Open Access
Add to BibSonomyAbstract
It remains a challenge to characterize the doping type in nanowires (NWs). We report in this paper a novel way to probe the doping type in GaN NWs by photoassisted kelvin probe force microscopy (KPFM), as a proper example showing that this approach is straight forward, effective and practical. Through illumination with super-bandgap light, photo-generated electrons in the n-region are swept away from the surface due to the electric field in the space-charge region, thus the holes move to the surface; while in contrast, electrons in the p-region will move to the surface. The fact that the quasi-Fermi level moves upwards in n-type while downwards in p-type identifies the doping type of GaN NWs, and is clearly revealed by the contact potential difference detected by photoassisted KPFM.
© 2017 Optical Society of America
1. Introduction
Semiconductor nanowires (NWs) with desired doping type is essential for realizing future nanoscale devices and circuits, such as field effect transistor [1–3], light emitting diodes [4], lasers and spin-orbit qubits [5,6]. However, in contrast to bulk semiconductors or films, many obstacles remain in both understanding and controlling doping of NWs. In particular, one of those challenges is the difficulty in characterizing the doping type in NWs. Unlike thin films, the small size of NWs prevents the usage of standard Hall effect measurements to determine the doping levels/types, carrier concentrations and mobility. Optical techniques, such as photoluminescence [7], cathodoluminescence [8] and Raman spectroscopy [7], can sometimes reveal the existence of certain impurities and provide indirect evidence of conduction type, but those approaches are unable to directly confirm the doping polarity. Beneficial from its high spatial resolution, Kelvin probe force microscopy (KPFM) [9] has been adopted to investigate the doping status of NWs. With KPFM applied, the differences of the probe contact potential on various parts of semiconductor NWs could be measured [10–12], such as the contact potential of GaAs p-n junctions [11] and GaN p-n junctions [12]. However, those works only got the results of the contrast of doping level between different parts in NWs, while the brightness contrast in KPFM signals induced by the work function difference cannot confirm the semiconductor’s doping type being either p-type or n-type. In fact weak and strong n-type in a same semiconductor can also result in the brightness contrast probed by KPFM.
In this work, we have used photoassisted KPFM for GaN NWs, which can be generalized as an example of this approach. In comparison with traditional semiconductors such as Si and GaAs, p-type doping in GaN via Molecular Beam Epitaxy (MBE) and its clarification are much more difficult. Besides, due to the lack of suitable contacts and complicated sample preparation (since NWs are short), Ohmic contact in GaN NW is also very hard to make. Therefore, the new and efficient approach is of great importance in probing GaN NWs. In this paper, KPFM measurements with and without super-bandgap illumination were applied to measure the contact potential differences (CPD), VCPD, between the tip and sample [13,14]. Super-bandgap illumination causes photo-generated electrons/holes to accumulate at the surface in either p or n-type regions, which decreases the band bending, and hence enlarges the CPD. The results demonstrate for the first time that KPFM under light illumination can unambiguously confirm the doping type of GaN NWs, and we believe that this approach can be extended to more general semiconductor NWs.
Figure 1(a) shows the energy band diagram of an ideal p-GaN NW and an n-GaN NW surfaces with a metal tip. Here, φtip, φn and φp are work functions of the tip, n-GaN NW and p-GaN NW, respectively. In the KPFM measurements, the CPD of the p-n junction in GaN NW is, Δp-n = (φp − φtip)− (φn − φtip). However, due to surface band bending mainly induced by the surface states and the spontaneous fields [15], the measured value of Δp-n is usually much smaller than the ideal one. That happens in some similar studies on other bulk and NW materials [10,11]. This principle of the surface states’ function is schematically indicated in Fig. 1(b) where the surface state energy positions are usually located inside the band gap, forming a separate band. For an n-type GaN, Ef (bulk) is higher than Ef (surface) beyond equilibrium, then the electrons will transfer from the bulk to surface [16]. At equilibrium, on the other hand, the Fermi level will be unified and the energy band will bend upward, making the work function between the tip and sample decreases. For a p-type one, those transferring electrons will, on the contrary, cause a downward band bending [16], and the work function between the tip and sample increases. As shown in Fig. 1(b), the CPD of a p-n junction, i.e., Δp-n’ = (φp’ − φtip )− (φn’ − φtip ), is much smaller than the ideal value of Δp-n. After all, the potential difference leads to the brightness contrast in KPFM measurement of a p-n junction [10,11]. However, the KPFM results alone can only tell the difference of the bright and dark regions, but not yet their doping polarity. Hereby we will show that only the light illumination induced shift that can confirm the doping type with no doubt.
