Shape Control of Colloidal Cu2-x S Polyhedral Nanocrystals by Tuning the Nucleation Rates.

Synthesis protocols for colloidal nanocrystals (NCs) with narrow size and shape distributions are of particular interest for the successful implementation of these nanocrystals into devices. Moreover, the preparation of NCs with well-defined crystal phases is of key importance. In this work, we show that Sn(IV)-thiolate complexes formed in situ strongly influence the nucleation and growth rates of colloidal Cu2-x S polyhedral NCs, thereby dictating their final size, shape, and crystal structure. This allowed us to successfully synthesize hexagonal bifrustums and hexagonal bipyramid NCs with low-chalcocite crystal structure, and hexagonal nanoplatelets with various thicknesses and aspect ratios with the djurleite crystal structure, by solely varying the concentration of Sn(IV)-additives (namely, SnBr4) in the reaction medium. Solution and solid-state 119Sn NMR measurements show that SnBr4 is converted in situ to Sn(IV)-thiolate complexes, which increase the Cu2-x S nucleation barrier without affecting the precursor conversion rates. This influences both the nucleation and growth rates in a concentration-dependent fashion and leads to a better separation between nucleation and growth. Our approach of tuning the nucleation and growth rates with in situ-generated Sn-thiolate complexes might have a more general impact due to the availability of various metal-thiolate complexes, possibly resulting in polyhedral NCs of a wide variety of metal-sulfide compositions.

Introduction

In recent years, the scientific community has achieved a high level of mastery over the size, shape, and composition of colloidal nanocrystals (NCs) and heteronanocrystals (HNCs) of Cd– and Pb–chalcogenide compositions.[1−4] However, their further deployment into applications has been hindered by toxicity concerns.[5−8] The properties of copper chalcogenides (Cu2–A with A = S, Se and Te) make them attractive alternatives for Cd- and Pb-based semiconductors for several applications.[6,8−10] For instance, Cu2–S is a direct p-type semiconductor with a band gap that depends on its stoichiometry (1.1–1.4 eV for x = 0–0.04; 1.5 eV for x = 0.2; 2.0 eV for x = 1).[11−13] The combination of a suitable band gap, high absorption coefficient (104 cm–1), low cost, and low toxicity has made Cu2–S a promising candidate for large scale and sustainable implementation into photovoltaic (PV) devices.[14,15] Moreover, Cu2–A NCs can hold both excitons and tunable localized surface plasmon resonances on demand.[9,13,16−18] This makes Cu2–A NCs promising materials for photovoltaics,[15] photocatalysis,[19] sensing,[20] and nanoplasmonics.[9,11,17,21,22]

The application of colloidal NCs in photovoltaic and nanoplasmonic devices requires strict control over the size, shape, and polydispersity of the NC ensemble because these characteristics are of crucial importance not only for the optoelectronic properties of the NCs themselves but also for the quality of the NC thin films obtained by solution-based deposition techniques.[23−25] Cu2–S NCs are of particular interest because a variety of shapes not attainable for other semiconductor NCs can easily be synthesized with narrow size and shape distributions. For example, hexagonal nanodisks, nanoplatelets, and nanosheets, among other polyhedral shapes, have been successfully synthesized.[6,26−29] Shape control over Cu2–S NCs may also have a more general impact because Cu+ ions in copper chalcogenides have been shown to be easily exchangeable by other cations.[6,30−37] This opens up the possibility of using (partial) topotactic cation exchange reactions to convert Cu2–A NCs into other compositions while preserving the size and shape of the parent NCs, thereby making NCs with novel functionalities.[33,36,37] However, the methods currently available to control the size and shape of Cu2–S NCs lack flexibility because different sets of physical-chemical parameters (concentrations, ligands, reaction temperatures, reaction times) have to be used for each different shape.[6,26,28] In this work, a novel methodology for the size and shape control of colloidal Cu2–S NCs is developed, which relies on changing just one single reaction variable: the concentration of Sn(IV) complexes that are used to control the nucleation and growth rates of Cu2–S NCs. In this way, hexagonal bifrustums, hexagonal bipyramids, and hexagonal nanoplatelets of various aspect ratios are synthesized with narrow size and shape distributions. Solid-state 119Sn NMR measurements indicate that Sn–thiolate complexes formed in situ control the nucleation and growth rates, which results in various polyhedral Cu2–S NCs. Our study provides insights into the formation of polyhedral Cu2–S NCs and possibly paves the way toward the development of synthetic protocols for polyhedral NCs of various compositions by deployment of in situ nucleation and growth controlling agents, such as Sn(IV)–thiolates.

Experimental Section

Colloidal Hexagonal Bipyramid-Shaped Cu2–S NCs

These NCs were synthesized based on the protocol described by Kuzuya et al.,[38] which was modified by adding SnBr4 to the reaction mixture. In a typical synthesis, 1.0 mmol CuCl (99 mg) and 0.5 mmol SnBr4 (219 mg) were mixed in 8 mL (33.6 mmol) of 1-dodecanethiol (DDT) and 2 mL (6 mmol) of oleylamine (OLAM). The flask was purged with N2, and the solution was gradually heated to 225 °C. At first, a creamy white substance was present, but around 80 °C, the solution turned slightly yellow. When further heated, a clear yellow solution was obtained around 130 °C. Nucleation and growth of Cu2–S started when the temperature reached 200 °C. Finally, the particles were allowed to grow at 225 °C for 1 h, and subsequently, the nanoparticles were precipitated by the repeated addition of a methanol/butanol solution and redispersion in toluene.

