- Email: [email protected]
Structural, electrical, and optical properties of CoxNi1-xO films grown by metalorganic chemical vapor deposition
Journal of Crystal Growth 414 (2015) 123–129
Contents lists available at ScienceDirect
Journal of Crystal Growth journal homepage: www.elsevier.com/locate/jcrysgro
Structural, electrical, and optical properties of CoxNi1-xO films grown by metalorganic chemical vapor deposition Teuku Muhammad Roffi n, Kazuo Uchida, Shinji Nozaki Department of Engineering Science, Graduate School of Informatics and Engineering, The University of Electro-Communications, 1–5-1 Chofugaoka, Chofu, Tokyo 182–8585, Japan
art ic l e i nf o
a b s t r a c t
Article history: Received 12 July 2014 Received in revised form 11 October 2014 Accepted 13 October 2014 Communicated by: T.F. Kuech Available online 22 October 2014
Thin films of cobalt-nickel oxide (CoxNi1-xO, x¼ 0.01, 0.02, 0.08, 0.17, 0.22, 0.35, 0.56, 0.72) were grown on Al2O3 substrate by atmospheric pressure metalorganic chemical vapor deposition (APMOCVD). The effect of the cobalt composition on the structural, morphological, optical, and electrical properties of the films was investigated. X-ray diffraction (XRD) analysis revealed that all of the films grew with a preferred orientation towards [1 1 1]NiO and a twinned structure. Cobalt was well dispersed in the NiO structure up to x¼ 0.08. CoxNi1-xO alloys were formed from x¼0.17 to x ¼0.22, while phase-separated NiO and CoxNi1 xO formed when xZ0.35. The bandgap of the CoxNi1 xO film was found to decrease with increasing cobalt composition. Four-point probe measurements showed that the resistivity of the film also decreased with increasing cobalt composition, reaching a minimum of 0.006 Ωcm. Hall measurements of the films revealed n-type conductivity. The correlation between the presence of cobalt in different ionization states and the observed decrease in resistivity as well as the type of conductivity in CoxNi1 xO is discussed. & 2015 Elsevier B.V. All rights reserved.
Keywords: Doping X-ray diffraction Metalorganic chemical vapor deposition Metals Oxides Semiconducting materials
1. Introduction Nickel oxide has drawn interest for its unique combination of optical transparency, electrical conductivity, and wide energy gap of 3.6–4 eV. These properties make NiO promising for use in transparent conducting oxides, UV LEDs, and sensors [1,2,3]. Although stoichiometric nickel oxide is known to be an insulator with resistivity around 1011–1013 Ωcm [4], nickel oxide normally contains native defects such as an excess of oxygen (NiOx) that results in p-type conductivity and allows the generation of uncompensated holes [5]. Attempts to improve the p-type conductivity of NiO, especially by lithium doping, have been reported in the literature [6,7]. Substitution of nickel by monovalent atoms such as lithium enables the generation of an excess of uncompensated holes, thereby improving the p-type conductivity. Other methods of modifying the conductivity of NiO have also been reported, such as the post-deposition annealing of sputter grown NiO [8] and ultraviolet laser irradiation of PLD grown NiO [9]. However, doping to produce n-type NiO is not yet well studied. In this work we have attempted to produce n-type conductivity in NiO by incorporating cobalt to form CoxNi1 xO.
n
Corresponding author. E-mail address: tmroffi@uec.ac.jp (T.M. Roffi).
http://dx.doi.org/10.1016/j.jcrysgro.2014.10.027 0022-0248/& 2015 Elsevier B.V. All rights reserved.
The effect of cobalt incorporation on the structural, electrical, and optical properties of CoxNi1-xO films is of significant interest. The possibility of engineering the bandgap of CoxNi1 xO films is also important, especially for the fabrication of devices with selective response in the UV region. CoxNi1-xO films with cobalt composition ranging from x¼ 0.01 to x¼ 0.72 were grown on Al2O3 substrates at 500 1C in a horizontal atmospheric pressure metalorganic chemical vapor deposition (APMOCVD) reactor. The cobalt compositions of the obtained films were determined by energy dispersive X-ray spectroscopy (EDS). X-ray diffraction analysis (XRD) in θ-2θ and ϕ scan modes and Raman spectroscopy were employed to determine the crystal structure of the films. Scanning electron microscopy (SEM) was used to analyze the surface morphology of the films. Bandgap energies were approximated from optical transmittance data measured by UV–vis-NIR transmittance spectroscopy. The resistivity of the films was measured using the four point probe method, while their mobility and carrier concentration were investigated using Hall effect measurements.
2. Experimental details Based on the previous successful epitaxial growth of NiO in our group, single crystalline (0 0 1)Al2O3 substrates were used in this work. Prior to growth, the substrates were degreased in an ultrasonic bath using acetone, ethanol, and deionized water for
124
T.M. Roffi et al. / Journal of Crystal Growth 414 (2015) 123–129
10 min each. CoxNi1 xO films were grown on the substrates by APMOCVD at 500 1C. The APMOCVD system comprised a horizontal reactor designed for the growth of oxide thin films. Allyl (cyclopentadienyl) nickel (C8H10Ni) was used as the nickel precursor and was kept at 30 1C to obtain a vapor pressure of 0.84 Torr. Bis(methylcyclopentadienyl) cobalt (Co(CH3C5H4)2) was used as cobalt precursor and was kept at 500 1C to obtain a vapor pressure of 0.3 Torr. Pure O2 gas was used as an oxygen source and N2 was used as the carrier gas. The bubbling flow rates of each metal precursor were adjusted to obtain the desired cobalt composition. Growth times were adjusted to maintain the same total amount of precursors transported throughout all of the reactions. The resulting samples were CoxNi1 xO with x of 0, 0.01, 0.02, 0.08, 0.17, 0.22, 0.35, 0.56, 0.72, and 1, as determined by EDS analysis. The thicknesses of the films varied from around 500 nm to 1300 nm as observed by field emission scanning electron microscopy (FESEM). The obtained samples were subjected to various analyses, including structural, optical, and electrical. The crystal structure and crystallographic orientation relationship between the CoxNi1 xO film and the Al2O3 substrate were analyzed by XRD in θ-2θ, rocking curve, and ϕ configurations with Cu Kα radiation (λ ¼1.54 Å). Further structure and compositional analyses were performed by Raman spectroscopy. The optical transmittance of the films was measured using UV–vis-NIR spectroscopy. The resistivity of the samples was mainly measured by the four-point probe method and further confirmed with van der Pauw configuration Hall effect measurements. For Hall measurements, 500nm-thick metal contacts consisting of Pt (100 nm), Ti (300 nm), and Au (100 nm) and having an area of 1 mm2 were deposited on each corner of the square sample films. The conductivity type, carrier concentration, and mobility of the films were then investigated. All measurements were performed at room temperature.
direction and an out-of-plane epitaxial relationship between the NiO film and the Al2O3 substrate, [1 1 1]NiO||[0 0 6]Al2O3. Next, ϕ scans were carried out to determine the in-plane crystallographic relationship between the substrate and the film. First, a ϕ scan was performed on the NiO sample at a tilt of 54.731 from the (111)NiO plane with Bragg condition θ and 2θ of 21.641 and 43.281, respectively, satisfying the orientation of the (2 0 0)NiO plane. Then, another ϕ scan was performed at a tilt of 42.311 from the (0 0 6)Al2O3 plane with Bragg condition θ and 2θ of 25.291 and 57.501, respectively, satisfying the orientation of the (1 1 6)Al2O3 plane. As shown in Fig. 2, six (200)NiO plane peaks appeared in the scan result at the same ϕ angle as the six peaks of (116)Al2O3. In cubic NiO, the (200)NiO plane has three-fold rotational symmetry around the [1 1 1]NiO axis. The existence of six (200)NiO plane peaks indicates that the epitaxial NiO film had two different crystallographic orientations (twinned) along the growth direction [1 1 1]NiO rotated by 601. A detailed microstructural study of such a twinned structure in epitaxially grown NiO on Al2O3 has been reported elsewhere [10]. The ϕ scan also showed that both the (200)NiO plane and (1 1 6)Al2O3 plane had the same rotational angle
3. Results and discussion 3.1. Structure and crystal phase XRD analyses in θ-2θ scanning mode were conducted to determine the structure and phase of the samples. To distinguish the peaks of the substrate from those of NiO, the XRD patterns of the Al2O3 substrate and x¼0 (NiO) film are presented in Fig. 1. The peaks at 411 and 901 belong to (0 0 6)Al2O3 (ICDD #46–1212), while the peaks at 371 and 791 belong to (1 1 1)NiO and (2 2 2)NiO (ICDD #47–1049), respectively. This indicates epitaxial growth of NiO towards the [1 1 1]NiO
Fig. 1. θ-2θ XRD scans of Al2O3 substrate and NiO film. The peaks at 411 and 901 belong to (006)Al2O3 while those at 371 and 791 belong to (1 1 1)NiO.
