Amaresh
Shukla
a,
Shouqi
Shao
b,
Sadie
Carter-Searjeant
a,
Sarah
Haigh
b,
David
Richards
a,
Mark
Green
*a and
Anatoly V.
Zayats
*a
aDepartment of Physics and London Centre for Nanotechnology, King's College London, Strand, London WC2R 2LS, UK. E-mail: a.zayats@kcl.ac.uk; mark.a.green@kcl.ac.uk
bDepartment of Materials, University of Manchester, Oxford Road, Manchester M13 9PL, UK
First published on 24th January 2023
Copper sulphide (covellite) nanoplatelets have recently emerged as a plasmonic platform in the near-infrared with ultrafast nonlinear optical properties. Here we demonstrate that the free-carrier density in CuS, which is an order of magnitude lower than in traditional plasmonic metals, can be further tuned by chemical doping. Using ion exchange to replace Cu with an increasing content of Zn in the nanoparticles, the free-hole density can be lowered, resulting in a long-wavelength shift of the localised plasmon resonances from 1250 nm to 1750 nm. The proposed approach provides new opportunities for tuning the plasmonic response of covellite nanocrystals as well as the carrier relaxation time which decreases for lower free-carrier densities.
Dopant-induced tuning of the plasmon resonance has been achieved in metal oxide nanoparticles, where the post addition of p-type dopant metal ions resulted in varying the free-carrier concentration and hence the plasma frequency.14 The control over a dopant radial distribution in metal oxide nanoparticles may also result in the variation of the plasmonic response when a core–shell geometry is formed.15
Copper monosulfide (CuS) is a naturally occurring layered chalcogenide with a free-carrier concentration of n ≈ 3 × 1027 m−3,16 which is an order of magnitude smaller than in noble metals, such as Au or Ag, making CuSan excellent candidate for plasmonic applications in the near-infrared. The anisotropic nature of the crystal lattice means that CuS nanoplatelets support a variety of LSP resonances in and around the second biological spectral window (1000 nm–1300 nm) and close to the optical communication wavelength range.16 The LSP wavelength of CuS nanocrystals can be tuned in the usual way through their size and shape,17–20 as well as by controlling crystal structure21 and chemical tuning.22 The low free-carrier density in vacancy doped copper chalcogenides is favorable for LSP control23via electrochemical charging/dischaging or chemical doping as has been shown for some other materials like Cu2−xSe or indium oxide.24,25 CuS has a partially filled valence band, making it p-type plasmonic material due to the holes related to the 3p and 3s orbitals of S and 3d orbitals of Cu.26 This opens up an opportunity to control the hole concentration in CuS by exchanging Cu ions with other suitable cations, thereby controlling the position of the LSP resonance.
Cation exchange in Cu2−xS nanoparticles occurs at high reaction temperatures with the Cu cations being replaced by another cation diffusing in the interstitial sites.27,28 This process has been used to transform Cu4SnS4 to Cu2ZnSnS4 nanoparticles29 and also to transform Cu2S nanocrystals to CuInS and CuInZnS nanocrystals by exchanging cations with (In3+) and (Zn2+).30 The prime objective of these and similar works was to create new nanoparticles with a required photoluminescence spectrum. However, there has been not much attempt to control the plasmonic properties of chalcogenide nanocrystals via cation exchange, focused on modifying the resulting carrier density. The recent report described the chemical doping of CuS nanoplatelets by the addition of Sb during synthesis, with the dopants located near the surface of the particles. The obtained LSP was tuneable in the range from ca. 650 nm to ca. 2100 nm by the addition of varying amounts of Sb.31
In this work, we use the ion interchange mechanism to modify the free-hole density in CuS nanocrystals replacing Cu-cation with Zn, in order to enable a widely tuneable LSP platform. The resulting Zn-doped CuS nanoparticles exhibit LSP resonances that depend on the doping, shifting from 1250 nm to 1750 nm with increasing Zn doping, as the hole concentration decreases. Thin films of these nanoparticles exhibit optical properties consistent with the observed plasma frequency changes. The control over a free-carrier density was also favorable for tuning the relaxation time of hot-carriers in nanocrystals,32 required for optimisation of this nanomaterial in photochemical and nonlinear optical applications.
