and better understanding of the photo-physical processes in charged QDs leading to their efficient use in future nanotechnology applications.

ELECTROCHEMISTRY
In electrochemistry the applied potential induces electron exchange between working electrodes and the sample under study 24 .By changing the potential of the working electrode with time (versus a reference electrode) while measuring the current that passes to the counter electrode, a cyclic voltammogram (CV) is obtained 25 , showing distinct bands at the potentials where charge carriers can enter the electronic states of the system 14,24,26 .Monodispersed CdSe QDs with a diameter of ~3 nm were used for sensitization of the TiO2 coated FTO (SI).Such TiO2-FTO system is analogous to the photoanode architecture of QD-sensitized solar cells 7 .CVs of CdSe QDs were measured in a conventional three electrode electrochemical cell configuration, where the QD-TiO2-FTO assembly serves as the working electrode, a leakless pseudo Ag│AgCl electrode is used as reference and a platinum wire acts as a counter electrode.Detail description is given in Supporting Information (SI).All the potentials in this work are reported vs. the Ag│AgCl pseudo reference.In equilibrium the Fermi levels of the semiconductors that are in contact with each other are considered to be equal, which will equilibrate the charges through QD-TiO2-FTO layers 27 .In that respect the TiO2-FTO layer can be considered to be as one system with a common Fermi level while CdSe QDs are separate semiconductors attached to the TiO2-FTO by a linker molecule.The reference electrode is considered to have a fixed potential, thereby applying a negative bias corresponds to an increase in the TiO2-FTO Fermi level.All potentials in this paper is reported versus Ag|AgCl pseudo reference (SI).At sufficiently high negative potentials, electrons in the conduction band (CB) of the TiO2-FTO can reach energies equal to or higher than of the CdSe CB, hence electrons can be injected into the QDs.When electrons are exchanged between QDs and TiO2-FTO, a current change can be detected in the CV which is represented in Figure 1A.For a better understanding of electrochemical measurement and changes in the current, it is essential to define the relation between energy levels of the QDs and the applied potentials.By identifying the electrochemical potential for the CB minimum (first excited state) of the QDs U1Se (in Volts), the electrochemical potential for the valence band energy levels Uh  ℎ can be expressed as 13 Uh= U1Se + ΔEopt/e + 1.8e/(4π εr ∞ ε0 r) , where ΔEopt is the energy of the optical transition from corresponding energy level in the valence band to 1Se, e is the elementary charge, εr ∞ the medium's relative permittivity at high frequencies, ε0 the vacuum permittivity, and r is the radius of the QDs.The approximate electrochemical potentials for the energy levels obtained from the CV and the discrete energy levels of 3 nm CdSe QDs with their corresponding optical transitions are depicted in Figure 1B and C 13,14,26 .Measurement was performed in a three-electrode cell where the CdSe QDs assembly on a TiO2 coated FTO electrode served as the working electrode, Ag/AgCl as the reference and a platinum plate was used as counter electrode.0.1 M solution of tetra butyl ammonium hexafluorophosphate in dichloromethane was used as the supporting electrolyte.A scan rate of 1 mV/s was used at room temperature.The inset represents a close-up of the current peaks at -1.5 V, 1.75 V and -1.9 V. B) calculated electrochemical potentials for 3 nm CdSe QDs energy levels by assuming conduction band (first excited state) potential as -1.75 V. C) discrete energy levels of 3 nm CdSe QDs 13,28 .Up and down arrows connecting optically allowed transitions for absorption and emission.In the bottom measured absorption and emission of the 3 nm CdSe QDs are also represented.D) energy levels of the conduction band and valence band of the TiO2 and the FTO.
In CV, Figure 1A, with increasing negative bias we see a steady raise of the current due to Faradaic processes 24 .At -1.5 V, a clear peak emerges since the Fermi level reaches the conduction band of TiO2 and, consequently, additional electrons can be injected into the CB of TiO2.At -1.75 V another peak indicates the electrochemical potential where electrons can be transferred from TiO2 to the 1Se level in the QDs.The next peak at -1.9 V is at too low potential to correspond to the next energy level, 1Pe, as illustrated in Figure 1B and C 29 .Accordingly, we interpret the peak as further charging of the QDs since the 1Se level can simultaneously accommodate two electrons.Due to the Coulomb repulsion, it is expected that addition of the second electron requires more energy which corresponds to a higher negative potential in electrochemistry.Since the features between -1.5 V and -1.75 V are a combination of the QDs and that of the TiO2-FTO, exact description of them solely based on CV measurement is not straightforward.In addition, as one can see from CV measurement in Figure 1A and as reported previously in literature, the electrochemical measurements on QDs have irreversible character 14,30,31 .This indicates that the charges injected into the QDs do not fully return on the time scale of the electrochemical cycle when the potential is reversed 30 .Previous studies have shown that charging can induce surface modifications which leads to the formation of traps in QDs without inducing significant changes in absorption and photoluminescence 22 .It has been also shown that it is possible to charge QDs without reducing the surface by using passivating ligand or by building core-shell structures to eliminate such surface species 21,22,32 .For precise scrutiny of the effect of external bias on the QD assembly, as the main goal of this study, combination of electrochemistry with spectroscopy 33 is used.Changes of the absorption spectra due to the state filling can further identify the energy levels that are involved in charging 13,34 .