Colloidal Polyhedral Cu2–S NCs of Various Morphologies

These NCs were synthesized based on the same method as described above only with the amount of SnBr4 varied (Cu:Sn ratio of 2:3 for low aspect ratio hexagonal nanoplatelets, 1:2 for high aspect ratio hexagonal nanoplatelets, 30:1 for small hexagonal bifrustums, and 1:1 for large hexagonal bipyramids).

Transmission Electron Microscopy (TEM) and Energy Dispersive X-ray Spectroscopy (EDS)

TEM and EDS measurements were performed on a Tecnai20F (FEI) microscope equipped with a Field Emission Gun, a Gatan 694 CCD camera, and an EDAX spectrometer. The microscope was operated at 200 kV. Acquisition time for EDS measurements was 30 s. Samples for TEM imaging were prepared by dropping a diluted nanocrystal solution in toluene on a carbon-coated polymer film copper grid (300 mesh). The solvent (toluene) was allowed to evaporate prior to imaging.

High-Resolution (Scanning) Transmission Electron Microscopy (HRTEM)

HRTEM measurements were performed on a double aberration-corrected cubed FEI Titan 50–80 electron microscope operated at 120 kV. HRTEM measurements were performed on FEI Osiris and FEI Tecnai electron microscopes operated at 200 kV. Electron tomography measurements were performed in high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) mode to get rid of any unwanted diffraction contrast present in TEM by using an aberration-corrected cubed FEI Titan 60–300 electron microscope and a double aberration-corrected cubed FEI Titan 50–80 electron microscope operated at 200 and 300 kV. The acquisition of all of the series was performed manually over a tilt range from −74° to +74° and a tilt increment of 2° by using a Fischione model 2020 single tilt tomography holder. The alignment of all the acquired series was performed by using cross-correlation and the reconstruction by using the simultaneous iterative reconstruction technique (SIRT) as implemented in the ASTRA toolbox.[39]

Solid-State 119Sn NMR and Solution 1H NMR

These experiments were performed using a Bruker Avance III spectrometer equipped with a 4 mm double resonance probe head at 11.7 T static magnetic field. All experiments on precipitated and dry nanocrystals were measured at room temperature and on solution samples at 308 K. Tin and proton field strengths for 90° pulses were 50 and 71 kHz, respectively. Spectral referencing was done using SnBr4 for tin and adamantane for protons. For solution 119Sn and 1H NMR measurements, no additional solvents were added. The crude reaction mixture was loaded in a capillary at elevated temperatures and subsequently cooled to room temperature before measurement. No precipitation was observed upon cooling.

X-ray Diffraction (XRD)

XRD patterns were obtained using a PW 1729 Philips diffractometer equipped with a Cu Kα X-ray source (λ = 1.5418 Å). Samples for XRD analysis were prepared by depositing purified NCs on a Si wafer substrate under an inert atmosphere. The purification procedure consisted of precipitating the NCs from a solution of NCs in toluene by adding anhydrous methanol (1:1 volume ratio). The sediment was isolated by centrifugation (3000 rpm, 15 min) and redispersed in chloroform. The concentrated solution of NCs was dropcasted on the Si wafer, and the chloroform was allowed to evaporate at RT, resulting in a concentrated NC solid.

1D Powder Electron Diffraction (PED)

PED patterns were obtained by radially integrating the 2D ED patterns acquired on a Tecnai-12 transmission electron microscope using a selected-area aperture. 2D ED patterns were acquired on areas containing a large number of nanocrystals to make the 1D PED patterns statistically valid.

Results and Discussion

Shape Control of Anisotropic Cu2–S NCs by SnBr4 Addition

Colloidal Cu2–S NCs were synthesized by heating a solution of CuCl in 1-dodecanethiol (DDT) and oleylamine (OLAM) to 225 °C. In this protocol, DDT has the combined roles of ligand, solvent, and sulfur source.[38,40,41] In the absence of SnBr4, large polydisperse nanoplatelets (∼100 nm lateral dimensions) are obtained (Figure a). The addition of SnBr4 to the reaction mixture has a dramatic impact on the size and shape of the product NCs, leading to a variety of polyhedral shapes (hexagonal bifrustums, hexagonal bipyramids, low and high aspect ratio hexagonal nanoplatelets) with small polydispersity, depending on the SnBr4 concentration (Figure ).

Figure 1

Representative transmission electron microscopy (TEM) images of Cu2–S nanocrystals synthesized with various SnBr4 concentrations under otherwise constant reaction conditions. The amount of SnBr4 was the only variable; all other reaction parameters were kept constant. (a) Polydisperse Cu2–S nanoplatelets synthesized in the absence of SnBr4. (b) Monodisperse Cu2–S bifrustum NCs (diameter: 17 nm) are obtained when a small amount of SnBr4 is added (Cu:Sn 30:1). (c) A Cu:Sn ratio of 2:1 yields hexagonal bipyramids (28 nm wide, 38 nm long). (d) Cu2–S NCs synthesized under Cu:Sn = 1:1. (e) Hexagonal nanoplatelets (25 nm thick, 50 nm wide) are obtained under Cu:Sn = 2:3. (f) Further increasing the Cu:Sn ratio to 1:2 leads to wider and thinner hexagonal nanoplatelets (80 nm by 10 nm).