A Zn precursor solution (1 M) was prepared by dissolving zinc diethylthiodicarbamate in 10 ml octadecene and maintained at 70 °C in nitrogen atmosphere under stirring. The precursor was injected into the CuS solution under stirring at room temperature and under a nitrogen atmosphere. The contents were then steadily heated and aliquots taken at defined temperatures points (90 °C, 120 °C, 150 °C, 180 °C, and 230 °C). Ethanol was added to the samples and centrifuged at 4500 rpm for 5 min. The precipitates were washed with methanol and dissolved in chloroform. Repetitive cycles of precipitation (by adding an equal volume of ethanol and centrifuging at 4500 rpm for 5 min) and washing were performed and then finally the samples were dissolved in chloroform.
In this way, CuS:
Zn nanoparticle samples with different concentrations of Zn were obtained from the base CuS nanoparticles. For comparison, original CuS nanoparticles were treated at the same temperatures and conditions but without the Zn precursor present in order to evaluate potential modifications of the nanoparticles due to the temperature. For detailed structural characterisation, we chose to examine the samples obtained at 90 °C and 230 °C.
Energy dispersive X-ray spectroscopy (EDXS) was performed on the Titan SREM using the Super-X EDXS system (collection angle 0.7 srad) and quantified using the Cliff Lorimer analysis without absorption correction (Fig. 2). STEM-EDXS elemental quantification (Table 1) of the CuS:
Zn samples obtained at both the lowest (90 °C) and highest temperature (230 °C) all showed Cu and Zn as well as the expected S content consistent with the single crystal lattice structure (∼48 at% at 90 °C and ∼46 at% at 230 °C). The Cu
:
Zn content for individual particles was ∼1
:
1 for samples obtained at 90 °C but showed large variation at 230 °C (from 1
:
4 to 3
:
2). The 230 °C particles appear to have no nonuniform compositional variations, whereas the 90 °C particles have a Zn rich surface layer. High-angle annular diffraction field (HAADF) STEM measurements (Fig. 2) and the EDX quantifications for the CuS
:
Zn samples obtained at both lowest (90 °C) and highest (230 °C) temperatures show the presence of Cu, Zn and S, all in significant proportions. At the same time, no shell structure around the nanoparticles is observed. (Such an outer ZnS shell over the original CuS nanoparticles might influence the LSP resonance.) Although the zinc dithiocarbamate was added to the reaction after the formation of the CuS particles, the deposition of the precursor did not result in a CuS/ZnS core/shell structure, rather cationic exchange appears to have occurred, resulting in what we term a CuS
:
Zn material. A Zn rich surface is observed for particle 6 (Fig. 1) – possibly resulting from faster ion exchange at a stacking fault (Fig. 2(g–m) and Table 1).
Nanoparticles | Cu at% | Zn at% | S at% |
---|---|---|---|
1 – 230 °C | 10.0 | 43.3 | 46.7 |
2 – 230 °C | 25.8 | 26.3 | 47.9 |
3 – 230 °C | 34.9 | 20.0 | 45.1 |
4 – 90 °C | 30.3 | 21.8 | 47.9 |
5 – 90 °C | 28.4 | 23.3 | 48.2 |
6 – 90 °C | 25.1 | 27.0 | 47.9 |
7 – 90 °C | 28.5 | 23.2 | 48.3 |
8 – 90 °C | 18.9 | 33.3 | 47.8 |
Nanoparticles prepared at 90 °C displayed an almost homogeneous distribution of Cu and Zn, with a slight predominance of zinc at the surface as might be expected in a lower temperature reaction (Fig. 2). The percentages of copper and zinc highlight that both elements varied in differing particles, both between 20% and 30% (Table 1), whilst the percentage of sulfur was relatively stable at 48%, highlighting that this was indeed cationic exchange and that the dithiocarbamate did not add further sulfur to the particle. Nanoparticles prepared at 230 °C displayed approximately similar elemental distributions, although occasional particles displayed a predominance of zinc with percentages significantly exceeding those of the parent copper (Fig. 2(a), particle 1, and Table 1). Again, the percentage of sulfur remained approximately constant at 45% to 48%; the higher synthesis temperature accounting for the lower sulfur percentage. This is consistent with previous reports on cation exchange in both CuS29,36 and Cu2−xSe37 systems.
