Chemical Physics Letters 457 (2008) 3–11

Contents lists available at ScienceDirect

Chemical Physics Letters
journal homepage:


Multiple exciton generation in semiconductor quantum dots
Arthur J. Nozik *
National Renewable Energy Laboratory, Golden, CO 80401, USA
Department of Chemistry and Biochemistry, University of Colorado, Boulder, CO 80309, USA

a r t i c l e i n f o a b s t r a c t

Article history: A review is presented of recent work on (1) the origin of the concept of enhanced multiple electron–hole
Received 30 January 2008 pair (i.e. exciton) production in semiconductor quantum dots (QDs), (2) various experiments based on
In final form 25 March 2008 time-resolved fs to ns spectroscopy (transient IR absorption, transient visible to near-IR bleaching due
Available online 7 April 2008
to state filling, terahertz spectroscopy, and time-resolved photoluminescence) that support the occur-
rence of highly efficient multiple exciton generation (MEG) in QDs, (3) thermodynamic analyses of the
theoretical enhancement of the conversion efficiency in solar cells that are based on MEG in QDs, (4)
MEG in QD arrays that can be used in QD solar cells, (5) theoretical models to explain MEG, and (6) some
recent controversy about the evidence for MEG.
Ó 2008 Elsevier B.V. All rights reserved.

1. Introduction The creation of more than one e–h+ pair per absorbed photon
has been recognized for over 50 years in bulk semiconductors; it
The efficient formation of more than one photoinduced elec- has been observed in the photocurrent of bulk p–n junctions in
tron–hole (e–h+) pair upon the absorption of a single photon is Si, Ge, PbS, PbSe, PbTe, and InSb [3–10], and in these systems is
a process not only of great current scientific interest, but is poten- termed impact ionization. However, impact ionization in bulk
tially important for optoelectronic devices (photovoltaic and pho- semiconductors is not an efficient process and the threshold pho-
toelectrochemical cells) that directly convert solar radiant energy ton energy required is many multiples of the threshold absorption
into electricity or stored chemical potential in solar-derived fuels energy. For important PV semiconductors like Si, which is over-
like hydrogen, alcohols, and hydrocarbons. This is because this pro- whelming dominant in the photovoltaic cells in use today, this
cess is one way to improve the efficiency of the direct conversion of means that impact ionization does not become significant until
solar irradiance to electricity or fuel (a process we term solar the incident photon energy exceeds 3.5 eV, an ultraviolet energy
photoconversion). Conversion efficiency increases because the ex- threshold that is beyond the photon energies present in the solar
cess kinetic energy of electrons and holes produced in a photocon- spectrum. Furthermore, even with 5 eV photons, impact ionization
version cell by absorption of photons with energies above the only generates a quantum yield of about 1.3. Hence, impact ioniza-
threshold energy for absorption (the bandgap in semiconductors tion in bulk semiconductors is not a meaningful approach to in-
and the HOMO–LUMO energy difference in molecular systems) crease the efficiency of conventional PV cells.
creates additional e–h+ pairs when the photon energy is at least However, for semiconductor nanocrystals (also called quantum
twice the bandgap or HOMO–LUMO energy, and the extra elec- dots (QDs)) the generation of multiple e–h+ pairs from a single
trons and holes can be separated, transported, and collected to photon becomes very efficient and the threshold photon energy
yield higher photocurrents in the cell. In present photoconversion for the process to generate two electron–hole pairs per photon
cells such excess kinetic energy is converted to heat and becomes can approach values as low as twice the threshold energy for
unavailable for conversion to electrical or chemical free energy, absorption (the absolute minimum to satisfy energy conservation);
thus limiting the maximum thermodynamic conversion efficiency. this effect allows the threshold to occur in the visible or near-IR
Since at present the most prevalent form of photoconversion cells spectral region.
are photovoltaic (PV) cells that generate solar electricity, we will In semiconductor QDs, the e–h+ pairs become correlated be-
only discuss these types of cells here. However, the fundamental cause of the spatial confinement and thus exist as excitons. There-
principles of the topics we present here are the same for cells that fore, we label the formation of multiple excitons in quantum dots
produce either electricity or fuel; the difference lies in the engi- multiple exciton generation (MEG); other authors prefer to use the
neering design of the two types of cells, and those differences are term carrier multiplication (CM). However, since free charge carri-
presented elsewhere [1,2]. ers do not exist in isolated QDs and they can only form upon dis-
sociation of the excitons and subsequent separation of the
electrons and holes in a device architecture, we prefer to use the
* Fax: +1 303 384 6655.
E-mail address: former term and will do so here.

0009-2614/$ - see front matter Ó 2008 Elsevier B.V. All rights reserved.

