Background

Recently, doped one-dimension (1D) semiconductor nanostructures are especially attractive for their excellent and unique optical and optoelectronic properties [1, 2], which were affected greatly by optical micro-cavity and dopant. 1D nanostructures doped with transition metal (such as Cr, Mn, Fe, Co, and Ni), which can find extensive application in spintronics and nanophotonics [35], show novel emission and interesting magnetic transport properties. For example, single crystalline Ga0.95Mn0.05As nanowires show temperature-dependent hopping conduction [6]. Cu-doped Cd0.84Zn0.16S nanoribbons show four orders of magnitude larger photocurrent than the undoped ones, demonstrating potential application in photoconductors and chemical sensors [7]. The emission of transition metal ion has specific wavelength, such as the emission of manganese (Mn) ion which is located generally at 585 nm. Moreover, 1D nanostructures can confine the coherent transport or transmission of photon to the definite direction, that is, 1D nanostructures can form optical micro-cavity easily and work as effective optical waveguide within a nanometer scale [8]. Recently, there is an increasing research interest on the optical micro-cavity and corresponding multi-mode emission spectra in doped 1D nanostructures [9]. Zou et al. observed multi-mode emission from doped ZnO nanowires due to F-P cavity effect [10]. Multi-mode emission was also observed in In x Ga1 - xN superlattice [11]. Except for the inorganic semiconductor nanostructures, organic nanofibers can also act as coherent random laser with multi-mode emission [12]. Recent research shows that the formation of multi-intracavities plays an important role in the multi-mode emission [13]. These multi-intracavities can couple to produce coherent emission. These confined cavities and multi-band emission of 1D nanostructures are affected strongly by synthesis parameter and deliberate doping. The optical properties of 1D nanostructures are sensitive to minute change of crystal quality, crystal defect, and dopant. The latter can introduce defect state and is therefore very important. So, it is necessary to investigate the direct correlation between dopant and optical properties within the nanometer scale.

ZnSe, a direct semiconductor with a bandgap of 2.63 eV at room temperature, shows excellent optical properties and potential application in light emitting diode and laser diode. 1D ZnSe nanostructures possess novel light emission property [14]. Recently, Vugt et al. observed the novel light-matter interaction in ZnSe nanowires, which can be used to tailor waveguide dispersion and speed of propagating light [15]. In this paper, we synthesize three Mn-ZnSe nanobelts using different dopant compounds. Transmission electron microscopy (TEM) and scanning near-field optical microscopy (SNOM) techniques were used to provide simultaneous investigation on the micro-structure and crystallinity, micro-PL spectrum, and mode-selected mapping image. Both near-bandgap emission and trapped-state emission of ZnSe are observed in Mn-ZnSe nanobelts obtained using Mn powder as dopant. However, the Mn ion transition emission cannot be observed in this ZnSeMn nanobelt. Using manganese chloride (MnCl2) as dopant, strong Mn ion transition emission and weak near-bandgap emission are observed. We can also observe the strong Mn ion transition emission and weak near-bandgap emission in the Mn-ZnSe nanobelts obtained using manganese acetate as dopant. More interestingly, the Mn ion transition emission can split into multi-mode emission due to multi-Fabry-Pérot cavity effect in the ZnSe Mn ( CH 3 COO ) 2 nanobelt. Raman spectrum was used to confirm the effective doping. These results are helpful in understanding the effect of dopant on the optical micro-cavities and multi-mode emission. These Mn-ZnSe nanostructures can find promising applications in multicolor emitter or wavelength selective photodetector.

Methods

The 1D Mn-ZnSe nanobelts were synthesized by a simple thermal evaporation method. Commercial grade mixed powder of ZnSe and Mn or MnCl2 or manganese acetate (Mn(CH3COO)2) with a weight ratio of 5:1 was used as source material. The obtained samples were labeled as ZnSeMn, ZnSe MnCl 2 , ZnSe Mn ( CH 3 COO ) 2 , respectively. The other synthesis processes are similar with our previous report [16]. The evaporation temperature, growth temperature, and growth time are set to 900°C, 600°C, and 45 min, respectively. A yellow product deposited on the silicon wafer after the furnace cools down to room temperature. For comparison, the pure ZnSe nanobelts were also synthesized using ZnSe powder as source material.

