Size- and morphology-dependent optical properties of ZnS:Al one-dimensional structures

The XRD patterns of the ZnS:Al samples are shown in Fig. 1. To determine the unit cell parameters, the XRD patterns were analyzed by a simple Pawley refinement method using the Topas Academic V5.0 software. The XRD images show that nearly all of the peaks agree with the ZnS wurtzite structure (JCPDS 36-1450), except for a small peak at 32.8°. Furthermore, the ZnS NW exhibits a space group of P63 mc, in which the main peaks at 2Θ = 28.52, 47.50, 26.89, 30.50, and 56.33° correspond to the directions of the (002), (110), (010), (011), and (112) planes, and the small peak at 32.8° agrees well with the (200) plane of ZnS zinc blende structure (JCPDS 65-0309). Because the (002) and (110) planes of the ZnS wurtzite structure overlap with the (111) and (022) planes of the ZnS zinc blende, the structures of the ZnS:Al samples can be regarded as a mixture of wurtzite and zinc blende structures with a dominant wurtzite ZnS. Additionally, the lattice constants were obtained from the Rietveld Refinement Treatment, as shown in Table 1. For the lattice constants, a = b = 3.8284 and c = 6.2573 Å were found for the undoped NWs; and a = 3.8264, 3.8259, 3.8255, 3.8250 Å and c = 6.2578, 6.2549, 6.2532, 6.2511 Å were found for #A1, #A2, #B1, and #B2, respectively. Compared to the undoped ZnS, the lattice constant a contracted by 0.197, 0.25, 0.286, and 0.342 %, and the constant c contracted by −0.054, 0.243, 0.408, and 0.623 % were found for samples #A1, #A2, #B1, and #B2, respectively. Except for the constant c of sample #A1, the lattice constants a and c decreased with increasing content of the Al source. A gradual decrease in the lattice parameters indicates that an increasing number of Al atoms are successfully incorporated into the ZnS host as the Al source content increases, which is consistent with the relations of Vegard’s law, as shown in Fig. 2. With the increase of Al³⁺ ions, the lattice parameters of the ZnS:Al host decreases because the ionic radius of Al³⁺ (54 pm) is smaller than that of Zn²⁺ (74 pm). Furthermore, an abnormal increase of the lattice constant c of sample #A1 can be regarded as the surface tensile induced by its typical morphology. Because the Al-dopant concentration is very low when using the thermal evaporation method and is hard to identify in the EDS because of its small quantities, the XRD Rietveld Refinement is used to provide evidence for the incorporation of atoms into the ZnS host. The quantitative elemental analysis from the EDS spectra for the ZnS:Al samples indicate that Zn and S are major elements; however, the S/Zn atomic ratio is less than one, indicating that sulfide vacancies are present in the ZnS:Al samples.

The morphologies and compositions of the products were characterized via FESEM, as shown in Figs. 3 and 4. Figure 3a, b represent the SEM images of sample #A1, whereas Fig. 3c, d represent the SEM images of sample #A2. For sample #A1, some of nanowires have a main bough and many needle-like twigs distributed alternatively along each side. The needle-like twig has a diameter of tens to one hundred nanometers at the base in contact with the bough, and the diameter of the tip of the needle is several to a few tens of nanometers. Additionally, there is a several to tens of nm separation between two neighboring needles, making the structure resemble a pine leaf. And some nanowires have a saw-like structure with a width of the order of tens of nanometers. For sample #A2, each NW resembles a firry leaf with a length ranging from tens to several hundred nanometers and a width of several hundred nm, but without the main bough and twigs, as shown in Fig. 3c, d. From Fig. 3 a–d one can find that the size of sample #A2 is larger than that of sample #A1. Sample #B1 has a microbelt shape with a width of 1–3 micrometers and a length in the tens of micrometers, as shown in Fig. 4c. Sample #B2 is shaped like a banana leaf and exhibits the same size as sample #B1 but with a greater thickness, as shown in Fig. 4d. Sample #B2 more closely resembles the bulk material.