Fig. 2. XRD ϕ scan of the NiO sample tilted 54.731 from the (111)NiO plane, satisfying the Bragg condition for the (200)NiO plane, and tilted 42.311 from the (0 0 6)Al2O3 plane, satisfying the Bragg condition for (1 1 6)Al2O3.
Fig. 3. XRD θ-2θ scans of CoxNi1 xO films. Samples with x ¼0 to x ¼0.72 showed peaks at 371 and 791, which weakened with increasing cobalt composition. The additional peaks at about 381, 821 and the series of peaks at 181, 591, and 1101 observed at higher cobalt compositions (xZ 0.35) are indicative of phase separation.
T.M. Roffi et al. / Journal of Crystal Growth 414 (2015) 123–129
125
Fig. 4. XRD ϕ scans of the CoxNi1 xO samples of x ¼0.08 (a), x ¼ 0.22 (b), and x ¼0.35 (c) tilted 54.731 from the (1 1 1)NiO plane, satisfying the Bragg condition for the (2 0 0)NiO plane. All samples exhibited six-fold symmetry, indicative of a twinned structure oriented along the growth direction.
around the [1 1 1]NiO axis. Therefore, the in-plane relationship can be established as [112]NiO||[110]Al2O3 and [211]Twinned-NiO|| [1 1 0]Al2O3. Unequal intensities of the peaks in the ϕ scan were also found in a previous report [9]. This was probably a result of lattice mismatch causing variation in lattice relaxation in the lateral direction. The presence of defect such as twin phase in the CoxNi1 xO films may also cause a higher variation of intensity compared to the sapphire substrate (Fig. 4). Fig. 3 shows that all the samples containing nickel (x ¼0 to x ¼0.72) exhibited peaks at 371 and 791, indicating the existence of cubic phase NiO oriented towards [1 1 1]NiO. The films also exhibit twinned structure oriented along the growth direction as indicated by the presence of six-fold symmetry in XRD ϕ scans (Fig. 4). The NiO structure remained dominant in the samples with x ¼0 to x ¼0.08, as indicated by the narrow FWHM in Fig. 5 and strong peaks. However, the structure of NiO was weakened at higher cobalt incorporation in the film, as indicated by peak broadening and intensity decrease. The effect of cobalt became stronger in films with x ¼ 0.17 and x¼ 0.22, the patterns of which exhibited a shift towards higher angle as well as FWHM broadening of the (1 1 1)NiO peak. The existence of additional 2θ peaks at about 381 and 821 and the series of 2θ peaks at 181, 591, and 1101 observed for the samples with higher cobalt composition (x Z0.35) is indicative of phase separation. These peaks were attributed to either NiCo2O4 (ICDD #20–781) or Co3O4 (ICDD #42–1467). Further phase characterization could not be accomplished through XRD analysis owing to the similarity between the structural parameters of NiCo2O4 and Co3O4. Hence, Raman scattering analysis was carried out to complement the XRD data, as described later. High resolution ω-2θ scans normalized to the Al2O3 peaks were carried out to investigate any peak shift caused by cobalt incorporation. As shown in Fig. 6, the peak positions of the CoxNi1-xO samples slightly shifted to lower values for cobalt compositions between x¼ 0 and x¼0.08, then shifted to higher values for higher cobalt compositions. The reason for this is as follows. The observed peak positions of the x¼0 (NiO) sample were higher than those of the ICDD data, indicating a smaller interplanar spacing than that expected for stoichiometric NiO. It is well known that as-deposited NiO may possess intrinsic defects comprising excess oxygen (nickel vacancies) allowing for p-type conductivity. Cobalt may have been incorporated in these nickel-vacant sites, resulting in a larger unit cell volume compared with that of as-deposited NiO that appeared as a slight peak shift towards lower angle in the XRD pattern. Both divalent and trivalent cobalt may exist in cases of nickel substitution [11,12], and this may have caused the observed slight shift towards higher angle owing to their smaller ionic radii (0.69Å, 0.65Å, and 0.545Å for Ni2þ , Co2 þ , and Co3 þ , respectively [13]). Films grown without any nickel exhibited peaks matching those of Co3O4 (ICDD #42–1467).
Fig. 5. FWHM of (1 1 1)NiO obtained from rocking curve scan data. The FWHM of (1 1 1)NiO became broader as cobalt composition increased. Inset shows ω scan result for the x ¼ 0 (NiO) sample, showing a FWHM of 0.1121.
Fig. 6. High resolution ω-2θ scan normalized to the Al2O3 peaks. The NiO peak position slightly shifted to lower value for x ¼ 0 to x ¼ 0.08, then shifted to higher value for higher cobalt compositions. The inset presents the trend in peak position with cobalt composition.
Raman scattering spectra of the samples are shown in Fig. 7. The lowermost plot is the Raman spectra of Al2O3, which exhibits peaks at 400, 560, and 740 cm 1. The peaks overlapping these Al2O3 peaks
126
T.M. Roffi et al. / Journal of Crystal Growth 414 (2015) 123–129
were not considered in further analysis. The peaks observed for the x¼0 (NiO) film at about 560, 925, 1100, and 1500 cm 1 match those of the literature [14] representing the LO, TOþLO, 2LO, and 2M modes. The TO and 2TO modes expected at about 440 and 740 cm 1 could not be observed owing to the presence of overlapping Al2O3 peaks. At low cobalt composition (x ¼0.01), no significant change in Raman spectra was observable. At x ¼0.02, two effects could be observed. First, the magnon excitation mode of NiO (2M) weakened and shifted towards lower wavenumbers. Second, additional Raman peaks appeared at about 520 and 620 cm 1, which could be addressed to CoO phase[15]. The existence of Co-O bonding is expected when cobalt substitutes into vacant nickel sites or substitutes nickel in the lattice, in agreement with the XRD analysis. Thus, although the coexistence of NiO and CoO structures could not be determined from the XRD data, their different vibrational modes made it possible to observe them by Raman spectroscopy, as explained above. The Raman spectra of samples with x ¼0.17 to x¼ 0.72 were identical to those reported in the literature for nickel-cobalt oxide series [16,17], with two major Raman peaks at about 465 and 655 cm 1 and a minor one at about 551 cm 1. The peak at about 550 cm 1 in our samples, however, appeared to be sensitive towards the cobalt composition. The peak at about 650 cm 1 has previously been attributed to a vibrational mode involving oxygen atom in octahedral sites in the spinel lattice [18]. Thus, it is suggested that the slight shift of this peak towards higher frequency in samples of higher cobalt composition was caused by Co3 þ substitution of Ni2 þ in the octahedral sites of the spinel structure. Co3 þ substitution of Ni2 þ at spinel octahedral sites would cause structural distortion owing to the smaller ionic radius of Co3 þ and shorter bond length, thus causing a shift of the Raman band to higher frequency. With the XRD results, it can be confirmed that the phases present in this sample were NiO and the series of nickel-cobalt oxides (NiCo2O4), not Co3O4. Consistent with the XRD results, the Raman peaks of samples grown without any nickel, at 480, 519, 617, and 689 cm 1, match those reported for Co3O4 in the literature [19,20]. As summarized in Table 1, cobalt was incorporated in the NiO lattice for CoxNi1 xO samples with xr0.08. Formation of CoxNi1 xO alloy occurred at x¼ 0.17 and x¼ 0.22. Attempts to incorporate a greater proportion of cobalt resulted in phase separation into NiO and NiCo2O4 phases, as evidenced for the samples with
Fig. 7. Raman Scattering of CoxNi1 xO films. The spectrum obtained for the x ¼0.02 sample was assigned to NiO and CoO, while the spectra of higher cobalt compositions (x¼ 0.17 to x¼ 0.72) were attributed to nickel-cobalt oxides.