XRD (D/max-5000, Rigaku Corporation, Tokyo, Japan), E-SEM (QUANTA 200, FEI, Hillsboro, OR, USA), energy dispersive X-ray spectroscopy (EDS; attached to SEM), and TEM (JEM-3010, JEOL Ltd., Tokyo, Japan) were used to examine the phase structure, crystallinity, and composition of the as-prepared nanobelts. Raman spectroscopy was performed in a confocal microscope (LABRAM-010, HORIBA Ltd., Kyoto, Japan) using He-Ne laser (632.8 nm) as excitation light source. The PL and corresponding mapping were obtained by SNOM (alpha 300 series, WITec GmbH, Ulm, Germany) with He-Cd laser (325 nm) as excitation source at room temperature. In all optical experiments, the excitation signal illuminated perpendicularly onto the sample surface.

Results and discussion

The XRD patterns of pure and doped ZnSe nanobelts are shown in Figure 1. All of the XRD pattern peaks of pure and doped ZnSe nanobelts are in agreement with the standard values (JCPDS card no. 37–1463), see Figure 1a. There are no diffraction peaks of Mn or MnSe in the doped samples. However, the diffraction peaks of doped ZnSe nanobelts shift to lower angle direction compared with pure sample (Figure 1b). Doping can cause a little change to lattice constant. Therefore, the present measurable shift of diffraction peak (about 0.05°) come from doped Mn because of the larger ionic radius of Mn2+ (0.80 Å) than that of Zn2+ (0.74 Å). Such shift of diffraction peak can also be observed in other doped nanostructures [1719]. Therefore, manganese can diffuse and dope into ZnSe nanobelts efficiently when MnCl2 or Mn(CH3COO)2 were used as dopants.

Figure 1
figure 1

XRD patterns. (a) Pure ZnSe, ZnSeMn, ZnSe MnCl 2 , and ZnSe Mn ( CH 3 COO ) 2 nanobelts. (b) Enlarged (111) diffraction peak of the four samples.

Figure 2a is a SEM image of pure ZnSe nanobelts, which deposited on the Si substrate randomly. The nanobelts have a length of hundreds of micro-meter, width of several micro-meter, and thickness of tens of nanometer. EDS (inset of Figure 2a) shows only Zn and Se elements (Si comes from the substrate). The atomic ratio of Zn to Se approaches to 1, demonstrating that pure ZnSe is stoichiometric. Figure 2b,c,d shows the SEM images of doped ZnSe nanobelts obtained using Mn, MnCl2, Mn(CH3COO)2 as dopants. The belt-like morphology of ZnSeMn is similar with that of pure ZnSe but shows a little difference from those of ZnSe MnCl 2 and ZnSe Mn ( CH 3 COO ) 2 . The insets of Figure 2b,c,d are the corresponding EDS images. We cannot detect the Mn element, and the ratio between Zn and Se deviates a little from 1 in ZnSeMn nanobelts. The dopant concentrations are 0.72% and 1.98% in ZnSe MnCl 2 ZnSe MnCl 2 and ZnSe Mn ( CH 3 COO ) 2 nanobelts, respectively. Mn powder is hard to be evaporated due to its high melting point. Therefore, little manganese can dope into the ZnSe nanobelts under the present evaporation temperature when Mn powder was used as the dopant. MnCl2 and Mn(CH3COO)2 have low melting points and are easy to be evaporated. So, manganese can dope into the ZnSe nanobelts effectively when MnCl2 or Mn(CH3COO)2 were used as dopants. The MnCl2 and Mn(CH3COO)2 were usually used as dopants in other semiconductor nanostructures [16, 17]. We mapped the elements to detect the distribution of Mn dopant in the nanobelt. Figure 2e shows the EDS mapping of ZnSe MnCl 2 nanobelt. The mapping profiles show that Mn, Zn, and Se elements distributed homogeneously within the nanobelt. Figure 2f is the EDS mapping of ZnSe Mn ( CH 3 COO ) 2 nanobelt, which shows that the distribution of Mn element is inhomogeneous. The minute inhomogeneous distribution of Mn can affect the optical property of the ZnSe Mn ( CH 3 COO ) 2 nanobelt greatly.

Figure 2
figure 2

SEM images and corresponding EDS and element mapping. (a) to (d) Pure ZnSe, ZnSeMn, ZnSe MnCl 2 , and ZnSe Mn ( CH 3 COO ) 2 nanobelts, respectively. The insets are the corresponding EDS images. (e) to (f) Element mapping of single ZnSe MnCl 2 cand ZnSe Mn ( CH 3 COO ) 2 nanobelts, respectively.

