Mat. Res. Soc. Symp. Proc. Vol. 693 © 2002 Materials Research Society

Measurement of the Effective Piezoelectric Constant of Nitride Thin Films and Heterostructures Using Scanning Force Microscopy B. J. Rodriguez1, D-J. Kim2, A. I. Kingon2, R. J. Nemanich1 Department of Physics1 and Department of Materials Science and Engineering2, North Carolina State University, Box 8202, Raleigh, NC, 27695-8202, USA ABSTRACT Piezoelectric properties of wurtzite AlN and GaN/AlN are investigated using scanning force microscopy (SFM). The magnitude of the effective longitudinal piezoelectric constant d33 of AlN and GaN/AlN thin films are measured and reported, and the d33 coefficients of these films are verified using an interferometric technique. Simultaneous imaging of the topography, and of the phase and magnitude of the piezoelectric strain is performed. Using a GaN film with patterned polarities, we demonstrate that polarity can be inferred from the phase image of the piezoelectric strain. We report d33=3±1 pm/V for AlN/SiC and 2±1 pm/V for GaN/AlN/SiC. Films grown by organo-metallic vapor phase epitaxy (OMVPE) on SiC, sputtered AlN films and films grown by molecular beam epitaxy (MBE) are characterized and compared. INTRODUCTION The piezoelectric properties of III-V nitrides are important for new heterostructure devices. As such, measurement of the piezoelectric properties of nitride thin films and heterostructures with nanometer scale resolution is of considerable interest for determining how defects and patterned structures affect the piezoelectric response (piezoresponse). Spontaneous and piezoelectric polarization-induced electric fields have been used to achieve a two dimensional electron gas (2DEG) at the interface of AlGaN/GaN heterostructures with sheet concentrations as high as 2x1013cm-2 in unintentionally doped high electron mobility transistors (HEMTs) [1]. In their wurtzite phase, InN, GaN and AlN are pyroelectric materials with spontaneous polarization along the [000 1 ] direction (PSP = -0.042, -0.034, and -0.090 C/m2) for InN, GaN and AlN, respectively [2]. Unlike ferroelectric materials, the orientation of this polarization cannot be changed by the application of an electric field. In GaN, the [0001] axis points from the Ga atom to the nearest neighbor N atom (cation to anion) in the positive z direction, and from Nface to Ga-face, since Ga will be on the top position of the {0001} bilayer. The orientation of spontaneous polarization is defined by convention such that the positive direction is also from cation to anion along the crystallographic c-axis. Since the sign of the calculated spontaneous polarization of the III-V nitrides is negative, this indicates that the spontaneous polarization points from the Ga-face towards the N-face in GaN (in the [000 1 ] direction). While spontaneous polarization is independent of strain, piezoelectric polarization is strain-induced [3]. The total polarization of a nitride layer is the sum of the spontaneous and the piezoelectric polarizations. In the case of a relaxed layer, the total polarization is equal to the spontaneous polarization. In the case of psuedomorphic AlGaN or InGaN on GaN, the magnitude of piezoelectric polarization increases with Al or In fraction. As shown by Ambacher, et al. [1] piezoelectric and spontaneous polarizations are parallel for tensile strain and are antiparallel for compressive strain. In the case of a Ga-face AlGaN/GaN heterostructure, the abrupt change of I9.9.1

