A Closer Look into the Traditional Purification Process of CdSe

Dec 1, 2015 - This paper describes how the postprocessing procedure for wurtzite CdSe quantum dots (QDs) 4.8 and 6.7 nm in diameter is affected by bot...
0 downloads 8 Views 1MB Size
Download PDF
Article pubs.acs.org/Langmuir

A Closer Look into the Traditional Purification Process of CdSe Semiconductor Quantum Dots Behtash Shakeri† and Robert W. Meulenberg*,‡ †

Department of Chemistry, University of Maine, Orono, Maine 04469, United States Laboratory for Surface Science and Technology and Department of Physics and Astronomy, University of Maine, Orono, Maine 04469, United States

‡

S Supporting Information *

ABSTRACT: This paper describes how the postprocessing procedure for wurtzite CdSe quantum dots (QDs) 4.8 and 6.7 nm in diameter is affected by both the choice of nonsolvent and the number of processing steps. Using a host of analytical techniques (ultraviolet−visible, photoluminescence, nuclear magnetic, X-ray photoelectron, and infrared spectroscopy, as well as thermogravimetric analysis), we find that control over the ligand type and surface density can be achieved simply by the number of washing steps used during the postprocessing procedure. Using multiple washing steps we can achieve colloidally stable solutions of QDs with organic mass fractions as low as 13% by mass. For CdSe QDs passivated with trioctylphosphine oxide (TOPO) and stearic acid (SA), essentially no TOPO is bound to the particle surface after three or four washing steps, with a plateau in the amount of SA being removed. The results can be explained using the L- and X-type ligand classification system for QDs, with L-type ligands (TOPO) removed in the early processing steps but the removal of X-type (SA) ligand stalling at a large number of washing steps due to charging of the QDs. Importantly, very little change is observed in the photoluminescence (PL) properties, suggesting that the choice of nonsolvent during postprocessing will allow the production of QD materials with very low organic content by mass but with good PL quantum yields.

■

INTRODUCTION

for charge-carrier relaxation and consequently increasing the PLQY.8 For CdSe QDs, many groups have reported fairly large PLQYs;9−11 however, in organically passivated systems, the PLQY degrades over time.10,11 Complicating this matter, both the synthetic methods and the QD workup (postprocessing) can have a distinct impact on the photoluminescent stability of nanomaterials.12−14 The typical postprocessing purification methods employed for QDs are performed by the addition of a minimal amount of nonsolvent to the QD dispersion to encourage QD flocculation followed by separation of the flocculated particles by centrifugation. Factors that determine the outcome of the purification process include both the concentration and solubility of the impurities and the composition of the ligands surrounding the QDs, as well as parameters related to the solvent and nonsolvent, such as the temperature, ratio, polarity, basicity, and dielectric properties. One simple solution to solving the problem of how the postprocessing methods may impact the desired properties of the QDs is to use the shelling approach,15 but it must be noted

1

The remarkable size-tunable optical properties of colloidal semiconductor quantum dots (QDs) is the pretext that has driven numerous research institutes as well as industrial R&D teams over the past few decades to focus their efforts toward exploring the physics and chemistry of these sterically stabilized materials. It is believed that exploitation of the quantum characteristics of QDs will lead to future generations of photonic2 and optoelectronic3 devices as well as long-lasting high-quality probes for biological imaging.4 The structure of the colloidal QDs consists of a semiconductor crystalline core that is surrounded by a layer of organic or inorganic molecules that could either be chemisorbed or physisorbed to the unpassivated crystal surface.5 When considering the organic passivants, the headgroup moieties provide electroneutrality for QDs by passivating the bare crystalline facets,6,7 while the tail end interacts with the environment in such a way that QDs can be suspended as fairly stable colloids for a significant amount of time without flocculation. In addition to maintaining charge balance and preventing QD coagulation, capping agents also can play an important role in controlling the photoluminescence quantum yield (PLQY) by passivating the surface trap states, which results in blocking the nonradiative pathways © XXXX American Chemical Society

Received: September 25, 2015 Revised: November 3, 2015

A

DOI: 10.1021/acs.langmuir.5b03584 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

ethanol and 10 mL of methanol. The mixture is shaken well and centrifuged for 6 min at 6000 rpm. The supernatant is decanted and the precipitate is dispersed in 25 mL of toluene and collected. This stock solution is the starting material for the successive experiments and will sometimes be referred to as “the bulk material” (not to be confused with “bulk” CdSe). This material has been minimally cleaned and exhibits both free and bound ligands (vide infra). Particle sizes, dispersity, and crystallinity are derived via UV−visible spectroscopy21 and X-ray diffraction. Both UV−vis and photoluminescence spectroscopy indicate particle sizes of 4.8 and 6.7 nm, and XRD verifies the wurtzite crystal structure (Figure 1).