Further characterization of these doped ZnSe nanobelt is performed by means of TEM operating at 300 kV. High-resolution TEM (HRTEM) can be used to describe the crystal quality and growth direction. Figure 3a is a TEM image of a ZnSeMn nanobelt. The morphology and size were consistent with those observed using SEM. Figure 3b is the corresponding HRTEM image. The well-resolved lattice fringes confirmed the single crystalline structure. The measured lattice fringe of 0.325 nm corresponds to the inter-planar distance of (111) plane as known from the bulk ZnSe crystal. Therefore, the growth direction of ZnSeMn nanobelt is designated to be [111]. The result also confirmed the fact that (111) is the most densely packed facet for fcc structure and is thus the most favorable facet for growth. Figure 3c is a TEM image of ZnSe MnCl 2 nanobelt. Figure 3d is the corresponding HRTEM image. The ZnSe MnCl 2 nanobelt shows a single crystalline structure (see the fast Fourier transform (FFT) image in the inset of Figure 3d). The measured lattice fringe is 0.325 nm. The angle between the lattice plane and the axis direction of the nanobelt is 71° (see in Figure 3d). Therefore, the growth direction of the nanobelt can also be designate to be part of the <111> family directions. Figure 3e is a TEM image of the ZnSe Mn ( CH 3 COO ) 2 nanobelt. Figure 3f is the corresponding HRTEM image. Similar with ZnSe MnCl 2 nanobelt, the ZnSe Mn ( CH 3 COO ) 2 nanobelt also shows a single crystalline nature and [111] growth direction. The HRTEM also indicates that there are a lot of defect states and impurities in the ZnSe Mn ( CH 3 COO ) 2 nanobelt (see the labeled cycle zone in Figure 3f).

Figure 3
figure 3

TEM and HRTEM images. (a) and (b) Single ZnSeMn nanobelt. (c) and (d) Single ZnSe MnCl 2 nanobelt. Insets in (d) are the calculated lattice fringe image and FFT. (e) and (f) Single ZnSe Mn ( CH 3 COO ) 2 nanobelt.

Raman spectroscopy can provide abundant structure information and is powerful for fast and non-destructive detection of dopant. Figure 4 shows the micro-Raman spectra of single pure and doped ZnSe nanobelt at room temperature. In the Raman spectrum of the pure ZnSe nanobelt (Figure 4a), the peaks at 205 and 249 cm-1 can be assigned to TO and LO modes of zinc blende ZnSe crystal, respectively [16]. Figure 4b is the Raman spectrum of the ZnSeMn nanobelt. Besides the LO and TO vibration modes of ZnSe, there is another mode at 285 cm-1 with weak intensity, which related to the defect state (stacking fault) in the doped ZnSe [20]. Figure 4c is the Raman spectrum of ZnSe MnCl 2 nanobelt. Besides the 201, 248, and 294 cm-1 vibration modes, there is another mode at 135 cm-1 which is not the intrinsic mode of ZnSe. The 135 cm-1 mode can be assigned to the TO impurity vibration modes of MnSe [21]. The presence of impurity vibration modes of MnSe confirms that Mn can dope into ZnSe nanobelts effectively with MnCl2 as dopant in the present synthesis parameters. However, the absence of impurity vibration modes of MnSe in ZnSeMn nanobelt demonstrates that the concentration of Mn2+ is too low, and the Mn powder is not the appropriate dopant. The vibration modes of the ZnSe Mn ( CH 3 COO ) 2 nanobelt are almost the same with those of the ZnSe MnCl 2 nanobelt (Figure 4d). The difference of these two Raman spectra is that the intensity ratio of ZnSe to MnSe mode is larger in the ZnSe Mn ( CH 3 COO ) 2 nanobelt. The four Raman spectra indicate varied vibration modes, which were affected by the dopant in the Mn-ZnSe nanobelt.

Figure 4
figure 4

Raman spectra. (a) Pure ZnSe, (b) ZnSeMn, (c) ZnSe MnCl 2 , and (d) ZnSe Mn ( CH 3 COO ) 2 nanobelt, respectively.