High-resolution transmission electron microscopy (HRTEM) (Figure ), electron diffraction (Figure S1), and X-ray diffractometry (Figure S2) measurements revealed that the product Cu2–S nanoplatelets have the djurleite crystal structure,[42] whereas the product Cu2–S bipyramids have the low-chalcocite crystal structure. This is evident in the HRTEM images and the corresponding fast fourier transform (FFT) patterns of hexagonal bipyramid NCs viewed along the [010] zone axis (Figure a,d and Figure S3) and of low aspect ratio nanoplatelets (Figure b,c,e,f). The observed spacings are in agreement (within a 1% error margin) with the djurleite crystal structure (lattice parameters a = 26.89 Å, b = 15.74 Å, and c = 13.57 Å)[42] for the low aspect ratio nanoplatelets and with the low-chalcocite crystal structure (lattice parameters a = 11.92 Å, b = 27.34 Å, and c = 13.44 Å)[42] for the bipyramid NCs. The high aspect ratio nanoplatelets also have the djurleite structure, whereas the hexagonal bifrustums have the low-chalcocite crystal structure (Figure S1).

Figure 2

Structural analysis of the product Cu2–S nanocrystals. High-resolution transmission electron microscopy (HRTEM) images of (a) hexagonal bipyramid Cu2–S NCs and (b,c) low aspect ratio hexagonal Cu2–S nanoplatelets viewed along the [100] direction (b) and along the [001] direction (c). The white square in c indicates the area of which the FFT pattern is displayed in panel f. (d) FFT pattern of the bipyramid NC displayed in panel a, showing characteristic diffraction spots of the djurleite crystal structure viewed along the [010] zone axis. (e) FFT pattern of the NC displayed in panel b, showing characteristic diffraction spots of the djurleite crystal structure viewed along the [100] zone axis. (f) FFT pattern of a selected area (white square) of the NC displayed in panel c, showing characteristic diffraction spots of djurleite crystal structure viewed along the [001] direction.

The morphology of the NCs was investigated by carrying out electron tomography measurements in high-angle annular dark field scanning transmission electron microscopy (HAADF-STEM) mode (Figure ). By combining the electron tomography results, which enabled the determination of the shape of the NCs, with the crystallographic information obtained by the HRTEM analysis (Figure above), the different facets can be indexed according to the monoclinic low-chalcocite (for bipyramids) and monoclinic djurleite (for nanoplatelets) crystal structures.[42] In this way, the lateral facets of the low-chalcocite bipyramids are indexed as {364} (8 facets), {502} (2 facets), and {302} (2 facets) (Figure b,c). A different indexation holds for the low- and high aspect ratio hexagonal nanoplatelets, which have the djurleite crystal structure. The side facets of the nanoplatelets are formed by 8 trapezoidal {221} and 4 {101} facets, and the top and bottom facets are the hexagonal {100} facets. It is thus evident that the addition of SnBr4 alters the size and crystal structure, and hence, the faceting of Cu2–S NCs in a concentration-dependent fashion. At the lowest Cu:Sn ratio (30:1), small (d = 17 nm) hexagonal bifrustums are obtained with the low-chalcocite crystal structure in which the top and bottom {102} facets and the {364} and {302}/{502} side facets have similar areas. Increasing the Cu:Sn ratio to 2:1 leads to large (28 nm by 38 nm), hexagonal bipyramidal low-chalcocite NCs in which the {102} facets are almost absent. Further increasing the Cu:Sn ratio results in djurleite hexagonal nanoplatelets of which the relative area of the {100} facets, the lateral dimensions, and the aspect ratio grow with the Cu:Sn ratio (25 nm thick and 50 nm wide for Cu:Sn = 2:3; 10 nm thick and 80 nm wide for Cu:Sn = 1:2). The formation mechanism for low-chalcocite bipyramids and djurleite nanoplatelets will be discussed in detail below.

Figure 3

Electron tomography reconstructions of polyhedral Cu2–S nanocrystals. (a) Low-chalcocite hexagonal bipyramid nanocrystals. (b,c) Facet indexation of a low-chalcocite hexagonal bipyramid (b) viewed along the [001] direction and (c) along the [010] direction. (d) Djurleite low aspect ratio hexagonal nanoplatelets viewed along the [100] direction. (e) Facet indexation of a djurleite low aspect ratio hexagonal nanoplatelet viewed from the top. (f) Two high aspect ratio hexagonal nanoplatelets viewed from the top.