UV–Vis absorption spectra were obtained by measuring the optical absorption spectra on a UV–Vis spectrophotometer (Cary-5000) with an integrating sphere to guarantee a better measurement result, as illustrated in Fig. 5. The absorption spectra of samples #A1, #A2, #B1, #B2, and undoped ZnS semiconductor can be extracted from the formula (αhv) = c(hv − E _g)^1/2, where c is a constant, E _g is the band gap, and the direct band gaps estimated from a plot of (αhv)² versus the photon energy hv were equal to 3.626, 3.586, 3.612, 3.608, and 3.611 for undoped, #A1, #A2, #B1, and #B2, respectively. The band gap values show that the Al-doped samples have a slightly smaller band gap in comparison to the undoped sample, indicating that some Al atoms have been doped into the ZnS hosts. Sample #A1, with an atomic ratio of Al to Zn source equal to 3 at%, has the lowest band gap, which can be explained by an increase of surface defects in addition to the Al dopant because of the relatively smaller size and the large surface to volume ratio.

The PL properties of the undoped and Al-doped nano- and micro-scale ZnS structures were investigated via PL excitation at 325 nm to further assess their doping and structure qualities, as shown in Fig. 6. For the undoped ZnS structure, a slightly asymmetric emission at 518 nm was observed, and the Gaussian fitting revealed a dominant emission at 514 nm with a weak emission at 550 nm. The peak at approximately 514 nm has previously been attributed to the S vacancies in ZnS nanowires (Zhang et al. 2005), or ascribed to the peak at 518 nm representing the transitions of electrons from S vacancy states to zinc vacancy states (Kumar et al. 2006). The peak at 550 nm originates from bulk defects (Lei et al. 2011). For the Al-doped ZnS, a broad and asymmetric emission spectrum at approximately 500 nm was observed, and the deconvoluted peaks at ~490 and ~540 nm can be identified using a Gaussian fitting, as shown in Fig. 6c–f. For the four Al-doped ZnS nanostructures, sample #A2 (with 5 at% Al content in the source) exhibited a stronger emission than the other three samples, which is consistent with other reports, in which the highest luminescence intensity corresponded to the Al-dopant concentration of 6 at% for the ZnS:Al films (Prathap et al. 2009). Herein, the enhanced emission at 493 nm can be ascribed to the Al-doped ZnS nanoparticles, presumably from the donor–acceptor pair (DAP) transition (Nagamani et al. 2012); the observation of which is considered to be an obvious evidence for the substitution of Al in the host lattice in ZnS:Al nanoparticles (Reddy et al. 2014). The green band at 530 nm may be related to some native defects of pure ZnS (Sotillo et al. 2012, 2014), such as zinc vacancies (Mitsui et al. 1996, Chen et al. 2011) or S vacancies (Ye et al. 2004; Jiang et al. 2012). A decrease in the emission intensity at 500 nm with Al-dopant concentration can be explained by the formation of defect complexes (Al donors) that induces the structural disorder caused by an increase in non-radiative recombination (Prathap et al. 2009).

Raman spectral measurements of the samples were conducted using an excitation wavelength of 532 nm with a 3 % incident laser light power of 300 mW to study the effects of size and morphology on the Raman scattering spectra. The room-temperature Raman spectra for undoped and different Al-doped ZnS materials are shown in Fig. 7a. For all of the samples, a sharp peak at 348 cm⁻¹ was clearly observed. For the sample #A1 in Fig. 7a, the following peaks were clearly identified: a mode at ~217 cm⁻¹ corresponding to the second-order longitudinal acoustic (2LA) phonon in ZnS (Cheng et al. 2009); a peak at 414 cm⁻¹ associated with the LO+TA phonon mode; two peaks at 639 and 669 cm⁻¹ attributed to the second-order optical phonon LO+TO and 2LO modes (Lin et al. 2004; Krol et al. 1988), respectively; two small peaks at 155 and 178 cm⁻¹ assigned to disorder-activated second-order acoustic phonons (Scocioreanu et al. 2012); and a broad peak at 276 cm⁻¹, which consists of the unresolved A1(TO) and E1(TO) phonon modes at 273.3 cm⁻¹ and the E2(TO) phonon mode at 286.0 cm⁻¹, as discussed in (Radhu and Vijayan 2011; Adu et al. 2006). Furthermore, for all samples, the peak at 348 cm⁻¹ can be fitted using three Lorentzian curves, as shown in Fig. 7b, where the decomposition of the peak at 348 cm⁻¹ from sample #A1 exhibits three peaks located at 347.60 (called as P1), 350.14 (P2), and 337.2 cm⁻¹ (P3) with full-width-half-maxima (FWHM) of 4.377, 3.733, and 9.786 cm⁻¹, respectively. The peaks at 347.60 (P1) and 350.14 cm⁻¹ (P2) can be regarded as the E₁(LO) and A₁(LO) vibration modes, in which the E₁ (LO) mode results from the lateral morphology of the ZnS nanowires parallel to the c-axis, as discussed in (Lu et al. 2005), whereas the peak at 337.16 cm⁻¹ is assigned to a surface phonon mode (Kim et al. 2012). For the four samples, an identical decomposition was performed with nearly the same magnitude of the E₁ (LO) and A₁ (LO) phonons and their FWHMs. As shown in Table 2, the surface phonon was also found to be approximately 337 cm⁻¹, except for sample #B2 in which the value was red-shifted approximately 4 cm⁻¹ and the FWHW was larger because of the disappearance of the surface effects for sample #B2. Compared to the Raman spectrum of bulk hexagonal ZnS (LO: 352 cm⁻¹) (Brafman and Mitra1968; Nilsen 1969), the peaks of the first-order LO phonons from these ZnS nanowires exhibit a shift toward lower energy.