is termed hot carrier conversion and produces higher photovoltages ciency value is achieved in semiconductors with bandgaps ranging if the hot carriers can be captured by high-energy electron acceptors from about 1.2. This calculation was been investigated. and the Auger recombination processes (rAuger) (see Fig. (2) detailed bal. ser in 1961 [11] to be about 31%. facial transfer across the semiconductor interface must all be fast netic energy equal to the difference between the photon energy compared to the rate of carrier cooling [18–20]. This process attained between electrons and phonons. gaps perfectly matched to the solar spectrum in a tandem cell. PV with exciton multiplication tion band for electrons and top of valence band for holes) and equilibrate with the phonons (see Fig.5 to 3.1.25]. 1. and we will also discuss experiments to observe MEG in photocurrent extracted from QDs in various cell configurations. this excess kinetic energy creates an effective temperature cation (impact ionization (rII)) must be greater than the rates of that can be much higher than the lattice temperature. Such cells are generally MEG is incorporated into the calculations through the energy- called tandem solar cells. the analysis is also valid for the Other approaches to exceed the Shockley–Queisser limit have conversion to chemical free energy [12.16–19]. 2.23].5 eV. the rate of e–h+ pair multipli- carriers. rier temperature upon photon absorption can be as high as 10 the rate of electron transfer of cooled electrons must be faster than times a lattice temperature of 300 K. (3) a maximum yield of one e–h+ pair per ab. Nozik / Chemical Physics Letters 457 (2008) 3–11 In this review. electron–phonon scattering and subsequent phonon emission. 2. fore these high-energy e–h+ pairs convert their excess kinetic sorbed photon is produced. and for cases where multiple e– h+ pairs per photon are created. agree with previous calculations [27–31] on ideal solar cells with a gap semiconductors and lower energy photons in the lower band. and Si QDs will be presented.2 to 1. 2. carrier relaxation via phonon emission. PbTe.8% efficiency has been reported in a multi band- unconcentrated solar irradiance into electrical free energy in a sin. The 31% maximum effi. they include multiband and impurity band solar cells [22. and The maximum thermodynamic efficiency for the conversion of a world-record 42. Although the gle-threshold bulk absorber was calculated by Shockley and Queis. The the ultimate conversion efficiency at one-sun intensity increases functional form for QY(E) to model the quantum yield for ideal to about 66%.13]. dependent quantum yield. while those with the rates of photogenerated carrier separation. and inter- energies above it create electrons and holes with a total excess ki. it was first proposed that conversion effi- based on several constraints: (1) the only loss of photogenerated ciencies above 31% in single bandgap solar cells could be achieved electrons and positive holes (charge carriers) is through radiative by capturing the excess energy of e–h+ pairs created by the absorp- recombination (this is termed the radiative limit). Besides hot carrier solar cells [14. For a Boltzmann ensemble of the photogenerated To achieve higher photocurrent. the the radiative recombination rate (rradiative) and rAuger [14. which may exceed unity over certain photon energy ranges. The solar spectrum contains photons with and/or donors. This difference provides the incen- tive to investigate MEG in QDs and in solar cells based on QDs. In order to achieve the higher photovoltages. [14]). so photons with energies priate work functions.). below the optimum bandgap are not absorbed. tandem cell efficiencies are very impressive. gap system at standard terrestrial conditions [15].16–19]. PbS. PV with bulk semiconductors in laboratory-scale PV cells with two tandem p–n junctions. such charge carrier cooling (rcooling).21]. experimental evidence for MEG in PbSe. Fig. we will discuss some recent controversial issues about MEG in QDs. tion of solar photons larger than the bandgap to do useful work be- ance is assumed. or by electrical contacts to the solar cell with appro- energies ranging from about 0. 2. Fi- nally. The results such that higher energy photons are absorbed in the higher band. Also. the reasons why MEG in QDs is much more effi- cient and has a lower photon energy threshold than does impac- tion ionization in bulk semiconductors will be explained. and the bandgap. division of this kinetic energy between electrons and holes is Additional approaches to high efficiency have been proposed determined by their effective masses. ther- ing the conversion efficiency to 31% is that the absorbed photon mophotovoltaic/thermophotonic cells [22]. and (4) thermal equilibrium is energy to heat through phonon emission [14. with the carrier having the and studied and all approaches are commonly called ‘third gener- lower effective mass receiving more of the excess energy [14]. theoretical models for the mechanism of MEG will be discussed. single bandgap having greater than unity QY resulting from impact gap semiconductors. In the Shockley–Queisser analysis. transport. and up-and-down con- energy above the semiconductor bandgap is lost as heat through version of the incident solar photons [24. the biggest loss factor limit. electron transfer (rET) of hot electrons. In 1978.4 eV. as the carriers relax to their respective bandedges (bottom of conduc. Theoretical thermodynamic conversion efficiency of radiant energy to free energy We start our review with a discussion of the maximum thermo- dynamic limits of the solar power conversion efficiency in PV cells based on conventional bulk semiconductors where one e–h+ pair per absorbed photon is produced. Actual efficiencies of about 32% have been reported MEG QD absorbers is given by a sum of step functions . The present approach The detailed balance model of Shockley and Queisser has been to reduce this loss and increase efficiency above the 31% limit used [26] to calculate the power conversion efficiency of single- has been to use a stack of cascaded multiple p–n junctions with gap and two-gap tandem solar conversion devices which employ descending bandgap values along the direction of the incident light QD absorbers capable of MEG after photon absorption. their cost is very high. ation solar cells’ [22]. carriers are termed ‘hot electrons and hot holes’ and the initial car. this reduces the overall heat loss due to ionization. for photon to carrier generation. 1). QY(E). In the limit of an infinite stack of band. In bulk semiconductors.J.4 A. Hot carrier relaxation/cooling in semiconductors (from Ref.

which have electrons and holes with equal effective masses. the threshold photon energy for I. including the inverse Auger process of exciton m¼1 multiplication. produces the same degree of slowed cool- ing by inhibiting the Auger cooling channel [38]. to become competitive and even dominant in QDs. very efficient multiple e–h+ results are shown in Fig. A.e. or III– V semiconductors because the maximum QY for I.e. The concept of enhanced MEG M = 1. 3. crystal momentum need not be conserved because momen- sorbed photon. ð1Þ of Auger processes. must compete with the rate of energy relaxation by phonon emission through electron– phonon scattering. occurs slowly. does not ex- ceed 1.8 eV (4. Multi excitons have been detected using several spectroscopic The rate of cooling of photoexcited excitonic states in semicon. in semiconductors prevents possible from the solar spectrum. M = 2. and the rate QYðEÞ ¼ hðE. as described below. becomes competitive with phonon scattering rates only when the kinetic energy of the electron is many times the bandgap energy (Eg) [33–35]. PbS. Cooling dynamics in quantum dots Si. The in QDs is indicated in Fig.I.mEg) is the Heaviside unit step function. InAs. and 25% at hm  4.I. channels for relaxation. mEg Þ.) cannot contribute to improved quantum yields in present solar cells based on Si. measurements which are consistent with each other. Nozik / Chemical Physics Letters 457 (2008) 3–11 5 sion controls the efficiency of exciton multiplication. and Mmax is the maximum multiplication of the threshold photon energy for I. Additionally. Maximum thermodynamic power conversion efficiency with no solar con. either by rapidly removing the photogenerated holes or avoiding their formation in the first place by injecting only electrons into the QDs [32]. exceeds that required for energy conservation alone because. the rate of I. This large blue shift photon absorbed (M). This effect is called a ‘phonon bottleneck’ and in experi- ments conducted with InP quantum dots the cooling time of hot electrons increased by about a factor of 10 (from 200–300 fs to 2–3 ps) if the Auger cooling channel was blocked. pair (multi exciton) creation by one photon has now been reported in six semiconductor QD materials: PbSe. for example. We denote MEG absorbers with makes the momentum uncertain).I. PbTe. Taking M = 1 the concomitantly increased e–h+ Coulomb interaction.6Eg). CdSe. CdTe. and M = Mmax as M1. and Mmax. method is to monitor the signature of multi exciton generation . to be only 5% (i. and 3. is greatly enhanced due to carrier confinement and where h (E.J.I. several phonons must be emitted simultaneously via electron–phonon scattering to satisfy energy conservation.36]. 2. Setting M = Mmax = Emax/Eg gives the maximum pos. However. photo- exciting PbSe and PbS QDs. Dynamical channels for photoinduced hot electron–hole pair. tum is not a good quantum number for three-dimensionally sible multiplication allowed by energetic constraints and the confined carriers (from the Heisenberg Uncertainty Principle the highest possible efficiency for a MEG device (i. crystal momentum (k) must also be conserved. 3. M2. efficiency was found Fig. M1 is the Shockley–Queisser result with 1 e–h+ pair/photon.4Eg) [7. It has been shown that the rate of I. 4. This arises be- cause the energy separation between quantized levels in QDs can be many times the typical phonon energy. total quantum yield = 105%) at hm  4 eV centration as a function of bandgap for different carrier multiplication values per (3. more. Indeed. The first ductor quantum dots compared to cooling through phonon emis. The slowed relax- ation/cooling of hot excitons via phonon emission allow other Fig.I. in Si the I.0 until photon energies reach the ultraviolet region of the spectrum. respectively. 4. In bulk semiconductors. About 95% of the efficiency gain with carrier materials such as bulk Si and GaAs from yielding improved solar multiplication is obtained with generating just 2 e–h+ pairs/photon. The observed transition between inefficient and efficient I. Further- gives the usual assumption of one e–h+ pair generated per ab.I. This re- quires simultaneous multi-particle scattering events which be- come highly improbable with increasing numbers of phonons emitted. M e–h+ pairs for well-defined location of the electrons and holes in the nanocrystal all photons that are M times Eg). in quantum dots the rate of electron relaxation through electron–phonon interactions can be significantly reduced X M because of the discrete character of the e–h+ spectra.I. conversion efficiencies [9. in addition to conserving energy. the rates of excited state exciton cooling can be slowed down in quantized semiconductor structures because of the existence of discrete quantized energy levels. such as MEG. and in order for high-en- ergy electrons and holes in higher excited quantized states in the QDs to cool by relaxing from a given quantum level to the next lower level. Although rates of hot carrier cooling are very fast in bulk semiconductors (sub-ps time scale). CuInxGa1xSe2. Electron–hole pair (exciton) multiplication in quantum dots Impact ionization (I.36].I. M2 is the result for 2 e–h+ pairs/photon.