that the shell changes the chemistry of the surface and therefore must be considered. If one wants to understand the chemistry of the core QD and how the photoluminescence is inherently related to the surface, one must pay careful attention to all aspects of QD synthesis, including postprocessing. It has been reported that postprocessing methods employing size-exclusion chromatography are effective in increasing the monodispersity of CdSe QD solutions.16,17 Using a polystyrene gel permeation chromatography (GPC) column, Shen et al. expanded application of chromatography toward isolation of CdSe QDs from the crude synthesis product.18 By comparing the resulting material from GPC with QDs purified via the traditional precipitation/redissolution (PR) method using methanol, it was found that GPC is fairly effective in the purification of the QDs. Experimental restrictions, however, such as column size and QD concentration, restrict the yield of this method to a few tens of milligrams using regular laboratory equipment. On the other hand, the traditional PR method, which the majority of articles have utilized as their main purification procedure, can be performed on a grams scale of sample. Another advantage of the PR method is that the solvent/nonsolvent system can be customized depending on both the NC surface structure and any possible impurities in the mixture in a such way that the purification process can occur with minimum effects on the QD properties. Despite the frequency of employment of the PR method, the number of publications that provide a detailed explanation of this process is fairly low and have only explored the effect of using methanol as a nonsolvent in the PR process.14,19 In this paper, we explore the effects that multiple washing steps during the postprocessing procedure have on the properties of CdSe QDs. We find that L-type ligand content is removed after three or four washing steps and that a plateau exists for the total amount of organic ligand that can be removed from the QD surface. We believe that this is due to a strong Coulomb interaction between the outgoing ligands and the charged quantum dot. Lastly, we find that while the postprocessing impacts the surface chemistry, very few effects are observe in the photoluminescence properties, suggesting that the nonsolvent ethanol used is an ideal agent to use in the workup of quantum dot materials.

■

Figure 1. UV−visible absorption spectra (red) and photoluminescence (black) for (A) 6.7 and (B) 4.8 nm CdSe QDs and (C) powder X-ray diffraction for 6.7 nm CdSe QDs.

Purification of the CdSe QDs. For further purification of the QDs, 30 mL of ethanol is further added to the stock solution, and the mixture is shaken and centrifuged for 2 min at 6000 rpm. The washing procedure is repeated two additional times. The precipitates are washed five more times using 25 mL of toluene (enough toluene to create a 20 mg/mL solution) and an equal volume of ethanol for each tube. Subsequently, the product is dispersed in 65 mL of toluene and finally centrifuged for an additional 1 min at 6000 rpm for removal of nondispersible agglomerates. The QD solution is stored in crimpsealed vials at 3−5 °C in the absence of light. In order to track the effects of purification, a 2 mL aliquot from each step is collected for further characterization. Thermogravimetric Analysis (TGA). TGA data were recorded on a Q500 thermogravimetric analyzer (TA Instruments) under N2 atmosphere using a heating rate of 10 °C/min starting from room temperature to 635 °C. During the measurement the temperature was first maintained at 100 °C for ∼120 min to ensure evaporation of any trapped solvent. The sample was prepared by dripping various volumes (40−120 μL) of the QD solution on platinum pans. Ultraviolet−Visible (UV−vis) Spectroscopy. The UV−vis spectra were recorded in toluene on a DU 530 spectrophotometer (Beckman Coulter) using a 1 cm path length quartz cell. The absorbance values of the first absorption maximum were used to calculate the QD molar extinction coefficients, concentrations, and diameter using the empirical formula reported by Peng and coworkers.21 Photoluminescence (PL) Spectroscopy. PL spectra were measured on a PTI-814 fluorometer (Photon Technology International) with a scan step of 0.5 nm using an excitation wavelength of λmax = 480 nm (E = 2.58 eV). In order to compare the PL quantum yield (QY) of different samples, all the resulting spectra were normalized on the basis of the absorption of the solution at the excitation wavelength (i.e., typically 480 nm).

EXPERIMENTAL METHODS

Materials. Cadmium carbonate (98%, Sigma-Aldrich), selenium (powder, 100 mesh, 99.5% trace metals basis, Sigma-Aldrich), tri-noctylphosphine oxide (TOPO) (99%, Sigma-Aldrich), tri-n-octylphosphine (TOP) (≥97% Strem), stearic acid (95%, Sigma-Aldrich), toluene (99.5%, VWR), methanol (99.9%, Fisher-Scientific), and ethanol (94−96%, Alfa Aesar) were all used as received without further purification. Synthesis of CdSe Quantum Dots. CdSe quantum dots were synthesized in two sizes by utilizing a typical hot-injection procedure customized by Webber et al.20 Briefly, 20.3 mmol of CdCO3 is stirred with stearic acid (30.0 g) and TOPO (30.0 g) at 100 °C under flowing N2 for 1.5 h and then held at 360 °C for 1 h under static N2. In a separate flask, the selenium precursor is prepared by dissolving selenium powder (29.1 mmol) in TOP (30.0 mL) under N2. The selenium precursor is quickly injected into the flask, followed by immediate injection of n-octadecene (30.0 mL). To grow the QDs to the desired size, the heating mantle is removed 1−2 min after the first injection to cool down and quench the reaction. Once the reaction mixture approaches room temperature, 30.0 mL of toluene is injected into the mixture to facilitate the purification process. The mixture is divided into six 45-mL centrifuge tubes, and the product remains are rinsed with 10 mL of toluene. Then to each tube are added 11 mL of B