polarization at the interface results in a positive sheet charge density, which is proportional to the difference between the polarizations in each layer. To compensate the positive sheet charge, free carriers in the GaN form a 2DEG near the interface. When an electric field is applied to a piezoelectric material, the material strains due to the converse piezoelectric effect. Several techniques [4-6], including scanning force microscopy (SFM) [7-13] have been used to measure these piezoelectric displacements. The SFM technique, which we refer to as piezoresponse force microscopy (PFM), is advantageous because the piezoelectric displacement can be measured at specific points on a sample. In addition, the tip can be rastered to generate images of the phase and of the magnitude of the piezoelectric response. As such, PFM can resolve nanometer variation in the piezoelectric properties of a sample, and is perfectly suited for investigating the piezoelectric properties near threading or screw dislocations, stacking faults and other defects, all of which affect device performance. Ideally with PFM, when an electric field is applied to a piezoelectric material, the tip accurately follows the piezoelectric motion; however, tip motion can be due to a combination of piezoelectricity, electrostriction, and electrostatic interactions between the tip and electric field [14]. Several studies have employed PFM to observe the piezoresponse of thin films. Measurements can be performed by applying a voltage between a conducting tip and a bottom electrode, or between a conducting tip in contact with a top electrode, and a bottom electrode. For example, Gruverman et al. [9] performed measurements of Pb(Zrx,Ti1-x)O3 (PZT) thin films by applying a voltage between a conducting tip and a back electrode. With this configuration, measurements are performed directly on the piezoelectric material, but the electric field generated by the PFM tip decreases quadratically from the tip [11]. Christman et al. [12] made measurements using a top electrode and demonstrated a sub-micron variation of ferroelectric and of piezoelectric properties in PZT capacitors. To our knowledge this paper represents the first attempt to study nitrides using PFM. The piezoresponse measured by PFM represents the piezoelectric strain induced in the film as a result of the application of an external field, and is proportional to the effective d33 of the film being studied. Several studies have used an interferometric technique to measure the piezoelectric coefficient d33 of nitride thin films [15,16]. C. M. Lueng et al. reported d33= 3.9±0.1 pm/V for AlN/Si(111) and d33= 2.65±0.1 pm/V for both GaN/AlN/Si(100) and GaN/AlN/Si(111) (all films were prepared by MBE) [15]. I. L. Guy et al. reported d33= 2.0±0.1 pm/V for polycrystalline GaN/Si(100) grown by laser assisted chemical vapor deposition (CVD); d33= 2.8±0.1 pm/V for single crystal GaN/SiC grown by hydride vapor phase epitaxy; and d33= 3.2±0.3 pm/V and d33= 4.0±0.1 pm/V for polycrystalline AlN/Si(100) grown by plasma assisted and laser assisted CVD respectively [16]. It should be noted that the interferometric technique is macroscopic, while PFM is a microscopic technique. EXPERIMENTAL DETAILS PFM measurements were performed in contact mode with a commercial AFM (PSI Autoprobe M5), a function generator (HP33120A) and a dual-phase lock-in amplifier (SR830) [13]. A schematic of the experimental setup is shown in figure 1. When a top electrode is employed, the electric field is well defined, and the excited region is large compared to the tip radius.

I9.9.2

Figure 1. Schematic of the PFM experimental setup. A modulation voltage is applied to the tip and to the top electrode, which causes the piezoelectric film to oscillate at the same frequency as the applied voltage. The relative polarity of the film is determined from the phase image; the oscillation of the film is either in-phase or out-of-phase with the driving voltage, corresponding to a Ga- or N-face film, respectively. Measurements are also performed without a top electrode by applying a voltage between the conducting tip and a back electrode. For the piezoelectric measurements described here, frequencies of 1kHz were employed when a top electrode was used. The 1kHz frequency is greater than most environmental noise frequencies and well below the tip resonances. The commercially obtained conducting tips (TM Microscopes) had a 1000Å layer of 0.1 Ω-cm p-type diamond over the Si tips. The conducting diamond tips are mounted on alumina squares and silver epoxy connects a wire to the chip. Diamond tips have been shown to be long lasting and suitable for scanning applications [17,18]. A double beam laser interferometric measurement, described elsewhere, is used to complement the PFM technique [19]. Samples were prepared by OMVPE on SiC, sputtering AlN on Si and on molybdenum, MBE on SiC, and plasma induced MBE (PIMBE) on sapphire. The PIMBE samples are of particular interest because films with lateral Ga-face (on AlN buffer) and N-face regions were prepared. Switching the polarity of the film is the only way to invert the direction of the spontaneous polarization; the polarity determines the location of the 2DEG in GaN/AlGaN/GaN heterostructures, and thus plays an important role in device design [1]. RESULTS AND DISCUSSION In order to determine the effective piezoelectric coefficient, once a piezoresponse magnitude image was obtained, a histogram was generated, and the peak value noted. In table 1, PFM results are summarized. Interferometry measurements were obtained on several samples, and there is a correlation between the PFM and the interferometry values. The PFM values are also in agreement with values reported in the literature. An average of similar GaN/AlN/SiC films yields d33= 2.0±1 pm/V, and for AlN/SiC, we observe d33= 3.0±1 pm/V.

