Controlling opto-electronic characteristics of ternary II–VI alloyed quantum dots: alcohol processing assay

For their ultra-wide color gamut, high efficiency, robustness, and solution processability, Cd-based alloy semiconductor quantum dots (AQDs) continue to proliferate by driving innovations in the fields of optoelectronics, photovoltaics, multiplex bio-imaging, and cancer research. Herein, non-toxic, low-cost isopropyl alcohol vapor-based oxidative treatment protocol is developed and applied to tune the light emission spectrum of crystalline core–shell CdSe1−xSx/ZnS quantum dots. As evidenced by the results of structural investigations, these AQDs when exposed to vapors produced ultrasonically from 10:1 isopropyl alcohol-to-water mix undergo an isotropic, diameter non-specific size reduction at the rate of ∼1.3 Å min−1. Nonlinear time-dependent spectral shifts, revealed experimentally, are consistent with the results of the effective-mass approximation treatment. The emission yields are seen to undergo an initial drop, yet to plateau as the etch time increases. The study opens a door to a soft, top-down monotonic tailoring of the light emission characteristics and opto-electronic response of stoichiometrically- and hierarchically-complex core–shell constructs in technologically-viable group II–VI nano-semiconductors as well as AQD-based catalytic conversion of organic compounds.


Introduction
Semiconductor quantum dots (QDs), also known as artificial atoms, are quasi-zero-dimensional physical systems with an ultimate degree of charge carrier and exciton confinement and highly refined, size-tunable optoelectronic response. A recent progress pertaining to the controlled synthesis of colloidal binary core-shell QDs and the development of diverse surface-capping strategies have helped propel technological innovations across many fields including opto-electronics, bio-imaging, cancer research, and quantum communication to new heights [1][2][3][4][5][6][7]. In contrast to their elemental and binary counterparts [8], ternary semiconductor QDs and nanocrystals offer an additional degree of tunability of their light emission spectrum. These alloyed quantum dots (AQDs) possess many desirable characteristics such as increased external electroluminescent yield [9] as well as photovoltaic efficiency [10] that make them sought after for application in QD-based photo-detecting, energy harvesting, and light emitting devices [11][12][13][14].
Group II-VI semiconductor crystals and their alloy counterparts possess a range of key technologicallyrelevant opto-electronic responses and were a subject of detailed investigations before [15][16][17][18]. Among different ternary alloys, nanocrystal quantum dots in CdSeTe and CdSeS were studies most intensively as their band-edge emissions fall in the visible part of the electromagnetic spectrum [19]. While a hot-injection method remains a primary route to producing binary colloidal AQDs, other variants including diffusion based post-growth modification and non-injection one-pot approaches have been additionally developed to synthesize substratefree AQDs [20,21].
Yet, many challenges lie ahead and are stipulated by the inability to produce mono-disperse, size-and shapespecific AQDs while simultaneously maintaining their core stoichiometry. The problem is in-part associated Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
with overall fast, yet dissimilar reaction rates of chalcogenide precursors [22]. In addition, CdSeS spectral emission range can be practically tuned within a narrow window of ∼125 nm.
To cover the entire visible range, i.e. ∼400-650 nm, the alloy content has to be tuned in conjunction with the quantum dot physical size [23]. Unlike bulk semiconductors, whose dimensions can be altered using both additive and subtractive methods, owing to emission quenching by the surface defects/traps the top-down processing routes have to be ruled out for nano-semiconductors. Moreover, due to Ostwald ripening, synthesis of colloidal QDs with the size in a low nm-range is not feasible. Combined, this necessitates a search for and development of complimentary, yet soft and relatively fast etching protocols to tailor the size and, in turn, the opto-electronic response of stoichiometrically-complex nano-semiconductors including AQDs.
On this front, the dissolution of simpler, i.e. CdSe colloidal quantum dots were reported earlier in 3-amino-1-propanol/water mixtures [20] and in organic solvents by controlling precipitation-dissolution dynamics with surface stabilizing ligands [21]. The amine-assisted etching of CdSe remains very slow and time non-monotonic yielding highly anisotropic, pyramidally-shaped CdSe nanocrystals with Cd-terminated facets. [38] Faster etch rate, on the order of ∼0.1 A min −1 were achieved with oleic acid [24]. Yet, the dissolution of CdSe was reported to be strongly size dependent with the size 'defocusing' and spectral broadening reported among the key drawbacks. As another drawback of amine-assisted etching, the temperature of the dissolution-growth reversal is to vary with the diameter which results in dissimilar QD size populations and undesirable color-splitting effects.
In this work, we develop and report a facile, i.e. one-step isotropic size-tuning of CdSeS/ZnS core-shell AQDs using an isopropyl alcohol/water based etching protocol. In contrast to other reported studies, the treatment is done in a vapor phase which obviates the need for AQD precipitation and re-suspension in the etch solutions. The vapor-treated AQD specimen are confirmed to retain their shape anisotropy and alloy content while remaining luminescent. The emission spectra are further found to exhibit a non-linear dependence on the etch time and the effects are successfully modeled/accounted for using the effective mass approximation framework as discussed below.