DOI: 10.1021/acs.langmuir.5b03584 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir Infrared Spectroscopy. The infrared absorption spectra were recorded using a ABB FTLA 2000 spectrometer equipped with an DTGS detector. In order to record solvent-free IR spectra, ∼0.1 mg of the colloidal sample was transferred on a preblanked KBr pellet and dried under a N2 atmosphere. Typically, for each spectrum, 100 scans were coadded at 4 cm−1 resolution. The FTIR spectra were collected and processed using GRAMS/AI spectroscopy software. Nuclear Magnetic Resonance (NMR) Spectroscopy. 1H NMR spectra were recorded at 400 MHz, 13C NMR spectra were recorded at 100 MHz, and 31P spectra were recorded at 161.6 MHz on a Varian Unity 9.4 T instrument at room temperature using a broad-band probe equipped with a z-gradient coil giving a maximum nominal gradient of 32.767 G cm−1. Proton chemical shifts were referenced to toluene-d8 at 2.09 ppm, 13C chemical shifts were referenced to CDCl3 at 137.86 ppm, and phosphoric acid was used as the external standard for 31P with chemical shift of 0.0 ppm. Samples were prepared using 5−10 mg of the sample dissolved in 1 mL of solvent in standard tubes, with 5 mm outer diameter and 178 mm length. In most cases, toluene-d8 has been used as the primary solvent (unless otherwise specified). In some cases, where the particle agglomeration restricts the complete evaporation of the primary solvent, the following procedure was performed to transfer the QDs into the deuterated solvent: (1) The primary solvent was evaporated to the least amount possible before agglomeration takes place. (2) A 0.5−1 mL portion of the deuterated solvent is added to the mixture and the evaporation imposed once more to minimize the presence of the primary solvent. (3) The sample is then prepared by dispersion of the remnants in the deuterated solvent. The dispersion/evaporation is repeated if the original solvent peaks appear too strong in the spectra. The 1H-DOSY experiments were carried out at 19−21 °C. The bipolar pulse-pair stimulated-echo (dbppste) pulse sequence was used for acquiring diffusion data with 100−250 ms diffusion delay and equal increments in gradient squared with rectangular gradient pulses of 1−3 ms duration. Sixteen transients of 11 975 complex points were averaged for each of 30 gradient increments ranging from 2 to 20 G cm1 nominal amplitude. Acquisition times were determined on the basis of T1 relaxations ranging from 2.0 to 4.0 s with a PW90 pulse ranging between 18 and 21 μs. The DOSY Toolbox package developed by M. Nilsson was used to process the data.22 X-ray Diffraction (XRD). The QD structure was studied via XRD by θ−2θ scans with Cu Kα radiation using line focus on a PANalytical X’PertPro Diffractometer. XRD samples were prepared by dropcasting a small amount of the QD solution onto a glass slide. X-ray Photoelectron Spectroscopy (XPS). XPS analysis was performed using an Mg K-α anode as the excitation source, and the emitted electrons were collected with a SPECS hemispherical analyzer operated at 25 eV pass energy. Any effects due to sample charging were accounted for by calibrating to the carbon 1s photoelectron peak (284.6 eV). XPS samples were prepared by drop-casting a small amount of the QD solution onto a silicon wafer.

of the QD PL. To this end, we chose to do all postprocessing of the crude QD synthesis product using ethanol (EtOH), as EtOH is slightly less acidic than MeOH. Figure 2 plots the

Figure 2. UV−visible absorption spectra for (i) zero, (ii) three, and (iii) six washes for (A) 4.8 nm and (B) 6.7 nm CdSe QDs.

UV−visible absorption spectra for QDs of both sizes with an increasing number of purification washes with EtOH. As evident from the UV−vis data, the EtOH washings do not affect the QD size or size dispersity. Both the first exciton peak position and the full width half-maximum (fwhm) are unchanged with an increasing number of EtOH washing steps. Using a multipeak fitting routine, we find that the first exciton peak position and fwhm are on average (for all washes) E6.7 = 1.951 ± 0.007 and fwhm6.7 = 118.67 ± 10 meV for the 6.7 nm sample and E4.8 = 2.057 ± 0.003 and fwhm4.8 = 157.00 ± 10 meV for the 4.8 nm sample. If no cooperative complexation mechanism occurs during washing with EtOH, then we would expect very little impact on the PL during EtOH postprocessing. In Figure 3, we plot the PLQY as a function of the purification step using EtOH.

■

Figure 3. PL quantum yield as a function of purification step. The PL data are normalized to unity by the bulk solution PL (which is a relative QY of 10% against Rhodamine 6G). For comparison, data from the work of Morris-Cohen et al.14 are presented, where MeOH was used as the nonsolvent during postprocessing.

