Limin Liuab,
Xianfeng Zhanga,
Li Jib,
Hanwen Lia,
Huijuan Yua,
Fangjie Xub,
Jianhua Hua,
Dong Yang*a and
Angang Dong*b
aState Key Laboratory of Molecular Engineering of Polymers and Department of Macromolecular Science, Fudan University, Shanghai 200433, China. E-mail: yangdong@fudan.edu.cn
bCollaborative Innovation Center of Chemistry for Energy Materials, Shanghai Key Laboratory of Molecular Catalysis and Innovative Materials, Department of Chemistry, Fudan University, Shanghai 200433, China. E-mail: agdong@fudan.edu.cn
First published on 15th October 2015
In the surface treatment of colloidal nanocrystals (NCs), S2− ions have been widely employed as metal-free atomic ligands to efficiently replace the original long hydrocarbon ligands. Prior studies exclusively show that S2− ions considerably quench the photoluminescence (PL) of semiconductor NCs (e.g., CdSe and PbS) during ligand exchange. Here we report that the influence of S2− treatment on the luminescent properties of CdSe NCs is highly dependent on the NC size. We observe an unexpected PL brightening phenomenon when small CdSe NCs (<4 nm) are subject to S2− treatment followed by incubation in the presence of air and light irradiation, whereas PL enhancement is not observed in large CdSe NCs (>4 nm) treated under the same conditions. Systematic characterization establishes the evolution of CdSe/CdS core–shell structures in small CdSe NCs arising from anion exchange between Se2− and S2−, which in conjunction with the subsequent incubation process accounts for the PL enhancement. Notably, 2.1 nm CdSe NCs treated with (NH4)2S exhibit a PL quantum yield (QY) as high as ∼40% after 2 days of incubation, which is comparable to that of conventional hydrophobic CdSe/CdS core–shell NCs synthesized at high temperatures. Our studies demonstrate that S2− ions can substantially substitute Se2− in small CdSe NCs in addition to replacing the surface-coating ligands, enabling highly luminescent, hydrophilic CdSe/CdS core–shell NCs at room temperature.
Among various small species recently developed for ligand exchange, S2− ions are emerging as a class of inorganic atomic ligands that are particularly suitable for surface treatment of semiconductor NCs.25–27 In this regard, S2− ions replace the original organic ligands, resulting into hydrophilic, S2−-passivated NCs that are dispersible in polar solvents having high dielectric constants such as formamide (FA).25 Notably, this facile ligand-exchange process can be accomplished within seconds with high exchange efficiencies, presumably due to the strong binding affinity between S2− and the partially coordinated metal cations at NC surface.26 Since the first demonstration by Talapin and coworkers,25 S2− ligands have been widely applied to the surface modification of semiconductor NCs.26–30 Moreover, CdSe and PbSe NCs treated with S2− have been used to construct electronic and optoelectronic devices with enhanced performance.28–30 Despite these benefits, S2−-ligand exchange has been demonstrated to be detrimental for the luminescent properties of semiconductor NCs.25–27 For instance, Talapin and coworkers reported that the PL quantum yield (QY) of 5.5 nm, OA-capped CdSe NCs dropped from 13 to 2% upon S2− treatment,25 and 4.2 nm, S2−-treated CdSe NCs exhibited an even lower QY at 0.7%.27 Similarly, Robinson and coworkers showed that the PL intensity of PbS NC films treated with (NH4)2S gradually decreased with treatment time.26 In addition, the PL lifetime was observed to decrease after S2− treatment by both groups,25,26 indicating that the reduced QY in CdSe and PbS NCs could be caused by non-radiative recombination channels arising from surface defects introduced by S2− ligands. Although the exact quenching mechanism is yet to be explored, prior studies unambiguously demonstrate that S2− treatment considerably quenches the PL of semiconductor NCs, which is undesirable for applications that require high quantum efficiencies such as light-emitting devices31 and biological labeling.32