We studied further the luminescence properties of the as-synthesized Mn-ZnSe nanobelts by commercial SNOM. The insets of Figure 5a are bright-field optical and dark-field emission images of a single representative pure ZnSe nanobelt under the excitation of He-Cd laser (325 nm). The emission is strong at the excitation spot. Figure 5a is the corresponding far-field PL spectrum. The band at 458 nm comes from the near-band edge emission of ZnSe, while the broad emission band at lower energy between 575 and 675 nm is attributed to the trapped-state emission [16]. Trapped-state and dangling bond, such as Zn vacancy and interstitial state, are easy to form in nanostructures due to the reducing size. Therefore, the trapped-state emission is usually observed even in pure nanostructures [22]. The insets of Figure 5b are the bright-field optical and dark-field emission images of a single ZnSeMn nanobelt. Figure 5b is a corresponding far-field PL spectrum. We can observe a near-band edge emission of ZnSe with low intensity located at 461 nm and the trapped-state emission at 625 nm. There is another strong emission band at 545 nm, which can be explained by the dislocation, stacking faults, and nonstoichiometric defects, as reported in reference [2325]. We cannot observe the Mn ion emission (such as 4T16A1 transition emission at 585 nm) which demonstrates that the Mn concentration is too low or there is no Mn doping into the ZnSeMn nanobelt. The insets of Figure 5c are the bright-field optical and dark-field emission images of ZnSe MnCl 2 nanobelt. Figure 5c is the corresponding far-field PL spectrum. Except for the weak near-bandgap emission and defect state emissions at 460 and 536 nm, there are two strong emission bands at 584 and 650 nm. The 584-nm band corresponds to d-d (4T16A1) transition emission of tetrahedral coordinated Mn2+ states [26]. The 650-nm band is from the Mn-Mn emission centers, which is similar with the phenomenon of the Mn dimers [27, 28]. The Mn-Mn emission only occurs when the Mn dopant concentration is high enough [29]. There is another weak emission band at 694 nm, which is believed to originate from the Mn2+ ions at the distorted tetrahedral sites or the octahedral sites, due to the high Mn content [30, 31]. Manganese ions on such lattice sites show a different crystal-field splitting between the states of 3d orbitals, and then a red-shifted emission band is observed [32]. The appearance of the Mn2+ emission demonstrates the efficient doping of Mn2+ ion into the ZnSe crystal. We further carried out PL mapping of each individual emission band to explore the distribution of Mn2+ ions (Figure 5e). We can see that the distribution of near-band edge emission and Mn2+ ion emission is homogeneous in the whole nanobelt (see in Figure 5c). Therefore, the Mn2+ ions distribute homogeneously in the nanobelt, consistent with the EDS mapping result. The insets of Figure 5d are the bright-field optical and dark-field emission images of the ZnSe Mn ( CH 3 COO ) 2 nanobelt. A portion of the in situ emission propagated through the nanobelt and emitted at the opposite end, indicating that the nanobelt can act as an effective optical waveguide. Figure 5d is the corresponding far-field PL spectrum, which contains a near-band edge emission and a broad emission band between 525 and 725 nm. Similar to the PL spectrum of ZnSe MnCl 2 nanobelt, the broad emission contains four bands: 541, 590, 637, and 689 nm (see the fitted red curve in Figure 5d). Therefore, the Mn2+ ion efficiently doped into the ZnSe matrix crystal with ZnSe Mn ( CH 3 COO ) 2 as dopant. Moreover, in contrast to the reported Mn2+ transition emission (see the PL of the ZnSe MnCl 2 nanobelt), the current Mn2+ emission band splits into many narrow sub-bands, that is, multi-mode emission. The PL mapping is carried out for individual sub-bands to explore the origin of the multi-mode emission and photon propagation process in the nanobelt (Figure 5f). We can see that the near-band edge emission distributes in the whole nanobelt. In contrast, the mapping images of the Mn2+ ion emission sub-bands show irregular light intensity distribution along the nanobelt (the bright and dark regions represent the maximum and minimum intensities of emission, respectively). Moreover, there is slight modification between these Mn2+ ion emission mappings, such as it is a bright region at the end of 599 nm band, while it is dark for 637-nm band at the same position. This is due to the cavity mode selection within the belt. The mapping images indicate that there are several optical micro-cavities within the single nanobelt. Usually, the two end facets act as reflecting mirrors to form one Fabry-Pérot cavity in 1D nanostructures. However, multi-cavities can emerge in single doped 1D nanostructure when a dopant with varied refractive indexes is incorporated into the matrix [13, 16]. In the HRTEM image (Figure 3f), we can clearly see some impurity and defect sites possibly related to the Mn dopant in the ZnSe Mn ( CH 3 COO ) 2 nanobelt. When the nanobelt was excited, a large number of photons propagate along the axis, in which some were absorbed, some were reflected or scattered by high refractive index domain, and some others passed through the segment boundary. These reflected photons propagate to another boundary and resonate at the boundary zones. So, different emission lines were selected to be observed in a single nanobelt. Combining the mapping images and multi-modes spectra, we can calculate the sub-cavity length L using the formula: Δ, where n is the refractive index (n = 2.67 for ZnSe), λ1 and λ2 are the resonant wavelengths, and Δλ is the mode spacing [16]. The calculated cavity lengths of the adjacent bands are 9 to 11 μm, which are much shorter than the actual length of the nanobelt, but very close to the lengths of bright region in the mapping images. Therefore, the photon oscillation within a single nanobelt does not always happen between the two ends, but rather in the smaller inner segments and then result in multi-cavity and multi-mode emissions. Such in situ PL spectrum and mapping indicate strong localization and oscillation of photon propagation along the longitudinal axis. This behavior is a typical coupled optical multi-cavity.

