Open Access
Add to BibSonomyAbstract
CdZnO/ZnO quantum well (QW) samples are grown on GaN and ZnO templates with plasma-assisted molecular beam epitaxy under different conditions of substrate temperature, Cd effusion cell temperature, and O2 flow rate for emission characteristics comparison. It is found that the Cd incorporation on the ZnO template is generally lower, when compared with that on the GaN template, such that the O2 flow rate needs to be reduced for stoichiometric CdZnO/ZnO QW growth on the ZnO template. Besides the wurtzite (wt) CdZnO structure, the rock-salt (rs) CdZnO structure exists in the CdZnO well layers when the total Cd content is high. The rs structure may dominate over the wt structure in photoluminescence intensity when the total Cd content is high. In either group of samples on the GaN and ZnO templates, the emission efficiency first increases and then decreases with increasing total Cd content. The low emission efficiency at low (high) Cd content is attributed to the weaker quantum confinement (the poorer crystal quality) of the QWs. The emission efficiencies of the QW samples on the GaN template are generally higher than those on the ZnO template. The carrier localization behavior in a CdZnO/ZnO QW, grown on either GaN or ZnO template, is significantly weaker than that in an InGaN/GaN QW. The strength of the quantum-confined Stark effect generally increases with increasing Cd content in either group of samples on the GaN and ZnO templates.
©2012 Optical Society of America
1. Introduction
Because of the larger exciton binding energy in ZnO (60 meV in ZnO versus 24 meV in GaN), the existence of excitons at room temperature in a ZnO-based compound is expected to result in a higher radiative recombination rate for device application, when compared with a GaN-based compound. For visible emission, CdZnO is a suitable compound for extending emission wavelength to the green range [1]. However, because CdO has a rock-salt (rs) structure, its miscibility in wurtzite (wt) ZnO is low, leading to the difficulty of forming a CdZnO thin film or CdZnO/ZnO quantum well (QW) structure of high crystal quality. In CdZnO, when the Cd content is high, the rs-CdZnO structure may affect the quality of the wt-CdZnO structure. Also, phase separation may occur [2–4].
CdZnO thin films have been widely grown with molecular beam epitaxy (MBE), metalorganic chemical vapor deposition (MOCVD) [5], sputtering [6], and pulsed laser deposition (PLT) [7] on ZnO, sapphire, and Si substrates. Generally speaking, among different growth techniques, so far, MBE can result in better crystal quality. It has been shown that with MBE, high-quality CdZnO thin films can be obtained after a high-temperature (~550 °C) post-growth annealing procedure with the sample grown at a low temperature (~150 °C) under an oxygen-rich condition on ZnO or sapphire substrate [8, 9]. Also, with MBE growth on a GaN template, which is deposited on a sapphire substrate, by using a ZnO buffer layer, a high-quality CdZnO thin film can be formed at relatively higher growth temperatures (400-600 °C) [10]. Based on the photoluminescence (PL) measurement of Cd-content-varied samples, the band gap of CdZnO, Eg, follows the equation of Eg(x) = 3.37 - 2.82x + 0.95x2, where x represents the Cd content [11]. Post-growth thermal annealing can improve the crystal quality of CdZnO, particularly when the Cd content is high and the rs structure mixes with the wt structure. It was demonstrated that after a thermal annealing procedure at 800 °C of a CdZnO thin-film sample on sapphire or Si substrate, the rs structure can disappear, leading to the blue shift of its PL spectrum [12, 13].
CdZnO/ZnO QW structures have also been widely studied. First, the carrier localization behavior due to Cd composition fluctuation in a CdZnO/ZnO QW structure grown on ScAlMgO4 substrate with PLD was reported [14]. Also, with low-temperature MBE growth on sapphire substrate and post-growth thermal annealing, efficient PL and stimulated emissions from CdZnO/ZnO QWs were demonstrated in the range from ultraviolet through green color [15–17]. Meanwhile, various thermal annealing conditions were tested to conclude that 650 °C for 15 min is an optimum condition for improving the emission efficiency of a CdZnO/ZnO QW structure [18]. Besides, it was found that the incorporation of deuterium atoms into a QW structure could significantly enhance the emission efficiency [19, 20]. The PLD growth of CdZnO/ZnO QW structures on sapphire substrate was undertaken to illustrate a smooth sample surface and efficient emission [21]. In such a sample, weak carrier localization was observed [22]. Finally, the MOCVD growths of multiple-QW structures of various well widths were performed to confirm the effect of quantum confinement in such a structure [23].
The fabrication of all-ZnO light-emitting diode (LED) is limited by the stable growth of p-ZnO. Although many efforts have been put on the growth of p-ZnO, so far, the concern of its stability is still a major issue in this community. p-ZnO formations by doping N [24, 25], P [26, 27], As [28–32], Sb [33, 34], Ag [35], and Bi [36] were reported. Co-doping of N plus As [37] and N plus Al [38, 39] for growing p-ZnO were also reported. With MBE growth on Si substrate, a p-ZnO/i-CdZnO/n-ZnO heterostructure LED was fabricated to emit blue light at 459 nm [40]. Here, the n-ZnO and p-ZnO were doped with Ga and Sb, respectively. Recently, a structure of n-MgZnO/Cd0.2Zn0.8O/Cd0.92Zn0.08/n-MgZnO/n-ZnO was grown on p-4H-SiC substrate to form an LED for emitting light around 500 nm in wavelength [41]. The screening of the quantum-confined Stark effect (QCSE) was observed. Besides the all-ZnO LED development, because of the small lattice mismatch (1.8%) between ZnO and GaN, p-GaN, which has been widely grown with MOCVD, has been proposed to replace p-ZnO for fabricating a hybrid LED. An n-MgZnO/CdZnO/p-GaN single-QW LED structure on sapphire substrate grown with MOCVD (p-GaN) and MBE (n-MgZnO/CdZnO) was formed to emit electroluminescence from 390 through 410 nm in peak wavelength [42]. Recently, an MBE-grown CdZnO/n-ZnO multi-QW structure and an n + -ZnO cap layer were deposited on MOCVD-grown p-GaN, which was coated on c-plane sapphire substrate, for fabricating a low-voltage blue-emitting LED [43]. The development of CdZnO/ZnO QW LED based on the growth on p-GaN is quite promising. However, the optical characteristics of CdZnO/ZnO QWs grown on GaN have not been well studied yet. In particular, although it was speculated that carrier localization might exist in a CdZnO/ZnO QW, a systematical study has not been reported. Also, it is interesting to compare the optical characteristics between the CdZnO/ZnO QWs grown on GaN and ZnO templates. Such information is useful for further developing a ZnO/GaN hybrid-growth LED.