SPECTROELECTROCHEMISTRY
The spectroelectrochemistry 35   The absorption spectra show two clear bands at around 550 nm for 1S3/2-1Se and 2S3/2-1Se transitions and at 450 nm for the 1S1/2-1Se transition.At all negative potentials, an increasing featureless background from 600 nm to 1000 nm (Figure S5) is due to absorption of the charges in TiO2 36 .Until the potential reaches -1.5 V, the Fermi level in the TiO2-FTO layer is lower than the QD conduction band, hence the induced absorption changes at 535 nm and 565 nm are due to the Stark shift 37 .At -1.6 V and higher negative bias, strong bleach of the transitions at 550 nm and 450 nm show that the electrons fill the QD 1Se state and consequently the absorption of all transitions to this level become partially bleached.At -1.8 V, the bleach amplitude becomes two times larger.These observations provide strong evidence that at -1.6 V the QDs are charged by one, and at -1.8 V by two electrons which fills the doubly degenerate 1Se state.These results demonstrate that spectroelectrochemistry provides a convenient direct manipulation tool for a controlled charge-injection to the QDs.In the following we will be using this method to prepare negatively charged QDs for the transient absorption measurements.We also point out that we are well aware of the limitations of the effective mass theory description and in reality the colloidal QDs have much more energy levels shown by atomistic calculations 38 .Such a quasi-continuum view is important for understanding the excited electron relaxations in QDs in the following.
However, the main spectral features of these two approaches are in good agreement justifying our interpretation of electrochemistry in terms of k•p effective mass theory and state filling.It is worth mentioning that even at -1.8 V the band edge transition is not fully bleached.This may be due to a subset of QDs with incomplete charging.It may also be the result of the size distribution of the QDs as the smaller dots are less likely to become fully charged.This means that in reality at -1.8V there is a mixture of states which is making precise distinction of the processes in the following analyses complicated and the proposed models need to be taken as simplification of the full complexity of the real situation.

TRANSIENT ABSORPTION SPECTROSCOPY
Combination of spectroelectrochemistry with time resolved femtosecond transient absorption spectroscopy (TA) allows in situ probing of the excited state dynamics in well-controlled charged QDs.A detailed description of the TA measurements is given in SI.Excitation at 400 nm was used to populate higher energy levels (2P1/2-1Pe transition) so that the absorption changes under different electrochemical potentials do not affect the excitation conditions.The initial hot electron relaxation provides valuable information about the details of the carrier cooling process.Decay associated spectra (DAS) were obtained from global fitting of the TA data and used for mapping of the population and depopulation pathways of the excited states.The DAS spectra without normalization are presented in the SI.The TA measurements at OCP (-0.25V vs. Ag/AgCl) are taken as reference.In the following the TA features are discussed in consecutive order.

OCP AND NEGATIVE BIAS
Relaxation of the initially excited hot electrons and holes occurs in sub-ps timescale 39 .All DAS components of TA consist of a negative peak at 550nm corresponding to a decay of the bleach signal related to the almost instantaneous state filling of 1S3/2-1Se and 2S3/2-1Se transitions.Since at OCP, after the laser excitation only one electron and hole are present in QDs, electron-electron scattering cannot take place hence the relaxation is induced only by electron-phonon scattering.
Since the phonon frequencies are significantly smaller than the level spacing in the effective mass description, the hot carrier cooling corresponds to relaxation through a quasi-continuum of the energy levels 38 .The small positive band at 600 nm in the 500 fs DAS is excitation-induced spectral shift and also reflects the hot carrier arrival to the band edge 40,41 .It follows with two slower processes corresponding to the initial electron injection (9 ps) to a charge transfer state where the electron still interacts with the hole and the full charge separation via electron diffusion to the bulk of the TiO2 (95 ps) 42 .The broad positive signal at longer wavelengths is attributed to the excited state absorption (ESA) of free electrons in the CB of TiO2 36 .Eventually, the photo-generated charges are being recovered within a timescale that is longer than our experimental limit noted as >10 ns in all cartoon illustrations of the decay processes in Figure 3.

