Mixed Mercaptocarboxylic Acid Shells Provide Stable Dispersions of InPZnS/ZnSe/ZnS Multishell Quantum Dots in Aqueous Media

Highly luminescent indium phosphide zinc sulfide (InPZnS) quantum dots (QDs), with zinc selenide/zinc sulfide (ZnSe/ZnS) shells, were synthesized. The QDs were modified via a post-synthetic ligand exchange reaction with 3-mercaptopropionic acid (MPA) and 11-mercaptoundecanoic acid (MUA) in different MPA:MUA ratios, making this study the first investigation into the effects of mixed ligand shells on InPZnS QDs. Moreover, this article also describes an optimized method for the correlation of the QD size vs. optical absorption of the QDs. Upon ligand exchange, the QDs can be dispersed in water. Longer ligands (MUA) provide more stable dispersions than short-chain ligands. Thicker ZnSe/ZnS shells provide a better photoluminescence quantum yield (PLQY) and higher emission stability upon ligand exchange. Both the ligand exchange and the optical properties are highly reproducible between different QD batches. Before dialysis, QDs with a ZnS shell thickness of ~4.9 monolayers (ML), stabilized with a mixed MPA:MUA (mixing ratio of 1:10), showed the highest PLQY, at ~45%. After dialysis, QDs with a ZnS shell thickness of ~4.9 ML, stabilized with a mixed MPA:MUA and a ratio of 1:10 and 1:100, showed the highest PLQYs, of ~41%. The dispersions were stable up to 44 days at ambient conditions and in the dark. After 44 days, QDs with a ZnS shell thickness of ~4.9 ML, stabilized with only MUA, showed the highest PLQY, of ~34%.


Introduction
Quantum dots (QDs) have attracted tremendous attention over the last 30 years. This interest stems from their captivating size-dependent optical properties caused by the quantum confinement of the electron-hole pair [1][2][3]. QDs show high photoluminescence quantum yields (PLQYs), small full width at half maximum (FWHM), and high photostability. Absorption and emission properties can be tuned via the QD size while maintaining the same chemical composition [4][5][6][7][8]. Due to these exceptional optical properties, QDs have found application in many fields, such as displays and solar cells [9][10][11][12].

Precursors
Tributylphosphine selenide (TBPSe): 3.16 g (40.00 mmol) of Se were weighed into a Schlenk flask with a neck and cock, Teflon stir bar, and septum. After Se addition, the flask was slowly evacuated and filled with argon. This process was repeated three times. Then, the flask was heated with a heat gun (max. 120 • C) under a vacuum for 10 min to remove residual water. Special care was taken to ensure that the Se (melting point (mp)~220 • C) did not melt. The flask was cooled to room temperature and flooded with argon.
Then, 11.05 mL (44.8 mmol) TBP was added under argon flow and continuous stirring. The reaction mixture was stirred for approx. 12 h at room temperature. The dissolution of the Se indicated a complete reaction. After the Se was dissolved completely, the solution was stirred for an additional 24 h at room temperature.
Trioctylphosphine sulfide (TOPS): 3.85 g (0.12 mmol) of S was weighed into a Schlenk flask with neck and cock, Teflon stir bar, and septum. After S addition, the flask was slowly evacuated and filled with argon. This process was repeated three times. The flask was heated with a heat gun (max. 95 • C) under a vacuum for 10 min to remove residual water. Special care was taken to ensure that the S did not melt (mp~115 • C). The vessel with the reactant was cooled to room temperature and flooded with argon.
60.00 mL (134.5 mmol) trioctylphosphine (TOP) was added under argon flow and continuous stirring. The reaction mixture was stirred for approx. 12 h at room temperature. The dissolution of the S indicated a complete reaction. After the S was dissolved completely, the solution was stirred for an additional 24 h at room temperature.
Zinc octanoate (Zn(Oct) 2 ): 8.14 g (0.10 mol) of zinc oxide and 28.84 g (0.20 mol) of octanoic acid were weighed into a round bottom flask with a Teflon stir bar and a reflux condenser. 200 mL of toluene was added and the reaction mixture was heated to reflux and kept for four hours. The reaction mixture was cooled to room temperature after the solution became clear and the toluene evaporated using rotary evaporation. The solid Zn(Oct) 2 was dried overnight in a vacuum oven.