In this paper, we compare the emission characteristics between the CdZnO/ZnO QW samples of various Cd contents. In particular, we compare those between the QW samples grown on GaN and ZnO templates with plasma-assisted MBE. Temperature-dependent and excitation-power-dependent PL measurements are undertaken for illustrating their different emission characteristics. Besides the wt-CdZnO structures, the rs-CdZnO structures exist in the CdZnO well layers of the high-Cd samples. The rs structures become more prominent and may dominate over the wt structures in PL emission when the total Cd content is high. The emission efficiencies of the QW samples on the GaN template are generally higher than those on the ZnO template. In section 2 of this paper, the conditions of sample preparation and characterization are described. The transmission electron microscopy (TEM) and X-ray diffraction (XRD) results are shown in section 3. Then, in section 4, the PL emission characterization results are presented. Discussions are made in section 5. Finally, conclusions are drawn in section 6.
2. Sample preparation and characterization conditions
In this study, 12 samples in total are prepared for comparison, including 11 CdZnO/ZnO QW samples and one InGaN/GaN QW sample. The InGaN/GaN QW sample (designated as sample L) is prepared for showing the behavior of carrier localization. The 11 CdZnO/ZnO QW samples can be classified into three categories, including five samples of different Cd contents grown on the GaN templates (samples A-E), five samples of different Cd contents grown on the ZnO templates (samples G-K), and one sample (sample F) grown on the ZnO template under the same growth conditions as sample B, which is prepared on the GaN template. In rows 2-5 of Table 1 , we list the different growth conditions of the 11 CdZnO/ZnO QW samples. In a GaN template, a 2-μm thick undoped GaN layer is deposited at 1000 °C after the growth of a 50-nm GaN buffer layer at 550 °C on c-plane sapphire substrate with MOCVD. In a ZnO template a 200-nm thick undoped ZnO layer is deposited on c-plane sapphire substrate with plasma-assisted MBE under the stoichiometry conditions of 600 °C in substrate temperature, 295 °C in Zn effusion cell temperature, and 1.5 sccm in O2 flow rate. In each of the CdZnO/ZnO QW samples, either on the GaN or ZnO template, three QWs are grown with plasma-assisted MBE. The conditions of 200 °C in substrate temperature, 295 °C in Zn effusion cell temperature, and 1.5 sccm in O2 flow rate are used for depositing the ZnO barrier layers (~12 nm in thickness for each barrier layer) in all the CdZnO/ZnO QW samples. In growing the CdZnO well layers in all the CdZnO/ZnO QW samples, the Zn effusion cell temperature is fixed at 295 °C. However, the substrate temperature, Ts, Cd effusion cell temperature, TCd, and O2 flow rate are varied among different samples, as shown in rows 3-5 of Table 1. Here, one can see that except sample K, the substrate temperature is fixed at 200 °C. The comparison between samples J and K illustrates the effect of reducing the substrate temperature. In row 4 of Table 1, we can see that the used Cd effusion cell temperatures for growing the well layers in the samples on the GaN and ZnO templates are quite different. However, the PL emission wavelengths of the two categories of sample are comparable, as to be discussed later, implying that the Cd incorporation efficiencies on the GaN and ZnO templates are different. The key difference of the growth condition between the two categories of sample is the O2 flow rate in growing the well layers. The samples on the GaN template (samples A-E) are grown under the oxygen-rich condition with 1.8 sccm in O2 flow rate. Those on the ZnO template (samples G-K) are grown under the metal-rich condition with 1.2 sccm in O2 flow rate. Sample F is prepared under the conditions exactly the same as those for sample B except that samples B and F are grown on the GaN and ZnO templates, respectively. As to be discussed later, sample F has very low emission efficiency, indicating that the metal-rich condition is preferred for growing CdZnO/ZnO QWs of high emission efficiency on a ZnO template. Finally, sample L is a standard InGaN/GaN QW LED structure. It consists of five periods of blue-emitting InGaN/GaN QW on a 2-μm n-GaN layer and is capped by a 20-nm p-AlGaN layer and then a 120-nm p-GaN layer.
Table 1. Growth conditions and characterization results of various samples. Here, ws represents the whole spectrum.
A double-crystal Bede system is used for the XRD measurements. The TEM investigation is performed using a Philips Tecnai F30 field-emission electron microscope with an accelerating voltage of 300 kV and a probe forming lens of Cs = 1.2 mm. The PL measurement is excited with a 325-nm HeCd laser. For temperature-dependent PL measurements, the excitation power is fixed at 5 mW.
3. Basic characterization results
Figures 1(a) -1(c) show the high angle annular dark field (HAADF) images in TEM observation of samples B, F, and H, respectively. Here, one can clearly see three QWs in samples B and H, as indicated by the arrows. However, the QW structure in sample F is unclear, confirming that the growth of the QWs under the conditions for sample F is unsuccessful. Figure 2 shows the (0002)-plane ω-2θ XRD patterns of samples A-E. For comparison, that of the GaN template (labeled by “GaN”) is also shown. In this figure, the major peak at 34.87 degrees corresponds to GaN (0002) in the template. The shoulders on the left contain the signals of the ZnO in the barriers and wt-CdZnO in the well layers. The ZnO (0002) signal is located around 34.54 degrees, as indicated by a vertical arrow and the short-dashed line. The position of the wt-CdZnO signal varies slightly within a small range around 34.43 degrees, as indicated by the dashed arrows and the dotted line for various samples. It is noted that in the curve of the GaN template, there is a peak feature around 33.39 degrees, as indicated by a slant arrow. This feature appears at the same position in every other curve of Fig. 2, as indicated by the long-dashed line. This feature is supposed to be caused by certain unknown structure in the buffer layer for fabricating the GaN template on sapphire substrate. In samples A-E, except this unknown GaN feature, the oscillating behaviors left and right to the aforementioned major peaks and shoulders originate from the periodical QW structures. However, in some of the samples, the periodical patterns on the left-hand side are unclear, indicating that the QWs are not well-shaped in those samples. This result can also be caused by the coexistence of the wt-CdZnO and rs-CdZnO structures in the samples. Particularly, in sample E, there is a feature around 33.21 degrees, which has been identified to be caused by the rs-CdZnO (111) structure in the sample [6], as indicated by a thick arrow. The rs-CdZnO structures can also exist in other samples with possibly lower Cd contents and smaller volume ratios. Thick arrows are also drawn to indicate the peak features possibly caused by the rs-CdZnO structure in samples C and D. They will be further discussed together with the PL measurement results later.