Our group has previously reported that the addition of SnBr4 to synthesis protocols for Cu2–S NCs dramatically affects the size and shape of the product NCs, yielding ultrathin (2 nm thick) Cu2–S nanosheets with well-defined shape and size (triangular or hexagonal; 100 nm to 3 μm wide) instead of nearly spherical small (9 nm diameter) NCs.[29,43] The effect was clearly shown to be due to the halides, such that Sn(IV) tetrahalides were only relevant as sources of sufficiently high halide concentrations in the growth solution.[29,43] To investigate the roles of the halide and Sn(IV) in the present case, we carried out control experiments in which either SnBr4 was replaced by Sn(OAc)4 or additional halides were added in the form of CuBr in the absence of any Sn(IV) compound. The results clearly show that, in contrast to our previous study on the formation of ultrathin Cu2–S nanosheets,[29,43] the effect in the present case is due to the Sn(IV) (Figure S4) because additional halides yield polydisperse nanoplatelets similar to those obtained under the standard reaction conditions (i.e., SnBr4 absent; Figure a above), whereas addition of Sn(OAc)4 resulted in polyhedral NCs similar to those obtained upon addition of SnBr4 (Cu:Sn = 2:1; Figure c above). It should be noted that the polydispersity of the polyhedral NCs obtained by adding Sn(OAc)4 is larger than that observed for the NCs formed in the presence of SnBr4. This can be ascribed to the fact that Sn(OAc)4 is less reactive toward DDT molecules than SnBr4 because acetate and Sn4+ are a strong Lewis base and acid, respectively, and are better matched in terms of hardness (both hard) than bromide and Sn4+ (soft and hard, respectively).[44] This suggests that the active Sn(IV) species responsible for the observed morphological changes results from a reaction between DDT and the added Sn(IV) salt. This also explains the different roles observed for Sn(IV) in the present work (active species) and in our previous study on the formation of Cu2–S nanosheets (namely, halide carrier)[29,43] because in the latter case a smaller excess of DDT (10-fold with respect to Cu and diluted in 1-octadecene) was injected in the reaction mixture, whereas in the present case, DDT is used as the solvent, resulting in a larger excess (30-fold with respect to Cu). This larger excess of DDT allows the formation of both Cu-DDT and Sn-DDT complexes, despite the lower reactivity of Sn(IV) toward DDT. As will become clear below in the NMR Spectroscopy section, another important chemical difference in the present reaction system is the presence of oleylamine, which deprotonates DDT, thereby further facilitating its reaction with SnBr4.

For gaining further insight into the role of SnBr4 in the shape control of the product Cu2–S NCs, energy dispersive X-ray spectroscopy (EDS) chemical mapping was used to quantify and locate the elements present in the NCs. The Cu:Sn:S ratio was found to be 1.81 ± 0.14:0.02 ± 0.01:1.00 regardless of the Cu:Sn ratio used in the reaction. Bromide was not detected in any sample. Elemental mapping shows that Sn is not homogeneously distributed across the ensemble of NCs but is concentrated on a few NCs that appear morphologically distinct from the majority of the ensemble (Figures S5 and S6). The Sn distribution in the Sn-poor NCs does not indicate any preference for a particular facet regardless of the shape of the NC (i.e., bipyramids or nanoplatelets in Figures S5 and S6, respectively). The signal from individual NCs was too low to allow the composition of the Sn-rich NCs to be reliably established, but their different shapes imply that they no longer have the djurleite or low-chalcocite crystal structures,[42] and may have adopted the cubic crystal structure characteristic of ternary CuSnS phases (e.g., zinc blende for Cu2SnS3 and cubic spinel for Cu2Sn3S7).[45−48] The Cu–Sn–S phase diagram is however quite rich (16 different phases, 13 of them metastable),[48] and therefore, the range of possible compositions for the Sn-rich NCs is very wide (from Cu9Sn2S9 to Cu2Sn3.75S8). Nevertheless, it is clear that in these NCs some degree of Sn(IV) interdiffusion has taken place, inducing crystal structure and morphology changes. Considering that only a very small fraction of the NCs is observed to be Sn-rich (<5%), we can conclude that the Sn(IV) diffusion rates in the Cu2–S NCs were negligible under the conditions prevalent in our experiments. This is consistent with previous studies in which significant Sn(IV) interdiffusion in Cu2–S NCs was only observed at temperatures as high as 240 °C.[47] It is worth noting that studies on the nucleation and growth of multinary Cu chalcogenide compounds (such as CuInS2, Cu2SnS3, and Cu2ZnSnS4) have shown that binary Cu2S NCs nucleate first, after which the other cations diffuse into the existing seeds to form the multinary NCs.[6,47,49,50] Therefore, we exclude separate nucleation events as a possible explanation for the observed Sn-rich NCs.

The results presented above clearly demonstrate that the impact of SnBr4 in the size and shape evolution of Cu2–S NCs synthesized by heating a solution of CuCl in DDT is due to in situ-generated Sn(IV) species and rules out Sn(IV) incorporation as the cause for the observed changes. To uncover the nature of the Sn(IV) complexes that were formed in situ and to verify whether they bind to the surface of the Cu2–S NCs, we carried out 1H and 119Sn NMR spectroscopic measurements.

Solution and Solid-State NMR Spectroscopy

119Sn NMR spectroscopy is particularly suited to elucidate the nature of the Sn complexes in solution and on the surface of the NCs. Overall, the fairly high natural abundance of 119Sn nuclei as well as the large range of isotropic chemical shifts (around 6500 ppm) that leads to clearly separated chemical shift regions in the spectra makes 119Sn NMR a promising tool for characterizing tin compounds. Combined with novel NMR methods like dynamic nuclear polarization (DNP), 119Sn NMR also has great potential for the investigation of nanomaterial surfaces.[51] Furthermore, former studies have shown that the chemical shift tensor parameters of tin sulfides are highly sensitive to coordination numbers and symmetry in the local environment of the tin atom.[52] Therefore, probing the chemical shift anisotropy (CSA) via solid-state NMR provides an opportunity to detect structural details.