I9.9.3

Sample Growth AFM Interferometry Electrode top/bottom Method d33 (pm/V) d33 (pm/V) Diameter (µm) AlN/Mo AlN/Si AlN/Si AlN/SiC AlN/SiC GaN/AlN/SiC GaN/AlN/SiC GaN/AlN/SiC GaN/AlN/SiC AlN/SiC AlN/SiC AlN/Si AlN/Si

Sputtered Sputtered Sputtered MBE MBE OMVPE OMVPE OMVPE OMVPE OMVPE OMVPE MBE MBE

1.6 ± 0.6 4.0 ± 1.8 6.0 ± 3.0 2.9 ± 1.3 4.6 ± 2.4 2.4 ± 0.2 2.1 ± 1.1 2.4 ± 1.3 2.8 ± 1.4 2.7 ± 1.3 2.8 ± 1.5 3.7 ± 1.9 1.9 ± 0.8

1.8 ± 0.2 3.2 ± 0.3

3.9 ± 0.2

2000 500 800 100 100 200 200 200 200 100 100 100 100

Table 1. PFM results. As discussed by Kholkin et al. [4] in reference to single beam interferometric measurements of piezoelectric samples bonded to a substrate, a potential source of error is wafer bending. We assume the displacement of the top electrode is equal to the entire piezoelectric displacement, but wafer bending may artificially inflate the piezoresponse, and would be a function of electrode diameter. Although not presented here, we observe an increase in the piezoresponse with increasing diameter above 200µm; however, as shown in table 1, measurements on electrodes of 100 and 200µm are consistent, indicating that wafer bending effects are insignificant for electrodes of 200µm diameter and below. In addition, the electrode may have an averaging effect upon the underlying vibration of the piezoelectric material. Charge screening near electrode/nitride interfaces and trapped charges may also affect the measurements and future studies are planned to explore these effects. To explore the imaging capability of this approach, a PIMBE-grown film with alternating concentric squares of Ga- and N-face GaN (the inner-most square is Ga-face) was examined. The Ga-face GaN was grown on an AlN buffer layer (~10 nm thick), while the N-face GaN was grown directly on the sapphire substrate [20]. In figure 2, (a) topography, (b) piezoresponse magnitude and (c) piezoresponse phase images are displayed. The scan was taken at 10kHz without a top electrode, and the images were acquired simultaneously. Note the contrast difference in the piezoresponse phase image (figure 2(c)); the Ga- and N-face regions have opposite polarity, thus the regions oscillate out-of-phase from one another. As presented in figure 2(c), the Ga-face ‘points down’ (i.e. ~ -60 degrees) and the N-face ‘points up’ (i.e. ~ +80 degrees) in the expected direction of spontaneous polarization. A histogram representation of the relative areas of differing polarity is shown in figure 3, demonstrating the phase difference to be 140 degrees. The fact that the oscillation is not exactly 180 degrees out-of-phase may be attributed to differences in the dielectric constants and the thicknesses of the layers. It is interesting to note that the piezoresponse magnitude is higher (as denoted by a lighter contrast in figure 2(b)) for the N-face GaN. We suggest that this results from the fact that while the spontaneous and the piezoelectric polarizations are antiparallel for both Ga-face and N-face GaN, the magnitude of the piezoelectric polarization for the N-face GaN is less than that of the Ga-face GaN. Thus, the total polarization is larger for the N-face GaN. Another possible I9.9.4

explanation is that the adsorption of compensating surface charge may be different for the Gaand N-face surfaces, possibly screening the piezoresponse in one case [21]. From figure 2(b) we also see a slightly higher piezoresponse at the boundary between Gaand N-face regions. It has been suggested [21] that the local charge density increases at an inversion domain boundary, possibly explaining the increased piezoresponse at the polarity boundary.

Figure 2. 25µm PFM scan of (a) topography, (b) piezoresponse magnitude and (c) piezoresponse phase for PIMBE GaN film with regions of alternating polarity.