Fig. 1 TEM HAADF images of samples B (a), F (b), and H (c). The QWs are indicated by the horizontal arrows.
Fig. 2 XRD patterns in the (0002) plane of samples A-E. For comparison, the XRD result of the GaN template is also shown. The major peaks correspond to the GaN signals. The ZnO features are indicated by the vertical short-dashed line. The wt-CdZnO features are indicated by the vertical dotted line. The unknown feature from the GaN template is indicated by the vertical long-dashed line. The rs-CdZnO features are indicated by the thick arrows.
Figure 3 shows the (0002)-plane ω-2θ XRD patterns of samples F-K. For comparison, that of the ZnO template is also shown. In this figure, the major peak at 34.42 degrees corresponds to ZnO (0002) in the template and ZnO barriers. The shoulders on the left in a small range around 34.31 degrees contain the signals of wt-CdZnO in the well layers, as indicated by the dashed arrows and dotted line. The increasing wt-CdZnO signal strength from sample G through I can be clearly seen. Compared with samples G-I, the Cd content in sample F is lower. Again, the oscillating features on both sides of the ZnO major peaks and wt-CdZnO shoulders in samples F-I are due to the periodical QW structures. Similar to the results in Fig. 2, the periodical oscillating patterns are not very clear in Fig. 3, indicating that the QWs are not well-shaped either in the samples grown on the ZnO template. Particularly, in sample F, the features on both sides of the major peak are unclear, indicating that the CdZnO/ZnO QW structure is not successfully fabricated in this sample. The broad humps between 32.5 and 33.75 degrees in samples J and K are supposed to originate from the rs-CdZnO structures [6]. The broad-hump features imply the poor crystal quality and possibly the large-range variation of Cd content in the rs structures. This XRD signal hump in sample J is narrower than that in sample K, indicating that the Cd content in the rs-CdZnO structure of sample J can be smaller and/or its volume ratio is smaller. The rs-CdZnO structure can also exist in other samples. However, it is difficult to identify their peak features in the curves other than samples J and K. By comparing the rs-CdZnO features in the curves of samples C-E (see Fig. 2), J, and K (see Fig. 3), one can see that the XRD feature widths of the rs-CdZnO structures in the samples grown on the ZnO template are significantly larger than those in the samples grown on the GaN template. Just like the rs-CdZnO features in Fig. 2, those in Fig. 3 will be further discussed with the PL measurement results later.
Fig. 3 XRD patterns in the (0002) plane of samples F-K. For comparison, the XRD result of the ZnO template is also shown. The major peaks correspond to the ZnO signals. The wt-CdZnO features are indicated by the vertical dotted line. The rs-CdZnO features are indicated by the thick arrows.
It is noted that although we can identify the rs-CdZnO feature peaks in the XRD pattern, it is not straightforward to tell their corresponding Cd compositions. In the ω-2θ scanning, the rs-CdO (111) results in a feature at 33.1 degrees [6]. Based on this reference, the Cd contents of the rs-CdZnO structures in our samples can be quite high. However, this speculation cannot be assured. It is also noted that because the lattice sizes of ZnO grown on different under-layers (GaN or sapphire) are slightly different, their ω-2θ XRD features are located at slightly different scan angles, as shown in Figs. 2 and 3.
4. Photoluminescence measurement results
Figures 4(a) and 4(b) show the normalized PL spectra of samples A-E and L at the low temperature (LT) of 10 K and room temperature (RT) of 300 K, respectively. As mentioned earlier, sample L consists of a blue-emitting InGaN/GaN QW structure for demonstrating the different emission behaviors from those of the CdZnO/ZnO QWs under study. Here, the slightly oscillating spectral shape is due to the un-filtered Fabry-Perot behavior in this sample. The long-wavelength tails in all the CdZnO/ZnO QW samples are caused by the defect emissions. In Fig. 4(a), although single-peak spectra are observed in samples A-E, a shoulder can be clearly seen in each sample except sample B. Similar shoulder structures can also be observed at RT, as shown in Fig. 4(b). Generally, among those samples, the spectral peak wavelength increases with increasing TCd or Cd atom supply at either LT or RT. These spectral peak wavelengths are listed in rows 6 and 7 of Table 1. Figures 5(a) and 5(b) show the normalized PL spectra of samples G-K and L at the LT and RT, respectively. Here, the ZnO emission features around 375 nm can be clearly seen. At LT, two spectral peaks can be seen in sample K, indicating two different emission origins in this sample. The short-wavelength peak is stronger at LT. However, at RT, the long-wavelength peak becomes stronger and the short-wavelength one appears as a shoulder. In each of other samples shown in Figs. 5(a) and 5(b), two different emission origins may also exist. Nevertheless, they are not as clear as that of sample K. The PL spectral peak wavelengths at LT and RT are also listed in rows 6 and 7 of Table 1.
The PL emission in each sample may consist of two contributions from the wt- and rs-CdZnO structures. To differentiate these two emission contributions, we use two Gaussian curves to fit the PL emission spectrum at each measurement temperature for each of samples C-E, J, and K in the photon energy domain. However, at relatively higher temperatures, the fitting may become difficult. Also, in samples A, B, and G-I, although the two-component fitting can be feasible, we will not discuss their separate PL contributions of wt and rs structures because the differentiation between the two PL contributions is unclear. Figures 6(a) -6(c) show the two-Gaussian fitting results of samples C-E, respectively, at LT. It is noted that in these three samples, between the two PL contributions, the stronger contributions are located at relatively longer wavelengths. Figures 7(a) and 7(b) show the two-Gaussian fitting results of samples J and K, respectively, at LT. In these two samples, the stronger contributions are located at relatively shorter wavelengths.