66 0.82. and it was shown that efficient MEG occurs also in >5 ps. [21]). = 3. 4. can be calcu- lated by the product of the absorption cross section.2 150 0.85 eV.I.7 nm diameter PbSe QDs. 0.6 A. photon energy expressed as the ratio of the photon energy to the QD bandgap (HOMO–LUMO energy) for three PbSe QD sizes and one PbS (dia. = 5.7 nm). In one type of TA experiment pump photon energy to the bandgap (Ehm/Eg) varied from 1.9. and Eg = 0.91 eV Ehν/Eg = 1.4 0. A statistical analysis of these data also showed that the the dynamics of the photoinduced mid-IR intraband absorption is QY begins to surpass 1. QYs above 200% indicated the formation of more than two Fig.72 eV Ehν/Eg = 4.90 0.82 eV (dia. the dependence of the MEG QY on the ratio of the at their lowest state (the bandedges). 1Se–1Pe) of the newly created excitons (see Fig. PbS [38] (see Fig. in a second type of experiment the Eg = 0. and the resulting ex- change in the Auger decay dynamics of the photogenerated exci. ra. the average number of photons absorbed per NC. In the latter case. Multiple electron–hole pair (exciton) generation (MEG) in quantum dots excitons per absorbed photon.3 200 0. the peak magnitude of the initial early time photoin. Solid symbols indicate data acquired using mid- infrared probe. [38] a QY value of 300% for 3. respectively).91 eV (dia. Nozik / Chemical Physics Letters 457 (2008) 3–11 sients are detected by probing either with a probe pulse exciting across the QD bandgap. and 0. they reported an excitation en- ergy threshold for the efficient formation of two excitons per photon at 3Eg. plete and the probe pulse is interrogating the exciton population In Ref. of sharply increasing MEG efficiency. by which time carrier multiplication and cooling are com.73 eV. 0.4 nm and 5.39]. Fig.91 eV 0.9 nm). 5b).5 nm.60 Quantum Yield (%) 0. open symbols indicate bandedge probe energy. [38. both experiments yield the same MEG QYs. 5. cited state provides both the electron and the hole with excess en- tons is related to the number of excitons created. respectively. = 4. [38]).6 b 300 Ehν/Eg = 5.91.Lumo) Ehν/Eg = 4.72 eV Ehν/Eg = 3. In the for.1 0. For pho- ton energies less the 2Eg.7 nm).7. the tran. indicating the formation of three excitons per pho- ton for every photoexcited QD in the sample. where PðmÞ ¼ hN 0 i exp ðhN 0 iÞ=m! and nNC is the concentration of NCs in the photoexcitation volume.5 0. The NREL research group reported (from Ref. corresponding to the onset mer case. 5. Schaller and Klimov reported a QY value of 218% (118% I.0 (see monitored after the pump pulse (Fig.05 250 0. 5b.8Eg. In Refs. When exciting a colloidal solution of NCs using a laser pulse- width short compared to exciton recombination dynamics. and Eg = 0. 5b) and in PbTe nanocrystals [39].0 lm for a sample of 5.82 eV Ehν/Eg = 3. 4. a 0. 5a). 5a). (b) QY for exciton formation from a single photon vs.25 0.0 at Ehm/Eg values greater than 2.72 eV (normalized at tail) Ehν/Eg = 4. or with a mid-IR probe pulse that monitors the first 1Se–1Sp intraband transition.25 0. 5.g. = 3. It was noted that the 2Ph–2Pe transition in the (e. as probe pulse is in the mid-IR and monitors the intraband transitions shown in Fig. QDs is resonant with the 3Eg excitation. The multiple exciton analysis relies only on data for delays is 2Eg [38].00 Eg (Homo . jp: hN0i = ra jp. ergy of 1Eg.0 for PbSe QD samples with Eg = 0. (a) Exciton population decay dynamics obtained by probing intraband (intraexciton) transitions in the mid-IR at 5. where. The number of NCs that have absorbed m photons can be determined by the Poisson distribution function.9 to the probe pulse monitors the interband bleach dynamics with exci. . This symmetric transition duced absorption change created by the pump pulse plus the (2Ph–2Pe) dominates the absorption at 3Eg. and the pump photon fluence. hN0i.72 eV (dia. Quantum yield results were independent of the probe energy utilized (from Ref. Evidence was also using transient (pump–probe) absorption (TA) spectroscopy [37– provided that the threshold energy for MEG by optical excitation 41]. The first report of exciton multiplication was presented by Sch- aller and Klimov [37] for PbSe NCs. efficiency) at 3. nm = nNC P(m).0 100 0 100 200 300 400 2 3 4 5 Time delay (ps) Ehν/Eg Fig. Eg is the absorption energy gap of the nano- crystal (HOMO–LUMO transition energy) (it is common practice to refer to the HOMO–LUMO energy difference in QDs as the QD bandgap even though the electronic states are discrete). only one exciton is produced per ab- sorbed photon. tation across the QD bandgap.J.91 eV PbS (0.91 eV 0. [38].9 nm diameter PbSe QDs at a photon energy of 4Eg.85 eV) Δα 0.

