Electro-Optical Properties of Quantum Dots with Copper Impurities
- Author(s): Fuhr, Addis;
- Advisor(s): Sautet, Philippe;
- et al.
In this work, I focus on understanding the single exciton absorption/emission mechanisms, multicarrier interactions, and charge transport mechanisms for quantum dots (QDs) containing Cu cations with a particular emphasis on CuxIn2-xSeyS2-y (CISeS) QDs. Specifically, I combine theory with experiment to determine the origin of their large Stokes shifts, broad absorption and emission spectral linewidths, long radiative lifetimes, reveal their multicarrier (e.g. trions and biexciton) interactions, and charge transport mechanisms. My theoretical toolkit predominantly utilizes density functional theory (DFT) calculations in which I predict that most of the physical properties in these systems are due to Cu impurities such as native defects (e.g. anti-site defects), or extrinsic dopants (e.g. substitution of Zn2+ cations in ZnSe with Cu1+, or Cu2+ cations). Each of these predictions are then experimentally tested using a combination of ultra-fast spectroscopy, magneto-optical spectroscopy, single-particle spectroscopy, and in-situ spectral electrochemistry measurements. And, in some cases, device studies are compared to theoretical/experimental predictions to further confirm my findings.
My studies conclude that the single- and multi-exciton absorption and emission mechanisms are dominated by three basic pathways: band-edge, Cu1+ defect, and Cu2+ defect excitation and recombination. Band-edge optical processes (both single and multi-exciton) for CISeS QDs match the general trends predicted for “typical” II-VI QDs such as sharp absorption, narrow emission, fast radiative lifetimes and Auger decay, and adherence to “universal volume scaling” relationships. The origin of these effects are small subensembles of CISeS QDs in experimental batches that have no defects, and correspondingly can be described by delocalized (valence band-to-conduction band) excitonic interactions. Cu1+ defects stem from anti-site swapping of Cu1+ and In3+ atoms into charge-balanced CuIn’’ + InCu defect pairs, and result in intra-gap “occupied” CuIn’’ defect states with the [Ar]3d10 electron configuration. In order to emit, these defects need to localize a hole either by trapping valence band (VB) carriers, or direct intra-gap excitation. This leads to a small nuclear reorganization and Franck-Condon shift prior to emission referred to as the “real Stokes shift,” or the energy difference between absorptive CuIn’’ defects and emissive Cu2+ defects. These Cu2+ defects have a hole in their electronic configuration ([Ar]3d9), which are thereby expected to lead to strong Jahn-Teller distortions, and shift the energy of the Cux impurity state (where x=1+ or 2+). This new, emissive Cu2+ state also has an “apparent Stokes shift” defined by the energy difference between Cu2+ emission and band-edge absorption. Cu2+ defects can also occur in the ground-state by charge-compensation of copper vacancies in VCu’ + CuCu defect pairs. In this case, the Cu2+, or CuCu impurities are “emission ready,” do not require localization of VB holes, and only the apparent Stokes shift is observed. Instead, VB holes are removed by non-emissive traps (e.g. dangling bonds), which block the faster band edge transition. However, considering that the hole removal rate for QDs with CuCu defects occurs at a slower rate than hole localization for QDs with CuIn’’ defects, or biexciton decay for defect-free QDs, at the multi-exciton level QDs with CuCu defects exhibit Auger dynamics that are indistinguishable from defect-free QDs.
For each of these described single exciton and multiiexciton interactions, ensemble spectral measurements usually represent the average of all of the described subensembles. However, subensembles with Cu2+ defects are more prevalent in highly Cu-deficient QDs due to an excess of Cu vacancies, or when the Fermi-level is below the Cux state. On the other hand, Cu1+ defects are more common in subensembles where the Fermi-level is above the Cux state, or in (near)-stoichiometric, or Cu-rich conditions where the concentration of Cu1+ and In3+ cations are comparable. As expected by this description, the relative population of each of these subensembles can thereby be controlled by altering chemical process conditions with the clearest process control parameters focusing on either Fermi-level modulation, or Cu:In ratios. In addition to each of these effects, regardless of the oxidation state of Cux impurities, their local chemical bonding environment significantly alters their ground state energy. This is due to the high covalency of Cu-X (where X=S, or Se) bonds in which local variations in electrostatics and bond geometry alter their crystal field splitting energy. Hence, spectral linewidths at the single particle level are intrinsically narrow, but ensemble spectra is broadened by the lack of positional control of Cux defects during QD synthesis. This leads to large variations in the emission energy at the single particle level, which average to the broad spectra observed in ensemble measurements. This finding indicates that while electron-phonon coupling may be stronger in these systems than “typical” II-VI structures, there are no fundamental limits to achieving narrow ensemble spectra. Hence, if synthesis, or device fabrication conditions are altered to control heterogeneity, narrow ensemble spectra can be achieved.
Finally, I conclude that Cux defects lead to “self-doped” structures, and correspondingly p-type transport in QD films with inert substrates (e.g. Au). However, In-related defects (e.g. InCu) lead to n-type transport, and heating in the presence of In substrates can alter the charge polarity from p- to n-type. Regardless of the charge polarity, much like the optical properties for CISeS QDs, charge transport can be described by delocalized band-edge carriers and localized impurity carriers. Delocalized carriers lead to “high mobility” states whereas localized carriers lead to “low mobility states.” For charge transport, delocalized/localized pathways are thermally coupled, and higher conductivity films can be realized by minimizing the energy separation between impurity and band-edge states, which allows for carriers to be thermally promoted from low-to-high mobility states, and increases the contribution of high mobility states to transport. In this case, the Cu/In ratio is of less importance than the Se/S ratio where a higher Se concentration shifts the band-edges closer to the impurity states, and leads to stronger thermal coupling between the low and high mobility states.