Fig. 1 (a) Energy band diagram illustrating how the potential profile is measured for a GaN p-n junction. Evac is the local vacuum level, Ef is the Fermi level, and Ev and Ec are the valence and conduction band edges of the GaN NW, respectively. (b) Energy band diagram of a p-n junction considering the surface states. (c) Band bending of p-n junctions in dark and under illuminations depicted by solid lines and dashed lines, respectively.
The incident photon energy is super-bandgap, while on illumination, it breaks the Fermi equilibrium in the p-n junction, forming quasi-Fermi levels for electrons and holes, respectively (Fig. 1(c)). The direction of the carrier movement depends on the surface band bending. In an n-GaN, the electric field in the space-charge region [17] causes the excess electrons to be swept away from the surface. Thus, the excited holes move to the surface, and this serves to reduce the density of the surface states and decreases the band bending (as shown by the dash lines in the left part of Fig. 1(c)). In contrast, in the case of a p-GaN the excited electrons move to the surface, resulting in attenuation of the band bending in an opposite way (as shown by the dash lines in the right part of Fig. 1(c)). As such, what KPFM measures under illumination are two new quasi-Fermi levels for electrons in n-type and holes p-type materials, respectively. That consequently induces a relative shift of the surface photovoltaic [18] between p and n-type GaN, and thus enlarging the signal of CPD. Therefore, these results will definitely identify the doping type of GaN NWs.
2. Experimental
A Bruker icon scanning probe microscope is employed for the KPFM measurement. The system is operated as a double-pass apparatus, where the first pass measures the sample topography and the second one measures the VCPD in a lift mode. The second pass is performed without mechanical drive of the cantilever, but instead, an adjustable bias voltage is applied to the tip. In KPFM, the electrostatic force between the tip and sample rather than the displacement current is detected. The NWs were ultrasonically dispersed in acetone and deposited on a Si wafer with a 200 nm thick layer of thermally grown SiO2. The measurement is conducted in air. A Pt-Ir coated silicon tip (SCM-PIT) with radius of 25 nm is used as the probe. The tip’s resonant frequency is 68 kHz. The tip oscillates at a distance of about 10 nm above the sample surface. The samples were illuminated from the front side during the measurement. A laser beam with wavelength of 266 nm corresponding to a photon energy above the band gap of GaN was chosen.
The GaN NWs were grown by plasma-assisted MBE on 2-inch n-doped (111) Si substrate with resistivity < 0.002 Ω-cm. The substrate was cleaned with a hydrofluoric acid solution to remove native oxides prior to being loaded into the MBE chamber. Then the Si wafer was degassed in situ for 30 min at 900°C. Following the annealing, the substrate temperature was lowered until the reconstruction transition from (1 × 1) to (7 × 7) occurs at around 830°C. Subsequently, 1min nitridation of the Si surface at 680°C leads to formation of a SixNy film. Catalyst-free growth of GaN NWs was then conducted under N-rich conditions, maintaining a constant N2 flux of 1.5 sccm. The growth temperature was maintained at 780°C and Ga flux at 1.2 × 10−7 Torr. Si-doped GaN NWs with a height of ~900 nm were firstly grown, and then followed by deposition of Mg-doped GaN with a height of 500 nm at 500°C with Ga flux at 6.6 × 10−8 Torr. The temperature of the Si effusion cell was at 1270°C, and the temperature of the Mg effusion cell was changed from 250°C to 375°C, which corresponds Mg beam-equivalent pressure (BEP) of ~6 × 10−10 to ~2 × 10−8 Torr, with a step of 25°C for different GaN NWs, in order to find the best Mg-doping condition.