Figure 5
figure 5

PL spectra and corresponding emission mapping images. (a) Pure ZnSe, (b) ZnSeMn, (c), ZnSe MnCl 2 , and (d) ZnSe Mn ( CH 3 COO ) 2 nanobelt, respectively. The insets are the corresponding bright-field optical and dark-field emission images. The red curve in (d) is the fitted PL spectrum. (e) The PL of each individual emission band in (c). (f) PL mapping images of individual emission sub-band in (d). The scale is 4 μm.

The growth conditions can be adjusted to obtain another ZnSe Mn ( CH 3 COO ) 2 nanobelt. Figure 6a is the SEM image and EDS of the ZnSe Mn ( CH 3 COO ) 2 nanobelt with lower Mn concentration (0.39%). Figure 6b is the dark-field emission image of single ZnSe Mn ( CH 3 COO ) 2 nanobelt with 0.39% Mn content, which also shows the optical waveguide characteristic. The inset is the corresponding bright-field optical image. Figure 6c is the corresponding far-field PL spectrum. The PL spectrum contains near-band edge emission of ZnSe with weak intensity and transition emission of Mn2+ with strong intensity. Compared with Figure 5d, the split of Mn2+ emission in Figure 6c is not evident. We can distinguish ambiguously that the Mn2+ emission split into many narrow sub-bands with a smaller periodic span (about 2 nm). The PL mapping is carried out for individual sub-bands to see if there are integrated multi-cavities in the nanobelt (Figure 6d). We can see that the band of 552 nm distributes homogeneously in the whole nanobelt. The sub-bands of 584, 630 and 670 nm distribute almost at two sides of the nanobelt. The excited photon emits at the side and end of the nanobelt usually after scattering at the boundary many times [33]. The optical multi-cavity phenomenon is not evident, although it still exists in the nanobelt due to the incontinuous emission intensity distribution at the two sides. The reduced Mn content can reduce the impurity and trapped state in the nanobelt and then affect the cavity quality greatly. Therefore, both dopant and micro-cavity play an important role in the multi-modes emission.

Figure 6
figure 6

Characterization of another ZnSe Mn ( CH 3 COO ) 2 nanobelt with low Mn2+concentration (0.39%). (a) SEM image and EDS. (b) Dark-field emission image. The inset is the corresponding bright-field optical image. (c) The corresponding PL spectrum. (d) The corresponding PL mapping images of individual emission sub-bands. The scale is 10 μm.

Conclusions

We synthesized pure and Mn-doped ZnSe nanobelts successfully using thermal evaporation method. Mn can dope effectively into ZnSe crystal when MnCl2 or Mn(CH3COO)2 were used as dopants in the source material. EDS mapping indicates that the distribution of Mn is inhomogeneous in the ZnSe Mn ( CH 3 COO ) 2 nanobelt. All of these doped nanobelts grew along the <111> direction. HRTEM demonstrates that there are a lot of defect states in the ZnSe Mn ( CH 3 COO ) 2 nanobelt. Raman spectra confirm that Mn2+ was doped into ZnSe MnCl 2 and ZnSe Mn ( CH 3 COO ) 2 nanobelts successfully. The optical properties are affected strongly by the concentration and spatial distribution of the dopant. Optical micro-cavity also plays an important role to the emission property. Nanobelt shows strong 4T16A1 transition emission of Mn2+. However, the 4T16A1 transition emission of Mn2+ in ZnSe Mn ( CH 3 COO ) 2 nanobelt splits into many narrow sub-bands due to the formation of integrated multi-Fabry-Pérot cavities, which can couple to produce coherent emission with selected wavelength and cavity mode. PL mapping confirms that there are several micro-cavities in the single ZnSe Mn ( CH 3 COO ) 2 nanobelt. Such doped nanobelts with integrated multi-micro-cavities and modulated emission wavelength can be optimized to fabricate nanophotonic devices and quantum coherent modulators.

Authors' information

WZ got his PhD degree in 2010. He is an assistant professor now. RL is an associate professor. DT and BZ are professors.