Fig. 6 Normalized PL spectra at LT and their two-component fitting curves for samples C (a), D (b), and E (c). The assignments of those fitting components to either wt or rs structures are labeled.
Fig. 7 Normalized PL spectra at LT and their two-component fitting curves for samples J (a) and K (b). The assignments of those fitting components to either wt or rs structures are labeled.
To interpret the emission origins of the two PL contributions in Figs. 6 and 7, we make two reasonable assumptions. First, the two PL contributions individually correspond to the emissions from the wt-CdZnO and rs-CdZnO structures in the QWs. Second, when the total Cd content is relatively lower, the emission intensity of the wt-CdZnO structure is relatively stronger than that of the rs-CdZnO structure. However, when the total Cd content is high, the emission of the rs-CdZnO structure may dominate. The discussions start with the broad-hump rs-CdZnO XRD feature between 32.5 and 33.75 degrees (peaked at ~33.2 degrees) in sample K, as shown in Fig. 3. This rs-CdZnO structure is supposed to emit photons for producing the PL peak at 482 nm (labeled by rs) in Fig. 7(b). Therefore, the PL peak at 435 nm (labeled by wt) in this figure originates from the wt-CdZnO structure in sample K. Similarly, the XRD feature around 33.18 degrees in sample E (see Fig. 2) corresponds to the rs-CdZnO structure in this sample, which emits photons to produce the dominating PL peak at 486 nm (labeled by rs), as shown in Fig. 6(c). This attribution is based on the similar XRD scanning angles (~33.2 degrees) of the rs features in samples E and K and the similar peak wavelengths (~484 nm) of their long-wavelength PL components. Therefore, the relatively weaker PL peak at 457 nm (labeled by wt) originates from the wt-CdZnO structure in this sample. Because the XRD pattern of sample D in Fig. 2 does not have a significant feature around 33.18 degrees, the PL peak at 480 nm (see Fig. 6(b)) of this sample, which is close to the 486 nm peak of sample E, cannot be attributed to the emission of the rs-CdZnO structure. This PL contribution must originate from the wt structure and the other PL peak at 455 nm comes from the rs structure. This rs PL feature of a shorter wavelength in sample D may correspond to the XRD feature around 33.55 degrees shown in Fig. 2 (indicated by a thick arrow). Following the similar argument, the XRD feature around 33.74 degrees (also indicated by a thick arrow) in sample C can be caused by the rs structure, which produces the PL peak at 423 nm (see Fig. 6(a)). In sample J, the broad XRD feature peaked at 33.26 degrees corresponds to its rs structure, which is supposed to produce the PL peak at 464 nm in Fig. 7(a). The spectral peak wavelengths of the fitted components of the wt and rs contributions at LT are listed in rows 8 and 9, respectively, of Table 1.
In both groups of sample on the GaN and ZnO templates, unless the total Cd content is high (samples E and K), the Cd composition of the wt structure and the corresponding PL peak wavelength increase with increasing total Cd content. Also, the emission from the wt structure is always stronger than that from the rs structure. When the total Cd content is relatively higher (such as in samples E, J, and K), the rs structure dominates over the wt structure in the QWs such that the Cd composition of the wt structure is reduced for emitting a PL peak at a relatively shorter wavelength, when compared with that of the rs structure (457 nm vs. 486 nm in sample E, 455 nm vs. 464 nm in sample J, and 435 nm vs. 482 nm in sample K). Also, the emission peak of the wt structure is blue-shifted when the total Cd content is increased (457 nm in sample E vs. 480 nm in sample D, and 435 nm in sample K vs. 455 nm in sample J). Meanwhile, the wt emission contribution can become relatively weaker, as shown in Fig. 6(c) for sample E at LT. The rs structure can result in relatively stronger emission at RT even though its contribution at LT is smaller, as shown in Fig. 5(b) for sample K. Although the PL peak wavelengths from the rs structures in samples J and K are longer than the corresponding wavelengths from the wt structures, this trend is not necessarily true in samples C-E.
Figures 8(a) and 8(b) show the PL spectral peak energies as functions of temperature of the wt and rs components, respectively, of samples C-E. For comparison, the corresponding data of samples A and B with the whole spectra (without splitting into the wt and rs components) are also plotted in both Figs. 8(a) and 8(b). Figure 8(c) shows the PL spectral peak energies as functions of temperature of samples G-I (the whole spectra) and the wt and rs components of samples J and K. The curve of sample L is also shown in Figs. 8(a)-8(c) for comparison. Because the two-component fitting may not be feasible at a high temperature, some of the curves terminate at certain temperatures lower than 300 K. However, all the curves have the data up to 240 K. The curve of sample L shows an S-shaped variation with temperature. This S-shaped behavior is regarded as a crucial indicator of the carrier localization phenomenon in an InGaN/GaN QW [44–46]. In this behavior, the PL peak energy blue shifts in an intermediate temperature range (between 90 and 210 K in the current case) between the low and high temperature ranges of red shift trends. The blue-shift trend is due to the elevation of carriers to the higher localized states before they escape from localization at a higher temperature. In Figs. 8(a)-8(c), a few curves, such as D-wt, D-rs, E-wt, J-wt, J-rs, and K-wt, show non-monotonically decreasing variation trends. Although the behaviors of carrier localization may exist in those wt or rs structures, the evidence is not strong. Generally speaking, the behavior of carrier localization due to composition fluctuations is significantly weaker in a CdZnO/ZnO QW, when compared with an InGaN/GaN QW.
Fig. 8 (a) and (b): PL spectral peak energies as functions of temperature of the wt (a) and rs (b) components in samples C-E. For comparison, the corresponding data of samples A and B with the whole spectra are also plotted. (c): Similar results for samples G-I and the wt and rs components of samples J and K. The result of sample L is also shown in parts (a)-(c) for comparison.