3. Results and discussion
It is shown that the GaN NWs for measurements are around 1.4-μm-long on average with a mean diameter of ~100 nm (Fig. 2). From the top view of the SEM image, certain level of coalescing can be observed; such coalescing may bring about some defects, thus would somehow further influence on the VCPD values. Fortunately, the influence is not serious since the KPFM measures the surface potential and here we just focus on the variation value of VCPD, not the absolute one. Six different vertically standing GaN NWs samples with varied Mg concentration were analyzed by KPFM. The sample grown with Mg cell temperature of 300°C was chosen as an example for detailed data representation and the corresponding results are shown in Figs. 3(a)−3(c), while Fig. 3(d) depicts a comparison of the results from all six NWs. From the top-view topography of the NWs shown in Fig. 3(a), it can be seen that most of the NW diameters are broadened due to tip-sample convolution. Figure 3(b) shows the corresponding VCPD mapping while the inset is a 3D potential that represents the average potential of the sample. Figure 3(c) is the vertical height and VCPD profiles along the marked line. According to the corresponding profiles across the marked lines (Fig. 3(c)), the NWs exhibit negligible difference in thickness. Besides, the top viewed SEM image in Fig. 2 has high density. The data of KPFM are thus reliable. From the theory of KPFM in prior, while a voltage is applied to the tip, we have VCPD = (Φtip − Φsample)/−e, with Φtip and Φsample being the work functions of the tip and sample, respectively, where e is the elementary charge. Therefore, lowering of VCPD means lowering of the Fermi level in the sample, or in other words, enhancing of the density of holes. As such, a negative shift in VCPD in the experiment can be attributed to an increase of the hole concentration. Based on characterizing a series of NWs with varying Mg doping levels, the average VCPD of NWs is shown in Fig. 3(d). It can be seen that when the Mg cell temperature during growth is 300°C, corresponding to Mg flux of 1 × 10−9 Torr BEP, the p-type concentration is the highest according to the measured VCPD values, and this correspondence was unknown until this work. In fact, Mg doping not only provide acceptors, but may also create defects. Over-doping effect usually happens when the Mg concentration is increased beyond a point. That is why there is an optimized Mg cell temperature which can produce the best p-type doping. Even though there could be more defects induced by Mg doping, they would not affect the general results by using KPFM to identify the conduction type.
Fig. 2 SEM images of the NW sample: (a) lateral view and (b) top view. The vertically aligned GaN NWs are around 40 nm in diameter on the bottom, and about 130 nm on the top due to the lower growth temperature while doping Mg. The density of the NW varies in the range of 1−2 × 107 cm−2.
Fig. 3 (a) AFM topography (1 × 1 μm2) and (b) the corresponding KPFM map of GaN NWs. The scale bar is 200 nm. (c) The vertical height and VCPD profiles along the marked dash line. (d) Average VCPD of NWs varying with the Mg cell temperature.