Table 1 Observed phases in relation to cobalt composition of the films. Cobalt Composition x
0 0.01 0.02 0.08 0.17 0.22 0.35 0.56 0.72
Phases NiO
CoO
v v v v
v v v
CoxNi1-xO
NiCo2O4
v v v v v
v v v
x¼ 0.35–0.72. The NiCo2O4 phase became dominant when x¼ 0.72, while samples grown without nickel were found to be Co3O4. 3.2. Appearance and morphology The x ¼0 (NiO) film was optically clear and grayish green in color. The films darkened with increasing cobalt composition up to x¼ 0.22; beyond this, the films were black. The film grown without any nickel (Co3O4) was slightly lighter in color, dark brown, and was optically clear. The surface morphology of the CoxNi1-xO films differed depending on their cobalt composition, as observed from the SEM images presented in Fig. 8. Coalesced triangular pyramidal grains with poorly defined boundaries were found in the x¼0 (NiO) and x¼0.08 films. This result supports the structural analysis by XRD and Raman spectroscopy, which found that the NiO phase remains with cobalt incorporation at low x. The films with higher cobalt composition exhibited segregated grains with more distinct boundaries. Segregation of the grains was most likely introduced by the phase separation of NiO and NiCo2O4. It was also evident that some of the pyramids were rotated by 601, indicative of twinned (111)NiO in the films. 3.3. Transmittance and bandgap energy approximation UV–vis-NIR transmittance spectroscopy measurements were performed to investigate the optical characteristics and approximate the bandgap of the films. The optical transmittance spectra of the CoxNi1 xO samples in the wavelength range 200–800 nm at room temperature are given in Fig. 9. As shown in the inset, the x¼ 0 (NiO) film showed high transparency in the visible region, indicated by an 88% average transmittance for the 800-nm-thick film. Although the transmittance decreased with increasing cobalt composition, it was maintained at a relatively high percentage up to x¼0.17, i.e., 71% average transmittance for a 700-nm-thick film. These transmittance results agree well with the XRD analysis of the films; incorporation of cobalt at this level introduced defect sites, indicated by the slight change in structural parameters, which scatter incident light. More significant deterioration in the transparency of NiO films was observed in the case of lithium doping [21] owing to its larger difference in ionic radius. Beyond x¼0.17, transmittance in the visible region dropped as a result of enhanced alloy formation. A shift of fundamental absorption edge towards lower wavelength (redshift) was observed with increasing cobalt composition, which is related to a decrease in optical bandgap. The optical bandgaps of the films were approximated using the classical relation for near edge optical absorption in semiconductors. αhν2 ¼ A hν Eg ; ð1Þ where A is a constant, α is the absorption coefficient, and hν and Eg represent the incident photon energy and the semiconductor bandgap energy, respectively. Plots of (αhν)2 against (hν) for the
T.M. Roffi et al. / Journal of Crystal Growth 414 (2015) 123–129
127
Fig. 8. SEM micrographs of CoxNi1-xO samples taken at 10 000 magnification. Coalesced triangular pyramidal grains with poorly defined boundaries were found in x¼ 0 (NiO) and x ¼ 0.08 films, while films with higher cobalt composition exhibited segregated grains with more distinct boundaries. Insets show micrographs taken at 20 000x magnification. All white bars represent 1μm length.
CoxNi1 xO films are presented in inset of Fig. 10. The bandgap decreased from 3.7 eV to 2.35 eV with increasing cobalt composition. The relation between cobalt composition and optical band gap can be approximated by Vegard's law: Eg;
Cox Ni1 x O
¼ x Eg;
Co3 O4
þ ð1 xÞEg; NiO b x ð1 xÞ;
ð2Þ
where Eg is the bandgap of the respective materials and b is the bowing parameter, a measure of the non-linearity of the bandgap as a function of composition. Although phase separation occurred for cobalt composition x¼ 0.35, 0.56, and 0.72, which makes the use of Vegard's law unusual in this case, the approximation is well suited and useful for further compositional design of CoxNi1−xO,
especially by the current growth method. The optical bandgap and cobalt composition was well fitted by a bowing parameter of 3.5, as shown in Fig. 10. It should be noted that the bandgaps of x¼ 0.56 and x ¼0.72 films were not shown because of undefined absorption edges due to low transmittance. However, the approximated bandgap for x¼ 0.35 where phase separation occurs could be determined and was found to follow Vegard's law. The incorporation of cobalt into NiO to form CoxNi1 xO was therefore able to decrease the bandgap of NiO considerably. This result indicates that selective UV-region response materials based on NiO can be manufactured using the present APMOCVD system by controlling the cobalt composition.
128
T.M. Roffi et al. / Journal of Crystal Growth 414 (2015) 123–129
Fig. 11. Change in resistivity of the films with cobalt composition. The resistivity decreased from 960 to 0.006 Ωcm with increased cobalt composition from x¼ 0 to x¼ 0.72.
Fig. 9. Transmittance spectra of the CoxNi1-xO samples. The highest transmittance of 88% was obtained for the 800-nm-thick x ¼0 (NiO) film. The incorporation of cobalt decreased the transmittance and caused red shift (towards higher wavelength) of the fundamental absorption edge.
annealing, ultraviolet irradiation, and varied oxygen pressure [8,9,22], which cause oxygen deficiency. The defect reaction can be expressed as follows, assuming complete ionization. ' 1 OX O ⇆vO þ 2e þ O2 ðgÞ 2
ð4Þ
In the case of cobalt incorporation into NiO films, it is reported that cobalt ions exist in the nickel lattice as Co2 þ and Co3 þ [11, 12]. While the substitution of homovalent cobalt (Co2 þ ) for nickel in the lattice may not result in any carrier generation, incorporation of aliovalent Co3 þ may generate carriers according to the following Kröger–Vink notation. 1 O2 ðg Þ þ Co2 O3 ⇆ 2CoNi þ 4OX0 þ 2e' 2
Fig. 10. Approximated bandgap energy as a function of cobalt composition follows Vegard's law with bowing parameter 3.5. Data points for x ¼ 0.56 and x ¼0.72 are not available since these films are not transparent. Inset shows the plot of (αhν)2 to (hν).