Figures 9(a) -9(c) show the integrated PL intensities as functions of temperature for the whole spectra of samples A-E, the wt components of samples C-E, and the rs components of samples C-E, respectively. For comparison, the curves for the whole spectra of samples A and B are also plotted in Figs. 9(b) and 9(c). Figure 10(a) shows the similar results for the whole spectra of samples G-K. Then, Fig. 10(b) shows the similar results for the whole spectra, the wt and rs components of samples J and K. With the integrated PL intensity at 10 K normalized to unity, the corresponding level at a high temperature can be regarded as the internal quantum efficiency (IQE) [47, 48]. However, it is noted that in some of the curves of Figs. 9(a)-9(c), 10(a), and 10(b), the integrated PL intensity in a low temperature range is higher than that at 10 K. This behavior is particularly strong in sample J. It is due to the flow of carriers, which are photo-excited in the ZnO barrier layers and other cladding layers, into the well layers for PL emission. In this situation, if we use the level of the normalized integrated PL intensity at a high temperature as the IQE, the IQE level is overestimated. The IQEs read at 300 K from the data of the whole spectra (ws), those read at 240 K from the data of the wt component, and those read at 240 K from the data of the rs component are listed in rows 10-12, respectively, of Table 1. In each IQE data set of two numbers here, the first number corresponds to the result read directly from the level of the curve at either 240 or 300 K. The second number is obtained through normalizing the first number by the maximum level in the individual curve when there are one or more data points higher than unity. The IQE of the whole spectrum (ws), either adjusted or unadjusted, in either group of sample first increases and then decreases with increasing TCd or total Cd content. The maximum whole-spectrum IQE is observed in sample C (H) among the samples grown on the GaN (ZnO) template. The relatively lower IQEs when the total Cd contents are low are due to the poorer quantum confinement because of the smaller potential difference between the CdZnO well and ZnO barrier layers. The decreased IQEs when the total Cd contents are high are caused by the poorer crystal quality, particularly when comparable amounts of the wt and rs structures are mixed in the QWs. The IQEs, either the total emission, the wt component, or the rs component, of the samples grown on the GaN template are generally significantly higher than those grown on the ZnO template. However, no conclusion can be drawn in comparing the IQEs between the wt and rs components. It is noted that the IQEs read at 300 and 240 K of the InGaN/GaN QW sample (sample L) are 33.3 and 44.6%, respectively.
Fig. 9 Normalized integrated PL intensities as functions of temperature for the whole spectra of samples A-E (a), the wt components of samples C-E (b), and the rs components of samples C-E (c). For comparison, the curves for the whole spectra of samples A and B are also plotted in parts (b) and (c).
Fig. 10 (a): Normalized integrated PL intensities as functions of temperature for the whole spectra of samples G-K. (b): Similar results for the whole spectra, the wt and rs components of samples J and K.
Figures 11(a) and 11(b) show the peak energy variations of PL spectra when the excitation power increases from 1 through 10 mW at RT for samples A-E and G-K, respectively. Because it is difficult to differentiate the wt and rs components of PL emission at RT in some of the samples, the peaks of the whole spectra are used for plotting Figs. 11(a) and 11(b). When the PL excitation power is increased, the carrier screening effect can reduce the QW potential tilt, which is caused by the QCSE, for increasing the effective band gap or blue-shifting the emission spectrum. The spectral blue-shift range of such an excitation power-dependent PL measurement can be used for indicating the strength of the QCSE in a QW sample [49–51]. The blue-shift ranges from1 through 10 mW in excitation power of those samples are listed in row 13 of Table 1. They represent the QCSE strengths of the dominating components (wt or rs) in those samples. Here, one can see that except sample I, the blue-shift range increases with increasing total Cd content in either group of sample. The blue-shift ranges of samples B-D and G-J are distributed in a small range of 6.2-7.0 meV. The QCSE strengths of those samples are supposed to be about the same. That of sample A (E or K) is significantly weaker (stronger). The PL emission of sample K at RT is dominated by the rs component. It is noted that the blue-shift range of the InGaN/GaN QW sample (sample L) is 10.5 meV, which is close to that of sample K.
Fig. 11 Variations of PL spectral peak energy with excitation power at RT for samples A-E (a) and G-K (b). For comparison, the result of sample L is also plotted in both parts (a) and (b).
5. Discussions
Several characteristic differences between the samples grown on the GaN and ZnO templates can be identified. First, by comparing samples B and I, which have the similar PL spectral peaks at LT and RT, one can see that the Cd effusion cell temperature for growing sample I (300 °C) is significantly higher than that for growing sample B (238 °C). The Cd effusion cell temperatures at 300 and 238 °C lead to the beam equivalent pressures of Cd atom in the MBE chamber at 5.5 x 10−7 and 1.1 x 10−7 torr, respectively. In other words, the supply of Cd atoms in growing sample B is significantly lower than that in growing sample I. It is noted that in growing sample F, the growth conditions are the same as those for sample B, except that they are grown on different templates. However, the QW quality of sample F is significantly poorer, when compared with samples B and I. Such a comparison result implies that the Cd incorporation on the ZnO template is lower than that on the GaN template. Therefore, the Cd effusion cell temperature needs to be higher for achieving a higher Cd atom supply. Meanwhile, the oxygen supply must be reduced for stoichiometric growth. In growing sample F on the ZnO template, the oxygen is over-supplied such that the growth of CdZnO becomes difficult. Because the lattice size of GaN is smaller than that of ZnO, the compressive strain in the CdZnO layer on the GaN template is stronger than that on the ZnO template such that the Cd incorporation on the GaN template is expected to be lower. Therefore, the relative lattice size of GaN and ZnO cannot be used for explaining the observed higher Cd incorporation on the GaN template. This issue requires further investigation. Nevertheless, it is speculated that the higher crystal quality of the GaN template can be one of the reasons for the higher Cd incorporation and higher QW crystal quality on the GaN template. With the higher QW crystal qualities in the samples grown on the GaN template, their emission efficiencies are higher.