Processing methods
Vapor-based treatment Highly-luminescent, ∼6-8 nm diameter, oleic acid-capped CdSe 0.83 S 0.17 /ZnS core-shell AQDs suspended in toluene (∼1 mg ml −1 ) were obtained commercially (Sigma Aldrich). Their morphology, crystal structure and atomic compositions were obtained and the results are given below. As a first step, a 1 μl aliquot of AQDs is drop-cast onto a glass slide. Once toluene evaporated (∼3-4 min) the slide was fitted vertically into a vial containing ∼25 ml of C 3 H 8 O:H 2 O solution (10:1) with AQDs located ∼1 cm above the liquid/air interface.
The solution was agitated and, in parallel, heated ultrasonically (Misonix 3000 with horn attachment, ∼9 W power). The QD samples were exposed to the etchant mist/vapors using ∼3 min time intervals for a total time of 15 min. At the end of each cycle, the slide was taken out to assess the spectral characteristics of QDs.
The emission spectrum of AQDs treated with ∼10:1 isopropyl alcohol to DI water mixture exhibited a consistent blue-shift with the etch time, figure 1. Yet, treatments that were carried out similarly but only with either pure DI water or (∼99.9%) C 3 H 8 O yielded no discernible change of the light emission characteristics. Longer exposures (>15 min) to ∼99.9% pure C 3 H 8 O resulted in a noticeable reduction of the luminescence intensity.
Liquid phase treatment Alternatively, the treatments were carried out in a liquid phase by drop-casting the same alcohol water solution directly onto AQDs whose temperature was maintained at ∼60°C with an external electrical (Joule) heater.
While the spectral trends (color shifts) were similar, the alternative pathway is found to be an order of magnitude slower compared to the sonication method. We attribute this to an increase in the oxygen levels and net higher local temperatures, the latter are likely to aid oleic acid cap removal thus speeding up the processing. A monotonic time-dependent blue-shift of the emission color of the AQDs is a primary indicator of the AQDs undergoing etching in the presence of the alcohol-water vapors. To verify the mechanism and to identify the chemical pathway, we carried out a set of the structural and optical property characterizations of the post processed AQDs, the results of which are presented and discussed next.

