Background

ZnTe bulk crystal has increasing importance for various advanced semiconductor applications such as solar cells, blue-green laser diodes, terahertz imaging, electro-optic detector, and holographic interferometry [1–7]. The success of doping control of ZnTe thin films also makes ZnTe a suitable candidate for various optoelectronic device materials [8, 9]. ZnTe is well known to exhibit a pronounced preference for p-type due to self-compensation via Zn vacancies [10, 11]. Among many options for p-type doping of ZnTe, nitrogen (N) doping [12–16] and copper (Cu) doping [17, 18] have been mostly studied. Even though N doping using a radiofrequency (RF) plasma source has been proved to offer better doping characteristics, Cu doping has also been widely accepted as a good candidate as it can be achieved using various inexpensive ways including chemical immersions, sputtering, electrochemical deposition, and thermal evaporation [19–23]. Among previous reports on Cu doping of ZnTe thin films, most of them mainly focus on their electrical properties as a function of the Cu concentration without performing their structural and chemical analysis. Akkad and Abdulraheem recently provided evidence for the formation of the ternary zinc-copper-telluride alloy films containing Cu concentration above ~4 at.% prepared using RF magnetron sputtering [24], indicating that a structural or phase change could occur as Cu concentration in ZnTe reaches the over-doped regime. In this study, we discovered the formation of one-dimensional (1D) surface modulation and Cu-Te nano-rods for heavily Cu-doped ZnTe thin films grown by MBE. These findings explain the unusual observations in reflection high energy electron diffraction (RHEED) patterns and apparent resistivity of these thin films upon the increase of Cu cell temperature. Our work demonstrates that a highly reactive dopant could lead to formation of nanostructures in the host matrix, which may generally be applied to the studies of heavy doping using other physical or/and chemical vapor deposition approaches.

Methods

In this work, ZnTe:Cu thin films were prepared in a VG V80H MBE system, which is equipped with in-situ RHEED to provide a real time monitoring of the thin film growth. Prior to the growth of these doped thin films, the growth of undoped ZnTe thin films on semi-insulating GaAs (001) substrates was firstly optimized by tuning the temperatures of both the Zn and Te effusion cells as well as the substrate. Growth temperatures were optimized at Tsub = 358 °C, TZn = 230 °C, and TTe = 283 °C. Using these optimized temperatures, a set of 20 ZnTe:Cu samples with thicknesses around 850 nm were fabricated with the Cu cell temperature ranging from 840 to 1030 °C with a step of 10 °C. The room temperature resistivity measurements were carried out using a HL5500PC (Bio-Rad) system with needle probes contacting the indium electrodes made onto the samples in Van der Pauw geometry. Atomic force microscopy (AFM) characterizations were conducted by a Dimension 3100 AFM with a NanoScope IIIa controller (Digital Instruments) using tapping mode. Cross-sectional and plan-view high resolution transmission electron microscope (HRTEM) images were obtained by a JEOL 2010F TEM. The high resolution x-ray diffraction (HRXRD) measurements were performed with the PANalytical Multipurpose X-Ray Diffractometer. Time-of-flight secondary ion mass spectrometer (TOF-SIMS) was carried out using a PHI7200 SIMS system with Cs as the primary ion source.

Results and Discussion

For TCu ≤ 860 °C, the observed RHEED patterns were similar to those of an optimized ZnTe thin film. Spotty patterns in the initial growth period were observed due to the large lattice mismatch between ZnTe and GaAs, and then they evolved into streaky and bright patterns. Starting from TCu ≥ 870 °C, streaky pattern as shown in Fig. 1a was observed when the \( \left[1\overline{1}0\right] \) direction of the sample was aligned with the e-beam. While the sample was rotated, these straight streaks started to bend and the centered streak then evolved into a half circle as seen in Fig. 1b at a rotation angle of 90°, that is when the [110] direction of the sample was aligned with the e-beam. For the further 90°–180° rotation, a reverse order of bending with opposite bending direction proceeded. Such unusual RHEED patterns were firstly reported by us for an annealed Fe/ZnSe bilayer [25], which illustrates that these patterns came from a surface with one-dimensional surface modulation with the lateral scale around a few nanometers along the \( \left[1\overline{1}0\right] \) direction due to the interaction between the Fe atoms and the ZnSe surface lattice. Thus, the similar patterns observed for the ZnTe:Cu thin films grown using TCu ≥ 870 °C is likely due to a similar interaction between Cu atoms and ZnTe surface lattice when the incorporated Cu atoms reach a certain concentration. It was further observed that such RHEED patterns got dimmer as TCu increases, though the overall patterns remain unchanged.