To observe the effect of changing substrate temperature, we reduce the substrate temperature from 200 to 170 °C in sample K for comparing its emission characteristics with those of sample J. As shown in Figs. 5(a), 5(b), 7(a), and 7(b), by reducing the substrate temperature, both the total Cd content and the Cd content difference between the wt and rs structures become larger such that the spectral peaks of the two emission components become further apart. In particular, because the Cd content of the rs structure is significantly increased, that of the wt structure is reduced to become smaller than that of sample J. With the increased total Cd content and the increased Cd content in the rs structure of sample K, its emission efficiency is reduced. However, its QCSE becomes stronger.
6. Conclusions
In summary, we have grown CdZnO/ZnO QW samples on the GaN and ZnO templates under different conditions of substrate temperature, Cd effusion cell temperature, and O2 flow rate with plasma-assisted MBE. It was found that the Cd incorporation on the ZnO template was relatively lower such that the O2 flow rate needed to be reduced for stoichiometric CdZnO/ZnO QW growth on ZnO. In the CdZnO well layers, the rs and wt structures coexisted in the samples of high Cd contents. In this situation, the rs structures might dominate the emission. In either group of samples on the GaN and ZnO templates, the emission efficiency first increased and then decreased with increasing Cd content. The low emission efficiency at low (high) Cd content was due to the weaker quantum confinement (the poorer crystal quality) of the QWs. The emission efficiencies of the QW samples on the GaN template were generally higher than those on the ZnO template. The carrier localization behavior in a CdZnO/ZnO QW, grown on either GaN or ZnO template, was significantly weaker than that in an InGaN/GaN QW. The QCSE strength generally increased with increasing Cd content in either group of samples on the GaN and ZnO templates.
Acknowledgment
This research was supported by National Science Council, Taiwan, The Republic of China, under the grants of NSC 99-2221-E-002-123-MY3, 100-2622-E-002-008-CC2, 100-2221-E-002-170, by NTU Excellent Research Project (10R80908-B), by Epistar Corporation, and by US Air Force Scientific Research Office under the contract of AOARD-11-4114.
References and links
1. R. P. Koffyberg, “Thermoreflectance spectra of CdO: Band gaps and band-population effects,” Phys. Rev. B 13(10), 4470–4476 (1976). [CrossRef]
2. J. Ishihara, A. Nakamura, S. Shigemori, T. Aoki, and J. Temmyo, “Zn1−xCdxO systems with visible band gaps,” Appl. Phys. Lett. 89(9), 091914 (2006). [CrossRef]
3. Y. Z. Zhu, G. D. Chen, H. Ye, A. Walsh, C. Y. Moon, and S. H. Wei, “Electronic structure and phase stability of MgO, ZnO, CdO, and related ternary alloys,” Phys. Rev. B 77(24), 245209 (2008). [CrossRef]
4. K. Sakurai, T. Takagi, T. Kubo, D. Kajita, T. Tanabe, H. Takasu, S. Fujita, and S. Fujita, “Spatial composition fluctuations in blue-luminescent ZnCdO semiconductor films grown by molecular beam epitaxy,” J. Cryst. Growth 237-239, 514–517 (2002). [CrossRef]
5. A. Y. Azarov, T. C. Zhang, B. G. Svensson, and A. Y. Kuznetsov, “Cd diffusion and thermal stability of CdZnO/ZnO heterostructures,” Appl. Phys. Lett. 99, 111903 (2011). [CrossRef]
6. R. Zhang, P. Chen, Y. Zhang, X. Ma, and D. Yang, “Effect of rapid thermal annealing on photoluminescence and crystal structures of CdZnO films,” J. Cryst. Growth 312(12-13), 1908–1911 (2010). [CrossRef]
7. H. S. Kang, S. H. Lim, J. W. Kim, H. W. Chang, G. H. Kim, J. H. Kim, S. Y. Lee, Y. Li, J. S. Lee, J. K. Lee, M. A. Nastasi, S. A. Crooker, and Q. X. Jia, “Exciton localization and Stokes’ shift in Zn1−xCdxO thin films depending on chemical composition,” J. Cryst. Growth 287(1), 70–73 (2006). [CrossRef]
8. S. Sadofev, P. Schäfer, Y. H. Fan, S. Blumstengel, F. Henneberger, D. Schulz, and D. Klimm, “Radical-source molecular beam epitaxy of ZnMgO and ZnCdO alloys on ZnO substrates,” Appl. Phys. Lett. 91(20), 201923 (2007). [CrossRef]
9. S. Sadofev, S. Blumstengel, J. Cui, J. Puls, S. Rogaschewski, P. Schäfer, and F. Henneberger, “Visible band-gap ZnCdO heterostructures grown by molecular beam epitaxy,” Appl. Phys. Lett. 89(20), 201907 (2006). [CrossRef]
10. J. J. Chen, S. Jang, F. Rena, S. Rawal, Y. Li, H. S. Kim, D. P. Norton, S. J. Pearton, and A. Osinsky, “Thermal stability of Ti/Al/Pt/Au and Ti/Au Ohmic contacts on n-type ZnCdO,” Appl. Surf. Sci. 253(2), 746–752 (2006). [CrossRef]
11. X. J. Wang, I. A. Buyanova, W. M. Chen, M. Izadifard, S. Rawal, D. P. Norton, S. J. Pearton, A. Osinsky, J. W. Dong, and A. Dabiran, “Band gap properties of Zn1−xCdxO alloys grown by molecular-beam epitaxy,” Appl. Phys. Lett. 89(15), 151909 (2006). [CrossRef]