In this work, we report systematic studies of S2−-ligand exchange of CdSe NCs with various sizes and reveal a striking size-dependent influence of S2− treatment on the luminescent properties of CdSe NCs. In contrast to the quenching results reported previously,25–27 we have found that S2− treatment can significantly enhance the PL of CdSe NCs having sizes less than 4 nm upon incubation in the presence of air and light irradiation, whereas PL enhancement is not observed in large CdSe NCs (>4 nm) treated under the same conditions. This unexpected brightening phenomenon is attributed to the evolution of CdSe/CdS core–shell structures in small CdSe NCs, in which the thin CdS shell is formed by anion exchange between Se2− and S2− occurring during ligand exchange, as consistently confirmed by absorption and PL spectroscopies, X-ray diffraction (XRD), and energy-dispersive X-ray spectroscopy (EDS). In addition, both air and light irradiation prove to be critical parameters that dictate the PL enhancement in S2−-treated CdSe NCs, presumably due to the photochemical annealing of structural defects initially presenting at CdSe/CdS interfaces.33,34 Notably, 2.1 nm CdSe NCs treated with (NH4)2S exhibit a PL QY as high as ∼40%, which is comparable to that of high-quality hydrophobic CdSe/CdS core–shell NCs obtained by high-temperature colloidal synthesis.34,35 Our studies establish that S2− treatment can be employed as a new route to synthesize highly luminescent CdSe/CdS core–shell NCs at room temperature.
(NH4)2S is chosen to demonstrate S2− treatment, although a similar size-dependent behavior is also observed when K2S is used for ligand exchange. For all samples, FTIR spectra confirm that the original organic ligands are nearly completely removed after 10 min of S2− treatment (Fig. S2†), suggesting that the ligand-exchange efficiencies are independent of the sizes, crystal structures, and organic capping ligands of CdSe NCs.
Transmission electron microscopy (TEM, Fig. 1) and the corresponding statistical analyses (Fig. S3†) indicate that the size of the S2−-treated CdSe NCs is almost identical to that of the original NCs. Fig. 2 shows the representative absorption and PL spectra of CdSe NCs before and after S2− treatment. As mentioned above, the PL spectra and QY of the S2−-treated CdSe NCs are measured after 2 days of incubation in the presence of air and room lights. For 7.7 and 4.3 nm CdSe NCs, the excitonic features in their absorption spectra do not change noticeably after (NH4)2S treatment (Fig. 2a and b). In addition, both samples retain their band-edge PL without obvious shift in peak position. However, both the PL QY and lifetime decrease upon ligand exchange (Table 1 and Fig. S4†), suggesting that S2− treatment degrades the surface passivation of 7.7 and 4.3 nm CdSe NCs. The decreased PL QY and lifetime are consistent with previously reported results based on 5.5 and 4.2 nm CdSe NCs,25,27 indicating that S2− treatment indeed has an adverse effect on the luminescent properties of CdSe NCs having sizes larger than 4 nm.
