Introduction

Copper sulfide Cu2−xS nanocrystals (NCs) degenerately doped with copper vacancy, exhibit localized surface plasmon resonance (LSPR) because of the collective oscillation of free holes with incident light [1,2,3]. Because of the much lower carrier (hole) density (Nh ≈ 1020–1021 cm−3) compared with noble metals (Ne ≈ 1023 cm−3), the Cu2−xS NCs exhibit LSPR responses in near-infrared (NIR) and mid-infrared (MIR) regions [4]. Unlike plasmonic metal such as Cu, Ag and Au with face-centered cubic (fcc) crystal phase, Cu2−xS possesses a variety of crystal phases varying from copper-abundant chalcocite Cu2S, djurleite Cu1.94S, roxbyite and digenite Cu1.8S to copper-poor covellite CuS phase [5]. Cu2−xS NCs with different crystal phases exhibit distinct LSPR response, for instance djurleite Cu1.94S nanodisks (NDs) have two broad LSPR peaks in NIR and MIR regions, while covellite CuS NDs have only one narrow peak in NIR region [6, 7]. Due to the unique crystal phase-dependent LSPR properties, djurleite Cu1.94S was applied for cathodes in all-vanadium redox flow batteries, and covellite CuS was utilized as a photothermal agent and an electron donor to promote the photocatalytic water splitting efficiency of TiO2 [8,9,10].

In addition to the factor of crystal phase, the LSPR properties of Cu2−xS NCs could be finely manipulated by particle size, morphology, hole density, as well as plasmon coupling [11,12,13,14]. In chalcocite Cu2S/Cu2−xS NDs, the LSPR peak position was tuned by ND size. The out-of-plane oscillation peak blue-shifted with increasing size, while the in-plane oscillation peak red-shifted [11]. Besides particle size, the LSPR peak position and intensity were adjusted by morphology in chalcocite Cu2S/Cu2−xS nanospheres (NSs) and nanorods (NRs) [12]. Furthermore, in chalcocite Cu2S NSs, the hole density was increased by iodine oxidation and decreased by decamethylcobaltocene reduction, respectively. Hole density severely affected the LSPR peak position, absorption intensity and peak width of chalcocite Cu2S NSs [13]. In assembled thin films, covellite CuS NDs exhibited much stronger plasmon coupling effect than that of chalcocite Cu2S NDs and digenite Cu1.8S NDs [14]. However, previous researches mainly focused on the LSPR tuning of chalcocite Cu2S phase, and research on other crystal phases has been rarely demonstrated. On account of crystal phase not only determine the hole density and copper ion mobility, but also the hole distribution in the sulfur lattices. The aforementioned tuning approaches for chalcocite Cu2S could not be simply applied to other crystal phases, since the influence factors of LSPR were correlated with each other. Therefore, systematic investigation on LSPR tuning of Cu2−xS NCs with a special crystal phase instead of the chalcocite Cu2S is fundamentally interesting.

We selected roxbyite Cu1.8S with moderate hole density as a model and systematically tuned their LSPR. Such investigation could provide deeper understanding of crystal phase-dependent LSPR and further development of potential applications of roxbyite Cu1.8S. Herein, uniform Cu1.8S NDs were wet-chemically synthesized and their LSPR properties such as peak position, absorption intensity and peak width were flexibly tuned. Particle size, hole density via chemical oxidation and reduction, self-assembly and disassembly in solution, as well as plasmon coupling in solid film, severely affected the LSPR of Cu1.8S NDs. Therefore, the large tunability of LSPR suggests roxbyite Cu1.8S ND is a potential candidate for plasmonic photocatalysis, photothermal therapy, two-photon photochemistry and many others in NIR region.

Experimental details

Synthesis of Cu1.8S nanodisks (NDs) with roxbyite phase

The synthesis of roxbyite Cu1.8S NDs was modified from a previous report [15]. Typically, 1.5 mmol of copper (II) stearate, 4.0 mmol of N,N-dibutylthiourea (DBTU) and 3.0 mL of oleylamine (as reducing agent, ligand and solvent) were added in a 100-mL test tube and heated at 80 °C at the rate of 5 °C/min under N2 atmosphere. After further stirring at 80 °C for 15 min, 30 min and 60 min, the mixture was cooled to room temperature and purified with toluene and ethanol by centrifugation and finally dispersed in toluene and/or chloroform (CHCl3). The prepared samples were stored in the glovebox for further characterizations.