Afterwards, the NWs were removed from the as-grown sample and transferred to an oxide-coated silicon wafer, lying down on the wafer. The KPFM measurement was focused on the sample with Mg cell temperature of 300°C as a typical one. The KPFM image exhibits an obvious brightness contrast between the n-region and Mg-doped region. The CPD line profile of the NW is displayed on the top half of Fig. 4(e). It is noticeable that the surface potential line profile looks closely correlated with the topography amplitude along the NW axis. To rule out the influence of topography on the surface potential, undoped NWs with the same growth condition as the doped ones are also characterized in a similar fashion. Figures 4(c) and 4(d) show the measured topography and 2D KPFM voltage image of the undoped GaN NW. The height of the undoped GaN NW and its corresponding CPD are displayed in the bottom half of Fig. 4(e). Here the CPD value at the central depletion region is normalized to be zero for all samples in order to have a better comparison. (as well as in Fig. 6) Those data show that there is no relationship between the topography amplitude and its CPD. As a matter of fact, the reason for the Mg-doped NW becomes fatten lies in the lower growth temperature while doping Mg. In the top half of Fig. 4(e), the value of CPD increases by 150 mV at the point 400 nm away from the left end of the NW, where the Mg doping ends. That depicts the variation of the doping property along the wire. Ideally, the amplitude of the CPD between the p and n sides of a semiconductor p-n junction should be at least half of the bandgap. But the measured result is much smaller than that, being 150 mV only, due to the effect of the surface band bending as shown in Fig. 1(b). During the measurement the background doping is not taken into account, because the GaN nanowires grown by MBE exhibit low background doping (usually leads to n-type), and with the p-type doping by Mg dopant, those created electrons will be compensated by holes and the variation of the Fermi level will be determined by net holes.
Fig. 4 (a) AFM topography of a single p-n junction and (b) corresponding 2D KPFM voltage image. (c) AFM topography of an undoped NW and (d) corresponding 2D KPFM voltage image, and the scale bar is 200 nm. (e) The vertical height and VCPD profile of the doped and undoped NWs along the marked line, respectively.
NWs with various Mg doping levels are then all measured. After making the Si doping concentration as a unified standard, the results are summarized in Fig. 5. Starting from the right side of Fig. 5, the graph shows a strong change of the CPD in almost middle of the NWs. Various Mg cell temperature means various density of holes, and thus resulting in different value of CPD in the p-region. From the inset of Fig. 5, while the CPD dependence on the Mg cell temperature is displayed, one can see that the CPD value of the NW with Mg cell temperature of 300°C is the largest. That further supports the result in prior that 300°C is the best doping temperature.
Fig. 5 VCPD of p-n junctions with varying the Mg cell temperature. Inset shows the CPD with varying temperature. For the convenience of comparison, the CPD value at the n region is unified.
Even though we access the best Mg doping temperature in such a way, the question that whether those parts having lowered CPD values in the NWs are p-type or just weak n-type can still not be clarified for sure. We will show that the light illumination induced shift does unambiguously confirm the the doping type of GaN NWs.
The measurements were performed by illuminating the sample with a laser beam (λ = 266 nm, the corresponding photon energy being over the samples’ bandgap). To study the evolution of CPD under UV illumination, the NW with most efficient Mg doping was chosen and the illumination light power was changed from 1 to 15 mW. Figure 6(a) and 6(b) show typical KPFM images for the same NW in dark and under 15 mW illumination, respectively. From Fig. 6(c) one can clearly see that with increasing light power, the value of CPD increases positively in the n-region while it increases negatively in the Mg-doped region. That is a remarkable sign of the doping type. Taking n-region as an example, the electron quasi Fermi level is related to the non-equilibrium electron density by
where NC is the conduction band density of states, Ec the bottom of the conduction band, and kB the Boltzmann constant, so that the quasi Fermi level can be expressed asis resulted from an upward shift of the corresponding Fermi level from the equilibrium value of Ef, and contributes to the measured CPD. While for a p-type material, the quasi Fermi level is shifted downward from the equilibrium Fermi level Ef. Therefore, the total value of the CPD contrast along the NW is significantly enlarged by the sum of the two shifts of the quasi-Fermi levels for electrons in the n-type material and that for holes in p-type part. Thus, the contrast of the CPD signal which reflects the p-n junction become more profound. It is also noticed from Fig. 6(c) that the measured depletion region’s width increases with increasing light power. That can be understood from the equation for the depletion region width Xm:where VD is the built-in voltage, V the applied voltage, and NA/ND the acceptor/donor concentrations, respectively. When a sample of p-n junction is illuminated by an above bandgap light beam, excess carriers will be excited and the potential in the depletion region will drag the excess electrons (holes) to n (p) regions, respectively, named as carrier extraction. That is equivalent to applying a reversed bias to the p-n junction, making a negative value of V in the above equation, and hence enlarge the width of the depletion region. The stronger the light power, the more remarkable this effect, and so that the depletion region becomes wider and wider with increasing light power. Obviously, the contrast of CPD from the NW under illumination is much more distinct compared with that in dark. As explained by Fig. 1(c), photoassisted KPFM measurements indeed unambiguously identify the doping type of this structure, and that approach, as far as we known, is being used for studying NWs (not limited for GaN NWs) for the first time.Fig. 6 (a) The KPFM image of p-n junction in dark and (b) under 15 mW illumination. The scale bar is 200 nm. (c) CPD of p-n junctions under illumination with varying light power.