RESULTS AND DISCUSSION Effects of Excessive Cleaning of Quantum Dots with Ethanol. There have been recent studies that investigate both how the choice of nonsolvent and multiple steps during the purification process can affect both the optical properties and the ligand surface density.5,14,23 In general, it is well-known that short-chain alcohols can effectively remove X-type ligands from the QD surface, but using methanol (MeOH) during the purification process leads to a strong reduction of the QD PLQY.14 Later work suggested that MeOH displaced the native ligands through either a proton-mediated X-type exchange or cooperative complexation mechanisms.13 It is desirable, then, to use nonsolvents that are acidic enough to promote desorption of X-type ligands during the purification process but not to promote a cooperative complexation mechanism that can remove surface-bound Cd-carboxylates and lead to a quenching

Unlike the reports using MeOH during the purification process, the PLQY measurement on EtOH-purified samples do not exhibit any degradation in the PLQY, even after many purification steps. Although both the UV−vis and PL results suggests that EtOH washing during postprocessing does not promote cooperative complexation, as observed from the unchanged first exciton peak and unchanged PLQY, we can use other techniques to attempt to probe the Cd:Se ratios. Figure S1 of the Supporting Information (SI) plots XPS survey, Cd 4d, and C

DOI: 10.1021/acs.langmuir.5b03584 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir Se 2p scans for unwashed 4.8 nm CdSe QDs and those washed eight times. The XPS scans exhibit the typical elements present for CdSe QDs: Cd, Se, P, O, and C. Although determination of C and O associated with the ligands is obscured by adventitious carbon and air oxidation, the amounts of Cd and Se should be directly assignable to the QDs. Analysis of the XPS data suggests no observable changes in the Cd:Se ratios with washings, but quantification with XPS, a primarily surface sensitive technique can be difficult. Although the XPS results are not 100% infallible, these results, plus the optical results, generally support the assertion that EtOH washing does not lead to a cooperative complexation reaction on the particle surface. Extracting quantitative information regarding the ligand concentration is much harder and can be done with careful experimentation,23 but we will turn to alternative methods to probe how the QD surface chemistry changes with multiple washings. Composition of the Quantum Dot Surface. Although using EtOH during postprocessing of the QD materials does not lead to any reduction of particle size or Cd:Se ratios, we have not addressed how well this nonsolvent acts in removing excess ligands from the QD surface. To address this, we first turn to FTIR spectroscopy. Figure 4A displays FTIR spectra

Figure 5. Effects of sequential purification on the asymmetric carboxylate peak energy and the carbonyl peak intensity for the 6.7 nm CdSe QD sample.

for the 6.7 nm CdSe QD sample. As quite evident from the plot, the two events are highly correlated, with a complete loss of intensity of the carbonyl peak at 1700 cm−1 at the maximum number of purification steps. As the carbonyl peak (free ligand) reduces in intensity during cleaning, we observe a shift of the carboxylate peak from ∼1540 to ∼1535 cm−1 with increasing purification. We attribute this behavior to an increase in the effective electronegativity of the Cd atoms with increasing purification step. It has been shown that the more electronegative the cation, the lower in frequency the asymmetric carboxylate peak will appear.29 In a sense, one can simplify the situation as the following: as we purify the CdSe QD sample, we reduce the bound/physisorbed ligand density from the surface, causing the Cd atoms, on average, to be more electronegative due to the reduction in Cd/COO− bonding. As TOPO is also used in the chemical synthesis, we also use FTIR to probe the phosphorus-containing species in the CdSe QD samples (Figure 4B). The peak at 1135 cm−1 in Figure 4B is uniquely observed in the unpurified sample spectrum and can be assigned to the PO stretch of free TOPO molecules. As the particles undergo further cleaning, the free TOPO is removed and a loss of this feature is observed. Concomitant with the loss of the free TOPO peak, two other features are observed at ∼1100 and 1015 cm−1. We assign these peaks to the symmetric (1015 cm−1) and asymmetric (1100 cm−1) stretches of the PO2− moiety. This feature will arise from phosphinic acid-like impurities that could be present in TOPO.30 We also note that some of these features could arise from bound TOPO, as a red-shift of the PO stretching frequency has been observed when TOPO is coordinated to the QD surface.31 To further probe our assignments from FTIR, we turn to nuclear magnetic resonance (NMR) spectroscopy. NMR is a powerful technique to probe bound and unbound ligands in QDs32 and can provide information complementary to FTIR. Figure 6 plots 1H NMR spectra for both free TOPO and stearic acid, as well as for a 6.7 nm CdSe QD that has been washed four times. Due to transversal internuclear dipolar relaxation, the nuclei on ligands bound to the QD surface undergo both peak shifting and line broadening.33 In the blue spectrum in Figure 6 , the triplet at ∼2.34 ppm is assigned to the α-CH2 of stearic acid, and it is shifted and broadened when stearic acid is