Characterizations

TEM observations were carried out using a JEM-1011 transmission electron microscope (JEOL) at an accelerating voltage of 100 kV. SEM observations were carried out using an S-4800 (HITACHI) Field-Emission Scanning Electron Microscope (FE-SEM) at an accelerating voltage of 5 kV. UV–Vis–NIR (300–2500 nm) absorption spectra of individual and coupled Cu1.8S NDs and corresponding multilayer films were measured using U-4100 spectrophotometer (HITACHI). The XRD patterns were taken on X’Pert Pro MPD (PANalytical) with CuKα radiation (λ = 1.542 Å) at 45 kV and 40 mA. In order to exclude the air oxidation effect on the samples, the XRD measurement should be carried out in inert atmosphere. Briefly, the samples deposited on the substrate were dried in glovebox and transported to N2-blanketed XRD holder for measurement immediately.

Results and discussion

Tuning of LSPR by particle size

An organic sulfur source DBTU was utilized to synthesize uniform roxbyite Cu1.8S NCs. Figure 1a–c shows the prepared hexagonal disk-shaped NCs at the reaction time of 15 min, 30 min and 60 min. The diameter of the nanodisks (NDs) varied from 9.3 to 13.4 nm and to 14.9 nm, while the thickness varied from 3.3 to 3.8 nm and to 4.0 nm. The larger increment in diameter demonstrates that the NDs were tightly enclosed by oleylamine (OAm) molecules in basal crystal planes which resulted in faster deposition of copper and sulfur species along the lateral directions. The OAm-capped NDs individually dispersed in toluene and chloroform (CHCl3), and no coupled ND assemblies were observed. The copper vacancies (free holes) in the Cu1.8S ND collectively oscillate with incident light which led to a broad and intense absorption in NIR region (Fig. 1d) [15]. Because of a retardation effect caused by uneven electric field across the particle, the LSPR peak of Cu1.8S NDs red-shifts from 1500 nm to 1660 nm and to 1706 nm with increasing aspect ratio [16, 17]. The growth of roxbyite Cu1.8S with reaction time did not decrease the hole density, which was derived from the case of digenite Cu2−xS NCs [18]. Large variation in LSPR response tuned by particle size suggests the Cu1.8S NDs are capable of size-dependent plasmonic applications such as photothermal therapy and photocatalytic organic synthesis [19, 20]. Noticeably, roxbyite Cu1.8S NDs exhibit single LSPR peak which is different from djurleite Cu1.94S and digenite Cu1.8S NDs with two distinct LSPR peaks [14]. The singular LSPR peak in roxbyite Cu1.8S NDs mainly originates from in-plane oscillation mode, resulting in enhanced electric field distribution at sharp corners of the ND (Fig. 1e) [21, 22].

Figure 1
figure 1

TEM images of prepared roxbyite Cu1.8S NDs by control of the reaction time at a 15 min, b 30 min and c 60 min. d UV–Vis–NIR absorption spectra of Cu1.8S NDs with different aspect ratios (aspect ratio = AR: diameter/thickness). e Scheme of in-plane collective oscillation of free holes with incident light. Blue arrow: anti-electric field in the ND. Red region at corners of ND: enhanced electric field induced by LSPR