3.4. Electrical properties Four-point-probe measurements were conducted at room temperature to investigate the resistivity of the samples as a function of cobalt composition (Fig. 11). The resistivity of the films was found to decrease from 960 Ωcm to 0.006 Ωcm with increasing cobalt composition from x ¼0 to x ¼0.72. However, the resistivity did not change much from x ¼0 to x ¼0.08. The effect of the cobalt composition on the resistivity can be explained by the carrier generation process in CoxNi1 xO. Although stoichiometric NiO is an insulator, it is well known that defects play major role in the conductivity of NiO. As-deposited NiO may contain native defects of excess oxygen which create cation deficiencies on NiO lattice sites that cause the generation of holes, as expressed by the following Kröger–Vink notation. 1 þ O2 ðg Þ ⇆ vNi '' þOX o þ 2h 2
ð3Þ
As expressed in Eq. (3), two holes are generated to maintain charge neutrality. Consistently, defects in NiO may also reverse its conductivity type. It is reported that NiO may obtain n-type conductivity by certain treatments such as post deposition
ð5Þ
As expressed in Eq. (5), two electrons are generated for every two trivalent cobalt ions substituting nickel in the lattice. Therefore, the decreasing resistivity observed with increasing cobalt composition is caused by increased carrier generation in CoxNi1 xO. This is well in agreement with the Hall measurement results (inset of Fig. 11) of the CoxNi1 xO samples with x ¼0.35, 0.56, and 0.72, which exhibit n-type conductivity. It is surmised that trivalent cobalt in CoxNi1 xO alloy behaved as a donor. We note that caution is needed in interpreting the Hall results owing to possible magnetic influence of the sample. It has been reported that although Co3O4 and NiO are p-type semiconductors, the conductivity of Ni1 xCoxO4/3 was not clear, with Hall measurements showing both p- and n-type conductivity in the same sample [23]. Indeed, in our Hall measurements only three samples exhibited consistent n-type conductivity, while the conductivity type of the rest of the samples could not be determined.
4. Conclusions Successful growth of CoxNi1 xO with a wide cobalt composition range of x¼ 0.01 to x¼ 0.72 has been achieved using APMOCVD. The properties of the films were found to vary with cobalt composition (x). All the films grown had a preferred orientation towards [1 1 1]NiO with a twinned structure. Cobalt was well incorporated in the NiO structure up to x¼ 0.08, CoxNi1 xO alloys were formed from x ¼0.17 to x ¼0.22, and phase separation of NiO and CoxNi1 xO occurred when x Z0.35. Films grown without any nickel were found to be composed of Co3O4. The bandgap of the CoxNi1 xO films decreased with increasing cobalt composition, and controllable bandgap engineering in the UV region was demonstrated. The resistivity of the films also decreased with increasing cobalt composition, reaching a minimum of 0.006 Ωcm. Some of the samples exhibited n-type conductivity, determined by Hall measurements. The existence of
T.M. Roffi et al. / Journal of Crystal Growth 414 (2015) 123–129
trivalent cobalt as a donor in the NiO structure is considered to be the cause of the n-type conductivity observed in the films. In the future, it would be of interest to investigate the thickness dependence of the films properties for each composition, because it may be relevant for device applications. Acknowledgment T. M. Roffi gratefully acknowledges a scholarship for his Ph.D. study from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan. References [1] H. Sato, T. Minami, S. Takata,, T. Yamada, Transparent conducting p-type NiO thin films prepared by magnetron sputtering, Thin Solid Films 236 (1) (1993) 27–31. [2] H. Ohta, M. Hirano, K. Nakahara, H. Maruta, T. Tanabe, M. Kamiya, T. Kamiya,, H. Hosono, Fabrication and photoresponse of a pn-heterojunction diode composed of transparent oxide semiconductors, p-NiO and n-ZnO, Appl. Phys. Lett. 83 (5) (2003) 1029–1031. [3] M. Tyagi, M. Tomar,, V. Gupta, Influence of hole mobility on the response characteristics of p-type nickel oxide thin film based glucose biosensor, Anal. Chim. Acta. 726 (2012) 93–101. [4] T. Dutta, P. Gupta, A. Gupta,, J. Narayan, High work function (p-type NiO1 þ x)/ Zn0. 95Ga0. 05O heterostructures for transparent conducting oxides, J. Phys. D.: Appl. Phys. 43 (10) (2010) 105301. [5] H. Rooksby,, M. Vernon, Lithium-and gallium-doped nickel oxide, Brit. J. Appl. Phys. 17 (9) (1966) 1227. [6] T. Dutta, P. Gupta, A. Gupta,, J. Narayan, Effect of Li doping in NiO thin films on its transparent and conducting properties and its application in heteroepitaxial pn junctions, J. Appl. Phys. 108 (8) (2010) (083715–083715). [7] W.-L. Jang, Y.-M. Lu, W.-S. Hwang,, W.-C. Chen, Electrical properties of Lidoped NiO films, J. Eur. Ceram. Soc. 30 (2) (2010) 503–508. [8] M. Tyagi, M. Tomar,, V. Gupta, Postdeposition annealing of NiOx thin films: A transition from n-type to p-type conductivity for short wave length optoelectronic devices, J. Mater. Res. 28 (2013) 723–732.
129
[9] P. Gupta, T. Dutta, S. Mal,, J. Narayan, Controlled p-type to n-type conductivity transformation in NiO thin films by ultraviolet-laser irradiation, J. Appl. Phys. 111 (1) (2012) (013706–013706). [10] K. Uchida, K. Yoshida, D. Zhang, A. Koizumi, S. Nozaki, High-quality single crystalline NiO with twin phases grown on sapphire substrate by metalorganic vapor phase epitaxy AIP ADVANCES, 2 042154 (2012), http://dx.doi.org/10. 1063/1.4769082. [11] G. Natu, P. Hasin, Z. Huang, Z. Ji, M. He,, Y. Wu, Valence band-edge engineering of Nickel Oxide Nanoparticles via Cobalt doping for application in p-type dyesensitized solar cells, ACS. Appl. Mater. Inter. 4 (11) (2012) 5922–5929. [12] Y. Gou, X. Liang,, B. Chen, Porous Ni–Co bimetal oxides nanosheets and catalytic properties for CO oxidation, J. Alloy. Compd. 574 (2013) 181–187. [13] J.A. Dean, Lange's Chemistry Handbook, vol. 4, McGraw-Hill, Inc, 1999. [14] N. Mironova-Ulmane, A. Kuzmin, I. Sildos,, M. Pärs, Polarisation dependent Raman study of single-crystal nickel oxide, Cent. Eur. J. Phys. 9 (4) (2011) 1096–1099. [15] H.C. Choi, Y.M. Jung, I. Noda,, S.B. Kim, A study of the mechanism of the electrochemical reaction of lithium with CoO by two-dimensional soft X-ray absorption spectroscopy (2D XAS), 2D Raman, and 2D heterospectral XASRaman correlation analysis, J. Phys. Chem. B. 107 (24) (2003) 5806–5811. [16] G. Calin, M. Irimia, C. Scarlat, M. Purica, F. Comanescu, F. Iacomi, Synthesis and characterization of nickel cobalt oxide thin films, in: Semiconductor Conference (CAS), 2010 International, 2010. [17] C.F. Windisch Jr, G.J. Exarhos,, S.K. Sharma, Influence of temperature and electronic disorder on the Raman spectra of nickel cobalt oxides, J. Appl. Phys. 92 (9) (2002) 5572–5574. [18] C.F. Windisch Jr, K.F. Ferris, G.J. Exarhos,, S.K. Sharma, Conducting spinel oxide films with infrared transparency, Thin Solid Film. 420 (2002) 89–99. [19] H. Zhou, L. Chen, V. Malik, C. Knies, D.M. Hofmann, K.P. Bhatti, S. Chaudhary, P. J. Klar, W. Heimbrodt, C. Klingshirn (others), Raman studies of ZnO: Co thin films, 204, Phys. Status Solidi A (2007) 112–117. [20] V. Hadjiev, M. Iliev,, I. Vergilov, The raman spectra of Co3O4, J. Phys. C.: Solid. State. Phys. 21 (7) (1988) L199. [21] Y.H. Kwon, S.H. Chun,, H.K. Cho, Semiconducting p-type MgNiO: Li epitaxial films fabricated by cosputtering method, J. Vacuum. Sci. Technol. A. 31 (4) (2013) 041501. [22] R. Molaei, R. Bayati,, J. Narayan, Crystallographic Characteristics and p-Type to n-Type Transition in Epitaxial NiO Thin Film, Cryst. Growth Des. 13 (12) (2013) 5459–5465. [23] C. Windisch Jr, G.J. Exarhos, K.F. Ferris, M.H. Engelhard,, D.C. Stewart, Infrared transparent spinel films with p-type conductivity, Thin Solid Film. 398 (2001) 45–52.