Figure 4. FTIR spectra between (A) 1400 and 1800 and (B) 970 and 1200 cm−1 for 4.8 nm CdSe QDs as a function of purification step.

between 1400 and 1800 cm−1 of the CdSe QD samples after each purification step. Considering all the probable species in the product, the two peaks at ∼1540 and ∼1440 cm−1 are peaks that originate from the asymmetric and symmetric stretch of carboxylate moiety, respectively, confirming the presence of stearic acid in the mixture in its deprotonated state.24 On the basis of prior studies, the wavenumber difference between the two bands (Δω) is an indicator of the type of interaction between the COO− headgroup with the metal atoms.25 When Δω ∼ 200−320 cm−1, a monodentate interaction is most likely, while Δω ∼ 140−190 cm−1 suggests a bridging bidentate interaction. Lastly, when Δω < 110 cm−1, the bonding motif is suggested to be of the chelating bidentate form.26 In the current study, the Δω ∼ 100 cm−1, which implies that the interaction of the stearate ligands with surface cadmium atoms is in the chelating bidentate form. This behavior is consistent with a prior report on oleate-capped nanocrystals.27 The peak at 1700 cm−1 originates from the CO stretch of the carboxylic acid moiety, allowing the identification of excess stearic acid in the product as well as the 1740 cm−1 peak that can originate from the stearic acid anhydride formed as a byproduct of the synthesis.28 Figure 5 plots both the carboxylate frequency and carbonyl integrated intensity as a function of purification step D

DOI: 10.1021/acs.langmuir.5b03584 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Figure 6. 1H NMR of (black) 6.7 nm CdSe QD in toluene-d8, (blue) stearic acid, and (red) TOPO in CDCl3. The asterisk marks a solvent peak and has been cutoff for clarity.

bound to the QD (Figure 6, black spectrum). The α-CH2 peak at ∼2.25 ppm is nearly obscured by the large solvent peak marked with an asterisk. Generally speaking, the 1H NMR resonances all undergo peak broadening and shifting when the ligands are bound to the QD surface. Using 1H NMR to probe TOPO on the QD surface is more challenging. The α-CH2 resonances appear at ∼1.56 ppm for free TOPO; a combination of peak broadening and shifting causes these features to be obscured by the CH2 resonances of the carbon chain. We therefore turn to 31P NMR to probe the phosphoruscontaining species in our materials. Figure 7 plots the 31P NMR spectra for two CdSe QD samples during different stages of the purification cycle; free TOPO is plotted for comparison. Consistent with prior work,34 the chemical shift for TOPO arises at ∼49 ppm. There is also a small component at ∼53 ppm that is consistent with the exhaustive and comprehensive study performed by the Buhro group.30 This peak most likely arises from small concentrations of 1-methylheptyl-n-octylphosphinic acid (MHOPA). For the “lightly cleaned” QDs, we observed two peaks, one peak at ∼53 ppm and one at ∼37 ppm, consistent with MHOPA and TOPSe,14 respectively. It is of note that no resonances can be ascribed to either free or bound TOPO species. This is consistent with our FTIR results, where we observe free TOPO only in the early stages of workup35 and we see a bound phosphorus species during the postpurification process. We claim that these species are phosphinic acid species, and the likely candidate is MHOPA. The evidence for a bound MHOPA-type species comes from both a sharp and broad peak component for the peak at ∼53 ppm (see the inset of Figure 7) reflecting both free and bound ligand species. For the heavily cleaned QD materials, the 31P NMR is featureless, suggesting that no phosphorus-containing species are present. We note, however, that the FTIR results

Figure 7. 31P NMR of (a) eight-times-washed and (b) three-timeswashed 6.7 nm CdSe QD in toluene-d8 and (c) free TOPO in CDCl3. The inset expands the region for the three-times-washed sample showing both a sharp and broad peak component, reflecting both free and bound ligand species.

show evidence for PO2− species in the heavily cleaned materials. This is related to the detection limits in NMR and is consistent with a prior report.14 It is therefore imperative to use a host of analytical techniques when probing QD surface chemistry. Lastly, we show 1H DOSY spectra (Figure S2, SI) to further verify the lack of free ligands in our heavily purified materials. The peaks derived from the unpurified QDs (shown in red) E