Fig. 1
figure 1

RHEED patterns of sample grown using TCu = 880 °C when e-beam aligned with a \( \left[1\overline{1}0\right] \) and b [110] direction. c Apparent resistivity as a function of Cu cell temperature. d–f AFM images of three samples grown under Cu cell temperature at 870, 940, and 1030 °C

Full size image

On the other hand, an unusual characteristic was also observed in the apparent resistivity of the as-grown ZnTe:Cu samples. Figure 1c displays the measured apparent resistivity of all 20 ZnTe:Cu samples, in which one can see that the apparent resistivity falls rapidly till TCu reaches 870 °C followed by a relatively slower decline as TCu further increases. For samples with TCu ≥ 990 °C, a characteristic sharp fall followed by a plateau region is observed. In the following paragraphs, we will describe various studies performed on the as-grown ZnTe:Cu thin films to understand the unusual characteristics on their RHEED patterns and apparent resistivity.

AFM surface imaging was conducted on a number of ZnTe:Cu thin films to reveal the underlying cause of their unusual RHEED patterns. The AFM images for samples grown using TCu ≤ 860 °C are similar to that observed on an ordinary MBE-grown thin film. As can be seen in Fig. 1d, the surface of the sample grown using TCu = 870 °C presents a [110] oriented 1D feature with non-uniform spacing around tens of nanometers, which is induced by the interaction between Cu atoms and the ZnTe surface similar to what we reported for the interaction between Fe atoms and the ZnSe surface [25], leading to the unusual RHEED pattern observed in samples grown using TCu ≥ 870 °C. Additionally, inspection of Fig. 1e, f reveals that besides the 1D feature with non-uniform spacing, some [110] oriented dented nano-trenches with width from a few tens up to a few hundreds of nanometers appears at samples grown using TCu ≥ 880 °C and their densities increase as TCu further increases. These nano-trenches explain why the unusual RHEED patterns got dimmer as TCu further increases from 870 °C since the addition of the nano-trenches will reduce the effective area of the background 1D feature.

In order to have a better understanding on the self-assembled mechanism and the chemical composition of the observed nano-trenches, cross-sectional TEM imaging together with energy-dispersive X-ray spectroscopy (EDS) were conducted on two samples grown using TCu = 940 and 1030 °C. Figure 2a, b displays the corresponding cross-sectional TEM images of an individual nano-trench in each of these two samples taken with a zone axis along the [110] direction. These images reveal that surprisingly, the nano-trenches have a triangular cross-sectional shape and from now on they are called nano-rods. The insets in Fig. 2a, b show the corresponding cross-sectional view with a lower magnification, indicating that the observed nano-rods exist only near the surface. Based on their cross-sectional TEM and AFM images, one can conclude that the nano-rods grow larger and denser as TCu increases. A plan-view TEM image of the sample grown using TCu = 1030 °C is displayed in Fig. 2c, which provides further evidence of the morphology of the aligned Cu-Te nano-rods. The results of EDS (see Additional file 1) performed on regions inside and outside the nano-rods indicate that the nano-rods are composed of mainly Cu-Te alloy with very rich Cu composition, roughly reaching a Cu:Te ratio close to 2:1.

Fig. 2
figure 2

Cross-sectional TEM images of samples grown using Cu cell temperature at a 940 °C and b 1030 °C. Insets show the corresponding low-magnification images. c Plan-view TEM image of sample grown using TCu = 1030 °C in which some nano-rods near the surface can be clearly seen

Full size image

Further confirmation of the composition of the nano-rods comes from the HRXRD results. Figure 3 displays the HRXRD profiles of samples grown using TCu = 860 and 1000 °C, showing that two extra peaks located at 27.4° and 64.5° are observed for the sample using TCu = 1000 °C. This indicates that some new phase(s) in the film was(were) formed at high Cu concentration, in fact the 2-theta values of both extra peaks match those of two reported diffraction peaks of hexagonal Cu-rich Cu-Te system [26]. It is worth pointing out that both TEM imaging and HRXRD results indicate that Cu-Te nano-rods are crystalline materials. As mentioned above, the Cu-Te nano-rods formed only near the surface region. This indicates that at high Cu concentration the incorporated Cu atoms prefer to migrate upward toward the surface of the thin film during growth. The results of a separated study using TOF-SIMS performed on several ZnTe:Cu samples grown using TCu = 840, 920, 990, and 1030 °C, as shown in Fig. 4a–d, support this claim. For sample grown using TCu = 840 °C, within the ZnTe:Cu layer, the concentration of CsCu + (black line) is nearly constant, which indicates that Cu atoms randomly substitute the Zn atoms and formed a uniform layer. For other three samples, Cu atoms are mainly distributed near the surface, becoming more obvious as TCu increases, which is consistent with the results obtained from TEM imaging and EDS analysis.