To study the changes of the Raman scattering intensity, we selected sample #B2 as a reference and determined that the ratios of the 1LO-integrated areas of #A1, #A2, and #B1 to that of #B2 are equal to 8.13, 3.56, and 2.52, respectively, indicating that sample #A1 exhibits greatly enhanced Raman scattering. Comparing samples #A1 and #A2, the greatest difference is in their size and morphological structures. For sample #A1, the pine leaf-like nanowires grow along a main bough with many needle-like twigs alternatively grown along the two sides. Additionally, there is a several to tens of nanometers separation between two neighboring needles, resulting that the diameter for the main bough and the twig is only several tens of nanometer. Even for the saw-like structure nanowires in sample #A1, the diameter is only approximately 100 nm. For sample #A2, although many ZnS:Al nanostructures grow along both sides of the z-direction in a firry leaf-like nanowire, there is no independent nanopattern; their sizes are far more larger than that of sample #A1, thus the surface to volume ratio decreases. Therefore, the greater enhancement of the Raman scattering observed from sample #A1 is mainly attributed to its typical morphology and small size of nanopatterns. The phenomenon of the effect of morphology on the Raman scattering was observed in (Fasolato et al. 2014), where the enhanced Raman spectra were attributed to the different cluster morphologies rather than to the different Ag nanoparticle diameters. Additionally, optimized high-density Si nanowire arrays have demonstrated a remarkable increase in Raman sensitivity compared to reference planar samples, which was explained as surface multiple scattering (Bontempi et al. 2014). Comparing the samples #A2, #B1, and #B2, the 1LO-integrated area gradually decreases, implying that as the Al-dopant concentration increases, the Raman scattering intensity decreases. This observation contradicts the results (Zhao et al. 2013) of Mn-doped CdS nanoparticles, which showed that the Raman scattering intensity increased with increasing dopant content before decreasing.

Actually, surface-enhanced Raman scattering (SERS) will take place when the energy of the incident laser light is close to the surface plasmon energy in noble metal nanoclusters which are intentionally introduced in NC arrays, or the energy of the incident laser light matches that of interband electronic transitions in the NCs. For nanostructure materials both quantum effects and surface effects are more important, and size- and shape-dependent Raman effects will be more obvious. Theoretically, the calculations (Zenidaka et al. 2011) of the SERS based on the optical near-field intensity enhancement on the metallic (plasmonic) and the nonmetallic (Mie scattering) nanostructured substrates with two-dimensional (2D) periodic nanohole arrays indicated that both the inter-void distance and the void diameter influence the optical intensity enhancement. Experimentally, the sensitivity of the field enhancement in the hot spots to the distance of the particles, as well as the frequency and polarization of the excitation laser have been reported (KNEIPP et al. 2006). In our studies, the enhanced Raman scattering from interband electronic transitions can be excluded since all the PL spectra centered at approximately 500 nm, and the strongest PL intensity is not for #A1 but for #A2. As reported in (Hu et al. 2013), the coupling strength of the exciton—phonon increases with increasing lateral size, which will result in the reduction of the Raman scattering. Therefore, the greatly enhanced Raman scattering for sample #A1 can be explained by the large surface multiple scattering and a surface polarization effect.