DOI: 10.1021/acs.langmuir.5b03584 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir appear in two separate diffusion ranges, (0−2) × 10−10 and (2− 4) × 10−10 m2 s−1, and can be interpreted as the presence of both free and bound ligands in the unpurified sample.36 Unlike the unpurified samples, the purified QD samples have only one set of diffusion peaks, suggesting only bound ligands in the analyte. Ligand Surface Density during Postprocessing. Shen et al. has estimated the oleate ligand density on two samples of 2.7 nm QDs using quantitative NMR measurements.18 One sample is the result of the QDs purified using size-exclusion chromatography (GPC), while the other sample is the result of two cycles of precipitation and redissolution using methanol (PR2). The paper has reported 135 (GPC) and 172 (PR2) oleate ligands per particle. Using these numbers, the inorganic content of the samples should be 48% (GPC) and 42% (PR2), while thermogravimetric analysis (TGA) results suggested a 56% (GPC) and 35% (PR2) inorganic content. Considering the QDs as a sphere, the total oleate ligands from TGA is ∼172 oleate molecules per QD (PR2) sample, which matches the amount estimated by NMR. TGA shows ∼94 oleate molecules per QD (GPC), which, although less than the number derived via NMR, is close enough to consider TGA as a reliable method to derive ligand surface density. Figure 8 plots normalized TGA curves of the studied CdSe QDs from each purification step. It is evident that with

organic content. This result in generally consistent with our FTIR and NMR results. On the basis of the above-mentioned ligand classification system, assuming the mass loss between 150 and 300 °C with TOPO and that between 300 and 500 °C is associated with stearate/phosphonic acid molecules, we can use TGA to estimate the ligand coverage of each sample. Figure 9 displays

Figure 9. Ligand coverage for (A) 4.8 nm and (B) 6.7 nm CdSe QDs per nm2 during various stages of purification.

the ligand coverage at each purification step for each size of CdSe QD sample. As is evident in the results, for both QD samples the neutral ligand coverage rapidly approaches zero while the X-type ligand coverage decreases and then plateaus with increasing purification step. The plateau observed in surface ligand density can be simply ascribed to an electrostatic argument. If one considers the Coulomb energy between a ligand and QD as a function of QD charge, an energy landscape can be built showing the difficulties in removing charged (Xtype) ligands once the QD charge gets too high (see the SI for more details). Due to the simplicity of this model, we cannot rigorously use this calculation to directly correlate to the number of ligands removed in our experiments. What we can say, however, is that these calculations support the notion that a plateau in the ligand surface density with increasing purification step (Figure 9) can be ascribed to energetic considerations between the ligand and QD.

Figure 8. Thermogravimetric analysis of (A) 4.8 nm and (B) 6.7 nm CdSe QDs in various stages of purification.

■

CONCLUSIONS We have examined how the choice of nonsolvent during postprocessing can influence the properties of CdSe quantum dots. We have found, in particular, that using ethanol during the postprocessing of CdSe QDs will (1) lead to stable colloidal suspensions with extremely low organic content by mass (less than 15%), (2) promote removal of both L- and X-type ligands without stripping the QD surface of cadmium atoms, and (3) show no degradation in the photoluminescence (PL) quantum yield. We also suggest that during the postprocessing of QDs, for weaker nonsolvents like ethanol, there is a limit to the amount of X-type ligand that can be removed from the QD surface due to charging of the QD upon ligand removal. We suggest that the strategies employed in this work represent a simple and rational launching point for the postprocessing of CdSe QDs with good PL properties and controlled ligand surface densities.

increasing washes (i.e., purification steps), the organic content decreases and remains nearly constant after about five washes. As expected, the smaller QDs carry a smaller inorganic fraction by mass due to two primary reasons: (1) the primary solvent was evaporated to the least amount possible before agglomeration takes place and (2) the larger curvature reduced the steric repulsion between adjacent surfactant molecules, resulting in higher packing density. The TGA curves presented in Figure 8 exhibit two noticeable mass loss regions: one between 150 and 300 °C and one between 300 and 500 °C. It has previously been suggested that the first region originates from the evaporation of neutral molecules like L-type ligands, which have relatively weak attachment to the QD surface, and the second region is derived from ionically bound ligands (e.g., X-type), which have stronger attachments to the surface atoms on the QDs.18 Both QD sizes initially begin with ∼45% organic content by mass. After one wash, the total organic content for the 4.8 and 6.7 nm QDs decreased to ∼23% and ∼13% by mass, respectively. In addition, the TOPO (i.e., L-type ligand) contribution to the organic mass decreased from ∼50 to less than 30% by mass after one wash. After three or four washes, a minimal contribution form the L-type ligand is present for the total

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.langmuir.5b03584. F

DOI: 10.1021/acs.langmuir.5b03584 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