12. L. Li, Z. Yang, Z. Zuo, J. H. Lim, and J. L. Liu, “Thermal stability of CdZnO thin films grown by molecular-beam epitaxy,” Appl. Surf. Sci. 256(14), 4734–4737 (2010). [CrossRef]
13. L. Li, Z. Yang, Z. Zuo, J. Y. Kong, and J. L. Liu, “Study of rapid thermal annealing effect on CdZnO thin films grown on Si substrate,” J. Vac. Sci. Technol. B 28(3), C3D13, D16 (2010). [CrossRef]
14. T. Makino, C. H. Chia, N. T. Tuan, Y. Segawa, M. Kawasaki, A. Ohtomo, K. Tamura, and H. Koinumac, “Radiative and nonradiative recombination processes in lattice-matched (Cd,Zn)O/(Mg,Zn)O multiquantum wells,” Appl. Phys. Lett. 77(11), 1632–1634 (2000). [CrossRef]
15. S. Sadofev, S. Kalusniak, J. Puls, P. Schäfer, S. Blumstengel, and F. Henneberger, “Visible-wavelength laser action of ZnCdO/(Zn,Mg)O multiple quantum well structures,” Appl. Phys. Lett. 91(23), 231103 (2007). [CrossRef]
16. S. Kalusniak, S. Sadofev, J. Puls, and F. Henneberger, “ZnCdO/ZnO – a new heterosystem for green-wavelength semiconductor lasing,” Laser Photon. Rev. 3(3), 233–242 (2009). [CrossRef]
17. S. Blumstengel, S. Sadofev, H. Kirmse, and F. Henneberger, “Extreme low-temperature molecular beam epitaxy of ZnO-based quantum structures,” Appl. Phys. Lett. 98(3), 031907 (2011). [CrossRef]
18. A. V. Thompson, C. Boutwell, J. W. Mares, W. V. Schoenfeld, A. Osinsky, B. Hertog, J. Q. Xie, S. J. Pearton, and D. P. Norton, “Thermal stability of CdZnO/ZnO multi-quantum-wells,” Appl. Phys. Lett. 91(20), 201921 (2007). [CrossRef]
19. W. Lim, D. P. Norton, S. J. Pearton, X. J. Wang, W. M. Chen, I. A. Buyanova, A. Osinsky, J. W. Dong, B. Hertog, A. V. Thompson, W. V. Schoenfeld, Y. L. Wang, and F. Ren, “Migration and luminescence enhancement effects of deuterium in ZnO/ZnCdO quantum wells,” Appl. Phys. Lett. 92(3), 032103 (2008). [CrossRef]
20. I. A. Buyanova, X. J. Wang, G. Pozina, W. M. Chen, W. Lim, D. P. Norton, S. J. Pearton, A. Osinsky, J. W. Dong, and B. Hertog, “Effects of hydrogen on the optical properties of ZnCdO/ZnO quantum wells grown by molecular beam epitaxy,” Appl. Phys. Lett. 92(26), 261912 (2008). [CrossRef]
21. W. F. Yang, B. Liu, R. Chen, L. M. Wong, S. J. Wang, and H. D. Sun, “Pulsed laser deposition of high-quality ZnCdO epilayers and ZnCdO/ZnO single quantum well on sapphire substrate,” Appl. Phys. Lett. 97(6), 061911 (2010). [CrossRef]
22. W. F. Yang, L. M. Wong, S. J. Wang, H. D. Sun, C. H. Ge, A. Y. S. Lee, and H. Gong, “Photoluminescence characteristics of ZnCdO/ZnO single quantum well grown by pulsed laser deposition,” Appl. Phys. Lett. 98(12), 121903 (2011). [CrossRef]
23. K. Yamamoto, M. Adachi, T. Tawara, H. Gotoh, A. Nakamura, and J. Temmyo, “Synthesis and characterization of ZnCdO/ZnO multiple quantum wells by remote-plasma-enhanced MOCVD,” J. Cryst. Growth 312(9), 1496–1499 (2010). [CrossRef]
24. X. Li, Y. Yan, T. A. Gessert, C. DeHart, C. L. Perkins, D. Young, and T. J. Coutts, “p-Type ZnO thin films formed by CVD reaction of diethylzinc and NO gas,” Electrochem. Solid-State Lett. 6(4), C56–C58 (2003). [CrossRef]
25. Z. Q. Fang, B. Claflin, D. C. Look, L. L. Kerr, and X. Li, “Electron and hole traps in N-doped ZnO grown on p-type Si by metalorganic chemical vapor deposition,” J. Appl. Phys. 102(2), 023714 (2007). [CrossRef]
26. S. Jang, J. J. Chen, B. S. Kang, F. Ren, D. P. Norton, S. J. Pearton, J. Lopata, and W. S. Hobson, “Formation of p-n homojunctions in n-ZnO bulk single crystals by diffusion from a Zn3P2 source,” Appl. Phys. Lett. 87(22), 222113 (2005). [CrossRef]
27. Z. G. Yu, P. Wu, and H. Gong, “Control of p- and n-type conductivities in P doped ZnO thin films by using radio-frequency sputtering,” Appl. Phys. Lett. 88(13), 132114 (2006). [CrossRef]
28. Y. R. Ryu, S. Zhu, D. C. Look, J. M. Wrobel, H. M. Jeong, and H. W. White, “Synthesis of p-type ZnO films,” J. Cryst. Growth 216(1-4), 330–334 (2000). [CrossRef]
29. D. C. Look, G. M. Renlund, R. H. Burgener, and J. R. Sizelove, “As-doped p-type ZnO produced by an evaporation/sputtering process,” Appl. Phys. Lett. 85(22), 5269–5271 (2004). [CrossRef]
30. V. Vaithinathan, B. T. Lee, C. W. Chang, K. Asokan, and S. S. Kim, “Characterization of As-doped, p-type ZnO by x-ray absorption near-edge structure spectroscopy,” Appl. Phys. Lett. 88(11), 112103 (2006). [CrossRef]
31. J. C. Sun, J. Z. Zhao, H. W. Liang, J. M. Bian, L. Z. Hu, H. Q. Zhang, X. P. Liang, W. F. Liu, and G. T. Du, “Realization of ultraviolet electroluminescence from ZnO homojunction with n-ZnO/p-ZnO:As/GaAs structure,” Appl. Phys. Lett. 90(12), 121128 (2007). [CrossRef]
32. C. Yuen, S. F. Yu, E. S. P. Leong, S. P. Lau, K. Pita, H. Y. Yang, and T. P. Chen, “Room temperature deposition of p-type arsenic doped ZnO polycrystalline films by laser-assist filtered cathodic vacuum arc technique,” J. Appl. Phys. 101(9), 094905 (2007). [CrossRef]