InPZnS Multi Shell QDs
InPZnS hybrid cores were synthesized via a protocol adapted from Reference [41]. In(Ac) 3 , zinc stearate (Zn(St) 2 ) or zinc octanoate (Zn(Oct) 2 ), and a stir bar were added to a three-neck-flask Nanomaterials 2020, 10, 1858 4 of 24 with a septum, reflux condenser connected with the vacuum line, and a thermometer. The mixture was heated to 150 • C under vacuum and stirred for 30 min. Then, the temperature was reduced to 120 • C under argon flow. 1-dodecanethiol and HOAc were injected and the mixture was stirred for 5 min. Subsequently, the temperature was reduced to 100 • C and P(TMSi) 3 solution (1 M in ODE) was rapidly injected into the hot reaction mixture. The mixture was stirred for 10 min., heated to 300 • C, and kept at that temperature for up to 30 min. Then, the reaction mixture was cooled to room temperature ( Table 1 shows the amounts of material used for sample preparation). The reaction mixtures of particles A to D were directly used for the synthesis of the two shells onto the QDs, as described below. Since samples E to J were used for high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) imaging, they were purified using the purification protocol described below. Table 1. Amounts of In(Ac) 3 , Zn(St) 2 , and Zn(Oct) 2 and volumes of 1-dodecanethiol, HOAc, and P(TMSi) 3 (1 M in ODE) of different QD-samples used for synthesis of the InPZnS hybrid cores.

Sample Name
In(Ac) 3  To synthesize the ZnSe shell, Zn(St) 2 was added to the reaction mixture. The mixture was heated to 100 • C and kept at that temperature until the Zn(St) 2 was melted. TBPSe (2 M in TBP) was added, the mixture was heated to 285 • C, and kept at that temperature for 6 min. Afterwards, the reaction mixture was cooled to room temperature (Table 2). To synthesize the ZnS shell on top of the ZnSe shell, Zn(St) 2 was added to the reaction mixture. The mixture was heated to 100 • C and kept at that temperature until the Zn(St) 2 melted. TOPS (2 M in TOP) was added, the mixture was heated to 250 • C, and was kept for 6 min. Afterwards, the reaction mixture was cooled to room temperature. This reaction step was adapted to generate different outer shell thicknesses (Table 3) and was performed one time for particles A, two times for particles B, three times for particles C, and five times for particles D. Step 1 Step 2 Step 3 Step 4 Step 5 For purification, 40 mL of acetone was added to the reaction mixture to remove organic impurities. Upon acetone addition, the QDs and excess ligand (stearate) precipitated. The mixture was centrifuged at 4500 rpm for 3 min and the supernatant was discarded. This procedure was repeated until the supernatant was colorless.
The precipitate was then washed with hexane extraction to remove the remaining excess ligand. The solid was mixed with 3 mL of hexane in a Falcon tube and centrifuged at 8000 rpm for 3 min. Afterwards the QDs remained in the hexane phase and the stearate ligands precipitated because of low solubility in hexane. The QD/hexane mixture was added to 40 mL of acetone and centrifuged at 8000 rpm for 3 min. The QDs precipitated and the colorless supernatant was discarded. The hexane extraction was repeated until the residue was nearly colorless.

Ligand Exchange
Ligand exchange with MCAs was done via a protocol adapted from Reference [28]. 20 mg of the QDs were dispersed in 2 mL of hexane in a Falcon tube. In a second Falcon tube, 1.9 mL of water and 1.15 mmol of MPA or/and MUA (MPA/MUA ratios see Table 4) were mixed. The QD dispersion was added to the MCA-solution. Then, 50 µL of a 1 M aqueous NaOH solution was added, and the mixture was vortexed and centrifuged at 3000 rpm for 2 min. This step was repeated until all QDs were transferred into the aqueous phase (as indicated by the color of the liquid phases). Figure 1 summarizes the entire synthesis and phase transfer process.  Purification after ligand exchange was achieved as follows. The aqueous QD dispersion was removed with a syringe and passed through a polyamide syringe filter (0.2 µm) into another Falcon tube. This aqueous QD dispersion was transferred to a dialysis cassette (Fisher Scientific Mw cutoff ~10,000 g/mol, volume 3 mL) and dialyzed against d.i. water.
The water was changed three times (after 1, 3, and 24 h). After the final water exchange dialysis was continued for 3 h. The aqueous QD dispersion was withdrawn from the cassette using a syringe and filled into a Falcon tube through a syringe filter (PA, pore size 0.2 µm). These QDs were stored in dispersion at room temperature in the dark. For the thermogravimetric analysis, some material was dried in a drying cabinet at 50 °C under vacuum until only the QD powder remained.