Contents lists available at ScienceDirect
Journal of Crystal Growth journal homepage: www.elsevier.com/locate/jcrysgro
Structural, electrical, and optical properties of CoxNi1-xO films grown by metalorganic chemical vapor deposition Teuku Muhammad Roffi n, Kazuo Uchida, Shinji Nozaki Department of Engineering Science, Graduate School of Informatics and Engineering, The University of Electro-Communications, 1–5-1 Chofugaoka, Chofu, Tokyo 182–8585, Japan
art ic l e i nf o
a b s t r a c t
Article history: Received 12 July 2014 Received in revised form 11 October 2014 Accepted 13 October 2014 Communicated by: T.F. Kuech Available online 22 October 2014
Thin films of cobalt-nickel oxide (CoxNi1-xO, x¼ 0.01, 0.02, 0.08, 0.17, 0.22, 0.35, 0.56, 0.72) were grown on Al2O3 substrate by atmospheric pressure metalorganic chemical vapor deposition (APMOCVD). The effect of the cobalt composition on the structural, morphological, optical, and electrical properties of the films was investigated. X-ray diffraction (XRD) analysis revealed that all of the films grew with a preferred orientation towards [1 1 1]NiO and a twinned structure. Cobalt was well dispersed in the NiO structure up to x¼ 0.08. CoxNi1-xO alloys were formed from x¼0.17 to x ¼0.22, while phase-separated NiO and CoxNi1 xO formed when xZ0.35. The bandgap of the CoxNi1 xO film was found to decrease with increasing cobalt composition. Four-point probe measurements showed that the resistivity of the film also decreased with increasing cobalt composition, reaching a minimum of 0.006 Ωcm. Hall measurements of the films revealed n-type conductivity. The correlation between the presence of cobalt in different ionization states and the observed decrease in resistivity as well as the type of conductivity in CoxNi1 xO is discussed. & 2015 Elsevier B.V. All rights reserved.
Keywords: Doping X-ray diffraction Metalorganic chemical vapor deposition Metals Oxides Semiconducting materials
1. Introduction Nickel oxide has drawn interest for its unique combination of optical transparency, electrical conductivity, and wide energy gap of 3.6–4 eV. These properties make NiO promising for use in transparent conducting oxides, UV LEDs, and sensors [1,2,3]. Although stoichiometric nickel oxide is known to be an insulator with resistivity around 1011–1013 Ωcm [4], nickel oxide normally contains native defects such as an excess of oxygen (NiOx) that results in p-type conductivity and allows the generation of uncompensated holes [5]. Attempts to improve the p-type conductivity of NiO, especially by lithium doping, have been reported in the literature [6,7]. Substitution of nickel by monovalent atoms such as lithium enables the generation of an excess of uncompensated holes, thereby improving the p-type conductivity. Other methods of modifying the conductivity of NiO have also been reported, such as the post-deposition annealing of sputter grown NiO [8] and ultraviolet laser irradiation of PLD grown NiO [9]. However, doping to produce n-type NiO is not yet well studied. In this work we have attempted to produce n-type conductivity in NiO by incorporating cobalt to form CoxNi1 xO.
n
Corresponding author. E-mail address: tmroffi@uec.ac.jp (T.M. Roffi).
http://dx.doi.org/10.1016/j.jcrysgro.2014.10.027 0022-0248/& 2015 Elsevier B.V. All rights reserved.
The effect of cobalt incorporation on the structural, electrical, and optical properties of CoxNi1-xO films is of significant interest. The possibility of engineering the bandgap of CoxNi1 xO films is also important, especially for the fabrication of devices with selective response in the UV region. CoxNi1-xO films with cobalt composition ranging from x¼ 0.01 to x¼ 0.72 were grown on Al2O3 substrates at 500 1C in a horizontal atmospheric pressure metalorganic chemical vapor deposition (APMOCVD) reactor. The cobalt compositions of the obtained films were determined by energy dispersive X-ray spectroscopy (EDS). X-ray diffraction analysis (XRD) in θ-2θ and ϕ scan modes and Raman spectroscopy were employed to determine the crystal structure of the films. Scanning electron microscopy (SEM) was used to analyze the surface morphology of the films. Bandgap energies were approximated from optical transmittance data measured by UV–vis-NIR transmittance spectroscopy. The resistivity of the films was measured using the four point probe method, while their mobility and carrier concentration were investigated using Hall effect measurements.
2. Experimental details Based on the previous successful epitaxial growth of NiO in our group, single crystalline (0 0 1)Al2O3 substrates were used in this work. Prior to growth, the substrates were degreased in an ultrasonic bath using acetone, ethanol, and deionized water for
124
T.M. Roffi et al. / Journal of Crystal Growth 414 (2015) 123–129
10 min each. CoxNi1 xO films were grown on the substrates by APMOCVD at 500 1C. The APMOCVD system comprised a horizontal reactor designed for the growth of oxide thin films. Allyl (cyclopentadienyl) nickel (C8H10Ni) was used as the nickel precursor and was kept at 30 1C to obtain a vapor pressure of 0.84 Torr. Bis(methylcyclopentadienyl) cobalt (Co(CH3C5H4)2) was used as cobalt precursor and was kept at 500 1C to obtain a vapor pressure of 0.3 Torr. Pure O2 gas was used as an oxygen source and N2 was used as the carrier gas. The bubbling flow rates of each metal precursor were adjusted to obtain the desired cobalt composition. Growth times were adjusted to maintain the same total amount of precursors transported throughout all of the reactions. The resulting samples were CoxNi1 xO with x of 0, 0.01, 0.02, 0.08, 0.17, 0.22, 0.35, 0.56, 0.72, and 1, as determined by EDS analysis. The thicknesses of the films varied from around 500 nm to 1300 nm as observed by field emission scanning electron microscopy (FESEM). The obtained samples were subjected to various analyses, including structural, optical, and electrical. The crystal structure and crystallographic orientation relationship between the CoxNi1 xO film and the Al2O3 substrate were analyzed by XRD in θ-2θ, rocking curve, and ϕ configurations with Cu Kα radiation (λ ¼1.54 Å). Further structure and compositional analyses were performed by Raman spectroscopy. The optical transmittance of the films was measured using UV–vis-NIR spectroscopy. The resistivity of the samples was mainly measured by the four-point probe method and further confirmed with van der Pauw configuration Hall effect measurements. For Hall measurements, 500nm-thick metal contacts consisting of Pt (100 nm), Ti (300 nm), and Au (100 nm) and having an area of 1 mm2 were deposited on each corner of the square sample films. The conductivity type, carrier concentration, and mobility of the films were then investigated. All measurements were performed at room temperature.
direction and an out-of-plane epitaxial relationship between the NiO film and the Al2O3 substrate, [1 1 1]NiO||[0 0 6]Al2O3. Next, ϕ scans were carried out to determine the in-plane crystallographic relationship between the substrate and the film. First, a ϕ scan was performed on the NiO sample at a tilt of 54.731 from the (111)NiO plane with Bragg condition θ and 2θ of 21.641 and 43.281, respectively, satisfying the orientation of the (2 0 0)NiO plane. Then, another ϕ scan was performed at a tilt of 42.311 from the (0 0 6)Al2O3 plane with Bragg condition θ and 2θ of 25.291 and 57.501, respectively, satisfying the orientation of the (1 1 6)Al2O3 plane. As shown in Fig. 2, six (200)NiO plane peaks appeared in the scan result at the same ϕ angle as the six peaks of (116)Al2O3. In cubic NiO, the (200)NiO plane has three-fold rotational symmetry around the [1 1 1]NiO axis. The existence of six (200)NiO plane peaks indicates that the epitaxial NiO film had two different crystallographic orientations (twinned) along the growth direction [1 1 1]NiO rotated by 601. A detailed microstructural study of such a twinned structure in epitaxially grown NiO on Al2O3 has been reported elsewhere [10]. The ϕ scan also showed that both the (200)NiO plane and (1 1 6)Al2O3 plane had the same rotational angle
3. Results and discussion 3.1. Structure and crystal phase XRD analyses in θ-2θ scanning mode were conducted to determine the structure and phase of the samples. To distinguish the peaks of the substrate from those of NiO, the XRD patterns of the Al2O3 substrate and x¼0 (NiO) film are presented in Fig. 1. The peaks at 411 and 901 belong to (0 0 6)Al2O3 (ICDD #46–1212), while the peaks at 371 and 791 belong to (1 1 1)NiO and (2 2 2)NiO (ICDD #47–1049), respectively. This indicates epitaxial growth of NiO towards the [1 1 1]NiO
Fig. 1. θ-2θ XRD scans of Al2O3 substrate and NiO film. The peaks at 411 and 901 belong to (006)Al2O3 while those at 371 and 791 belong to (1 1 1)NiO.