Full size image

Tuning of LSPR by hole density via chemical oxidation and reduction

In addition to particle size, the LSPR of roxbyite Cu1.8S NDs was tuned by air-induced oxidation and 3-mercaptopropionic acid (MPA)-induced reduction. The freshly prepared, a certain amount of Cu1.8S NDs were dispersed in 5.0 mL toluene and then exposed to air with magnetic stirring. Time evolution of LSPR absorptions was recorded afterward (Fig. 2a). The LSPR peak gradually blue-shifted, and the absorption intensity increased with oxidation time. This variation of LSPR can be ascribed to an increase in hole density, which has been well investigated in Cu2−xSe NCs [23]. Oxygen extracted Cu+ in the crystal lattice of roxbyite Cu1.8S and oxidized the Cu+ into Cu2+ species. Recently, Kaseman et al. reported the air-induced oxidation of OAm-capped Cu2−xSe NCs. Two distinct Cu2+ surface species assigned to Cu2+ bound to OAm (Cu2+–OAm) and Cu2+ located in CuO domain were identified [24]. It is likely that the present OAm-capped roxbyite Cu1.8S NDs experienced similar Cu+/Cu2+ transformation under air oxidation. The ND size before and after air oxidation is almost same. A decrease in particle size leads to a blueshifted LSPR peak as well as weakened absorption intensity. In addition, an increase in hole density leads to a blueshifted LSPR peak but enhanced absorption intensity [25]. Therefore, the variation of LSPR response under air oxidation is caused by an increase in hole density rather than a decrease in particle size. The hole density varied from Nh1 = 4.47 × 1021 cm−3 to Nh2 = 5.88 × 1021 cm−3 upon air-induced oxidation. The details of hole density calculation are shown in the supplementary information. Besides LSPR variation, the band gap of Cu1.8S NDs slightly increased possibly due to a Moss–Burstein effect (Fig. 2b) [26]. Tunable hole density of Cu1.8S NDs by simple oxidation is a unique property beyond plasmonic metals. Because of the intense absorption in NIR wavelengths, the oxidized Cu1.8S NDs were applied as a plasmonic sensor. The sensitivity factor was estimated to be 558 nm/RIU (Fig. S1), which is much larger than that of gold nanoparticles in visible wavelengths [27].

Figure 2
figure 2

a Time evolution of UV–Vis–NIR absorption spectra of roxbyite Cu1.8S NDs under air oxidation. b Scheme of air oxidation-induced copper vacancy generation and band gap variation in Cu1.8S ND. C.B. conduction band, V.B. valence band. Red and blue double-headed arrow: band gap of Cu1.8S ND before and after air oxidation, respectively

Full size image

Afterward, the oxidized roxbyite Cu1.8S NDs dispersed in 5.0 mL toluene were treated with 5.0 μL of MPA by shaking with hand under N2 atmosphere (Fig. 3a). The amphiphilic MPA molecules were generally utilized in ligand exchange with the hydrophobic –SH group tightly capping the metal sulfide NC surface, while the hydrophilic –COOH group gifting the NC favorable dispersity in aqueous solutions [28]. In the present study, the proton in –SH group provides MPA reducing ability to fill the copper vacancies in the crystal lattice of roxbyite Cu1.8S. The LSPR peak gradually red-shifted, and the absorption intensity decreased with reduction time. This variation of LSPR is due to a decrease in hole density (Nh3 = 5.20 × 1021 cm−3 after MPA reduction), which has been demonstrated in Cu2−xSe NCs [23]. The decrease in hole density arising from the pinning effect near the ND surface by MPA molecules was excluded. It is suggested that the MPA molecules reduced surface Cu2+ species of the Cu1.8S ND through the formation of disulfide and/or coordination of MPA to surface copper ions [29, 30]. The reduced copper species filled the holes in the valence band of Cu1.8S, and the band gap was slightly decreased (Fig. 3b).

Figure 3
figure 3

a Time evolution of UV–Vis–NIR absorption spectra of roxbyite Cu1.8S NDs under MPA reduction. b Scheme of MPA reduction-induced hole filling and band gap variation in Cu1.8S ND. C.B. conduction band, V.B. valence band. Red and blue double-headed arrow: band gap of Cu1.8S ND before and after MPA reduction, respectively. c TEM image of MPA-reduced Cu1.8S NDs. d XRD patterns of as-synthesized, air oxidized and MPA-reduced Cu1.8S NDs