33. L. J. Mandalapu, Z. Yang, F. X. Xiu, D. T. Zhao, and J. L. Liu, “Homojunction photodiodes based on Sb-doped p-type ZnO for ultraviolet detection,” Appl. Phys. Lett. 88(9), 092103 (2006). [CrossRef]
34. W. Guo, A. Allenic, Y. B. Chen, X. Q. Pan, Y. Che, Z. D. Hu, and B. Liu, “Microstructure and properties of epitaxial antimony-doped p-type ZnO films fabricated by pulsed laser deposition,” Appl. Phys. Lett. 90(24), 242108 (2007). [CrossRef]
35. H. S. Kang, B. D. Ahn, J. H. Kim, G. H. Kim, S. H. Lim, H. W. Chang, and S. Y. Lee, “Structural, electrical, and optical properties of p-type ZnO thin films with Ag dopant,” Appl. Phys. Lett. 88(20), 202108 (2006). [CrossRef]
36. F. X. Xiu, L. J. Mandalapu, Z. Yang, J. L. Li, G. F. Liu, and J. A. Yarmoff, “Bi-induced acceptor states in ZnO by molecular-beam epitaxy,” Appl. Phys. Lett. 89(5), 052103 (2006). [CrossRef]
37. A. Krtschil, A. Dadgar, N. Oleynik, J. Blasing, A. Diez, and A. Krost, “Local p-type conductivity in zinc oxide dual-doped with nitrogen and arsenic,” Appl. Phys. Lett. 87(26), 262105 (2005). [CrossRef]
38. J. G. Lu, Z. Z. Ye, F. Zhuge, Y. J. Zeng, B. H. Zhao, and L. P. Zhu, “p-type conduction in N–Al co-doped ZnO thin films,” Appl. Phys. Lett. 85(15), 3134–3135 (2004). [CrossRef]
39. J. G. Lu, Z. Z. Ye, G. D. Yuan, Y. J. Zeng, F. Zhuge, L. P. Zhu, B. H. Zhao, and S. B. Zhang, “Electrical characterization of ZnO-based homojunctions,” Appl. Phys. Lett. 89(5), 053501 (2006). [CrossRef]
40. L. Li, Z. Yang, J. Y. Kong, and J. L. Liu, “Blue electroluminescence from ZnO based heterojunction diodes with CdZnO active layers,” Appl. Phys. Lett. 95(23), 232117 (2009). [CrossRef]
41. K. Yamamoto, A. Nakamura, J. Temmyo, E. Muñoz, and A. Hierro, “Green electroluminescence from ZnCdO multiple quantum-well light-emitting diodes grown by remote-plasma-enhanced metal–organic chemical vapor deposition,” IEEE Photon Technol. Lett. 23(15), 1052–1054 (2011). [CrossRef]
42. J. W. Mares, M. Falanga, A. V. Thompson, A. Osinsky, J. Q. Xie, B. Hertog, A. Dabiran, P. P. Chow, S. Karpov, and W. V. Schoenfeld, “Hybrid CdZnO/GaN quantum-well light emitting diodes,” J. Appl. Phys. 104(9), 093107 (2008). [CrossRef]
43. S. Y. Ting, H. S. Chen, W. M. Chang, J. J. Huang, C. H. Liao, C. Y. Chen, C. Hsieh, Y. F. Yao, H. T. Chen, Y. W. Kiang, and C. C. Yang, “MBE-grown CdZnO/ZnO multiple quantum-well light-emitting diode on MOCVD-grown p-type GaN,” IEEE Photon. Technol. Lett. 24(11), 909–911 (2012). [CrossRef]
44. I. H. Ho and G. B. Stringfellow, “Solid phase immiscibility in GaInN,” Appl. Phys. Lett. 69(18), 2701–2703 (1996). [CrossRef]
45. Y. S. Lin, K. J. Ma, C. Hsu, S. W. Feng, Y. C. Cheng, C. C. Liao, C. C. Yang, C. C. Chuo, C. M. Lee, and J. I. Chyi, “Dependence of composition fluctuation on indium content in InGaN/GaN multiple quantum wells,” Appl. Phys. Lett. 77(19), 2988–2990 (2000). [CrossRef]
46. Y. C. Cheng, E. C. Lin, C. M. Wu, C. C. Yang, J. R. Yang, A. Rosenauer, K.-J. Ma, S.-C. Shi, L. C. Chen, C.-C. Pan, and J.-I. Chyi, “Nanostructures and carrier localization behaviors of green-luminescence InGaN/GaN quantum-well structures of various silicon-doping conditions,” Appl. Phys. Lett. 84(14), 2506–2508 (2004). [CrossRef]
47. S. Watanabe, N. Yamada, M. Nagashima, Y. Ueki, C. Sasaki, Y. Yamada, T. Taguchi, K. Tadatomo, H. Okagawa, and H. Kudo, “Internal quantum efficiency of highly-efficient InxGa1−xN-based near-ultraviolet light-emitting diodes,” Appl. Phys. Lett. 83(24), 4906–4908 (2003). [CrossRef]
48. A. Sasaki, S. Shibakawa, Y. Kawakami, K. Nishizuka, Y. Narukawa, and T. Mukai, “Equation for internal quantum efficiency and its temperature dependence of luminescence, and application to InxGa1-xN/GaN multiple quantum wells,” Jpn. J. Appl. Phys. 45(11), 8719–8723 (2006). [CrossRef]
49. S. F. Chichibu, A. C. Abare, M. S. Minsky, S. Keller, S. B. Fleischer, J. E. Bowers, E. Hu, U. K. Mishra, L. A. Coldren, S. P. DenBaars, and T. Sota, “Effective band gap inhomogeneity and piezoelectric field in InGaN/GaN multiquantum well structures,” Appl. Phys. Lett. 73(14), 2006–2008 (1998). [CrossRef]
50. T. Takeuchi, C. Wetzel, S. Yamaguchi, H. Sakai, H. Amano, I. Akasaki, Y. Kaneko, S. Nakagawa, Y. Yamaoka, and N. Yamada, “Determination of piezoelectric fields in strained GaInN quantum wells using the quantum-confined Stark effect,” Appl. Phys. Lett. 73(12), 1691–1693 (1998). [CrossRef]
51. C. F. Huang, C. Y. Chen, C. F. Lu, and C. C. Yang, “Reduced injection current induced blueshift in an InGaN/GaN quantum well light-emitting diode of prestrained growth,” Appl. Phys. Lett. 91(5), 051121 (2007). [CrossRef]