Fig. 2. XRD ϕ scan of the NiO sample tilted 54.731 from the (111)NiO plane, satisfying the Bragg condition for the (200)NiO plane, and tilted 42.311 from the (0 0 6)Al2O3 plane, satisfying the Bragg condition for (1 1 6)Al2O3.
Fig. 3. XRD θ-2θ scans of CoxNi1 xO films. Samples with x ¼0 to x ¼0.72 showed peaks at 371 and 791, which weakened with increasing cobalt composition. The additional peaks at about 381, 821 and the series of peaks at 181, 591, and 1101 observed at higher cobalt compositions (xZ 0.35) are indicative of phase separation.
T.M. Roffi et al. / Journal of Crystal Growth 414 (2015) 123–129
125
Fig. 4. XRD ϕ scans of the CoxNi1 xO samples of x ¼0.08 (a), x ¼ 0.22 (b), and x ¼0.35 (c) tilted 54.731 from the (1 1 1)NiO plane, satisfying the Bragg condition for the (2 0 0)NiO plane. All samples exhibited six-fold symmetry, indicative of a twinned structure oriented along the growth direction.
around the [1 1 1]NiO axis. Therefore, the in-plane relationship can be established as [112]NiO||[110]Al2O3 and [211]Twinned-NiO|| [1 1 0]Al2O3. Unequal intensities of the peaks in the ϕ scan were also found in a previous report [9]. This was probably a result of lattice mismatch causing variation in lattice relaxation in the lateral direction. The presence of defect such as twin phase in the CoxNi1 xO films may also cause a higher variation of intensity compared to the sapphire substrate (Fig. 4). Fig. 3 shows that all the samples containing nickel (x ¼0 to x ¼0.72) exhibited peaks at 371 and 791, indicating the existence of cubic phase NiO oriented towards [1 1 1]NiO. The films also exhibit twinned structure oriented along the growth direction as indicated by the presence of six-fold symmetry in XRD ϕ scans (Fig. 4). The NiO structure remained dominant in the samples with x ¼0 to x ¼0.08, as indicated by the narrow FWHM in Fig. 5 and strong peaks. However, the structure of NiO was weakened at higher cobalt incorporation in the film, as indicated by peak broadening and intensity decrease. The effect of cobalt became stronger in films with x ¼ 0.17 and x¼ 0.22, the patterns of which exhibited a shift towards higher angle as well as FWHM broadening of the (1 1 1)NiO peak. The existence of additional 2θ peaks at about 381 and 821 and the series of 2θ peaks at 181, 591, and 1101 observed for the samples with higher cobalt composition (x Z0.35) is indicative of phase separation. These peaks were attributed to either NiCo2O4 (ICDD #20–781) or Co3O4 (ICDD #42–1467). Further phase characterization could not be accomplished through XRD analysis owing to the similarity between the structural parameters of NiCo2O4 and Co3O4. Hence, Raman scattering analysis was carried out to complement the XRD data, as described later. High resolution ω-2θ scans normalized to the Al2O3 peaks were carried out to investigate any peak shift caused by cobalt incorporation. As shown in Fig. 6, the peak positions of the CoxNi1-xO samples slightly shifted to lower values for cobalt compositions between x¼ 0 and x¼0.08, then shifted to higher values for higher cobalt compositions. The reason for this is as follows. The observed peak positions of the x¼0 (NiO) sample were higher than those of the ICDD data, indicating a smaller interplanar spacing than that expected for stoichiometric NiO. It is well known that as-deposited NiO may possess intrinsic defects comprising excess oxygen (nickel vacancies) allowing for p-type conductivity. Cobalt may have been incorporated in these nickel-vacant sites, resulting in a larger unit cell volume compared with that of as-deposited NiO that appeared as a slight peak shift towards lower angle in the XRD pattern. Both divalent and trivalent cobalt may exist in cases of nickel substitution [11,12], and this may have caused the observed slight shift towards higher angle owing to their smaller ionic radii (0.69Å, 0.65Å, and 0.545Å for Ni2þ , Co2 þ , and Co3 þ , respectively [13]). Films grown without any nickel exhibited peaks matching those of Co3O4 (ICDD #42–1467).
Fig. 5. FWHM of (1 1 1)NiO obtained from rocking curve scan data. The FWHM of (1 1 1)NiO became broader as cobalt composition increased. Inset shows ω scan result for the x ¼ 0 (NiO) sample, showing a FWHM of 0.1121.
Fig. 6. High resolution ω-2θ scan normalized to the Al2O3 peaks. The NiO peak position slightly shifted to lower value for x ¼ 0 to x ¼ 0.08, then shifted to higher value for higher cobalt compositions. The inset presents the trend in peak position with cobalt composition.
Raman scattering spectra of the samples are shown in Fig. 7. The lowermost plot is the Raman spectra of Al2O3, which exhibits peaks at 400, 560, and 740 cm 1. The peaks overlapping these Al2O3 peaks
126
T.M. Roffi et al. / Journal of Crystal Growth 414 (2015) 123–129
were not considered in further analysis. The peaks observed for the x¼0 (NiO) film at about 560, 925, 1100, and 1500 cm 1 match those of the literature [14] representing the LO, TOþLO, 2LO, and 2M modes. The TO and 2TO modes expected at about 440 and 740 cm 1 could not be observed owing to the presence of overlapping Al2O3 peaks. At low cobalt composition (x ¼0.01), no significant change in Raman spectra was observable. At x ¼0.02, two effects could be observed. First, the magnon excitation mode of NiO (2M) weakened and shifted towards lower wavenumbers. Second, additional Raman peaks appeared at about 520 and 620 cm 1, which could be addressed to CoO phase[15]. The existence of Co-O bonding is expected when cobalt substitutes into vacant nickel sites or substitutes nickel in the lattice, in agreement with the XRD analysis. Thus, although the coexistence of NiO and CoO structures could not be determined from the XRD data, their different vibrational modes made it possible to observe them by Raman spectroscopy, as explained above. The Raman spectra of samples with x ¼0.17 to x¼ 0.72 were identical to those reported in the literature for nickel-cobalt oxide series [16,17], with two major Raman peaks at about 465 and 655 cm 1 and a minor one at about 551 cm 1. The peak at about 550 cm 1 in our samples, however, appeared to be sensitive towards the cobalt composition. The peak at about 650 cm 1 has previously been attributed to a vibrational mode involving oxygen atom in octahedral sites in the spinel lattice [18]. Thus, it is suggested that the slight shift of this peak towards higher frequency in samples of higher cobalt composition was caused by Co3 þ substitution of Ni2 þ in the octahedral sites of the spinel structure. Co3 þ substitution of Ni2 þ at spinel octahedral sites would cause structural distortion owing to the smaller ionic radius of Co3 þ and shorter bond length, thus causing a shift of the Raman band to higher frequency. With the XRD results, it can be confirmed that the phases present in this sample were NiO and the series of nickel-cobalt oxides (NiCo2O4), not Co3O4. Consistent with the XRD results, the Raman peaks of samples grown without any nickel, at 480, 519, 617, and 689 cm 1, match those reported for Co3O4 in the literature [19,20]. As summarized in Table 1, cobalt was incorporated in the NiO lattice for CoxNi1 xO samples with xr0.08. Formation of CoxNi1 xO alloy occurred at x¼ 0.17 and x¼ 0.22. Attempts to incorporate a greater proportion of cobalt resulted in phase separation into NiO and NiCo2O4 phases, as evidenced for the samples with
Fig. 7. Raman Scattering of CoxNi1 xO films. The spectrum obtained for the x ¼0.02 sample was assigned to NiO and CoO, while the spectra of higher cobalt compositions (x¼ 0.17 to x¼ 0.72) were attributed to nickel-cobalt oxides.