Full size image

It is noticeable that MPA reduction also resulted in copper and sulfur atoms at the sharp corners of hexagonal ND migrating to inner crystal lattices of Cu1.8S and thus the morphology evolution to rounded plates (Fig. 3c). This observation was similar to a previous report which showed the removal of sulfur and Cu atoms moving to new lattice positions in covellite CuS, leading to the crystal phase and morphology evolution after 1-dodecanethiol (1-DDT) treatment [31]. The air-induced extraction of Cu+ in the crystal lattice and MPA-induced reductive filling of holes were confirmed by XRD measurement (Fig. 3d). Air oxidation of freshly synthesized roxbyite Cu1.8S NDs caused a slight shift of (00\( \bar{8} \)) peak to higher diffraction angle, while the subsequent MPA reduction recovered the diffraction angle. According to the Bragg equation, an increase in diffraction angle suggests the lattice contraction of the Cu1.8S ND, which resulted from the extraction of Cu+ induced by air. On the contrary, a decrease in diffraction angle means the lattice expansion of the Cu1.8S ND, which resulted from the filling of holes or insertion of coppers induced by MPA. Furthermore, the obvious shift of (00\( \bar{8} \)) diffraction peak in XRD pattern implies higher mobility of Cu+ in (00\( \bar{8} \)) crystal plane than that in other crystal planes such as (1\( \overline{26} \)), (4 \( \overline{36} \)) and (800) in the roxbyite Cu1.8S. The variation of LSPR response by air-induced oxidation and MPA-induced reduction were ceased after a period of time (Fig. S2). This “LSPR fixing effect” is consistent with a previous report [32]. In addition to LSPR response, the XRD peak shift of Cu1.8S NDs was also ceased after the completion of oxidation and reduction process (Fig. S3).

Tuning of LSPR by self-assembly and disassembly in solution

Plasmon coupling is an effect caused by the Coulomb interaction between adjacent NCs in an assembly, which induces large variation of LSPR spectrum, intensity, as well as the local electric field spatial distribution, and polarization [33, 34]. Therefore, this effect provides a facile approach to tune the plasmonic properties of metals and semiconductors. In order to trigger the self-assembly, a certain amount of roxbyite Cu1.8S NDs dispersed in 5.0 mL CHCl3 were sonicated with 5.0 μL of MPA under N2 atmosphere. The LSPR peak of Cu1.8S NDs red-shifted from 1596 to 1682 nm, and the absorption intensity increased after MPA treatment (Fig. 4a). This variation of LSPR response was derived from a shoulder-to-shoulder self-assembly of Cu1.8S NDs (Fig. 4c). The interparticle distance changed from 2.4 nm (Fig. 1c) to 1.5 nm by MPA treatment, and thus, a strong shoulder-to-shoulder in-plane plasmon coupling effect was expected. This fashion of self-assembly occurred probably due to the hydrogen bond between two –COOH groups of the MPA molecules capped on the shoulders of the Cu1.8S ND.

Figure 4
figure 4

UV–Vis–NIR absorption spectra of roxbyite Cu1.8S NDs treated with a MPA, b 1-DDT and c additional OAm. TEM images of c shoulder-to-shoulder assembled Cu1.8S NDs, d face-to-face assembled Cu1.8S NDs and e disassembled Cu1.8S NDs

Full size image

In addition to MPA, the Cu1.8S NDs were also sonicated with 5.0 μL of 1-DDT under N2 atmosphere. On the contrary, the LSPR peak of Cu1.8S NDs blue-shifted from 1702 to 1460 nm and the absorption intensity dramatically decreased (Fig. 4b). This variation of LSPR response was due to the face-to-face self-assembly of Cu1.8S NDs (Fig. 4d), which was caused by the hydrophobic interactions provided by 1-DDT molecules (ligand exchange from OAm to 1-DDT) on the basal ND surface [35]. The interparticle distance in the disk array was around 2.1 nm after 1-DDT treatment, and thus, a face-to-face in-plane plasmon coupling effect was expected. The interparticle distance was adjusted by linear thiols with different hydrocarbon chain lengths such as 1-hexanethiol and 1-hexadecanethiol. Besides the plasmon coupling, the dramatic decrease in absorption intensity was caused by the precipitation of long Cu1.8S ND arrays. Afterward, the face-to-face assembled Cu1.8S NDs were disassembled by sonication with 5.0 μL of OAm. The LSPR peak slightly red-shifted from 1460 to 1520 nm, and most of disk arrays were separated (Fig. 3b, e). It is likely that the 1-DDT molecules on the ND surface were replaced by OAm and the nonlinear OAm caused the separation of disk arrays. This observation was consistent with a previous report which demonstrated the ligand exchange from 1-DDT to OAm, leading to the disassembly of supercrystal structures of Cu1.97S NCs [36]. Hence, organic capping ligand triggered self-assembly and disassembly of Cu1.8S NDs is a facile route to tune their LSPR properties. The treatment of roxbyite Cu1.8S NDs with MPA, 1-DDT and OAm did not cause the shift of XRD peak, but it altered the relative peak intensity because of different orientations of NDs on the substrate.