Results and analysis
We relied on HITACHI H-9000NAR High Resolution Transmission Electron Microscope (HRTEM) to assess the effect of the alcohol etching on the crystal order, morphology, size distribution, and aspect ratio of the AQDs comparatively, i.e. before and after the treatment. Figure 3 shows representative HRTEM images of as-received AQDs and the samples processed for 15 min. The electron diffraction patterns were additionally obtained and the results confirmed that the crystal structure of AQDs was similar to that of wurtzite bulk CdSe (supplementary materials is available online at stacks.iop.org/MRX/7/075008/mmedia), though the former possess slightly reduced lattice constants: a=4.12±0.32 and c=6.72±0.04 Å.
HRTEM images of larger ensembles of AQDs were collected to quantify the impact of the etch time on the diameter and aspect ratio distributions. According to figure 2, for the etch time of ∼15 min the major diameter shrank by ∼2.0 nm whereas the diameter aspect ratio changed by −4.6% , i.e. from 1.30 to 1.24, thus confirming that the alcohol etch rate remains isotropic.
The light emission characteristics of AQD samples were assessed next by carrying out room temperature cwphotoluminescence (PL) spectroscopic tests. The samples were excited with light generated by a 450 W Xenon lamp (excitation wavelength of ∼365 nm; optical power density of ∼5 mW cm −2 ) and the emission spectra were collected with an Olympus 51X microscope and dispersed by another double-grating monochromator onto a  photomultiplier tube operating in a photon counting mode and serving as a detector. The instrument resolution was limited to 2 nm and the spectra were acquired in the range of ∼400-680 nm.
The emission spectra of the processed AQDs were obtained by Gaussian decomposition/fitting that helped identify and remove background PL that consisted of two static PL bands with peaks centered at ∼500 and 556 nm (supplementary materials) and attributed to the etch reaction by-products. The stand-alone individual narrow-width bands seen in figure 4(a) are excitonic emission bands of the AQDs. These AQD bands exhibit characteristic and very consistent blue shifting and slightly broaden as etch time increases. While initially the integral PL intensity decreases, it quickly stabilizes and remains almost unchanged for the treatment durations of up to 15 min, figure 4(b).
The observed initial drop of the quantum yield (QY) is generally expected as even a partial removal of the ZnS shell is to result in re-activation and increase of the ratio of the surface defects that act as non-radiative recombination centers [19,25]. Assuming a typical shell thickness on the order of a monolayer, i.e. ∼3.1 Å for optimized core-shell QD configurations [25,26] and the mean etch rate of ∼0.7 Å min −1 , etch time of ∼4 min should be sufficient to strip ZnS. The removal of the shell is to coincide with the onset of the non-radiative recombinations-a primary cause of the luminescence loss for semiconductor nano-crystals. In our case, the intensity loss is already evident at ∼3 min, figure 4(b) and attributed to a larger etch rate of the defect-prone ZnS shell. Furthermore, as the QD diameter reduces, the QY is to drop due to a rise of the surface-to-volume ratio.
Surface defects that act as PL quenches do play due to a large surface-to-volume ratio a more pronounced role in small diameter QDs, but only in uncapped ones. To our surprise and contrary to prior literature reports, the PL yields remain high for the treated samples as well. This is most likely due to the fact that etching recipe remains mild.
In contrast, the QY remained almost time-independent for etch durations >3 min, figure 4(b), in part due to an increase in the oscillator strength of the excitonic transitions in smaller size 0D systems.
The time-color interplay (figure 1) is next analyzed for bare AQDs (etch time >3 min) within the framework of effective mass approximation (EMA) [27,28]. The band gap energy of bulk alloyed CdSe 1−x S x can be obtained based on the following equation, the confinement effects prevail and are to control spectral shifts. In this regime, the photon emission energy of the AQDs is obtained by treating Coulomb e-h interactions perturbatively, with the emission energy given by where the second and third terms stand for the electron and hole confinement and Coulomb electrostatic potential energies, respectively. Approximating the electron and hole effective masses with their bulk values for CdSe: m e =0.13m 0 and m h =0.45m 0 , and using  r =6 for Cd 1 Se 0.83 S 0.17 [30], we next obtain and plot the emission energy versus radius in figure 5. According to the results presented in figure 2, the size distribution, and, more specifically, diameter broadening is to vary only slightly with etching. Furthermore, as the temperature and, in turn, the net etch rate remain constant throughout the treatments, the radii of all AQDs are expected to shrink linearly with time. Given that the confinement energy prevails over the Coulomb energy (the last term in the above equation), both the spectral shift and the excitonic peak broadening are predicted to exhibit highly nonlinear time evolutions, which are readily evident in figure 5, the inset. This further serves as a confirmation that the light emission/ spectrum shifts are primarily controlled by the energy quantization effects in our AQDs with a mean diameter of up to ∼7 nm. It should be noted that the results additionally point to a very negligible, if any, role of the radial alloy content gradient that can be present and induced by unbalanced precursor reactivities within the AQD cores [19,21,22].
The two background, i.e. ∼500 and 556 nm PL peaks revealed by the Gaussian spectral decomposition only appear in the processed samples and therefore are attributed to the reaction by-products. The intensity and width of the 500 nm peak remain constant for all treatment durations, whereas the intensity of the 556 nm peak slightly increased with the processing time. The bands are consistent and assigned to a defect-associated luminescence originating from ZnO, CdS and CdO surface defect states, respectively. More specifically, the 500 nm emission involves oxygen vacancies of ZnO by-product. These oxygen vacancies are known to form deep-donor recombination states within the band gap of ZnO. The origin of the 556 nm band is traced to the surface defect states of CdS and CdO [31][32][33][34].
Oxidative degradation and etching of CdSe QDs could be linked to a molecular oxygen. Albeit possible, this pathway is to yield SeO , 2 SO 2 and, more importantly, CdO as a surface capping layer. The latter is however to arrest the etch process, in contrast to what was observed by us experimentally. Since treatments with the water alone were equally ineffective, AQD etching pathways that involve water most likely are thermodynamically unfavorable [35].
Surprisingly, the AQD alcohol-based etching didn't proceed without water either. Due to an increased presence of defects, surface of Se-rich CdSeS is to carry a partial positive charge [36]. This is to render the AQD surface hydrophilic and to aid the surface absorption of water and, in turn, alcohol molecules from the mist. According to prior work [37], a surface-supplied positive charge, i.e. hole, h + is integral to photo-oxidation/ conversion of isopropyl alcohol into acetone which is accompanied by the formation of hydrogen radicals on CdS surface. Owing to a large surface-to-volume ratio and superior light absorption characteristics of spherically-shaped QDs, the latter mechanism can be particularly beneficial to both catalytic conversion of various organic compounds as well as etching of CdS nano-structures.
Depicted below are the key surface reactions that are likely, among others, to control the isopropyl alcohol etching of the core-shell AQDs in our case:  In the above scheme, e − and h + are either thermally or photo-generated ZnS-and CdS-supplied charges. The role of water is thus limited to assisting with (a) absorption of CH CHOH Cd , 2 and -SO 4 2 etch products. The metal cations are to eventually react with the oxygen and to precipitate as solid ZnO and CdO. It is worth noting that many secondary reactions such as those involving, for instance, S, Se and H while not shown, are not to be ruled out.

Summary
One-step, highly non-toxic and isotropic isopropyl-water based vapor etching protocol was developed and applied to continuously down-tune the size of highly heterogeneous CdSe 1−x S x /ZnS core-shell QDs. In contrast to many other, prior-developed top-down semiconductor processing routes, the composition and shape anisotropy of AQDs remain unaffected by the etching. The water is argued to play an important but mostly secondary role: a dissolution and removal of the reaction intermediates/by-products which will otherwise impede the etching. To restore emission yields, over-growth of ZnS or other similar cladding layer should be considered as a final post-etch step. The protocol can be used for precise tailoring of opto-electronic characteristics of group II-VI core-shell nano-semiconductors as well as AQD-based catalytic conversion of organic compounds.