In the following, three of our Cu2–S NC morphologies were analyzed with solid-state 119Sn NMR, namely, the hexagonal bipyramids (Cu:Sn = 2:1), the low aspect ratio nanoplatelets (Cu:Sn = 2:3), and the high aspect ratio nanoplatelets (Cu:Sn = 1:2). The reference for solid-state 119Sn NMR measurements was SnBr4 with a sharp peak at chemical shift δ −638 ppm. None of the measured NC samples showed a detectable amount of SnBr4 in the spectra. Instead, in the cases of an excess of Sn compared to Cu, the spectra were dominated by a sharp signal at 144 ppm under static conditions as well as magic angle spinning (MAS) at 5 kHz (black lines in Figure b, signal at 144 ppm indicated with a red star; see also Figure S7). The fact that there is no line narrowing after spinning at 5 kHz indicates that these compounds have a symmetrical surrounding, meaning they are not on the NC surface.[52] Additionally, less intense and broader signals in the chemical shift ranging between 0 and 80 ppm were observed for the high aspect ratio Cu2–S nanoplatelets obtained by mixing Cu(I) and Sn(IV) salts in a 1:2 ratio (Figure b, indicated with blue stars). In the presence of a higher Cu amount compared to that of Sn (Cu:Sn 2:1, bipyramids), one broad peak centered at −208 ppm was detected (top of Figure b, indicated with a black star; see also Figure S7c). Here, 5 kHz spinning resulted in significant line narrowing (top of Figure b), indicating an asymmetrical surrounding. This signal was therefore assigned to a Sn compound bound to the NC surface. Comparison with the chemical shift values reported by Kovalenko et al.[53,54] for Sn sulfocomplexes ([Sn2S6]4– and [SnS4]4– at δ 56.3 and 70–75 ppm, respectively) lead to the conclusion that the Sn(IV) species formed in situ in our experiments are of a different nature. This is further supported by the fact that the anisotropic line broadening (around 60 ppm, top of Figure b) of the 119Sn NMR spectrum obtained for hexagonal bipyramids (Cu:Sn of 2:1) is far less pronounced than reported in the literature for salts containing [Sn2Sn6]4– ions.[55]

Figure 4

Solid-state and solution 119Sn NMR spectra of polyhedral Cu2–S nanocrystals and prenucleation reaction products. (a) Schematic representation of the Sn(IV)–thiolate complexes formed by stepwise replacement of bromide for deprotonated thiols. The chemical shift values for each structure are indicated at the top of the panel. (b) Solid-state 119Sn NMR spectra of (top panel) hexagonal Cu2–S bipyramids (with and without 5 kHZ MAS) and (bottom panel) low aspect ratio Cu2–S nanoplatelets and high aspect ratio Cu2–S nanoplatelets. The symbols refer to the Sn(IV)–thiolate complexes shown in panel a. (c) Solution 119Sn NMR spectra of CuCl and SnBr4 mixed in the same concentration and ratio used to synthesize the nanoplatelets (blue lines) and of different equivalents of SnBr4 in DDT/OLAM (red lines).

We thus propose a mechanism where thiol molecules replace bromides from SnBr4 to form Sn–thiolate complexes. This replacement is a stepwise process, where one Br is replaced by a thiol in each step (Figure a). Tin(IV) methylthiolate, (MeS)4Sn, is known to show a sharp signal at 160 ppm.[56,57] A shift to slightly lower ppm values is expected if the methylthiolate groups are replaced by thiolate groups containing longer alkyl chains,[56,57] such as dodecylthiolate in the present work. The sharp signal at 144 ppm is thus assigned to (C12H25S)4Sn (indicated by a red star in panel a). This interpretation is supported by earlier findings, which showed that related sharp signals with a lack of prominent sidebands under slow spinning speeds (around 3 kHz) and low magnetic fields correspond to salts containing [SnS4]4– compounds.[52] The clear contribution of anisotropic chemical shielding (CSA) interactions to the broad signal at −208 ppm indicates a nonisotropic environment of the central tin atom (Figure a, indicated by a black star). The lower ppm value compared to the sharp peak at 144 ppm suggests the presence of remaining Br atoms because the quadrupolar nature of 79Br and 81Br would further contribute to the line broadening. Previous studies have shown that successive replacement of Br via alkyls in tin compounds leads to a large, nonlinear shift toward higher ppm values (e.g., MeSnBr3, −165 ppm; Me2SnBr2, 70 ppm; Me3SnBr, 128 ppm).[54] Bearing these values in mind, we attribute the signal at −208 ppm to the intermediate compound (C12H25S)SnBr3.

Further, 119Sn NMR was also used to analyze the compounds present in solution prior to the onset of Cu2–S NC nucleation (Figure c). To this end, CuCl and SnBr4 were mixed in different Cu:Sn ratios (2:1 for hexagonal bipyramids, 2:3 for low aspect ratio nanoplatelets, and 1:2 for high aspect ratio hexagonal nanoplatelets, blue lines in Figure c; see also Figure S7) in a DDT/OLAM solution and heated to 180 °C. This temperature is slightly below the onset of Cu2–S NC nucleation (∼220 °C as evidenced by a color change from yellow to brown), which gives rise to the formation of all precursors without (significant) formation of Cu2–S nuclei. The dominating peak at 144 ppm is present in solution as well (blue lines in Figure c), and a second Sn complex is observed around 80 ppm (indicated with a blue star). Furthermore, solution samples without Cu+ ions (red lines in Figure c; see also Figure S7) show the same signal at 144 ppm (indicated with a red star). This finding demonstrates that the formation of the compound responsible for the peak at 144 ppm is independent of the presence of Cu+ or NCs, so this peak can be unambiguously ascribed to unbound (C12H25S)4Sn. However, additional peaks in the range of 80–0 ppm show a clear dependency on the presence of Cu+ (peaks indicated with a blue star; only observed when Cu(I) and Sn(IV) are simultaneously present). This suggests that a thiolate complex containing both Sn(IV) and Cu(I) is present prior to the onset of Cu2–S nucleation. Heterometallic polynuclear Sn(IV)–Cu(I) complexes in which thiolates act as ligands are known to be stable both in solution and as crystalline solids.[58,59] It is thus plausible that similar complexes are also formed during the heating of the reaction medium to the reaction temperature. Although the chemical shift values are slightly different, the signals observed in the prenucleation solutions at 80 ppm and in the solid-state samples of high aspect ratio nanoplatelets (Cu:Sn ratio = 1:2) at 80–0 ppm might have the same origin. Different packing in the solid state and different coordination modes in solution could be a possible explanation for the differences, as has already been reported for other tin(IV) compounds.[60]