Measurements
UV-Vis spectra were recorded on a Perkin Elmer Lambda 19 spectrometer (Perkin Elmer, Waltham, MA, USA) from 280 to 850 nm with a resolution of 1 nm. The QDs were dispersed in toluene before and in ultrapure water after ligand exchange. All the experiments were done using quartz cuvettes with a 10 mm path length.
Photoluminescence spectra and absolute quantum yields were recorded on a Hamamatsu C9920-02 (Hamamatsu Photonics, Hamamatsu City, Japan). The QDs were dispersed in toluene before the ligand exchange and in ultrapure water after the ligand exchange. The dispersions were diluted until they showed an absorbance of 0.1 at 350 nm. Spectra were recorded from 300 to 950 nm with a resolution of <2 nm with an excitation wavelength of λexc = 350 nm (built in 150 W Xenon light source) Thermogravimetric analysis (TGA) was performed on a Netzsch TG 209 F1 Iris thermomicrobalance (Netzsch-Gerätebau GmbH, Selb, Germany) and a TA Instruments Q500 (TA Instruments, Inc., Newcastle, DE, USA) from 25 °C to 550 °C at a heating rate of 10 K/min under nitrogen flow in an aluminum crucible. The QD samples were prepared by placing a drop of the QD in aqueous dispersion in the crucible and drying at 50 °C under a vacuum prior to the measurements.
Zeta potential was determined with a Zetasizer Nano ZS (Malvern Instruments, Malvern, UK) using a disposable capillary cell (DTS1070) and an equilibration time of 20 s. For all the measurements, 10 µL of the aqueous QD dispersion, obtained after dialysis, were diluted by a factor of 10 and directly measured afterwards. Purification after ligand exchange was achieved as follows. The aqueous QD dispersion was removed with a syringe and passed through a polyamide syringe filter (0.2 µm) into another Falcon tube. This aqueous QD dispersion was transferred to a dialysis cassette (Fisher Scientific M w cutof 10,000 g/mol, volume 3 mL) and dialyzed against d.i. water. The water was changed three times (after 1, 3, and 24 h). After the final water exchange dialysis was continued for 3 h. The aqueous QD dispersion was withdrawn from the cassette using a syringe and filled into a Falcon tube through a syringe filter (PA, pore size 0.2 µm). These QDs were stored in dispersion at room temperature in the dark. For the thermogravimetric analysis, some material was dried in a drying cabinet at 50 • C under vacuum until only the QD powder remained.

Measurements
UV-Vis spectra were recorded on a Perkin Elmer Lambda 19 spectrometer (Perkin Elmer, Waltham, MA, USA) from 280 to 850 nm with a resolution of 1 nm. The QDs were dispersed in toluene before and in ultrapure water after ligand exchange. All the experiments were done using quartz cuvettes with a 10 mm path length.
Photoluminescence spectra and absolute quantum yields were recorded on a Hamamatsu C9920-02 (Hamamatsu Photonics, Hamamatsu City, Japan). The QDs were dispersed in toluene before the ligand exchange and in ultrapure water after the ligand exchange. The dispersions were diluted until they showed an absorbance of 0.1 at 350 nm. Spectra were recorded from 300 to 950 nm with a resolution of <2 nm with an excitation wavelength of λ exc = 350 nm (built in 150 W Xenon light source) Thermogravimetric analysis (TGA) was performed on a Netzsch TG 209 F1 Iris thermomicrobalance (Netzsch-Gerätebau GmbH, Selb, Germany) and a TA Instruments Q500 (TA Instruments, Inc., Newcastle, DE, USA) from 25 • C to 550 • C at a heating rate of 10 K/min under nitrogen flow in an aluminum crucible. The QD samples were prepared by placing a drop of the QD in aqueous dispersion in the crucible and drying at 50 • C under a vacuum prior to the measurements.
Zeta potential was determined with a Zetasizer Nano ZS (Malvern Instruments, Malvern, UK) using a disposable capillary cell (DTS1070) and an equilibration time of 20 s. For all the measurements, Nanomaterials 2020, 10, 1858 7 of 24 10 µL of the aqueous QD dispersion, obtained after dialysis, were diluted by a factor of 10 and directly measured afterwards.
Transmission electron microscopy (TEM) images were obtained on a JEOL JEM-1400 Plus (Jeol GmbH, Freising, Germany) operated at 120 kV. The QD samples were dispersed in toluene to a concentration of 0.01 wt%. TEM specimens were prepared by depositing a drop of the dispersion on carbon-coated copper grids and were dried in the air.
High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) images were acquired using an FEI Titan Themis microscope operating at 300 kV. Specimens were prepared by depositing a drop of the QD dispersions on a carbon-coated copper grid and were dried in the air.
Powder X-ray diffraction (XRD) measurements were done on a Siemens D5000 equipped with a closed Eulerian cradle (Bruker-AXS) (Bruker, Billerica, USA)in reflection geometry using Ni-filtered Cu K α radiation (λ = 1.5418 Å). Range was 20 to 90 • 2θ and step size was 0.01 • . Specimens were prepared by depositing the dry powders on Kapton foil.