6). J0 is the photon fluence. the MEG efficiency would be the following equation can be fit to the data: greatly reduced because of the reduction of quantum confinement. 6. ported for isolated dots suspended in solvents. and the QY is assumed intrinsic region of a p–i–n structure in which extended states. MEG in PbSe QD arrays determined value. d is fixed formed from the coupled QDs. and the energy for MEG (0. QY is the number of excitons minescence (TRPL) to monitor the effects of multi excitons on the created per excited QD. A. terahertz (THz) spectroscopy to probe the in. (2) (gray lines potential device geometry [21]. and then the films were excited same trend is observed regarding single exciton and the biexciton above the threshold to obtain the information on exciton decay lifetimes. While this type of close electronic coupling is wavelength for all measurements. PbTe [39]. InAs [43. The separated charges must then above and below the MEG threshold pumping conditions (see drift and/or diffuse to the electrodes before recombining. This is a particularly interesting ation efficiency. A second. 7 along with previously reported [38] values critical to determine if MEG is preserved in such QD films and to for PbSe QDs in solution versus energy gap multiple. it is work is plotted in Fig. The cou- the carrier mobility of QD films [46–50] by reducing the interdot pled film has an exciton generation efficiency of 164% at 4Eg. some of these optical experiments use time-resolved photolu. aside from the biexciton lifetime scaling with volume. The QY for the films used in this necessary for the efficient extraction of carriers from a film. The QY agrees well with the first sample.45. and in hydrazine-treated films (right).J. These results were As for isolated QDs. pared to isolated QDs in solution. the ability to achieve effective electronic coupling between DT T0 j0  r  QY  r QDs in a QD film without reduction of MEG is very encouraging t¼3ps Rpop ¼   ¼ ð2Þ for the development of novel high-efficiency solar cells employing DT 1  expðj0  rÞ T0 t¼750ps close-packed arrays of QDs. allow for the delocalized photogen. cross section at the pump wavelength. to be 100%. and corresponds to the overall shorter molecules after film formation can drastically increase average efficiency of exciton generation in an excited QD. creased far infrared absorption of multi excitons. Nozik / Chemical Physics Letters 457 (2008) 3–11 7 Additional experiments observing MEG have been reported for where Rpop is defined as the ratio of exciton populations at 3 and CdSe [40. The determination of MEG Thus. In addition to 750 ps after excitation. from the Distinct excitonic features are still evident in these electronically slightly lower Eg of the coupled film while using the same pump coupled QD arrays. exhibiting appreciable electron delocalization resulting in reason- bination is complete (750 ps) is plotted versus photon fluence and ably good charge carrier transport. to some extent. ables the direct determination of the absorption cross section (r) of the QDs in the films at the pump wavelength without using a pre- 5. 6). dynamics to determine the MEG QY. [46]). best-fit value for the QY was found to be 148% at 4Eg for the QDs in Exchanging bulky capping ligands used in the QD synthesis with TCE as well as in the untreated film.65 eV/810 nm).46] to deduce exciton gener. In the sub-MEG threshold case. This spacing [51] while retaining relatively highly-passivated surfaces. after a post-film-fabrication soak treatment in 1 M hydra- using TA relies on the fact that multi exciton Auger recombination zine to electronically couple QDs in QD films. A related discovery presented below is Fig. The erated carriers to traverse the film and reach the contacts. no reduction of MEG is much faster than single exciton recombination. This analysis technique not only provides a reliable spectroscopy to observe the PL red shift and line shape changes way to accurately determine the QY of exciton generation it also en- due to multi excitons. the decay dynamics of single excitons in the all repeated using a smaller size of QDs with larger Eg (0. and quasi CW e7503ðpsÞ=s1 ). and Si [45]. a fit of Eq. Above the MEG threshold (green lines in Fig. the QDs treated and an electronically coupled film compared with that of a must be electronically coupled such that charge separation occurs solution of QDs in TCE from the same synthesis (see Fig. The QY on a timescale longer than MEG (1013 s) but shorter than the can be obtained by calculating the ratio of the QY from the fits biexciton lifetime (1010 s).42. ulation over the time frame of the experiment (in this case. 6) is applied where only r and d vary. at its sub-MEG threshold value while the QY is allowed to vary. more efficiency was found in the electronically coupled QD films com- simple analysis can be used [37. The ratio of the normalized change in transmission and important result because one might expect that in QD arrays soon after the excitation pulse (3 ps) to that after all Auger recom.42].   Thus. . an ordered 3D QD film forms the in Fig. in untreated PbSe QD films(middle). which may have some uncertainty. slight increase in the coupled film arises.90 eV) from QD films is first determined by photoexciting below the threshold another synthesis. r is the absorbance TA. 6). Ratio of exciton population at 3–750 ps (Rpop) after excitation with pump energy of <2Eg (squares) and 4Eg (circles) vs pump fluence for PbSe QDs in solution (left).44]. Using this procedure the MEG efficiency was measured in an un- For high-efficiency in certain MEG QD solar cell designs. The fits to these data are described in the text (from Ref. The QY results understand how the reduced quantum confinement of the excitons for coupled QD films are similar to what has been previously re- affects the MEG quantum yield (QY). In one Fig. and d is the decrease in single exciton pop- PL decay dynamics.