![]() | ||
Fig. 1 TEM images of different-sized CdSe NCs before (left column) and after (right column) (NH4)2S treatment: (a) 7.7 nm, (b) 4.3 nm, (c) 3.1 nm, (d) 2.1 nm. |
NC size (nm) | PL QY (%) | PL lifetime (ns) | ||
---|---|---|---|---|
Before treatment | After treatment | Before treatment | After treatment | |
7.7 | 0.7 | 0.3 | 16 | 9 |
4.3 | 1.2 | 0.6 | 29 | 17 |
3.1 | 1.9 | 9.4 | 38 | 48 |
2.1 | 1.6 | 39.8 | 43 | 58 |
It is interesting to note that the influence of S2− treatment on the optical properties of CdSe NCs is different when the NC size gets smaller. As shown in Fig. 2c, the excitonic feature is red shifted by 8 nm in the absorption spectra of 3.1 nm CdSe NCs, with the PL peak shifted from 581 to 589 nm. More pronounced spectra shifting is observed when we further decrease the size of CdSe NCs. In the case of 2.1 nm CdSe NCs (Fig. 2d), a large red shift of ∼40 and ∼30 nm is observed in absorption and PL spectra, respectively. More importantly, the PL quantum efficiencies of both NC samples are dramatically improved, which is in sharp contrast to the quenching phenomenon described above. The PL QY of 3.1 nm CdSe NCs is increased from 1.9 to 9.4%, while 2.1 nm CdSe NCs exhibit a QY as high as ∼40% upon S2− treatment, representing a 25-fold increase over the initial QY (∼1.6%, Table 1). The improved PL quantum efficiencies are also evident by comparing the luminescent photographs of NC solutions before and after ligand exchange (Fig. 2c and d, insets). In addition to the improved QY, (NH4)2S treatment also leads to an increase in PL lifetime for 3.1 and 2.1 nm CdSe NC samples (Table 1 and Fig. S4†). Given the PL quenching results obtained from 7.7 and 4.3 nm CdSe NCs, it is thus concluded that the influence of (NH4)2S treatment on the luminescent properties of CdSe NCs is highly dependent on the NC size. Using K2S for ligand exchange also improves the PL QY for small CdSe NCs (<4 nm) while quenching the PL for larger ones (>4 nm), although the PL enhancement is not as significant as the case of (NH4)2S treatment (Table S1 and Fig. S5†).
To understand the origin of this size-dependent behavior, XRD is performed to analyze the structural changes of CdSe NCs induced by S2− treatment. Fig. 3 shows the XRD patterns of S2−-treated CdSe NCs with varying sizes as well as the standard JCPDS stick patterns of CdSe and CdS bulk phases. The XRD patterns of individual CdSe NC samples before and after S2− treatment are also shown in Fig. S1.† Despite different crystal structures, the diffraction peaks of 7.7 and 4.3 nm CdSe NCs do not shift upon ligand exchange, matching the corresponding CdSe bulk phases (Fig. 3, green and pink curves). In contrast, the diffraction peaks of 3.1 nm, S2−-treated CdSe NCs systematically shift toward larger angles and lie in between those inherent to CdSe and CdS bulk phases (Fig. 3, blue curve). An even further shift is observed in the case of 2.1 nm, S2−-treated CdSe NCs, which display an XRD pattern nearly matching the CdS bulk phase (Fig. 3, red curve). As has been proposed previously, such a pronounced shift in XRD patterns could be ascribed to the formation of CdSexS1−x alloys37 or CdSe/CdS core–shell structures.38 However, the large red shifts observed in absorption and PL spectra in Fig. 2c and d rule out the alloy formation, because ternary alloyed CdSexS1−x NCs are expected to exhibit absorption and PL spectra that are blue shifted with respect to those of CdSe NCs due to the widened band gap energy.39,40 Therefore, XRD combined with absorption and PL spectroscopies strongly suggests the formation of CdSe/CdS core–shell structures in small S2−-treated CdSe NCs.41,42 It is noteworthy that our findings are in agreement with the results recently reported by Dukovic and coworkers,43 who demonstrated by experiments and calculations that CdTe NCs treated with chalcogenide ions (S2−, Se2−, and Te2−) could be regarded as core–shell NCs with thin ligand shells.