Tuning of LSPR by plasmon coupling in multilayer films

Plasmon coupling of roxbyite Cu1.8S NDs not only emerge in solution, but also in the multilayer films. In order to tune the LSPR of Cu1.8S NDs, two films of Cu1.8S ND and Cu1.8S ND array were assembled through dip coating method. Dip coating was repeated several times to obtain the desired films. Because of uniform size and morphology, the Cu1.8S NDs formed intimately interconnected films with small portion of vacancies (Fig. 5a). Differently, Cu1.8S ND arrays randomly oriented on the substrate (Fig. 5b). It should be noted that the variation of dielectric environment of NDs and ND arrays in the film affected the LSPR of Cu1.8S. Indeed, Cu1.8S ND and Cu1.8S ND array films exhibited redshifted and broadened LSPR peak compared to that of colloidal dispersions in CHCl3. The LSPR peak of Cu1.8S ND red-shifted from 1688 nm in colloidal dispersion to 1752 nm in multilayer film (Fig. 5c). The slight redshift can be attributed to an increase in dielectric constant of surrounding medium and a weak in-plane plasmon coupling. Diversely, the LSPR of Cu1.8S ND array red-shifted from 1528 nm in colloidal dispersion to 1920 nm in multilayer film (Fig. 5d). In addition to the increase in dielectric constant of surrounding medium, the large redshift of LSPR peak mainly resulted from a strong in-plane plasmon coupling. This was consistent with a previous report concerning covellite CuS NDs [14]. The strong plasmon coupling may lead to an intense electric field on the surface of Cu1.8S ND array film, which is desired to promote two-photon absorption of laser dye and then the subsequent estimation of electric field enhancement factor by hole-based LSPR of Cu1.8S NDs [37].

Figure 5
figure 5

SEM images of multilayer films assembled from a individual Cu1.8S NDs and b Cu1.8S ND arrays. UV–Vis–NIR absorption spectra of c Cu1.8S ND colloidal dispersion and multilayer film and d Cu1.8S ND array colloidal dispersion and multilayer film

Full size image

Conclusions

In summary, high-quality roxbyite Cu1.8S NDs were synthesized and their LSPR were readily tuned by particle size, hole density via chemical oxidation and reduction, self-assembly and disassembly in solution and plasmon coupling in multilayer film. Firstly, the LSPR peak of Cu1.8S NDs red-shifted with increasing particle size. Secondly, the LSPR peak blue-shifted and became stronger and sharper under air oxidation, while it red-shifted and became weaker and wider under MPA reduction. Thirdly, during self-assembly induced by MPA and 1-DDT in CHCl3, the LSPR peak of Cu1.8S NDs red-shifted and became stronger in shoulder-to-shoulder fashion of plasmon coupling, while it blue-shifted and became weaker in face-to-face fashion, respectively. Furthermore, in the disassembly of Cu1.8S ND arrays, the LSPR peak slightly red-shifted. Finally, the plasmon coupling effect occurred in both Cu1.8S ND and Cu1.8S ND array multilayer films, which resulted in a redshift and broadening of LSPR peak. This coupling effect was more significant in Cu1.8S ND array film than that of Cu1.8S ND film. The flexible tunability of LSPR in roxbyite Cu1.8S NDs guarantees potential plasmonic applications in photocatalysis, photothermal therapy, two-photon photochemistry and many others in NIR spectral wavelengths.