8 of fast multi while the quantum confinement is not sufficient to produce a large exciton Auger recombination serves as a signature for MEG. For Si NCs. Fig.86 eV. Left: TA photoexciting below and above the MEG threshold for 9. (2) the PA amplitude steadily in- creases with photon energy with an onset greater than 2Eg. is very improbable.5 nm Si QDs. In contrast.5Eg [36]. MEG has been independently state-filling-induced bleach via this interband transition. and the quantum yield per NC is absorbed. well below the effective bandgap. the appearance in Fig. the probe was mainly at 0. absorption at the bandgap. photons being emitted within the biexciton lifetime. TA data below the threshold showed that the biexciton decay times for three different Si NC sizes depended linearly on the QD volume in agreement with the Auger recombination (AR) mechanism [52]. However. but the presence of the new fast decay The dotted lines are guides to the eye. this means that sorbs at most one photon.5 nm and 3. Photoinduced transient absorption (TA) dynamics.28 to 1 eV. the other alternative process other than MEG can be envisioned that confinement is still enough to produce efficient MEG. The amplitude of the AR component increases as the photon energy increases past the energy threshold for efficient MEG to oc- cur. 8 shows the decay dynamics when hN0i is held constant at 0.4 at 4.6 ± 0. is consistent with the following experimentally determined obser- vations: (1) the dynamics are the same as AR (same decay rate and 5.9Eg (above the MEG threshold).8 A. No confinement kinetic energy with an attendant large blue shift.8 nm Si QDs (from Ref. [46]). By photoexciting above the energy conservation threshold for that MEG can still be efficient in relatively large QDs of Si (5 nm ra. a new decay channel observed at high pump fluences was confirmed to be non- radiative AR. and Silicon’s indirect band structure yields extremely weak linear (3) it remains at low excitation fluences. 8. multi exciton AR serves as a metric for MEG. and thus one cannot readily probe a In addition to the TA experiments. the first efficient MEG in Si NCs was reported using dynamical signature of MEG. for bulk Si the threshold for impact ionization der conditions where absorption of more than one photon per NC is 3.1Eg. Nozik / Chemical Physics Letters 457 (2008) 3–11 In the Si QD experiments. MEG (>2Eg) and at low intensity so that each photoexcited NC ab- dii. as in the TA experiments. . 7. eled with only one adjustable parameter.5 solar intensity. Instead. g.5 at different pump wavelengths for the 9. Quantum yield of exciton generation for PbSe QD films and QD solutions.5Eg and the QY rises to only 1. it was verified that the pho- toinduced absorption dynamics did not depend on the probe en- ergy over a broad range from 0. The data were mod- untreated films are identical (from Ref. the 3. MEG in Si QDs dependence on NC volume).J. The black crosses are the decay dynamics for pump energies of 1. Highly effi. TRPL measurements are extremely cient MEG in nanocrystalline Si at lower photon energies in the vis.1. A quasi-CW measurement also observed PL (QY) of excitons produced per absorbed photon reached 2. challenging due to the long single exciton lifetimes relative to typ- ible region has the potential to increase power conversion ical TRPL laser repetition rates (50–250 kHz) and the large differ- efficiency in Si-based PV cells toward a thermodynamic limit of ence between biexciton and single lifetimes which results in few 44% at standard AM1.2 spectra indicative of the multi-excitonic state in InAs [43] NCs un- at 3. respectively. Time-resolved photoluminescence toinduced intraband absorption changes. Note that the QYs for QDs in solution and component at times <300 ps is clearly evident. The quasi- Fig. the MEG efficiency.3Eg pump are noisy.8 nm sam- ples. which is about equal to the Bohr radius if Si). (TRPL) single photon counting experiments measured both the In Ref.4 ± 0.5Eg (below the MEG threshold) and the red crosses are for photon energies of 3. The data at long times (>300 ps) in Fig. as well as transient intraband absorption spectroscopy and the threshold the biexciton PL spectrum under conditions where only one photon photon energy for MEG was 2. so that each photoexcited NC ab- sorbs at most one photon. When photoexciting above the energy conservation threshold for MEG (>2Eg) at low intensity.7 and 1. CdSe [42] and InAs QDs [43]. reported in the literature with two different PL measurements in the exciton population dynamics is probed by the method of pho. 8 for Fig. Right: TA photoexciting below and above the MEG threshold for 3.3 and 2.4Eg. Thus. [45]). [45].

5 nm sample the y-axis inter. efficient MEG by a single photon in NCs is explained by the most 10% (for the 3. and disagreement toward a QY of 3. state. further research is needed to the case of 3.1. The density matrix approach is the only self-consistent method that takes into account the diverse processes responsible for MEG in NCs.0 at 3. which allows simultaneous consideration of an arbi- We estimate an onset of about 2. The present obser. Nex.9–1. WnEx is the multi exciton efficient MEG. tance between the electronic particles in the QD. tio of the bandedge biexciton and single exciton populations [54]: . at low fluence does not increase the uncertainty of the derived QY values. The theory has been applied to PbSe NCs where the onset of very efficient MEG occurs at photon ener- gies of 3Eg corresponding to a 2Ph to 2Pe transition [54] (but note that the threshold for the onset of (weak) MEG is 2Eg). t. tion of the NCs. e is the dielectric constant in the QD. The best fit parameters have fitting errors of at [54]. Other theoretical ments in Si NCs are complicated by the factor of 100 to 700 in models for MEG discussed below would have to be consistent with the ratio of absorption cross sections for pump wavelengths below this important feature. Theory of MEG Fig. as possible to 2Eg. and (3) dissociate the excitons into free. 3) for the highest possible PV efficiency of where s is the pump–probe delay time and t2 is the biexciton life. Hence the second method described that is sufficiently strong to produce a large blue-shift of the band- above for PbSe QD films was invoked to acquire high quality QY gap. pared to the bulk.42. Red triangles (inverted triangles) are data for bulk Si c1 6 1/c2  1/Ubi  1/Uex. 42% utilizing cells with MEG yielding just two excitons per photon time. fitting Eq.4Eg for MEG in Si QDs. and the ratio of these rates determines what a single photon creates: a single exciton or multi excitons. understand how to: (1) make the QY rise very steeply with photon indicating very efficient MEG. must be slower than both the energy relaxation rate of the multi ment regime and lack a large spectral blue shift. Nbi.5 when exciting below 2Eg. is not necessary for and W1Ex is the initial single exciton state.8 nm sample) and only 5% for the 9. thus. 1/ Fig. MEG QY for 9. [54].J. 9 presents the MEG QYs for Si QDs compared to literature data [36] for bulk Si obtained from photocurrent measurements Several theoretical models explaining the enhancement of MEG in p–n junctions. while the intercept is 2. to the single exciton.4 at 4. and very strong confine.5Eg. 6.5Eg and the threshold is at about 2. This approach ultimately yielded the ra- from Ref. This bandgap is close to the optimum QD bandgap toexcitation (s > 0) and after AR is complete (s > 3t2) is determined. In a novel theory of MEG only 1. which is about different dephasing rates for these states. The steady-state solution of the density matrix ment regime is defined as a > 3aB [53]. The absence of the need for confinement and above the MEG threshold.4Eg. The exciton Bohr radius (aB) for Si is 4. The very weak confine. trix approach.5 nm and 3.9 nm. W C ¼ hW1Ex je2 =erjWnEx i ð3Þ This result implies that very strong confinement (a  aB). of 0.54–56]. [36]. To calculate the efficiency of MEG. However. and collect them in the external circuit of the QY significantly (within the fitting error).1 eV (see Fig. close to the bulk bandgap. The scatter in the data increases as energy after the MEG threshold. Si QDs with a diameter of 9. Hence the scatter with a very high-efficiency. the question of MEG efficiency is reduced to the question of the state relaxation. respectively) but much shorter than the lifetime of the bandedge excitons (1/Uex and 1/Ubi).5 nm diameter of the Si NCs. equations yields the following conditions for efficient MEG: (1) ment is defined as a < aB. i. It was reported [45] that for a 9.e.5 nm formation of a coherent superposition of single and multi exciton sample. (2) establish the threshold as close the pump fluence decreases. For a strongly coupled superposition of single and multi exciton states. and short pulse excita- equal to the 9. A. This initial single e–h+ pair state (exciton) is cou- pled with the multi exciton states and creates a superposition of single and multi excitons. means that the bandgap at which efficient MEG can occur lies data with high SNRs. Nozik / Chemical Physics Letters 457 (2008) 3–11 9 CW method would also suffer from these limitations. in this second method the intensity bandgap of 1. at a threshold of 2Eg [26]. and r is the vectorial dis- duce stronger Coulomb coupling of the electrons and holes com.9 ± 0.5 nm Si NCs are more the energy relaxation rate of a single exciton (c1) initiated by light appropriately characterized as being in the intermediate confine.20 eV—only 80 meV larger than bulk Si—and shows dependence of the ratio of the exciton populations just after pho. The single and multi exciton components of the superposition have different decay rates. a single photon can only gener- ate a single electron–hole pair which is not an eigenstate of the multielectron Hamiltonian that describes the superposition of multi excitons. interaction between the single and multi exciton states el of MEG and for practical applications to solar photon conversion. efficient MEG. However. exciton (c2) and the rate of Coulomb coupling between the two vation of efficient MEG in indirect-gap semiconductor NCs is an states (WC/⁄). WC is the matrix element of the Coulomb important result that has significance both for the theoretical mod. 10). Many of these models depend upon en- is to be compared to a threshold of 3. but only sufficient confinement is needed to pro. as evi- denced by a large blue-shift of the bandgap. fitting the data set to Eq. where. The model uses a time-dependent density ma- sulted in fitting errors that were smaller than the reproducibility. 9. the analysis yields a QY of 1. 9. states (see Fig.3Eg excitation. it is necessary to compare the populations of the ground biexcitons. during time.8 nm diameter Si QDs vs ratio of photon energy to QD bandgap (from Ref.5 nm have a As previously described. The MEG QYs for Si QDs rise much more rapidly in quantum-confined NCs have been published. Under optical excitation of NCs. TA measure.9 for ble PV efficiencies in actual solar cells.5Eg for bulk Si and a QY of hanced Coulomb coupling in the QDs. In general. sepa- (2) but excluding the lowest two points does not change the value rated electrons and holes. as discussed in Ref. (2) to each individual data set re. [45]). trary strength of coupling between single and multi exciton states. which is much longer than the relaxation times of the excited states (1/c1 and 1/c2. to approach the highest possi- cept is 1. this remains [38.