SINGLE AND DOUBLE CHARGED QDs
When an external voltage is being applied to the photoanode, at potentials up to -1.5 V no electrons are electrochemically injected into the QDs, but the excess electrons in the TiO2-FTO do affect the excited state dynamics in the QDs as well as charge transfer to TiO2 due to long range Coulomb interaction in QDs 43,44 .At -1.0 V the Fermi level is still below the CB of the TiO2 and the electrons can only enter the TiO2 shallow traps 45 .The timescale of the hot electron relaxation is not influenced by this, while the photo-excited electron injection from the QDs to the TiO2 and the following diffusion of the injected electron away from the QD vicinity into the bulk of TiO2 are both significantly slowed down from 9ps to 80 ps and from 95ps to 900 ps, respectively.At -1.5 V, when electrons fill the lowest levels of the TiO2 conduction band, the charge separation process is further slowed down up to ns.The fast component that was earlier related to the hot carrier cooling, has also become significantly slower representing a combination of the cooling and some other slower process.
When the applied negative bias reaches -1.75 V, the Fermi level of the TiO2-FTO is shifted higher than the QD 1Se level which becomes populated by one electron.Excitation of such charged QDs creates trions.The hot electron relaxation in the trion is shortened to 370 fs due to the electronelectron scattering, which is now possible because of the extra electron 41 .Since at -1.75 V, the electron density in the CB of the TiO2 is high, the electron injection from QDs to TiO2 is very unlikely to occur.The 5.8 ps component is interpreted as a combined effect of Coulombic repulsion between the electrons and polaron formation 40 .The repulsion is likely to push the extra electron to the surface of the QDs.Such a new charge distribution is followed by nuclear rearrangement forming a polaron and thereby stabilizing the charge distribution 46 .Observation of strong Stokes shift of the fluorescence in electrochemically single and double charged CdSe QDs gives a strong evidence for polaron formation and for the related nuclear rearrangement 20 .The following 510 ps component is assigned to negative trion Auger recombination (see Figure 4 A and C) which is in good agreement with the previous reports 15,18 .After the Auger recombination, the QDs return back to the initial unexcited charged state, but the DAS analysis still shows a 12 ns component (10% of the TA signal SI Table S2) with a clear ground state bleach at the main absorption band.This minor long component corresponds to the QDs which have remained uncharged.
At -2 V, since QDs become double charged, the corresponding excited states are called tetrons, three electrons and a hole 47 .The hot electron relaxation (380fs) has about the same timescale as in case of the negative trions.The relaxation due to polaron formation becomes longer since stronger electrostatic repulsion can lead to larger charge redistribution and the corresponding rearrangement of the QDs' lattice needs more time to reach the new minimum energy level.As expected, the Auger process in the tetron is slightly shorter (415ps) than in trion.

CONCLUSION
In conclusion, spectroelectrochemistry is used to directly manipulate and monitor charging of the QD photoanode assembly in a well-controlled manner.Combination of spectroelectrochemistry with in situ TA provided a convenient tool for study of the excited state dynamics of the QDs under different potentials mimicking realistic working conditions in optoelectronic device.At low potentials the electrons cannot reach the CB of the TiO2 but populate the shallow traps which appear as constant featureless background over the whole visible spectral range.Next, the electrons start filling the CB of TiO2 without entering the QDs and the photo-induced electron transfer from the QDs to TiO2 is significantly slowed down.At even higher negative potentials, -1.75 V and -2.0 V, spectroelectrochemistry confirmed that the QDs are charged by one and two electrons, respectively.Now the hot electron cooling process becomes faster due to the electron-electron scattering.In the charged QDs a process in the picosecond timescale occurs, which is attributed to polaron formation.In addition, Auger relaxation with 500 ps time constant is observed in our

II. QD synthesis
For the synthesis of CdSe nanoparticles, a conventional hot injection method was used with slight modification [1].For a typical synthesis of CdSe nanoparticles, 1500 mg of CdO was dissolved in 7 ml of OLEA and 50 ml of ODE at 270°C in a three-neck flask.Selenium precursor was prepared by sonication of Se powder in 10 ml of ODE for 10 min in an Ar purged flask then 0.5 mL of Trioctylphosphine was added and then stirred until it became a transparent solution.
When the Cd precursor solution became clearly transparent the temperature was lowered to 240°C, then the Se solution was quickly injected at 240°c and stirred for 2 min.Then the flask was removed from the heater and the hot solution was quickly poured into a metallic bucket, which was placed in a cold bath filled with dry ice.This immediate cooling proved to be very efficient to keep the size distribution of the nanoparticles very narrow.For purification a modified method was used as given elsewhere [2], dichloromethane was used as solvent instead of toluene.The reason for using DCM is that it has a higher density, so it is a very good solvent for extraction of CdSe nanoparticles from the mixture of Cd-oleate in methanol and acetone.
Since the presence of excess oleic acid can complicate the ligand exchange, nm, shows a narrow emission band at 575 nm with FWHM 30 nm, which is comparable with the reported literature data [3].The size distribution of the QDs has a significant effect on the electrochemistry and charge transfer [4] behavior of the system under study.