Table 1 Observed phases in relation to cobalt composition of the films. Cobalt Composition x
0 0.01 0.02 0.08 0.17 0.22 0.35 0.56 0.72
Phases NiO
CoO
v v v v
v v v
CoxNi1-xO
NiCo2O4
v v v v v
v v v
x¼ 0.35–0.72. The NiCo2O4 phase became dominant when x¼ 0.72, while samples grown without nickel were found to be Co3O4. 3.2. Appearance and morphology The x ¼0 (NiO) film was optically clear and grayish green in color. The films darkened with increasing cobalt composition up to x¼ 0.22; beyond this, the films were black. The film grown without any nickel (Co3O4) was slightly lighter in color, dark brown, and was optically clear. The surface morphology of the CoxNi1-xO films differed depending on their cobalt composition, as observed from the SEM images presented in Fig. 8. Coalesced triangular pyramidal grains with poorly defined boundaries were found in the x¼0 (NiO) and x¼0.08 films. This result supports the structural analysis by XRD and Raman spectroscopy, which found that the NiO phase remains with cobalt incorporation at low x. The films with higher cobalt composition exhibited segregated grains with more distinct boundaries. Segregation of the grains was most likely introduced by the phase separation of NiO and NiCo2O4. It was also evident that some of the pyramids were rotated by 601, indicative of twinned (111)NiO in the films. 3.3. Transmittance and bandgap energy approximation UV–vis-NIR transmittance spectroscopy measurements were performed to investigate the optical characteristics and approximate the bandgap of the films. The optical transmittance spectra of the CoxNi1 xO samples in the wavelength range 200–800 nm at room temperature are given in Fig. 9. As shown in the inset, the x¼ 0 (NiO) film showed high transparency in the visible region, indicated by an 88% average transmittance for the 800-nm-thick film. Although the transmittance decreased with increasing cobalt composition, it was maintained at a relatively high percentage up to x¼0.17, i.e., 71% average transmittance for a 700-nm-thick film. These transmittance results agree well with the XRD analysis of the films; incorporation of cobalt at this level introduced defect sites, indicated by the slight change in structural parameters, which scatter incident light. More significant deterioration in the transparency of NiO films was observed in the case of lithium doping [21] owing to its larger difference in ionic radius. Beyond x¼0.17, transmittance in the visible region dropped as a result of enhanced alloy formation. A shift of fundamental absorption edge towards lower wavelength (redshift) was observed with increasing cobalt composition, which is related to a decrease in optical bandgap. The optical bandgaps of the films were approximated using the classical relation for near edge optical absorption in semiconductors. αhν2 ¼ A hν Eg ; ð1Þ where A is a constant, α is the absorption coefficient, and hν and Eg represent the incident photon energy and the semiconductor bandgap energy, respectively. Plots of (αhν)2 against (hν) for the
T.M. Roffi et al. / Journal of Crystal Growth 414 (2015) 123–129
127
Fig. 8. SEM micrographs of CoxNi1-xO samples taken at 10 000 magnification. Coalesced triangular pyramidal grains with poorly defined boundaries were found in x¼ 0 (NiO) and x ¼ 0.08 films, while films with higher cobalt composition exhibited segregated grains with more distinct boundaries. Insets show micrographs taken at 20 000x magnification. All white bars represent 1μm length.
CoxNi1 xO films are presented in inset of Fig. 10. The bandgap decreased from 3.7 eV to 2.35 eV with increasing cobalt composition. The relation between cobalt composition and optical band gap can be approximated by Vegard's law: Eg;
Cox Ni1 x O
¼ x Eg;
Co3 O4
þ ð1 xÞEg; NiO b x ð1 xÞ;
ð2Þ
where Eg is the bandgap of the respective materials and b is the bowing parameter, a measure of the non-linearity of the bandgap as a function of composition. Although phase separation occurred for cobalt composition x¼ 0.35, 0.56, and 0.72, which makes the use of Vegard's law unusual in this case, the approximation is well suited and useful for further compositional design of CoxNi1−xO,
especially by the current growth method. The optical bandgap and cobalt composition was well fitted by a bowing parameter of 3.5, as shown in Fig. 10. It should be noted that the bandgaps of x¼ 0.56 and x ¼0.72 films were not shown because of undefined absorption edges due to low transmittance. However, the approximated bandgap for x¼ 0.35 where phase separation occurs could be determined and was found to follow Vegard's law. The incorporation of cobalt into NiO to form CoxNi1 xO was therefore able to decrease the bandgap of NiO considerably. This result indicates that selective UV-region response materials based on NiO can be manufactured using the present APMOCVD system by controlling the cobalt composition.
128
T.M. Roffi et al. / Journal of Crystal Growth 414 (2015) 123–129
Fig. 11. Change in resistivity of the films with cobalt composition. The resistivity decreased from 960 to 0.006 Ωcm with increased cobalt composition from x¼ 0 to x¼ 0.72.
Fig. 9. Transmittance spectra of the CoxNi1-xO samples. The highest transmittance of 88% was obtained for the 800-nm-thick x ¼0 (NiO) film. The incorporation of cobalt decreased the transmittance and caused red shift (towards higher wavelength) of the fundamental absorption edge.
annealing, ultraviolet irradiation, and varied oxygen pressure [8,9,22], which cause oxygen deficiency. The defect reaction can be expressed as follows, assuming complete ionization. ' 1 OX O ⇆vO þ 2e þ O2 ðgÞ 2
ð4Þ
In the case of cobalt incorporation into NiO films, it is reported that cobalt ions exist in the nickel lattice as Co2 þ and Co3 þ [11, 12]. While the substitution of homovalent cobalt (Co2 þ ) for nickel in the lattice may not result in any carrier generation, incorporation of aliovalent Co3 þ may generate carriers according to the following Kröger–Vink notation. 1 O2 ðg Þ þ Co2 O3 ⇆ 2CoNi þ 4OX0 þ 2e' 2
Fig. 10. Approximated bandgap energy as a function of cobalt composition follows Vegard's law with bowing parameter 3.5. Data points for x ¼ 0.56 and x ¼0.72 are not available since these films are not transparent. Inset shows the plot of (αhν)2 to (hν).