4. Conclusion
In summary, it is demonstrated that the doping type in GaN NWs can be revealed via the photoassisted KPFM measurement scanning along growth direction on single GaN NW. For the n-type part of a p-n junction in GaN NW, the quasi Fermi level is shifted upward from the equilibrium Fermi level Ef ; On the contrary, for the p-type part, the quasi Fermi level is shifted downward from the equilibrium Fermi level Ef. Therefore, the potential profile reflecting p-n junctions becomes more and more remarkably in contrast with increasing laser power, in comparison with that in dark. Despite the uncertainty in determining the quantitative values of the doping concentration in NWs, the measured results greatly help to demonstrate the doping type of GaN NW. This approach is certainly not only very significant for GaN NWs as they are especially hard to be characterized, but definitely also suitable for studying other semiconductor NWs, except some special surface band bending cases such as InN and InGaN with high In composition [19–21].
Funding
This work was partially supported by the National Key Research and Development Program of China (No. 2016YFB0400100), the National Natural Science Foundation of China (Nos. 61225019, 61376060, 61428401 and 61521004), Science Challenge Project (No. JCKY2016212A503), NSAF (No. U1630109), the CAEP Microsystem and THz Science and Technology Foundation (No. CAEPMT201507) and the Open Fund of the State Key Laboratory on Integrated Optoelectronics.
References and links
1. Y. Cui and C. M. Lieber, “Functional Nanoscale Electronic Devices Assembled Using Silicon Nanowire Building Blocks,” Science 291(5505), 851–853 (2001). [CrossRef] [PubMed]
2. Ž. Gačević, D. López-Romero, T. Juan Mangas, and E. Calleja, “A top-gate GaN nanowire metal–semiconductor field effect transistor with improved channel electrostatic control,” Appl. Phys. Lett. 108(3), 033101 (2016). [CrossRef]
3. K. Maeda, N. Okabayashi, S. Kano, S. Takeshita, D. Tanaka, M. Sakamoto, T. Teranishi, and Y. Majima, “Logic Operations of Chemically Assembled Single-Electron Transistor,” ACS Nano 6(3), 2798–2803 (2012). [CrossRef] [PubMed]
4. H. M. Kim, T. W. Kang, and K. S. Chung, “Nanoscale Ultraviolet-Light-Emitting Diodes Using Wide-Bandgap Gallium Nitride Nanorods,” Adv. Mater. 15(78), 567–569 (2003). [CrossRef]
5. J. C. Johnson, H. J. Choi, K. P. Knutsen, R. D. Schaller, P. Yang, and R. J. Saykally, “Single Gallium Nitride Nanowire Lasers,” Nat. Mater. 1(2), 106–110 (2002). [CrossRef] [PubMed]
6. S. Nadj-Perge, S. M. Frolov, E. P. Bakkers, and L. P. Kouwenhoven, “Spin-orbit qubit in a semiconductor nanowire,” Nature 468(7327), 1084–1087 (2010). [CrossRef] [PubMed]
7. M. G. Kibria, F. A. Chowdhury, S. Zhao, M. L. Trudeau, H. Guo, and Z. Mi, “Defect-engineered GaN:Mg nanowire arrays for overall water splitting under violet light,” Appl. Phys. Lett. 106(11), 113105 (2015). [CrossRef]