■

Displacement and Binding. J. Am. Chem. Soc. 2013, 135, 18536− 18548. (14) Morris-Cohen, A. J.; Donakowski, M. D.; Knowles, K. E.; Weiss, E. A. The Effect of a Common Purification Procedure on the Chemical Composition of the Surfaces of CdSe Quantum Dots Synthesized with Trioctylphosphine Oxide. J. Phys. Chem. C 2010, 114, 897−906. (15) Hines, M. A.; Guyot-Sionnest, P. Synthesis and Characterization of Strongly Luminescing ZnS-Capped CdSe Nanocrystals. J. Phys. Chem. 1996, 100, 468−471. (16) Wang, M. F.; Dykstra, T. E.; Lou, X. D.; Salvador, M. R.; Scholes, G. D.; Winnik, M. A. Colloidal CdSe Nanocrystals Passivated by a Dye-Labeled Multidentate Polymer: Quantitative Analysis by Size-Exclusion Chromatography. Angew. Chem., Int. Ed. 2006, 45, 2221−2224. (17) Krueger, K. M.; Al-Somali, A. M.; Falkner, J. C.; Colvin, V. L. Characterization of Nanocrystalline CdSe by Size Exclusion Chromatography. Anal. Chem. 2005, 77, 3511−3515. (18) Shen, Y.; Gee, M. Y.; Tan, R.; Pellechia, P. J.; Greytak, A. B. Purification of Quantum Dots by Gel Permeation Chromatography and the Effect of Excess Ligands on Shell Growth and Ligand Exchange. Chem. Mater. 2013, 25, 2838−2848. (19) Kalyuzhny, G.; Murray, R. W. Ligand Effects on Optical Properties of CdSe Nanocrystals. J. Phys. Chem. B 2005, 109, 7012− 7021. (20) Webber, D. H.; Brutchey, R. L. Ligand Exchange on Colloidal CdSe Nanocrystals Using Thermally Labile tert-Butylthiol for Improved Photocurrent in Nanocrystal Films. J. Am. Chem. Soc. 2012, 134, 1085−1092. (21) Yu, W. W.; Qu, L. H.; Guo, W. Z.; Peng, X. G. Experimental Determination of the Extinction Coefficient of CdTe, CdSe, and CdS Nanocrystals. Chem. Mater. 2003, 15, 2854−2860. (22) Nilsson, M. The DOSY Toolbox: A New Tool for Processing PFG NMR Diffusion Data. J. Magn. Reson. 2009, 200, 296−302. (23) Morris-Cohen, A. J.; Malicki, M.; Peterson, M. D.; Slavin, J. W. J.; Weiss, E. A. Chemical, Structural, and Quantitative Analysis of the Ligand Shells of Colloidal Quantum Dots. Chem. Mater. 2013, 25, 1155−1165. (24) Oomens, J.; Steill, J. D. Free Carboxylate Stretching Modes. J. Phys. Chem. A 2008, 112, 3281−3283. (25) Cros-Gagneux, A.; Delpech, F.; Nayral, C.; Cornejo, A.; Coppel, Y.; Chaudret, B. Surface Chemistry of InP Quantum Dots: A Comprehensive Study. J. Am. Chem. Soc. 2010, 132, 18147−18157. (26) Wu, N.; Fu, L.; Su, M.; Aslam, M.; Wong, K. C.; Dravid, V. P. Interaction of Fatty Acid Monolayers with Cobalt Nanoparticles. Nano Lett. 2004, 4, 383−386. (27) Chen, O.; Yang, Y.; Wang, T.; Wu, H.; Niu, C.; Yang, J.; Cao, Y. C. Surface-Functionalization-Dependent Optical Properties of II-VI Semiconductor Nanocrystals. J. Am. Chem. Soc. 2011, 133, 17504− 17512. (28) Marshbanks, T. L.; Jugduth, H. K.; Delgass, W. N.; Franses, E. I. Structure and Composition of LB Films of Cadmium Sterate by FTIR Spectroscopy and XPS. Thin Solid Films 1993, 232, 126−138. (29) Ibrahim, M.; Nada, A.; Kamal, D. E. Density Functional Theory and FTIR Spectrscopic Study of Carboxyl Group. Indian J. Pure Appl. Phys. 2005, 43, 911−917. (30) Wang, F.; Tang, R.; Kao, J. L.-F.; Dingman, S. D.; Buhro, W. E. Spectroscopic Identification of Tri-n-octylphosphine Oxide (TOPO) Impurities and Elucidation of Their Roles in Cadmium Selenide Quantum-Wire Growth. J. Am. Chem. Soc. 2009, 131, 4983−4994. (31) von Holt, B.; Kudera, S.; Weiss, A.; Schrader, T. E.; Manna, L.; Parak, W. J.; Braun, M. Ligand exchange of CdSe nanocrystals probed by optical spectroscopy in the visible and mid-IR. J. Mater. Chem. 2008, 18, 2728−2732. (32) Hens, Z.; Martins, J. C. A Solution NMR Toolbox for Characterizing the Surface Chemistry of Colloidal Nanocrystals. Chem. Mater. 2013, 25, 1211−1221. (33) Sachleben, J. R.; Colvin, V.; Emsley, L.; Wooten, E. W.; Alivisatos, A. P. Solution-State NMR Studies of the Surface Structure

X-ray photoelectron and DOSY spectra for the studied materials (Figures S1 and S2) and an explanation of the electrostatic model and the electrostatic energy plotted as a function of the distance the ligand is from the QD surface (Figure S3) (PDF)

AUTHOR INFORMATION

Corresponding Author

*Phone: 1-207-581-2245. E-mail: robert.meulenberg@maine. edu. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by the National Science Foundation under Grant No. DMR-1206940. The authors would like to thank Dr. George Bernhardt and Mr. David Labrecque for assistance with experiments.