3.4. Electrical properties Four-point-probe measurements were conducted at room temperature to investigate the resistivity of the samples as a function of cobalt composition (Fig. 11). The resistivity of the films was found to decrease from 960 Ωcm to 0.006 Ωcm with increasing cobalt composition from x ¼0 to x ¼0.72. However, the resistivity did not change much from x ¼0 to x ¼0.08. The effect of the cobalt composition on the resistivity can be explained by the carrier generation process in CoxNi1 xO. Although stoichiometric NiO is an insulator, it is well known that defects play major role in the conductivity of NiO. As-deposited NiO may contain native defects of excess oxygen which create cation deficiencies on NiO lattice sites that cause the generation of holes, as expressed by the following Kröger–Vink notation. 1 þ O2 ðg Þ ⇆ vNi '' þOX o þ 2h 2
ð3Þ
As expressed in Eq. (3), two holes are generated to maintain charge neutrality. Consistently, defects in NiO may also reverse its conductivity type. It is reported that NiO may obtain n-type conductivity by certain treatments such as post deposition
ð5Þ
As expressed in Eq. (5), two electrons are generated for every two trivalent cobalt ions substituting nickel in the lattice. Therefore, the decreasing resistivity observed with increasing cobalt composition is caused by increased carrier generation in CoxNi1 xO. This is well in agreement with the Hall measurement results (inset of Fig. 11) of the CoxNi1 xO samples with x ¼0.35, 0.56, and 0.72, which exhibit n-type conductivity. It is surmised that trivalent cobalt in CoxNi1 xO alloy behaved as a donor. We note that caution is needed in interpreting the Hall results owing to possible magnetic influence of the sample. It has been reported that although Co3O4 and NiO are p-type semiconductors, the conductivity of Ni1 xCoxO4/3 was not clear, with Hall measurements showing both p- and n-type conductivity in the same sample [23]. Indeed, in our Hall measurements only three samples exhibited consistent n-type conductivity, while the conductivity type of the rest of the samples could not be determined.
4. Conclusions Successful growth of CoxNi1 xO with a wide cobalt composition range of x¼ 0.01 to x¼ 0.72 has been achieved using APMOCVD. The properties of the films were found to vary with cobalt composition (x). All the films grown had a preferred orientation towards [1 1 1]NiO with a twinned structure. Cobalt was well incorporated in the NiO structure up to x¼ 0.08, CoxNi1 xO alloys were formed from x ¼0.17 to x ¼0.22, and phase separation of NiO and CoxNi1 xO occurred when x Z0.35. Films grown without any nickel were found to be composed of Co3O4. The bandgap of the CoxNi1 xO films decreased with increasing cobalt composition, and controllable bandgap engineering in the UV region was demonstrated. The resistivity of the films also decreased with increasing cobalt composition, reaching a minimum of 0.006 Ωcm. Some of the samples exhibited n-type conductivity, determined by Hall measurements. The existence of
T.M. Roffi et al. / Journal of Crystal Growth 414 (2015) 123–129
trivalent cobalt as a donor in the NiO structure is considered to be the cause of the n-type conductivity observed in the films. In the future, it would be of interest to investigate the thickness dependence of the films properties for each composition, because it may be relevant for device applications. Acknowledgment T. M. Roffi gratefully acknowledges a scholarship for his Ph.D. study from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan. References [1] H. Sato, T. Minami, S. Takata,, T. Yamada, Transparent conducting p-type NiO thin films prepared by magnetron sputtering, Thin Solid Films 236 (1) (1993) 27–31. [2] H. Ohta, M. Hirano, K. Nakahara, H. Maruta, T. Tanabe, M. Kamiya, T. Kamiya,, H. Hosono, Fabrication and photoresponse of a pn-heterojunction diode composed of transparent oxide semiconductors, p-NiO and n-ZnO, Appl. Phys. Lett. 83 (5) (2003) 1029–1031. [3] M. Tyagi, M. Tomar,, V. Gupta, Influence of hole mobility on the response characteristics of p-type nickel oxide thin film based glucose biosensor, Anal. Chim. Acta. 726 (2012) 93–101. [4] T. Dutta, P. Gupta, A. Gupta,, J. Narayan, High work function (p-type NiO1 þ x)/ Zn0. 95Ga0. 05O heterostructures for transparent conducting oxides, J. Phys. D.: Appl. Phys. 43 (10) (2010) 105301. [5] H. Rooksby,, M. Vernon, Lithium-and gallium-doped nickel oxide, Brit. J. Appl. Phys. 17 (9) (1966) 1227. [6] T. Dutta, P. Gupta, A. Gupta,, J. Narayan, Effect of Li doping in NiO thin films on its transparent and conducting properties and its application in heteroepitaxial pn junctions, J. Appl. Phys. 108 (8) (2010) (083715–083715). [7] W.-L. Jang, Y.-M. Lu, W.-S. Hwang,, W.-C. Chen, Electrical properties of Lidoped NiO films, J. Eur. Ceram. Soc. 30 (2) (2010) 503–508. [8] M. Tyagi, M. Tomar,, V. Gupta, Postdeposition annealing of NiOx thin films: A transition from n-type to p-type conductivity for short wave length optoelectronic devices, J. Mater. Res. 28 (2013) 723–732.
129
[9] P. Gupta, T. Dutta, S. Mal,, J. Narayan, Controlled p-type to n-type conductivity transformation in NiO thin films by ultraviolet-laser irradiation, J. Appl. Phys. 111 (1) (2012) (013706–013706). [10] K. Uchida, K. Yoshida, D. Zhang, A. Koizumi, S. Nozaki, High-quality single crystalline NiO with twin phases grown on sapphire substrate by metalorganic vapor phase epitaxy AIP ADVANCES, 2 042154 (2012), http://dx.doi.org/10. 1063/1.4769082. [11] G. Natu, P. Hasin, Z. Huang, Z. Ji, M. He,, Y. Wu, Valence band-edge engineering of Nickel Oxide Nanoparticles via Cobalt doping for application in p-type dyesensitized solar cells, ACS. Appl. Mater. Inter. 4 (11) (2012) 5922–5929. [12] Y. Gou, X. Liang,, B. Chen, Porous Ni–Co bimetal oxides nanosheets and catalytic properties for CO oxidation, J. Alloy. Compd. 574 (2013) 181–187. [13] J.A. Dean, Lange's Chemistry Handbook, vol. 4, McGraw-Hill, Inc, 1999. [14] N. Mironova-Ulmane, A. Kuzmin, I. Sildos,, M. Pärs, Polarisation dependent Raman study of single-crystal nickel oxide, Cent. Eur. J. Phys. 9 (4) (2011) 1096–1099. [15] H.C. Choi, Y.M. Jung, I. Noda,, S.B. Kim, A study of the mechanism of the electrochemical reaction of lithium with CoO by two-dimensional soft X-ray absorption spectroscopy (2D XAS), 2D Raman, and 2D heterospectral XASRaman correlation analysis, J. Phys. Chem. B. 107 (24) (2003) 5806–5811. [16] G. Calin, M. Irimia, C. Scarlat, M. Purica, F. Comanescu, F. Iacomi, Synthesis and characterization of nickel cobalt oxide thin films, in: Semiconductor Conference (CAS), 2010 International, 2010. [17] C.F. Windisch Jr, G.J. Exarhos,, S.K. Sharma, Influence of temperature and electronic disorder on the Raman spectra of nickel cobalt oxides, J. Appl. Phys. 92 (9) (2002) 5572–5574. [18] C.F. Windisch Jr, K.F. Ferris, G.J. Exarhos,, S.K. Sharma, Conducting spinel oxide films with infrared transparency, Thin Solid Film. 420 (2002) 89–99. [19] H. Zhou, L. Chen, V. Malik, C. Knies, D.M. Hofmann, K.P. Bhatti, S. Chaudhary, P. J. Klar, W. Heimbrodt, C. Klingshirn (others), Raman studies of ZnO: Co thin films, 204, Phys. Status Solidi A (2007) 112–117. [20] V. Hadjiev, M. Iliev,, I. Vergilov, The raman spectra of Co3O4, J. Phys. C.: Solid. State. Phys. 21 (7) (1988) L199. [21] Y.H. Kwon, S.H. Chun,, H.K. Cho, Semiconducting p-type MgNiO: Li epitaxial films fabricated by cosputtering method, J. Vacuum. Sci. Technol. A. 31 (4) (2013) 041501. [22] R. Molaei, R. Bayati,, J. Narayan, Crystallographic Characteristics and p-Type to n-Type Transition in Epitaxial NiO Thin Film, Cryst. Growth Des. 13 (12) (2013) 5459–5465. [23] C. Windisch Jr, G.J. Exarhos, K.F. Ferris, M.H. Engelhard,, D.C. Stewart, Infrared transparent spinel films with p-type conductivity, Thin Solid Film. 398 (2001) 45–52.