The most convincing result would be a measurement of a gime. Although. If the measured QY for the tion of multi excitons. while 1/sex and 1/sbi are the rates of the intraband photoinduced absorption in the mid-IR. i. scattering. meaning that multiple excitons are essentially created extremely fast upon absorption of high-energy photons [38. in all optoelectronic QD devices. (4) shows that efficient MEG. WC/⁄  c2. separated e–h+ pair recombination. and quasi-CW of these latter 3 rates are slow compared to the rates of relaxation and Coulomb photoluminescence spectroscopy) and they all report results on coupling (from Ref. this new model also explains the lower threshold photon energy for MEG at 2Eg. In conclusion. and charge transport be much faster compared to exciton or value of (c2/c1). is possible only if the relaxation rate of photocurrent is >1. c1.0.10 A. V is the overall rate of decay of the interband bleaching probed with bandgap excitation. only 3 degenerate states are shown upon absorption of a 3Eg photon (2Ph–2Pe): the initial single 2Ph2Pe exciton. For simplicity. all reports published on observing MEG in QDs have populations experience oscillations with time (quantum beats). However. and attendant greatly increased density of surface trapping and slows down the formation of a coherent superposition.J. The most direct method for where P1–2 is the probability of the population of the excited biex. Nbi/Nex = 4WC2/(⁄2 c1 c2). weak coupling of symmetric e–h+ pairs created by light with phonons suppresses their relaxation and would result in efficient MEG because the condition c1  c2 is fulfilled. there have been excited biexcitons (the latter enhances electron–phonon coupling). which show that the QY de- pends only weakly on size and is a function primarily of only one parameter: Ehm/Eg. Fig.0 would require that the other It was found theoretically [54] that the biexciton relaxation rate losses in the device.e. In both cases. The latter loses could be ac- and PbSe NCs with asymmetric e–h pair configuration are 10 to 40 counted for by measuring the QY for photocurrent based only on times stronger than that of their coupling with symmetric e–h+ the photon flux absorbed in the QDs (this is defined as the internal . all these spectroscopic measurement are nevertheless N bi =N ex ¼ ðc2 =c1 ÞP 12 ð4Þ indirect methods of measuring MEG. determining how many e–h+ pairs are created per absorbed pho- citon state via its Coulomb coupling to the single exciton state ex.0 with respect to the incident photon flux (defined as the much larger than unity if WC/⁄  c1 [54]. MEG. structure added to the new non-radiative recombination channels portional to their coupling with phonons. then there can be no doubt whatsoever that the biexciton is much faster than that of the exciton (c2  c1). new and efficient non-radiative recombination Detuning of the resonance between single and multi exciton states. an increase of c2/c1 always leads to an increase of QY. 7. MEG photocurrent the 4th state (1Sh1Se1Sh1Se1Sh1Se) is not represented and electronic states between the 1S and 2P states are also not shown. Theoretically. ton is to count the electrons in photocurrent measured in an exter- cited by a photon. the predominant genera. exciton interactions with the polar optical phonons that are sensing the total charge are very weak [58]. a mechanism for MEG was introduced in Ref. Nbi/Nex is then determined by the ratio of the biexciton creation when the QDs are configured to allow charge extraction of the dis- rate. as discussed above. assuming re-radiation. nal circuit connected to an optoelectronic device consisting of QDs Eq. the MEG QY will not show a size dependence if it is assumed that the ratio c1/c2 does not depend on size. and absorption in the non-QD elements of the device that the relaxation rate for various e–h+ pair configurations is pro. [57]). channels are opened that arise from the charge delocalization for example from a significant size dispersion of the NC population. be. can be efficient even for a weak coupling re. and the NC thus retains its local neutrality even after exciton creation. As a result. The relaxation been based on spectroscopic measurements. This result requires that in the cause P12 < 1. recombination sites. however. its value is always <1. their To date. 1Sh1Se2Pe1Se). charge distributions of the optically created electron and hole com- pensate each other almost exactly at each point of the NC. time-resolved photoluminescence. (2WC/⁄)2 c2. and can be QY > 1. For strong coupling. MEG with six different semiconductors that are consistent with one other. where the MEG QY (cooling) rate of the single exciton state (1/c1) is slower than the relaxation rate of is determined by analyzing the effects of the formation of multiple the biexciton states (1/c2) because of the asymmetric charge distribution in the excitons on light. The predicted size independence of the average number of e–h+ pairs generated via excitation of the same optical transition is consistent with experimental results [37–39. the transition probability ap. transient excited single exciton to the ground state.41]. All spectroscopy. such as reflection. The 3 degenerate states are coupled by the Coulomb interaction (WC) and form a coherent superposition of 3 states. This is because in the latter case. 10. These calculations show arising from charge collection do not exceed the increase in exciton that polar interactions of intrinsic semiconductor phonons in CdSe population created through MEG. transmission. Nozik / Chemical Physics Letters 457 (2008) 3–11 pair configurations (nLenLh). and the two excited biexciton states (2Ph1Se1Sh1Se. In this case. [54] that invokes a coherent superposition of multiple excitonic states. For simplicity. MEG is occurring within the QDs. It also is quite obvious that the coupling of asymmetric e–h configurations with phonons of organic molecules at the NC surface [59] should be significantly stronger as well. sociated excitons. This preferen- several different types of spectroscopic measurements (transient tial dephasing of the coherence for the excited biexcitons results in the final for- mation of a biexciton from a single photon. The relative population external QY (EQY)). charge separa- proaches 1 and the population ratio approaches its maximum tion. to the direct single exciton relaxation rate. where light is absorbed only in the QDs.54]. QD device the time scale for exciton dissociation. At the same time. An EQY > 1. is much faster than the single exciton relaxation rate. Theoretical model for MEG. transient THz decay of the relaxed (cooled) single exciton and biexcitons to the ground state.