■

REFERENCES

(1) Alivisatos, A. P. Perspectives on the Physical Chemistry of Semiconductor Nanocrystals. J. Phys. Chem. 1996, 100, 13226−13239. (2) McDonald, S. A.; Konstantatos, G.; Zhang, S. G.; Cyr, P. W.; Klem, E. J. D.; Levina, L.; Sargent, E. H. Solution-Processed PbS Quantum Dot Infrared Photodetectors and Photovoltaics. Nat. Mater. 2005, 4, 138−U14. (3) Caruge, J. M.; Halpert, J. E.; Wood, V.; Bulovic, V.; Bawendi, M. G. Colloidal Quantum-Dot Light-Emitting Diodes with Metal-Oxide Charge Transport Layers. Nat. Photonics 2008, 2, 247−250. (4) Michalet, X.; Pinaud, F. F.; Bentolila, L. A.; Tsay, J. M.; Doose, S.; Li, J. J.; Sundaresan, G.; Wu, A. M.; Gambhir, S. S.; Weiss, S. Quantum Dots for Live Cells, In Vivo Imaging, and Diagnostics. Science 2005, 307, 538−544. (5) Hassinen, A.; Moreels, I.; De Nolf, K.; Smet, P. F.; Martins, J. C.; Hens, Z. Short-Chain Alcohols Strip X-Type Ligands and Quench the Luminescence of PbSe and CdSe Quantum Dots, Acetonitrile Does Not. J. Am. Chem. Soc. 2012, 134, 20705−20712. (6) Gomes, R.; Hassinen, A.; Szczygiel, A.; Zhao, Q.; Vantomme, A.; Martins, J. C.; Hens, Z. Binding of Phosphonic Acids to CdSe Quantum Dots: A Solution NMR Study. J. Phys. Chem. Lett. 2011, 2, 145−152. (7) Liu, H.; Owen, J. S.; Alivisatos, A. P. Mechanistic Study of Precursor Evolution in Colloidal Group II-VI Semiconductor Nanocrystal Synthesis. J. Am. Chem. Soc. 2007, 129, 305−312. (8) Oszwaldowski, S.; Roberts, K. P. Comparative Study on Coating CdSe Nanocrystals with Surfactants. Microchim. Acta 2013, 180, 1341−1350. (9) Qu, L.; Peng, X. Control of Photoluminescence Properties of CdSe Nanocrystals in Growth. J. Am. Chem. Soc. 2002, 124, 2049− 2055. (10) de Mello Donega, C.; Hickey, S.; Wuister, S.; Vanmaekelbergh, D.; Meijerink, A. Single-Step Synthesis to Control the Photoluminescence Quantum Yield and Size Dispersion of CdSe Nanocrystals. J. Phys. Chem. B 2003, 107, 489−496. (11) Smith, A. M.; Duan, H. W.; Rhyner, M. N.; Ruan, G.; Nie, S. M. A Systematic Examination of Surface Coatings on the Optical and Chemical Properties of Semiconductor Quantum Dots. Phys. Chem. Chem. Phys. 2006, 8, 3895−3903. (12) Akdas, T.; Distaso, M.; Kuhri, S.; Winter, B.; Birajdar, B.; Spiecker, E.; Guldi, D. M.; Peukert, W. The Effects of Post-Processing on the Surface and the Optical Properties of Copper Indium Sulfide Quantum Dots. J. Colloid Interface Sci. 2015, 445, 337−347. (13) Anderson, N. C.; Hendricks, M. P.; Choi, J. J.; Owen, J. S. Ligand Exchange and the Stoichiometry of Metal Chalcogenide Nanocrystals: Spectroscopic Observation of Facile Metal-Carboxylate G

DOI: 10.1021/acs.langmuir.5b03584 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir and Dynamics of Semiconductor Nanocrystals. J. Phys. Chem. B 1998, 102, 10117−10128. (34) Wang, F.; Tang, R.; Buhro, W. E. The Trouble with TOPO; Identification of Adventitious Impurities Beneficial to the Growth of Cadmium Selenide Quantum Dots, Rods, and Wires. Nano Lett. 2008, 8, 3521−3524. (35) It is difficult to compare directly the wash steps reported in the TGA and FTIR experiments with those performed in the NMR experiments. To preprare the NMR sample, a few extra steps are required in order to concentrate the solute and deuterate the solvent. Therefore, it is difficult to examine one or two wash samples via NMR and explains why no evidence for unbound TOPO is seen with NMR. (36) Hassinen, A.; Moreels, I.; de Mello Donega, C.; Martins, J. C.; Hens, Z. Nuclear Magnetic Resonance Spectroscopy Demonstrating Dynamic Stabilization of CdSe Quantum Dots by Alkylamines. J. Phys. Chem. Lett. 2010, 1, 2577−2581.

H

DOI: 10.1021/acs.langmuir.5b03584 Langmuir XXXX, XXX, XXX−XXX