Calculation of Core Sizes and Amounts of Shell Precursor
The absorption size dependencies for the Cd-based [42] and InP-based QDs from optical absorption and XRD data have been established [43]. In the current study, we used an extended approach, which also took into account TEM data [3,43,44]. The QD core size was calculated by assuming a spherical QD shape and the resulting shell precursor quantities were used for the syntheses of the QDs in this publication.
However, TEM images showed that the QDs were smaller than those calculated from XRD and emission spectra. Therefore, six QD core samples of different absorption wavelengths, and thus core sizes, were measured by HAADF/STEM to modify the calculation of the core size via the first exciton absorption maxima. Figure 2 shows the absorption and emission spectra of particles E to J. Nanomaterials 2020, 10, x FOR PEER REVIEW 7 of 25 Transmission electron microscopy (TEM) images were obtained on a JEOL JEM-1400 Plus (Jeol GmbH, Freising, Germany) operated at 120 kV. The QD samples were dispersed in toluene to a concentration of 0.01 wt%. TEM specimens were prepared by depositing a drop of the dispersion on carbon-coated copper grids and were dried in the air.
High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) images were acquired using an FEI Titan Themis microscope operating at 300 kV. Specimens were prepared by depositing a drop of the QD dispersions on a carbon-coated copper grid and were dried in the air.
Powder X-ray diffraction (XRD) measurements were done on a Siemens D5000 equipped with a closed Eulerian cradle (Bruker-AXS) (Bruker, Billerica, USA)in reflection geometry using Ni-filtered Cu Kα radiation (λ = 1.5418 Å). Range was 20 to 90° 2θ and step size was 0.01°. Specimens were prepared by depositing the dry powders on Kapton foil.
Chemical composition of the powders was determined via inductively-coupled plasma optical emission spectroscopy (

Calculation of Core Sizes and Amounts of Shell Precursor
The absorption size dependencies for the Cd-based [42] and InP-based QDs from optical absorption and XRD data have been established [43]. In the current study, we used an extended approach, which also took into account TEM data [3,43,44]. The QD core size was calculated by assuming a spherical QD shape and the resulting shell precursor quantities were used for the syntheses of the QDs in this publication.
However, TEM images showed that the QDs were smaller than those calculated from XRD and emission spectra. Therefore, six QD core samples of different absorption wavelengths, and thus core sizes, were measured by HAADF/STEM to modify the calculation of the core size via the first exciton absorption maxima. Figure 2 shows the absorption and emission spectra of particles E to J.  For the InPZnS hybrid core system used here, an optimized method for the calculation of the QD size vs. optical absorption correlation was developed. For this purpose, cores with different sizes were evaluated using HAADF-STEM images. Figure A1a  For the InPZnS hybrid core system used here, an optimized method for the calculation of the QD size vs. optical absorption correlation was developed. For this purpose, cores with different sizes were evaluated using HAADF-STEM images. Figure A1a,b (Appendix A) shows representative HAADF/STEM images of samples E and J. Analysis of the electron diffraction pattern ( Figure A1c,d) yields interplanar spacings of 3.3 Å, 2.0 Å, and 1.7 Å, which corresponds to the (111), (220), and (222) or (311) planes of the InP zinc blende structure in sample E. In the case of sample J, the values for d(hkl) varied within 3.1-3.4 Å (111), 1.9-2.0 Å (220), and 1.6-1.7 Å (222) or (311). We attributed this ambiguity to the enhanced presence of ZnS in the larger QDs (see Tables A1-A3, Appendices B-D) [45,46].
The true particle diameters were determined using GIMP (www.gimp.org) and ImageJ (www.imagej.nih.gov), assuming a spherical QD shape. The QDs were identified by hand using GIMP and the surface areas were calculated by ImageJ for each QD marked. Furthermore, particle sizes were also determined from the corresponding maxima of first exciton absorption. The first exciton absorption peak of the QDs was determined from the minimum of the second derivative of the absorption curve [47]. Combining the sizes obtained from HAADF-STEM images and exciton peaks, a calibration curve was obtained, as shown in Figure 3 (detailed data are shown in Table A4, Appendix E).  (see Tables A1-A3, Appendixs B-D) [45,46]. The true particle diameters were determined using GIMP (www.gimp.org) and ImageJ (www.imagej.nih.gov), assuming a spherical QD shape. The QDs were identified by hand using GIMP and the surface areas were calculated by ImageJ for each QD marked. Furthermore, particle sizes were also determined from the corresponding maxima of first exciton absorption. The first exciton absorption peak of the QDs was determined from the minimum of the second derivative of the absorption curve [47]. Combining the sizes obtained from HAADF-STEM images and exciton peaks, a calibration curve was obtained, as shown in Figure 3 (detailed data are shown in Table A4, Appendix E). Equation (1) shows how the size of the InP based cores can be heuristically related to the absorption peak (red line in Figure 3). Using the lattice parameters of (bulk) zinc selenide (ZnSe) and zinc sulfide (ZnS), the thickness of one ZnS or ZnSe monolayer (ML) can be calculated. One ZnSe ML is 0.33 nm and one ZnS ML is 0.31 nm thick [48]. Assuming a 100% conversion of the starting materials, the amount of precursor needed for ZnS/ZnSe monolayer formation on the InPZnS QD core could be calculated. With the additional assumption of a uniform ML growth, the theoretical shell thicknesses can be calculated and thus the total size of the QDs can be determined. For all the QDs synthesized, the shell thicknesses were calculated from the precursor quantities used.  Equation (1) shows how the size of the InP based cores can be heuristically related to the absorption peak (red line in Figure 3). Using the lattice parameters of (bulk) zinc selenide (ZnSe) and zinc sulfide (ZnS), the thickness of one ZnS or ZnSe monolayer (ML) can be calculated. One ZnSe ML is 0.33 nm and one ZnS ML is 0.31 nm thick [48].