7 (1957) 91. he is also a member of the American Chemical [22] M. Phys. [19] F. Chem. [58] S. Shabaev. 78 (1997) 5014. S. Abram. B 35 (1987) 8113. Schaller. A. Ben-LuLu et al.D.J.T. A. Phys.0.S. Queisser. Office U. U. Besides the lack of published reports of an EQY or IQY > 1. Czech. Chem.B.J. and has delivered over 250 invited talks at universities.. J. Efros. A. Ross. Semiconductor Electrochemistry. and symposia. Sov. Würfel. An. Wehrenberg. M. 72 (1992) 3554. Chashchin. P.M. A. K. cay dynamics upon excitation with >3Eg photons compared to pho. H. Queisser.H.0 for [36] M. N. Physica E 14 (2002) 115. Koc.A. Nozik. Harrison.J. Solar Energy Research Institute (SERI). Werner.J. Brand. Semiconductor Nanocrystals: from Basic Principles to Applications.J. Beattie. Qing Song) and theorists at the Na. D. [62] that for some QD materials MEG efficiencies may de. Martí. Hess. Solids 8 (1959) 223.0. Dr. Werner.J.J.C. Chem. conferences. 128 (2006) 3248.V. Nozik. J. J. Rev. J. F. Schaller. Agranovich. photoelectrochemistry of semiconductor-molecule interfaces. Phys. Chem. Beard. 72 (1994) 3851. 52. Phys. P. He has published over Aver’yanov. Soc. Heckler. Office of Basic Energy Sciences. C 111 (2007) 4146. Queisser. 53 (1982) 3813. Appl. Appl. [42] to explain the discrepancy in the TRPL results 20192. J. [14] A. [43] J. A. [26] M. some concern about MEG is that the TRPL measurements of exciton [40] R. MEG in InAs core-shell QDs using TA.A.D. which is diffi- [25] T. Rev. Sol. J. 45 (1966) 1. Beard et al. Lett. Taniguchi. unique effects in nanostructures to advanced approaches for solar photon conver- [8] A. Werner.J. A.J. EQY or IQY for photocurrent is >1. Luque. the excitation energy for the [44] R. 46 (1967) 4590. Efros and Andrew Shabaev). Chem. Kolodinski. ences. Wiley-VCH. Bude. Nozik is a Senior Research Fellow at the acknowledge the support of the US.J. 63 (1993) tures. Klimov. Nano Lett. E. However. Schaller. Yu.S.A. R. [62]. 6 (2006) 424. Schmitt-Rink. Arthur J. Dr. Guyot-Sionnest. A. Energy Mat. [15] Press Release.M. Phys.J. the Electrochemical Society. Talapin. A A295 (1980) 453. Kunzner. val Research Laboratory (Alexander L. Memming. research at the Materials Research Center of the Allied [4] V. Lockwood. Biosciences and Geosciences under Contract No. Werner. [62] G. Matt Law. Wittchen. A. Ivakhno. Guyot-Sionnest. Wehrenberg.J. Chem. Phys. Chem. Before joining NREL in 1978. B 110 (2006) 25332.L. Wang. Urban. Brendel.S. Vavilov. Rev. Sov. holds 11 [11] W. [50] D. Phys. Solids 8 (1959) 219. Phys. S.E. Phys. The authors of Ref. [46] J. Nature 311 (1984) 21. Chem. Beard. J. M. in the QD excited state decay dynamics. C. Nature 271 (1978) 137.H. Tauc. 2006 issue. Law. 2007.V. Society. Song. J. Klimov. Ross. size quantization effects in semiconductor quantum dots and quantum wells.H. J. 2005. Sydney. Note added in proof [52] D.K. photocatalysis. ACS Nano 2 (2008) 271. Chem. Phys. But the IQY requires knowledge of the optical and [23] A. Chem. Badescu. Murphy et al. Phys.M.J. 6 (2006) 2191. J. are in progress.L. Luther. photogenerated carrier relaxation dynamics in various semiconductor struc- [9] S. J. Appl. C. R. Pietryga. Mic´ic´. De Vos. Nozik. Chem. M. C. Appl. workshops. Cells 51 (1998) 413. J.A. be reproduced. B 52 (1995) 11319. Talapin. Klimov.J. [43] report that they cannot repeat their THz [53] A. Landsberg. Murray. and Mössbauer spectroscopy. Hanna. B. Schaller. [60] J. Nano Lett. Experiments by the [48] B. J. Shabaev. V. [63] J. Phys. Soc. S. M. Phys.64] also report they find no Publishers. Phys. Brendel. Chem.N. Joey Luther. Justin John. [13] R. dielectric constants of the heterogeneous QD device. Appl. in press.J. Christensen. O. he conducted [3] S. P. Chemical Corporation (now Honeywell. B 109 (2005) authors of Ref. the applications of [7] O. Phys. Nano Lett. Third Generation Photovoltaics. London Ser. B. Kovalev. Rev.S. 2003. Hamaguchi. [6] V. Shevchenko. Koch. Rev. M. Cells 41-2 (1996) some promising advances in device architecture have been made 419. Williams. surements from QD device structures show an EQY or an IQY with [27] J. Pijpers et al. J. Phys.C. Yu. Blackburn. Miller. 2002.I. In 2007 he was appointed the Scientific Director of the 99GO10337 to NREL. 92 (2004) 186601. Colorado School of Mines. Appl. We Dr. J. Phys. 2008 Eni Award and the 2002 Research Award of the Electrochemical Society. Chemla. Nozik. 