For further investigation, we performed 1H NMR measurements on the solution and solid-state samples (Figures S8 and S9). All three spectra for solution samples show huge proton densities at chemical shift areas characteristic for alkyl chains (3.5–0 ppm) and a peak at around 5 ppm, which typically corresponds to an R-NH3+ species (Figure S8). This implies that the amine group of OLAM deprotonates the thiol headgroup of DDT, resulting in nucleophiles (C12H25S–) that are more reactive toward both Sn(IV) and Cu(I). This is consistent with our experimental observation that the presence of OLAM increases the reaction rates. Furthermore, a peak is observed around 7.5–8 ppm, which can be ascribed to the presence of Sn(IV) because it is absent in the case of the Cu-DDT/OLAM solution (i.e., when no Sn(IV) compound is present in the reaction medium).

Copper sulfide NCs prepared with thiols as sulfur source are known to be resistant to ligand exchange procedures.[61] A recent study by Turo and Macdonald[61] provided strong evidence that, when thiols are used as sulfur source, they are effectively integrated in the NC (“crystal-bound” thiols), forming the terminal sulfur layers of the crystal and occupying high coordination number sites. These strongly bound thiols explain why ligand exchange is difficult. It is interesting to note that oleylamine capped Cu2–S NCs can only be obtained in the absence of DDT (i.e., using sulfur powder as S-source and OLAM as ligand).[61] Furthermore, OLAM is easily exchanged by DDT through postsynthetic ligand exchange.[61] The higher affinity of thiols for surface Cu+ ions can be understood considering that they have similar chemical hardness (both soft with η ≈ 6 eV),[44] whereas amines are harder Lewis bases than thiols. Nevertheless, amines have been shown to strongly bind to the surface of multinary Cu chalcogenide NCs, such as CuInS2 and Cu2ZnSnS4 NCs,[61−63] probably because these materials contain harder Lewis acids (In3+, Zn2+, Sn4+). In these works, surface-bound amines are characterized by broad signals in the 1H NMR spectra. Our solid-state and solution 1H NMR spectra (Figures S8 and S9) provide no evidence for surface-bound amines because the only broad signal observed corresponds to R-NH3+ species around 5 ppm. Broadening of thiol signals are hard to assign due to the high proton density corresponding to free thiol molecules, which give rise to signals at comparable chemical shifts. This makes it difficult to discriminate between free, surface-bound, and crystal-bound DDT.[61] Nevertheless, considering the evidence provided in ref (61), we assume that the surface of the polyhedral Cu2–S NCs prepared in the present work is capped by DDT (possibly crystal bound), whereas the amines have the adjuvant role of deprotonating the thiol headgroup (as discussed above).

Mechanism

On the basis of the findings discussed above, we propose a mechanism for the impact of SnBr4 and other Sn(IV) salts (e.g., Sn(OAc)4, SnCl4, Sn(acetylacetonate)Cl2) on the size, shape, and crystal structure evolution of Cu2–S NCs formed by heating Cu salts in DDT. First, the thiolate complexes of both Cu(I) and Sn(IV) (Cu(DDT) and Sn(DDT)4, respectively) are formed by replacement of the native anions by DDT. This reaction is facilitated by amines (OLAM in the present case), which deprotonate the thiol, forming a nucleophile that is sufficiently reactive to displace the native anions bound to Sn(IV). Our results (see NMR Spectroscopy section above) show that the replacement of Br by DDT occurs in a stepwise fashion, one Br at a time, until the fully substituted (C12H25S)4Sn complex is formed. The stability of Cu(I)–DDT is known to be higher than that of Cu(II)–DDT,[64] and therefore, when Cu(II) salts are used, the formation of the Cu–DDT complex is preceded by the reduction of Cu(II) to Cu(I) by oxidation of DDT to didodecyl disulfide.[41]

Copper(I) thiolates are very useful as single-source precursors for the synthesis of Cu2–S NCs and have been extensively used for that purpose, both in solventless- and solution-based routes using either hot-injection or heating protocols.[6,26,29,38,40,41,43,65,66] The rate limiting step in the formation of Cu2–S NCs from Cu(I)–thiolates has been shown to be the thermally induced cleavage of the C–S bond,[40] which is catalyzed by the Cu(I) atoms[67] so that only DDT molecules directly bound to Cu undergo thermolysis. For [Cu–S] monomers to be formed, several C–S bonds must be cleaved because each Cu atom is coordinated to four DDT molecules.[64] It is interesting to note that the observations discussed above imply that the reaction temperature used in our study (225 °C) is sufficiently high to lead to thermolysis of Cu–thiolate, followed by nucleation and growth of Cu2–S NCs, but is too low to induce significant thermolysis of the C–S bonds of the Sn(IV) thiolate complexes because [SnS] species are not observed. Instead, our study demonstrates that the active species are Sn(DDT)4 and/or the partially substituted complexes, such as Sn(DDT)Br3. The high thermal stability of Sn(DDT)4 is in line with the high decomposition temperatures reported for tin(IV) thiolates and dithiocarbamates, which require temperatures ranging from 250 to 375 °C to yield SnS2.[68]