8. X. Zhou, M. Y. Lu, Y. J. Lu, S. Gwo, and S. Gradečak, “Correlation of doping, structure, and carrier dynamics in a single GaN nanorod,” Appl. Phys. Lett. 102(25), 253104 (2013). [CrossRef]
9. M. Nonnenmacher, M. P. O’Boyle, and H. K. Wickramasinghe, “Kelvin probe force microscopy,” Appl. Phys. Lett. 58(25), 2921–2923 (1991). [CrossRef]
10. E. Koren, N. Berkovitch, and Y. Rosenwaks, “Measurement of active dopant distribution and diffusion in individual silicon nanowires,” Nano Lett. 10(4), 1163–1167 (2010). [CrossRef] [PubMed]
11. S. Vinaji, A. Lochthofen, W. Mertin, I. Regolin, C. Gutsche, W. Prost, F. J. Tegude, and G. Bacher, “Material and doping transitions in single GaAs-based nanowires probed by Kelvin probe force microscopy,” Nanotechnology 20(38), 385702 (2009). [CrossRef] [PubMed]
12. S. R. Ryu, S. D. G. Ram, S. J. Lee, H. Cho, S. Lee, T. W. Kang, S. Kwon, W. Yang, S. Shin, and Y. Woo, “Vertical current-flow enhancement via fabrication of GaN nanorod p–n junction diode on graphene,” Appl. Surf. Sci. 347, 793–798 (2015). [CrossRef]
13. G. Elias, T. Glatzel, E. Meyer, A. Schwarzman, A. Boag, and Y. Rosenwaks, “The role of the cantilever in Kelvin probe force microscopy measurements,” Beilstein J. Nanotechnol. 2, 252–260 (2011). [CrossRef] [PubMed]
14. G. H. Buh, H. J. Chung, J. H. Yi, I. T. Yoon, and Y. J. Kuk, “Electrical characterization of an operating Si p-n-junction diode with scanning capacitance microscopy and Kelvin probe force microscopy,” J. Appl. Phys. 90, 443 (2001).
15. W. E. Spicer, P. W. Chye, P. R. Skeath, C. Y. Su, and I. Lindau, “New and unified model for Schottky barrier and III–V insulator interface states formation,” J. Vac. Sci. Technol. 16(5), 1422–1433 (1979). [CrossRef]
16. Z. Zhang and J. T. Yates Jr., “Band bending in semiconductors: chemical and physical consequences at surfaces and interfaces,” Chem. Rev. 112(10), 5520–5551 (2012). [CrossRef] [PubMed]
17. L. Kronik and Y. Shapira, “Surface photovoltage spectroscopy of semiconductor structures: at the crossroads of physics, chemistry and electrical engineering,” Surf. Interface Anal. 31(10), 954–965 (2001). [CrossRef]
18. L. Kronik and Y. Shapira, “Surface photovoltage phenomena: theory, experiment and application,” Surf. Sci. Rep. 37(1-5), 1–206 (1999). [CrossRef]
19. T. D. Veal, P. H. Jefferson, L. F. J. Piper, C. F. McConville, T. B. Joyce, P. R. Chalker, L. Considine, H. Lu, and W. J. Schaff, “Transition from electron accumulation to depletion at InGaN surfaces,” Appl. Phys. Lett. 89(20), 202110 (2006). [CrossRef]
20. L. R. Bailey, T. D. Veal, P. D. C. King, C. F. McConville, J. Pereiro, J. Grandal, M. A. Sánchez-García, E. Muoz, and E. Calleja, “Band bending at the surfaces of In-rich InGaN alloys,” J. Appl. Phys. 104(11), 113716 (2008). [CrossRef]
21. C. Kuo, S. Lin, K. Chang, H. Shiu, and L. Chang, “Is electron accumulation universal at InN polar surfaces?” Appl. Phys. Lett. 98(5), 052101 (2011). [CrossRef]