92 (2002) 4117. J. Phys. 100 (1996) 13061. Lett. J.J. chaired and organized many international and national conferences. Semicond. V.. P. and symposia. Lett. J. Phys. Nozik has been a Senior Editor of The Journal of Physical Chemistry from 1993 to [18] D. K. Solid State 14 (1972) 481. Science 310 (2005) 86. 51 (1980) 2158. Phys. and the publications they co-authored and cited here. Appl. [51] J.5 eV) that may induce the involvement of surface species [47] D. Phys. A. Appl. Appl. H. 7 (2007) 2506. including multiple exciton generation from a single photon. M. [64] M. Phys. Klimov. Jeong.. B 76 (2007) 081304(R). G. sion. Dr.P. Wehrenberg.A. J. 200 papers and book chapters in these fields. J. Petruska. Lett. Shchetinin. Ellingson. Controversial aspects of MEG [33] J. A. Nozik. Phys. Phys.H. J.H. D: Appl. Rev. S. in press.S. Many research groups are working on this problem and [28] W. 8 (1974) 192.L. Am. new Center for Revolutionary Solar Photoconversion under the Colorado Renewable Energy Collaboratory. Phys. 32 (1961) 510. Appl. Phys. 92 (2002) 1668. Kolodinski. Phys. Bridge Printery. He has served on numerous scientific review and advisory panels.V. [12] R. Phys. multiplication in CdSe and CdTe QDs reported in Ref. [31] P.L. [37] R. Nozik is a [20] A. Lett. Green. [42] R. . Inc).C. Shockley. 85 (1999) 8186. 1 (2005) 189. a QY > 1. P. J. Phys. Nozik. R.L.. Nat. DOE National Renewable Energy Laboratory (NREL) of Science. Annu. Williams. H. University of Delaware. and received several awards in solar energy research. and Professor Adjoint in the Department of Chemistry and Biochemistry at the University of Colorado. Baryshev. Phys. Efros. private communication. Nair.J. written or edited 5 books. Weinheim.J.L. Nozik. Murphy. D. Division of Chemical Sci. Tsybeskov (Eds. Phys. another recently reported result [62] that has raised [39] J. Klimov. J. Nozik. Y. V. Sykora. C. Yu. Spirkl. J. tons with energy <2Eg. Pingrong Yu. Rev. J. [17] R. optical. [55] R. It appears from these reports [54] A. Sykora. 100 (2006) 074510. Chem. D. Efros. Williams.I. References Nozik received his BChE from Cornell University in 1959 and his PhD in Physical Chemistry from Yale University [1] A.M. New York. Memming. 2001. Sol. In Ref. Green. J. J. M. Thesis. J.S. L. 128 (2006) 3241. Queisser. Nozik.E. (2008). [49] B. Randy Ellingson. R. Trupke. Nozik. M. Department of Energy. in: A. Nozik. Appl. 52 (2001) 193. Phys. J. This review is based on the work of the NREL team (Matt Beard. R. Green.J. DEAC36. A. Ellingson et al. photoelectrochemical energy conversion. [61] J. J. [34] H. [45] M. tons (>5. p. 2405. Sol.J.T.I. H. V. V.I. Soc. Efros.H. Trans. [35] D. Nozik. magnetic and [10] N.J. Phys. Mark Hanna. 79 (1996) 2473. [60. experiments with CdSe and CdTe QDs require rather energetic pho. Am. Guyot-Sionnest. Phys. J. V. J. J.61] that may allow a definitive determination of whether the [30] A. H. son. (2008). 87 (2005) 253102. Franceschetti. Zunger. Würfel. Phys. Schaller.B. Rev. Ma. [38] R.C. and the Materials Research Society. Perkins. 5 (2005) 865. Phys.J. As a result.J. D. Nozik’s scientific career appeared in the December 21. Chem..M. Bawendi. Kharionovskii. Desoete. B 110 (2006) 25455.A. V. Acknowledgments [59] P. Schaller. for the Advancement of Science.I. J.R. 125 (2003) 7806. Jung.I. including the [16] F. A Special Festschrift Issue of The Journal of Physical Chemistry honoring Dr. Jim Murphy. Q. July 23. T. F. Pietryga. I.S. J. Philos. J. [42] cannot [41] R.D. Queisser. Trupke.S.T. to date no photocurrent mea..P.L. Murray. A. M. Kluwer Academic/Plenum results with their same InAs QDs [63. Nano Lett. the authors find no change in the PL de. B 61 (2000) 4485. Wolf. 8. Pijpers et al.B. then known as the [2] R. in 1967. Phys.D. Appl. 83 (1998) 4213. Science 300 (2003) 1277.L. [24] T.D. Phys. Nozik. J. M. Phys.T. Diener.C. A. Energy Mater. and Ref. [56] A. D. Chem.E. Nozik / Chemical Physics Letters 457 (2008) 3–11 11 QY (IQY)). Chem. Boudreaux. Nozik’s research interests include [5] J. [57] A.). [32] J. Boulder. 6 (2006) 2856. H. Luther. cult to obtain accurately. Am. Ross.G. M. 7 (2007) 3469. C. Fellow of the American Physical Society and a Fellow of the American Association [21] A. Polisski. Sol. 35 (2002) 1236. pend upon their specific surface chemistry. cussed above).J. Ries. photocurrent extracted from optoelectronic QD devices (as dis. [29] R. V. Chem. 123 (2005) 074709. 2007. 47 (1976) 689. Nano Lett. Phys. D. Solids 24 (1962) 1049. B 107 (2003) 102. Guyot-Sionnest. Nano Lett. A.J.J. patents. Klimov. Soc. Nano Lett. electrical properties of solids. Appl. H.