To understand how the Sn(DDT)Br (x = 1–4; y = 0–3) complexes generated in situ affect the size and shape evolution of the Cu2–S NCs, we have to consider the formation mechanism of colloidal Cu2–S NCs upon heating of a solution of CuCl in excess DDT. As discussed above, Cu–DDT is initially formed, which is followed at sufficiently high temperatures by thermolysis of the C–S bonds of the Cu–DDT complex, thereby forming [Cu–S] monomers. As a result of the high activation energies associated with the C–S bond cleavage, the precursor to monomer formation becomes the rate-limiting step in the formation of Cu2–S NCs from Cu–DDT.[6,40] This is in line with the general mechanism proposed for the formation of colloidal NCs of several metal chalcogenides (e.g., CdSe, PbSe, PbS)[1,69−71] and implies that tuning the precursor conversion kinetics can dramatically affect the nucleation and growth rates, allowing the size, shape, and crystal structure of the nanocrystals to be controlled. This size control strategy has been recently illustrated by Owen and co-workers,[72] who used the reactivity of substituted thiourea precursors to tune the size of NCs of a number of metal sulfides, including Cu2–S. The authors demonstrated that increasing the thiourea reactivity produces a higher concentration of smaller NCs as a result of faster monomer formation rates and, consequently, faster nucleation rates.

We propose that the dramatic impact of the Sn(DDT)Br complexes on the formation of Cu2–S NCs is also due to changes in the nucleation rates. However, in contrast to the examples discussed above, in the present case the nucleation is directly affected by the Sn(DDT)Br complexes, which increase the activation energy for nucleation, thereby making it more difficult (Figure a). This additional nucleation barrier is imposed by the interaction between the Sn(DDT)Br complexes and the [Cu–S] monomers, which transiently form heterometallic polynuclear Sn(IV)–Cu(I) thiolate complexes, as observed in the NMR measurements discussed above. This also implies that the higher the concentration of Sn(DDT)Br complexes, the higher the nucleation barrier (Figure a). The monomer formation rates are not affected because they depend only on the thermolysis rates of the C–S bonds of the Cu–thiolate complexes. As a result, nucleation becomes the rate-limiting step, which impacts both the size and shape of the product Cu2–S NCs.

Figure 5

Tuning the nucleation and growth rates of Cu2–S NCs by Sn–thiolate complexes. (a) La Mer plots of nucleation and growth of Cu2–S nanocrystals (NCs) without (blue) and with Sn(IV) additives (red). The addition of Sn(IV)–thiolate complexes to Cu2–S NC synthesis results in an increase in the nucleation barrier without affecting the monomer formation rates. As a result, nucleation and growth are well separated, resulting in polyhedral NCs with narrow size and shape distributions. (b,c) Schematic representations of the two Cu2–S formation mechanisms: (b) without Sn additives, polydisperse Cu2–S nanoplatelets are obtained, and (c) with Sn additives, well-defined polyhedral Cu2–S NCs are formed.

In the absence of Sn(DDT)Br complexes, the activation energy for nucleation is low, but the monomer supply is limited by the C–S thermolysis rates. The polydisperse ensemble of relatively large nanoplatelets obtained in the absence of SnBr4 (Figure a) suggests that the heating rates employed in our experiments are not sufficiently fast to induce a sudden burst of C–S thermolysis and monomer formation, thereby resulting in relatively few nucleation events spread over a wide temperature range. The addition of a small concentration of SnBr4 (which is converted in situ to Sn(DDT)Br complexes) already significantly increases the activation energy for nucleation, delaying it until sufficiently high temperatures have been reached. The concentration of monomers produced by thermolysis of the Cu–DDT complexes will then be high, and a burst of nucleation followed by fast growth and depletion of the monomers becomes possible, leading to a high concentration of relatively small and monodisperse NCs (Figure b, 17 nm hexagonal bifrustums). Further increase in the concentration of added SnBr4 leads to a higher concentration of in situ-generated Sn(DDT)Br complexes, which make the nucleation rates increasingly lower while keeping the monomer formation rates unaffected. As a result, the size of the product Cu2–S NCs increases with increasing Sn(DDT)Br (Figure ) because fewer nuclei are formed under a constant monomer supply (Figure b,c). This also affects the morphology of the NCs because facet development during growth and the final shape adopted by a colloidal NC are dictated by a balance between several driving forces,[1] as will be discussed below.

The growth rate of a given NC facet depends on its free energy and on the total concentration of monomers available for growth.[1] At low monomer activities, the overall growth rates are slow, and therefore, the differences between different crystallographic facets are not significant (thermodynamically controlled growth regime).[1] Consequently, the NC will grow toward an equilibrium shape that minimizes its total free energy the most, which implies that a relatively isotropic shape exposing low free-energy facets will be favored. In contrast, at high monomer activities, the overall growth rates become fast, allowing the high free-energy facets to grow faster than the low free-energy ones, outcompeting them for the monomer supply. This leads to anisotropic morphologies (kinetically controlled growth regime).[1] Surfactants (ligands) modify the free energy of specific facets through preferential binding, thereby depressing their growth rates relative to the facets that are less densely capped. This affects the shape evolution under both growth regimes.