QD Synthesis and Characterization
Equation (1): calculation of the size of QD cores using the first exciton absorption peak. Assuming a 100% conversion of the starting materials, the amount of precursor needed for ZnS/ZnSe monolayer formation on the InPZnS QD core could be calculated. With the additional assumption of a uniform ML growth, the theoretical shell thicknesses can be calculated and thus the total size of the QDs can be determined. For all the QDs synthesized, the shell thicknesses were calculated from the precursor quantities used. Table 5 shows the composition and basic characterization data of the different QDs. The shell thicknesses in the numbers of monolayers were calculated using the position of the first exciton absorption peak and the amount of TBPSe and TOPS, assuming 100% conversion (Section 2.4).  Figure 4 shows the absorption and emission spectra of QD materials A, B, C, and D redispersed in toluene after purification. The absorption spectra show shoulders (first exciton absorption peak) at 610, 579, 575, and 582 nm, respectively. Between B, C, and D both the absorption and emission peaks show a shift of up to 7 nm from B to D, correlating with the particle size and shell thicknesses. B, C, and D also show Stokes shifts of 30 to 31 nm. A Stokes shift of 38 nm from QD material A can be observed. The PLQYs decrease with increasing emission wavelength. Samples B, C, and D show PLQYs of 60%, 52%, and 42%, respectively. In contrast, the PLQY of A is only 31%. The FWHM values show a small increase from A to D. This correlates with increasing batch size and presumably with a slight increase in size distribution (less homogeneous nucleation) as the reaction volumes increase from samples A to D.    TGA data of particles A to D, as well as Zn(St)2, are shown in Figure 5a. TGA data of particles A and B show two weight losses, similar to Zn(St)2. In sample A, the onsets of these weight losses are observed at lower temperatures (~210 and ~295 °C), as compared to Zn(St)2, while sample B (~230 and 330 °C) shows onsets in approximately the same temperature range as Zn(St)2 at ~235 and ~330 °C. TGA data of particles C and D are different in that they only show one weight loss with an onset at ~350 °C. At 550 °C the residual mass of A is 51% indicating an organic fraction (i.e., ligand and free ligand that could not be removed during washing) of approx. 49%. Particles B, C, and D have a residual mass between 60% and 61% at 550 °C, indicating an organic fraction of approx. 39% to 40%. The nanometer size of the particles is further confirmed by transmission electron microscopy (TEM). Figure A2 (Appendix F) shows representative TEM images of the same samples, C and D. TEM indicates that the particles have roughly spherical shapes and diameters of <10 nm.  The nanometer size of the particles is further confirmed by transmission electron microscopy (TEM). Figure A2 (Appendix F) shows representative TEM images of the same samples, C and D. TEM indicates that the particles have roughly spherical shapes and diameters of <10 nm. Table 6 shows the ICP OES data obtained for C and D. The atomic contents of the elements In, P, Zn, S, and Se were normalized to the atomic content of In. C and D have the same P content in relation to In. In contrast, QDs C have double the Se content, indicating the presence of the first shell, consisting of ZnSe, which may be slightly thicker than in the QDs D. However, the amount of Zn and S was significantly higher in D than in C. Overall, ICP OES showed a ca. 3.45 times more than ZnS was present in sample D, which is consistent with the precursor concentrations used in the synthesis.