The accompanying X-ray diffraction (XRD) data for the nanoparticles show a diffraction pattern consistent with the previous reports of covellite structure of CuS nanoparticles, with no apparent contribution from ZnS (Fig. 3). This suggests that the overall material maintains the covellite crystalline structure following cationic exchange with Zn. No discernible differences were observed in the XRD patterns taken from the nanoparticles prepared at different temperatures, confirming that the particles did not grow significantly further in size at the elevated temperature. A slight shift (≈1°) to a lower reflection angle was observed at the higher synthesis temperatures, consistent with a slight increase in a lattice parameter.
In the control experiments, in order to discount the possibility of the shape or size modification of CuS nanoparticles under treatment at increased temperatures, the Zn precursor during heating was replaced with pure sulfur. No appreciable LSP shift was observed under these conditions (Fig. 4(b)). A very small reduction of the FWHM can be related to Ostwald ripening inducing a smaller size dispersion for the annealed nanoparticles.38
The dependence of the LSP position on the Zn concentration in the nanoparticles is indicative of the changes in the hole density in the nanoparticles upon Zn doping. The hole concentration decreases when Zn substitutes Cu, resulting in the lower plasma frequency and, therefore, the long-wavelength shift of the LSP resonance which depends on the plasma frequency.
The detailed modelling of the LSP in anisotropic nanoparticles requires numerical treatment of the plasmonic response with the known permittivity of the material. In order to obtain qualitative understanding of the LSP shift and estimate the changes to the free-carrier concentration, we model the nanoplatelets as oblate spheroids and neglect the anisotropy of the material. This shape approximation was shown to provide a relatively good correspondence with the dominant hexagonal shape of the nanoplatelets for the polarisation of light parallel to the platelets.16 The approximation of materials as isotropic can be justified by the dominant contribution of εx, the in-plane component of the permittivity of nanoplatelets, to the LSP response of CuS nanodisks.16
In this approximation, the polarisability of the subwavelength spheroidal nanoparticle along the principal axis is given by39
![]() | (1) |
n = AωLSP2 | (2) |
![]() | ||
Fig. 5 Hole density in CuS![]() ![]() |
As can be seen from the observed LSP behaviour, the carrier density for CuS:
Zn nanoparticles decreases almost 2 fold with the increasing temperature of synthesis, when more Zn substitutes Cu in the nanoparticles.
The near-infrared transmission spectra of the films of the nanoparticles with different Zn concentration show similar spectral behaviour with the absolute transmission strongly dependent on the doping (Fig. 6). The transmission spectra can be approximated considering the thin films of the nanoparticles as an effective medium. Assuming that the nanocrystals in the films are oriented in the same way with the z-axis normal to the interface, so that the same εx component of the permittivity is responsible for the optical response at normal incidence, the transfer matrix model41 for a thin film on a thick glass substrate can be used to evaluate the transmission spectra. Large number of free parameters (up to 10) is required in order to account for both the Drude and Lorenz parts of the permittivity due to a complex band structure of the material, even if the thickness of the film is fixed and a plasma frequency is taken from Fig. 5. This prevents reliable quantitative fitting and determination of the effective parameters due to overparametrisation. Nevertheless, the transmission spectra obtained with this model indicate qualitative agreement with the experimental spectra, when using relative differences in the carrier concentration in eqn (2) for the Drude part of the effective permittivity.
This journal is © The Royal Society of Chemistry 2023 |