Colloidal Cu2–S NCs synthesized by thermolysis of Cu–DDT complexes have been shown to adopt a nearly spherical morphology at early growth stages or under slow growth conditions (i.e., under thermodynamic control) and a hexagonal nanoplatelet morphology under fast growth conditions (i.e., high concentrations and/or high temperatures).[6,26,29,38,40,41,43,65,66] This implies that the free energies of the facets in the [010] and [001] directions (namely, {101} and {221} for djurleite) are higher than that of the {100} facet. The different shapes of the Cu2–S NCs obtained in the presence of different concentrations of SnBr4 can thus be understood as a direct consequence of the impact of the in situ-generated Sn(DDT)Br complexes on the nucleation rates. As discussed above, the nucleation rates decrease with increasing concentration of Sn(DDT)Br complexes, which leads to an increasingly higher concentration of monomers available for growth, thereby enhancing the growth rates and favoring the formation of nanoplatelets with increasingly larger aspect ratios. The observation of a polydisperse ensemble of nanoplatelets in the absence of SnBr4 (Figure a) can be ascribed to a combination of slow nucleation rates (limited by the monomer formation rates) spread over a wide temperature range and fast growth rates. The difference in the crystal structures of the differently shaped polyhedral Cu2–S NCs (see above) can also be rationalized from this perspective, which implies that the low-chalcocite structure is thermodynamically favored, whereas the djurleite structure is kinetically favored. This is consistent with the fact that low-chalcocite is the thermodynamically stable crystal structure of Cu2S below 105 °C.[42]

Tin(IV) compounds (namely, Sn(acetylacetonate)Cl2 and SnCl4) have been previously reported to affect the shape of Cu2–S NCs obtained by heating Cu(II)acetylacetonate in DDT, leading to the formation of either nanosheets or nanodisks under conditions that would otherwise yield spherical NCs.[73−75] This effect has been tentatively ascribed to in situ-generated [SnS] species that were presumed to act as selective surfactants, thereby influencing the relative growth rates of different facets and thus altering the NC morphology.[73−75] No evidence was provided, however, for the presence of such species, either in the reaction medium or at the surface of the product Cu2–S NCs. It is possible that the Sn(DDT)Br complexes may also act as surfactants, but the NMR spectroscopy and elemental mapping results discussed above do not provide any evidence supporting to the notion that the formation of nanoplatelets is induced by selective binding of these complexes to the top and bottom facets because Sn was detected at very low concentrations and randomly distributed, regardless of the NC shape, and no Sn compounds were observed bound to the surface of the nanoplatelets. Surface-bound Sn(DDT)Br3 complexes were observed, however, in the solid-state 119Sn NMR spectra of the bipyramid-shaped Cu2S NCs. This suggests that Sn(DDT)Br complexes preferentially bind to the {302}/{502} and {364} facets, possibly because these facets have a higher free-energy and are less densely capped with DDT molecules than the top and bottom {102} facets. This adsorption is not strong enough, however, to significantly depress the growth rates of the {302}/{502} and {364} facets, but slows it sufficiently with respect to the {102} facets to allow for the formation of bipyramid NCs, which require growth both in the [100] direction and in the [010] and [001] directions. This mild down-modulation of the {302}/{502} and {364} growth is nevertheless insufficient to counteract the dramatic increase in the growth rates brought about by further increasing the concentration of the Sn(DDT)Br complexes because this results in lower concentration of nuclei and a large increase in the monomer concentration available for growth.

Conclusions

We have shown that in situ-formed Sn(IV)–thiolate complexes can be used as shape-directing agents that modulate the nucleation and growth rates of polyhedral Cu2–S NCs. Several anisotropic polyhedral NCs were obtained with narrow size and shape distributions (e.g., hexagonal bifrustums and bipyramids and hexagonal nanoplatelets with various aspect ratios) by solely changing the concentration of the additive SnBr4, which is converted in situ to Sn(IV)–thiolate complexes through a stepwise replacement of bromide by deprotonated thiols. The crystal structure of the product Cu2–S NCs is observed to depend on the concentration of Sn(IV)–thiolate complexes, resulting in monoclinic low-chalcocite and monoclinic djurleite in the low- and high-concentration limits, respectively. Our results rule out that the impact of the Sn(IV)–thiolate complexes on the formation of Cu2–S NCs is due to Sn incorporation in the growing NCs or to a surfactant effect. Instead, the Sn(IV)–thiolate complexes increase the Cu2–S nucleation barrier without affecting the precursor conversion rates. This influences both the nucleation and growth rates and leads to a better separation between nucleation and growth, thereby decreasing the ensemble polydispersity. It also dictates the final shape and structure of the product Cu2–S NCs by affecting the balance between the nucleation and growth rates under constant monomer formation rates. In the low-concentration limit, the nucleation rates are relatively fast, leading to a high concentration of nuclei and a low concentration of monomers available for growth. This results in slow growth under thermodynamic control. The nucleation rates decrease with increasing concentration of Sn(IV)–thiolate complexes, which leaves an increasingly higher concentration of monomers available for growth. This progressively enhances the growth rates, thereby shifting the growth to the kinetically controlled regime.

The use of inorganic ligands as shape-directing agents has not been extensively studied yet and could possibly lead to novel NC morphologies and improved size control strategies. This could be beneficial for implementation of NCs in, for example, photovoltaic and photonic devices for which well-defined building blocks are required to form highly ordered thin layers of individual NCs. In combination with topotactic cation exchange reactions, this could boost the importance of the synthetic strategy developed in this work, giving rise to tailor-made NCs and NC solids with novel functionalities, which may prove beneficial for a number of applications.