Phase Transfer of QDs with Different Shell Thicknesses
The following section describes the influence of the shell thicknesses on stability during ligand exchange. Figure 6a shows the absolute PLQYs, Figure 6b shows the ratios of the PLQYs of the QDs dispersed in water vs. a reference sample in toluene, Figure 6c shows the emission peak wavelengths, and Figure 6d shows the FWHMs of samples A, B, C and D before and after ligand exchange with MPA and MUA (data from Table A5, Appendix G).
Upon ligand exchange, a PLQY loss was observed for each shell configuration. The QDs coated with MPA show higher PLQY losses than QDs coated with MUA. The highest PLQY losses of 33% and 34%, respectively, were observed for samples B and C after ligand exchange with MPA. D QDs coated with MPA showed a smaller reduction and A had the smallest reduction of the PLQY with 19% and 13%, respectively. The MUA coated QDs B and C showed the highest absolute PLQYs after ligand exchange (45% and 44%). The MUA coated sample D had an absolute PLQY of 39%. The PLQY losses of 15%, 8%, and 4% for B, C and D, stabilized with MUA, were significantly lower compared to QDs coated with MPA ( Figure 6a). Overall, when considering the individual PLQYs normalized to the reference PLQY, the particles coated with MUA showed that thicker shells lead to less pronounced PLQY losses (Figure 6b).
In addition, a red shift of the emission wavelengths of 1, 11, 12, and 4 nm could be observed after the ligand exchange with MPA and a corresponding shift of 1, 12, 4, and 3 nm was observed after the ligand exchange with MUA for samples A, B, C and D, respectively (Figure 6c). In contrast to the PLQY and emission wavelength, the FWHM did not change in a systematic way. All the FWHM values varied between ±5 nm, as compared to the reference FWHM (Figure 6d). All FWHM changes were quite small.
QDs C showed the lowest quantum efficiency losses, smallest shift of emission peak wavelength, smallest change of FWHM, and the highest absolute PLQY after ligand exchange with MUA. QDs C were therefore chosen for further investigation of the ligand exchange reaction with MPA:MUA mixtures. For comparison purposes, the results of the D QDs coated with mixtures of MPA and MUA are shown in Figure A3 Figure A3 (Appendix H). Figure 7 shows the optical properties of C QDs before (data from Table 5) and after ligand exchange with nine different MPA/MUA ratios (data from Tables A6 and A7, Appendix I and J). The errors were estimated based on an investigation of three samples, each for the ligand exchange with only MPA, MPA:MUA = 1:1, and only MUA. The largest error of approx. 4% for the PLQY was determined for the samples treated with MUA only. The errors of the emission peak wavelengths and FWHM values were approx. 2.5 and 1 nm, respectively. A clear trend of increasing PLQY from, 12% to 43%, before dialysis, and from 12% to 38% after dialysis, was observed with an increasing MUA content between a MPA:MUA ratio of 1:0 and 1:1. A constant PLQY could be seen between a MPA:MUA ratio of 1:1 and 0:1.

Phase Transfer of QDs Using Mixtures of MPA and MUA
A comparison of the data suggests two regimes before and after dialysis. First, the PLQY of the samples after dialysis seemed to be slightly higher than before dialysis up to a MPA:MUA ratio of 2:1. Starting at a MPA:MUA ratio of 1:1, this effect was reversed. Here, the samples after dialysis  Figure 7 shows the optical properties of C QDs before (data from Table 5) and after ligand exchange with nine different MPA/MUA ratios (data from Tables A6 and A7, Appendices I and J). The errors were estimated based on an investigation of three samples, each for the ligand exchange with only MPA, MPA:MUA = 1:1, and only MUA. The largest error of approx. 4% for the PLQY was determined for the samples treated with MUA only. The errors of the emission peak wavelengths and FWHM values were approx. 2.5 and 1 nm, respectively. A clear trend of increasing PLQY from, 12% to 43%, before dialysis, and from 12% to 38% after dialysis, was observed with an increasing MUA content between a MPA:MUA ratio of 1:0 and 1:1. A constant PLQY could be seen between a MPA:MUA ratio of 1:1 and 0:1.