![]() | ||
Fig. 3 XRD patterns of (NH4)2S-treated CdSe NCs with different sizes. The XRD stick patterns of bulk CdSe (top) and CdS (bottom) phases are provided for comparison. |
EDS is employed to quantify the composition of CdSe NCs before and after ligand exchange (Fig. S6†). All the S2−-treated CdSe NC samples are subject to repeated washing and centrifugation to remove residual S2− ions, and to ensure more accurate measurements, EDS spectra are collected from multiple regions and the average Cd, Se, and S atomic percentages are used to determine the composition of each sample. As shown in Table 2, the atomic ratio of Cd/Se in original CdSe NCs is close to 1:
1, while S2− treatment leads to a decrease in Se atomic percentage accompanied by an increase in S atomic percentage for all CdSe NC specimens. The considerably increased Cd/Se ratios upon ligand exchange suggest that Se2− ions in NC lattices have been partially substituted by S2−, presumably through an anion-exchange reaction.44 Note that in all cases, the atomic ratio of Cd/(Se + Stotal) after S2− treatment is still close to 1
:
1, implying that the content of the surface-bound S2− ligands (SL) contributes insignificantly to the total S content (Stotal). The S atomic percentage in NC lattices (SNC) arising from anion exchange can be roughly calculated (Table 2), considering that the atomic ratio of Cd/(Se + SNC) in S2−-treated CdSe NCs is equal to the initial Cd/Se ratio before S2− treatment. The SNC values determined are then plotted as a function of NC diameters in Fig. 4, which clearly shows that the degree of anion exchange is highly sensitive to the size of NCs. For example, ∼43.57% of Se2− ions in 2.1 nm CdSe NCs undergo anion exchange during S2− treatment, whereas only ∼8.51% of Se2− ions in 7.7 nm CdSe NCs are replaced by S2− under the same conditions (Table 2). The higher degree of anion exchange in smaller CdSe NCs could be attributed to their higher surface to volume ratios, as anion exchange is expected to proceed from NC surface through inward diffusion, similar to the case of cation-exchange reactions.45,46 The degree of anion exchange resulted from K2S treatment is also dependent on the size of CdSe NCs (Fig. S7†).
NC size (nm) | Before treatment | After treatment | dCdSe (nm) | ||||||
---|---|---|---|---|---|---|---|---|---|
Cd (%) | Se (%) | Cd (%) | Se (%) | Stotala (%) | SNCb (%) | SLc (%) | Sered (%) | ||
a The total content of S.b The atomic percentage of S in CdSe NCs arising from anion exchange.c The atomic percentage of S in surface-coating ligands.d The percentage of Se2− replaced by S2− by anion exchange: Sere = SNC/(SNC + Se).e The thickness of CdS shells. | |||||||||
7.7 | 51.02 | 48.98 | 49.90 | 43.83 | 6.27 | 4.07 | 2.20 | 8.51 | 0.11 |
4.3 | 49.98 | 50.02 | 48.07 | 41.73 | 10.19 | 6.38 | 3.81 | 13.26 | 0.10 |
3.1 | 52.60 | 47.40 | 50.83 | 28.56 | 20.61 | 17.24 | 3.37 | 37.64 | 0.23 |
2.1 | 51.13 | 48.87 | 46.89 | 25.29 | 27.82 | 19.53 | 8.29 | 43.57 | 0.18 |
![]() | ||
Fig. 4 Plots of atomic percentages of SNC and Se as a function of NC diameters upon (NH4)2S treatment. |
The thickness of CdS shells can be estimated by comparing the volume of the CdSe core and the CdS shell,47 assuming that CdS shells formed on the surface of CdSe NCs are smooth and continuous (Fig. S8†). As listed in Table 2, the thickness of CdS shells in 7.7, 4.3, 3.1, and 2.1 nm CdSe NCs is determined to be 0.11, 0.10, 0.23, and 0.18 nm, respectively. Given that a monolayer of CdS is about 0.21 nm in thickness,47 we speculate that S2− treatment leads to the formation of a monolayer CdS shell in small CdSe NCs (2.1 and 3.1 nm), whereas in the case of large CdSe NCs (4.3 and 7.7 nm), the resulting CdS shell is not thick enough to support a complete monolayer. The incomplete CdS coating also explains the PL quenching phenomenon observed in large CdSe NCs. We should point out that even in the case of small CdSe NCs with a large S content, the slight contrast difference between CdSe and CdS makes it hard to distinguish the core and the shell by TEM.