Fig. 3
figure 3

HRXRD results of samples grown using Cu cell temperature at 860 and 1000 °C

Full size image
Fig. 4
figure 4

SIMS profiles of four ZnTe:Cu thin films using different Cu cell temperatures at a TCu = 840 °C; b TCu = 920 °C; c TCu = 990 °C; and d TCu = 1030 °C

Full size image

Having revealed the composition of the nano-rods, now we can address the unusual characteristics observed in the apparent resistivity data. It is well known that Cu-doped ZnTe thin films or bulk materials are p-type due to the substitutional incorporation of Cu atoms into the Zn lattice sites [24]. The observed initial rapid fall of the resistivity up to TCu = 870 °C is attributed to such a substitutional doping. As TCu further increases, the apparent resistivity experiences a more gradual decline up to TCu = 970 °C, which might be resulted from the appearing of saturation for the substitutional doping while the Cu-rich Cu-Te nano-rods with relatively low density may only lower down the apparent resistivity moderately. This is because even though Cu-rich Cu-Te alloy system has long been known as a highly degenerated p-type semiconductor [27], the ZnTe:Cu host lattice with relatively high resistivity is still dominated in size. However, further increase in TCu leads to a quantum leap in the transport mechanism. It is likely that for samples grown using TCu = 990 °C, the density of the Cu-Te nano-rods has reached a certain value and quantum tunneling for the hole carriers among neighboring nano-rods arises, leading to the observed sharp fall of the apparent resistivity.

Figure 5 addresses the formation mechanism and the geometrical shape of the Cu-Te nano-rods based on a phenomenological perspective. At high TCu condition, only part of the incoming Cu atoms will be incorporated as a substitutional dopant in the ZnTe lattice, the remaining Cu atoms would prefer to migrate upward toward the surface during the growth. When the as-grown thin film reaches a certain thickness, the concentration of the extra Cu atoms coming from the up-migration as well as from the source flux reaches a threshold value for the formation of Cu-rich Cu-Te nano-rods in some favorable sites. It is worth pointing out that the two embedded sides of the triangular cross-section of the nano-rods are found to be the members of the {111} plane family of ZnTe, which is based on the measured angle between these two sides and the surface normal of the sample, as shown in the insets of Fig. 2a, b. This observation is similar to the geometry of the ZnSe nanotrenches induced by mobile Au-alloy droplets as reported by us earlier [28]. Thus, it is expected that the formation of Cu-Te nano-rods shares some similarities with the formation of the ZnSe nanotrenches. The triangular cross-sectional shape of the Cu-Te nano-rods indicates that the growth of the Cu-Te nano-rods must proceed with a simultaneous dissociation of the neighboring as-grown ZnTe:Cu to expose the {111} planes that have low surface energy [29, 30]. This is well supported by the fact that the bonding strength of Cu-Te is 230.5 ± 14.6 kJ mol–1 while that of Zn-Te is 117.6 ± 18.0 kJ mol–1 [31]. The stronger bonding strength of Cu-Te over Zn-Te explains why the growth of Cu-Te nano-rods is more preferable accompanying with the dissociation of the neighboring ZnTe:Cu lattice when the concentration of extra Cu atoms is high enough. It is also worth mentioning that the triangular cross-sectional shape of the nano-rods also implies that the continuous growth of the nano-rods should require more and more Cu atoms. However, as the external Cu flux is fixed for a fixed Cu cell temperature and the Cu source from up-migration is also expected to be limited, one should expect that the growth of the nano-rods may proceed with reducing growth rate. In fact, this is likely to be the cause of their dented surfaces as seen by both the AFM and TEM images in Fig. 1e, f and Fig. 2b, respectively. Interestingly, as shown in Fig. 2b, the top edges near the dented surface of the nano-rods (indicated by solid arrows) are covered with a thin layer of Cu-Te thin film, which can be explained by the expected faster growth rate along the kink sites on the top edges since it is well known that a kink site has a lower energy than a site at the top surface of a nano-rod.

Fig. 5
figure 5

Schematic formation mechanism of 1D surface modulation and the Cu-Te nano-rods with a–c represent the progress order of the formation

Full size image

Conclusions

In summary, we have carried out the MBE growth of a set of ZnTe:Cu thin films. It was found that 1D feature was observed in the RHEED patterns of thin films grown using TCu ≥ 870 °C, though this feature got dimmer with increasing Cu incorporation. The apparent resistivity of these films was observed to have an unusual sharp fall followed by a plateau for TCu ≥ 990 °C. Various post-growth characterizations including AFM, TEM, HRXRD, and SIMS reveal that at high Cu incorporation, substitutional doping reaches saturation and the remaining Cu atoms proceed with an up-migration. The accumulated Cu atoms at the surface, being supplied by both the up-migration and the source flux, modify the surface to form 1D surface modulation as well as forming Cu-Te nano-rods near the surface by dissociating their neighboring ZnTe:Cu lattice. We have provided a phenomenological model regarding the self-assembled formation mechanism of 1D surface modulation and nano-rods as well as illustrated how these nanostructures are related to the observed unusual RHEED patterns and apparent resistivity. Further studies on the exact Cu composition in the nano-rods as well as their crystal structure are underway and will be reported elsewhere.