Phase Transfer of QDs Using Mixtures of MPA and MUA
A comparison of the data suggests two regimes before and after dialysis. First, the PLQY of the samples after dialysis seemed to be slightly higher than before dialysis up to a MPA:MUA ratio of 2:1. Starting at a MPA:MUA ratio of 1:1, this effect was reversed. Here, the samples after dialysis showed a slightly lower PLQY compared to before dialysis (approx. up to 10%). Second, the PLQY did not vary significantly with a MPA:MUA ratio < 1:1. The only outlier seemed to be the sample which was treated with MUA only, where the PLQY dropped significantly by an absolute value of~10%, from 44% to 34% after dialysis, as compared to a rather slight decrease by~6% observed for the other samples. However, although the data points seem to indicate a clear trend for both regimes, this has to be proven Nanomaterials 2020, 10, 1858 13 of 24 with a more exact measurement. As we assumed a rather large error of 4% in the PLQY based on the reproducibility experiment, the error bars slightly overlapped and a final conclusion could not be made at this point. In addition, a trend could be observed in the peak emission wavelength for the different ratios of MPA to MUA. The lower the MPA:MUA ratio, the smaller the shift of the peak emission wavelength becomes. For the FWHM, a less systematic change in the region of 69 nm ± 4 nm was observed.
Zeta potential measurements of the aqueous QD dispersions just described ( Figure A4a,c) showed that the zeta potential was always below −30 mV. As a result, the dispersions were stable [51].
TGA data of same materials are shown in Figure A4b,d. A rough trend can be identified as the data show an increasing residual mass with an increasing amount of MPA ligands, i.e., less organic material, which proves that ligands are substituted during the process described above.
showed a slightly lower PLQY compared to before dialysis (approx. up to 10%). Second, the PLQY did not vary significantly with a MPA:MUA ratio < 1:1. The only outlier seemed to be the sample which was treated with MUA only, where the PLQY dropped significantly by an absolute value of ~10%, from 44% to 34% after dialysis, as compared to a rather slight decrease by ~6% observed for the other samples. However, although the data points seem to indicate a clear trend for both regimes, this has to be proven with a more exact measurement. As we assumed a rather large error of 4% in the PLQY based on the reproducibility experiment, the error bars slightly overlapped and a final conclusion could not be made at this point. In addition, a trend could be observed in the peak emission wavelength for the different ratios of MPA to MUA. The lower the MPA:MUA ratio, the smaller the shift of the peak emission wavelength becomes. For the FWHM, a less systematic change in the region of 69 nm ± 4 nm was observed.
Zeta potential measurements of the aqueous QD dispersions just described ( Figure A4a,c) showed that the zeta potential was always below −30 mV. As a result, the dispersions were stable [51].
TGA data of same materials are shown in Figure A4b,d. A rough trend can be identified as the data show an increasing residual mass with an increasing amount of MPA ligands, i.e., less organic material, which proves that ligands are substituted during the process described above.

Long Term Stability Test
After dialysis, the QD samples were stored in Falcon tubes under air and in the dark. The optical properties of these samples were recorded after 9, 15, 37, and 44 days, respectively. Figure 8 shows that the PLQY generally decreases within the first nine days. The only exception is the sample with only MUA, which shows a stable PLQY over the whole time period. The highest PLQY reduction was recorded for particles coated with MPA:MUA at 100:1, 2:1, and 1:2 with losses of 11%, 12%, and 13%, respectively. After the first PLQY drop, all PLQYs remain roughly constant with further losses smaller than 4%. No PLQY drops below~50% of the initial PLQY after 44 days.
properties of these samples were recorded after 9, 15, 37, and 44 days, respectively. Figure 8 shows that the PLQY generally decreases within the first nine days. The only exception is the sample with only MUA, which shows a stable PLQY over the whole time period. The highest PLQY reduction was recorded for particles coated with MPA:MUA at 100:1, 2:1, and 1:2 with losses of 11%, 12%, and 13%, respectively. After the first PLQY drop, all PLQYs remain roughly constant with further losses smaller than 4%. No PLQY drops below ~50% of the initial PLQY after 44 days.