It is also noteworthy that the reduced core size arising from the formation of CdS shells in small CdSe NCs can lead to a blue shift in absorption and PL spectra, the magnitude of which may not be overcompensated by that of the red shift caused by the CdS shell formation. In other words, the large red shift should be caused by some other reasons in addition to the formation of CdSe/CdS core–shell structures. We note that CdSe NCs, especially 3.1 and 2.1 nm CdSe NCs, tend to aggregate or even fuse after S2− treatment (Fig. 1c and d). Such NC aggregation and fusion could lead to a significant red shift, due to the enhanced interparticle interactions. We speculate that the magnitude of the red shift induced by the core–shell formation as well as NC aggregation and fusion eventually overcompensates that of the blue shift caused by the decreased core size, leading to the observed red shift in absorption and PL spectra.
To further investigate the brightening mechanism, a series of control experiments are carried out, in which the S2−-treated CdSe NCs are exposed to different environments for 12 h before PL measurements. As shown in Fig. 5b, NC samples incubated in dark do not show apparent PL brightening regardless of the presence of air (Fig. 5b, pink and green curves), suggesting that light irradiation is an important parameter for PL enhancement. On the other hand, when the sample incubated in a N2-purged glovebox is exposed to light irradiation, only a slight improvement in PL intensity is observed (Fig. 5b, blue curve), implying that air also plays a key role in PL brightening. Given that significant PL enhancement is achievable only when the S2−-treated CdSe NCs are exposed to both air and light irradiation (Fig. 5b, red curve), we surmise that the complex interplay between these two parameters dictates the PL enhancement.
Photoannealing is a commonly used strategy to improve the PL QY of core–shell nanostructures. For example, Buhro and coworkers reported that during surface treatment of CdTe nanowires with ethanethiol, photoannealing in the presence of O2 promoted the formation of an epitaxial CdS shell on the nanowire surface by anion-exchange reactions between Te2− and S2−.47 Likewise, Peng and co-workers observed PL brightening upon exposing CdSe/CdS core–shell NCs to light irradiation and oxygen.34 In our work, the evolution of CdS shells is found to be independent of air or light irradiation, as anion exchange between Se2− and S2− takes place even in the absence of air and light. However, PL brightening could be observed only when the S2−-treated CdSe NCs are exposed to both air and light irradiation, as shown in Fig. 5b. Based on these observations, we speculate that light irradiation in the presence of air triggers structural reorganizations at CdSe/CdS interfaces through a photochemical process. This enables the elimination of structural defects originally presenting at CdSe/CdS interfaces, resulting in an ideal type-I heterostructure as schematically illustrated in Fig. 5c. We emphasize that such a minor structural change may not be detected by XRD. The presence of a considerable amount of structural defects at CdSe/CdS interfaces is reasonable if one considers that anion exchange takes place at room temperature, such that the resulting CdS shells may not be smooth and well-defined due to the presence of unexchanged Se2− and/or lattice mismatch between CdSe and CdS. In addition, light irradiation could also induce surface reconstructions, contributing to the PL enhancement by decreasing the density of surface trap states.14,34 The improved interface and surface properties are also consistent with the increased PL lifetime as described above.
Finally, it is worth to mention that the PL enhancement in S2−-treated CdSe NCs can be accelerated by additional thermal annealing during the course of incubation. Fig. 6a shows the PL spectra collected from 2.1 nm, S2−-treated CdSe NCs which have been incubated at different temperatures for 12 h. Compared with the PL intensity of CdSe NCs incubated at room temperature (20 °C, Fig. 6a, black curve), incubation at 50 °C leads to a higher PL intensity without noticeable shift in PL peak position (Fig. 6a, red curve). In addition, the time required to achieve the maximum PL intensity is reduced to 1 day (Fig. 6b). Presumably, the accelerated PL enhancement is attributed to the facilitated interfacial and/or surface reconstructions induced by thermal annealing. Note that although the PL enhancement can be further accelerated by increasing the annealing temperature to 80 °C, the pronounced red shift (∼13 nm) in PL spectra suggests the occurrence of NC aggregation or ripening (Fig. 6a, blue curve).
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra18192e |
This journal is © The Royal Society of Chemistry 2015 |