III. Film preparation
By a conventional ligand exchange method, the capping ligand was changed to 3-MPA.The FTO slides were coated with TiO2 paste, sintered at 480°C and cooled overnight [5].One should consider that annealing of the TiO2 can have significant effect on electrode properties.TiO2 mesoporous layer was to achieve large surface area for QDs anchoring and thereby obtaining sufficient optical density for measurements.Films were prepared by soaking TiO2 coated FTO slides in a MPA capped QDs solution at basic pH>10 for more than 24 h, then washed with distilled water and heated to 110°C.Films were stored in a vacuum desiccator filled with Ar over drying agents to make sure it will be dry by the time of the experiments.Films were kept in the dark, to avoid light soaking and photodegradation was observed for samples under illumination due to charge transfer [6].

VI. Transient absorption spectroscopy
Output pulses of regenerative amplifier Spitfire Pro, 796 nm, 6mJ, 100fs were used to generate both pump and probe light.Second harmonic of the fundamental laser wavelength generated by BBO crystal was used as pump pulses with central wavelength 400 nm to ensure that the photoexcited electrons reach higher energies of the conduction band and are not disturbed by the electrochemically injected electrons.Output of a NOPA (Topas) was used to generate 1300nm which by CaF2 crystal generate broadband white light as probe covering the range from 350 nm to 1200 nm.After setting a specific potential and reaching equilibrium, spectral evolutions of the photoexcited QDs were recorded to track the excited state dynamics.
The data are collected as average of 1000 laser pulses at each delay point.In our measurement the negative signal corresponds to the ground state bleach and stimulated emission while the positive signal is excited state absorption.In order to avoid the multiexciton effects on the dynamics, a very low photon flux was used (50µW at 400µm spot size).For Transient absorption spectroscopy we used bulk-electrolysis or chronoamperometric method.In this method a potential in set for the whole period of the measurement.Only after the current is stabilized, we performed laser measurements.For example, in Fig. SI 4., we can see that after 100 second the current does not have any rapid changes and laser measurement can be started.Earlier than that the conditions would not be stable enough for the measurements because the charging of the redox species is not equilibrated.Global fitting was used as main data analyses method to model the time dependence of the experimental data as a sum of exponential decays by using Glotaran software.Instrument response function was considered to be a Gaussian with FWHM 60fs in the fitting model.In the following we present the decay associated spectra for each data set as obtained from fitting.The onset of the bleach occurs on the timescale of our apparatus response function and we will not be further interpreted and considered.The onset of the bleach occurs on the timescale of our apparatus response function and we will not be further interpreting and considering dynamics faster than 170fs.The charge carriers are mainly originate from the direct excitation of the CdSe by the 400 nm laser, therefore the instant onset of the signal within the apparatus response function at the red part of the spectrum(for example 850nm given in below plot).Since no clear rising component of this relatively week signal can be extracted, we conclude that the electron injection to TiO2 is hot electron transfer [10]. 43
measurements were carried out by following the absorption spectra at different negative potentials during the CV scan.The initial spectrum was taken at open circuit potential (OCP) corresponding to -0.25 V vs. Ag│AgCl and the potential dependent spectral changes are obtained by subtracting this initial spectrum from the measured absorption spectra at each potential as represented in Figure 2.

Figure 3 :
Figure 3: A, B and C: Summary of TA measurements and their corresponding fitted exponents

Figure 4 :
Figure 4: A and B. summary of TA measurement and their corresponding fitted exponents with

1 H
NMR was used to confirm the success of the washing process (see Fig SI 1.).The synthesized CdSe QDs were characterized to ensure the quality, size distribution and shape of the nanoparticles.The absorption spectrum shows a very distinctive excitonic peak around 555 nm.By calculating the concentration of the QDs in solution and measuring the absorbance at the excitonic peak, we evaluated the size of the QDs to be 3.5 nm.The SEM image shows spherical nanocrystals with a relatively narrow size distribution, see Fig. SI 1. Photoluminescence of the QDs excited at 400

Figure S1. 1 H
Figure S1. 1 H NMR of washed and unwashed QDs, TEM image of the QDS scale bar 5nm

Figure S2 .
Figure S2. the schematic band alignment of 3nm CdSe QDs on TiO2 in electrochemical cell

Figure S3 .
Figure S3.CV of blank FTO and TiO2 coated FTO

Figure S4 .
Figure S4.Change in steady state absorption of QDs under application of negative bias.The

Figure S11 . 42 Figure
Figure S11.540nm kinetic traces for negative potentials as dotted plots with symbols with their

Figure S14 .
Figure S14.850nm kinetic traces for negative potentials as dotted plots with symbols with their

Table S1 :
Summary of decay lifetimes with their corresponding contribution to the negative signal in TA measurements