Discussion
As stated in the introduction, the quest for new, Cd-free QDs poses various challenges, starting (1) with the synthesis of suitable materials on to (2) the search for particles with equal or better optical properties than those known from Cd-based QDs. A further challenge (3) arises when facing the problem of QD dispersion in aqueous (rather than organic) media. This article introduced a series of improved and highly promising QDs that (1) are entirely Cd-free, (2) show very good PLQY stability in spite of a ligand exchange, and (3) can be dispersed in water with the dispersions remaining stable for over a month without a significant loss in optical performance.
The synthesis is, although a multistep process, rather straightforward ( Figure 1) and produces pure and uniform products ( Figure 5). In particular, there is no need to isolate or purify individual intermediate products. Rather, the process works by simple adjustment of precursor ratios and reaction temperature (changes) producing quite uniform QDs. Moreover, the synthetic process can be adapted to directly generate different shell thicknesses. This greatly simplifies the work flow compared to earlier approaches [44,52,53].
The ligand exchange introduced here (Figure 1) is quite straightforward as well. It is based on simple ligands that are (1) commercially available in large amounts, cheap, and available in many different variants, and (2) can in principle be obtained from renewable resources such as fats and oils. Moreover, the ligand exchange is quite effective and the resulting dispersions are stable for up to

Discussion
As stated in the introduction, the quest for new, Cd-free QDs poses various challenges, starting (1) with the synthesis of suitable materials on to (2) the search for particles with equal or better optical properties than those known from Cd-based QDs. A further challenge (3) arises when facing the problem of QD dispersion in aqueous (rather than organic) media. This article introduced a series of improved and highly promising QDs that (1) are entirely Cd-free, (2) show very good PLQY stability in spite of a ligand exchange, and (3) can be dispersed in water with the dispersions remaining stable for over a month without a significant loss in optical performance.
The synthesis is, although a multistep process, rather straightforward ( Figure 1) and produces pure and uniform products ( Figure 5). In particular, there is no need to isolate or purify individual intermediate products. Rather, the process works by simple adjustment of precursor ratios and reaction temperature (changes) producing quite uniform QDs. Moreover, the synthetic process can be adapted to directly generate different shell thicknesses. This greatly simplifies the work flow compared to earlier approaches [44,52,53].
The ligand exchange introduced here (Figure 1) is quite straightforward as well. It is based on simple ligands that are (1) commercially available in large amounts, cheap, and available in many different variants, and (2) can in principle be obtained from renewable resources such as fats and oils. Moreover, the ligand exchange is quite effective and the resulting dispersions are stable for up to approximately 6 weeks without loss of stability, flocculation, aggregation, or particle degradation ( Figure 8).
The two-phase ligand exchange is used in many protocols for CdSe [51,[54][55][56][57][58], but so far there are only a few systematic studies on InP-based QD systems [28,59,60]. For example, Tamang et al. found a decrease in PLQY with an increasing chain length of the MCA stabilizers [28]. In contrast, our MUA modified QDs show higher PLQYs compared to MPA modified QDs, which shows a similar trend compared to Cd based QDs [61]. This could indicate that the ligand passivation is more efficient with MUA than with MPA. Presumably the thicker ligand shell provided by the MUA provide better protection against the aqueous phase. Moreover, the optical properties can be influenced by the shell thickness ( Table 5). The modification of the particle surface with the MCA ligands does alter the optical properties, but they remain such that the resulting particles are interesting for applications in aqueous environments.
Finally, this article also provides an updated and extended model approach for the correlation of the QD size with the optical absorption properties. This is a potentially very helpful tool for more accurate QD size determinations from optical absorption data.

Conclusions
This article describes the synthesis and optical properties of a series of core/shell/shell QDs with complex compositions and their transfer into the aqueous phase via a ligand exchange reaction with high reproducibility. The key novelty of the study is the proof that mixtures of cheap and abundant ligands are efficient stabilizers for luminescent QDs in aqueous media; this study is in fact the first investigation showing that ligand mixtures are powerful tools for improving the dispersion stability of QDs. Careful selection of the stabilizing ligands and their mixtures enable the preservation of the optical properties of the QDs and the production of highly stable QD dispersions. Before dialysis, PLQYs of up to 45% ± 4% in aqueous dispersion were achieved. Depending on the exact composition of the ligand shell and even after dialysis particles still showed PLQYs of up to 44% ± 4% in aqueous dispersion. All QDs showed a good long term stability in air and in the dark. Particles stabilized with only MUA showed no drop in PLQY over 44 days of storage. Overall, the QDs are thus interesting for applications in aqueous environments and further efforts need to focus on improving QD quality and additional surface functionalization (e.g., with antibodies or aptamers).  Appendix B  Appendix B    Appendix G