Figure 3. Histogram of piezoresponse phase image (figure 2(c)). CONCLUSION We report a SFM method for measuring the piezoresponse of nitrides. PFM-measured d33 values for AlN/SiC and GaN/AlN/SiC agree with double beam interferometry and reported values. We also demonstrate PFM imaging of Ga- and N-face polarity GaN, a new and important result towards understanding the roles of spontaneous and of piezoelectric polarization in the III-V nitrides. We intend to employ PFM to study defects and inversion domains in mixed films, and to explore piezoresponse as a function of strain in heterostructures. I9.9.5

ACKNOWLEDGEMENTS The authors acknowledge helpful discussions with Dr. Alexei Gruverman. The authors would also like to acknowledge Drs. Robert Davis, Peter Miraglia, and Amy Roskowski for OMVPE samples, Jung Won-Cho and Dr. Jerome Cuomo for sputtered AlN samples, Vadim Lebedev for MBE AlN/Si, and Drs. O. Ambacher and R. Dimitrov for PIMBE GaN. This work is supported by the Office of Naval Research MURI on Polarization Electronics Contract No. N00014-99-10729, under the direction of Dr. Colin Wood. REFERENCES 1.

2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21.

O. Ambacher, J. Smart, J. R. Shealy, N. G. Weimann, K. Chu, M. Murphy, W. J. Schaff, L. F. Eastman, R. Dimitrov, L. Wittmer, M. Stutzmann, W. Rieger, and J. Hilsenbeck, J. Appl. Phys. 85, 3222 (1999). A. Zoroddu, F. Bernardini, P. Ruggerone, V. Fiorentini, Phys. Rev. B 64, 045208 (2001). F. Bernardini, V. Fiorentini, and D. Vanderbilt, Phys. Rev. B 56, R10024 (1997). A. Kholkin, C. Wuetchrich, D. V. Taylor, N. Setter, Rev. Sci. Instrum. 67, 1935 (1996). H. Birk, J. Glatz-Reichenbach, Li-Jie, E. Schreck, and K. Dransfield, J. Vac. Sci. Technol. B 9, 1162 (1991). Lie-Jie, C. Baur, B. Koslowski, and K. Dransfield, Physica B 204, 318 (1995). T. Hidaka, T. Maruyama, M. Saitoh, N. Mikoshiba, M. Shimizu, T. Shiosaki, L. A. Wills, R. Hiskes, S. A. Dicarolis, and J. Amano, Appl. Phys. Lett. 68, 2358 (1996). A. Gruverman, O. Auciello, and H. Tokumoto. Annu. Rev. Mater. Sci. 28, 101 (1998) P. Guethner and K. Dransfield, Appl. Phys. Lett. 61, 1137 (1992). A. Gruverman, O. Auciello, and H. Tokumoto, J. Vac. Sci. Technol. B 14, 602 (1996). A. Gruverman, O. Auciello, and H. Tokumoto, Integr. Ferroelectr. 19. 49 (1998). J. A. Christman, S.-H. Kim, H. Maiwa, J.-P. Maria, B. J. Rodriguez, A. I. Kingon, and R. J. Nemanich, J. Appl. Phys. 87, 8031 (2000). J. A. Christman, R. Woolcott, Jr., A. Kingon, R. J. Nemanich, Appl. Phys. Lett. 73, 3851. K. Franke, J. Besold, W. Haessler, and C. Seegebarth, Surf. Sci. Lett. 302, L283 (1993). C. M. Lueng, H. L. W. Chan, C. Surya, and C. L. Choy, J. Appl. Phys. 88, 5360 (2000). I. L. Guy, S. Muensit, and E. M. Goldys, Appl. Phys. Lett. 75, 4133 (1999). P. Niedermann, W. Haenni, N. Blanc, R. Christoph, and J. Burger, J. Vac. Sci. Technol. A 14, 1233 (1996). S. J. O'Shea, R. M. Atta, and M. E. Welland, Rev. Sci. Instrum. 66, 2508 (1995). D.-J. Kim, PhD. Thesis, North Carolina State University, 2001. R. Dimitrov, V. Tilak, M. Murphy, W. J. Schaff, L. F. Eastman, A. P. Lima, C. Miskys, O. Ambacher, and M. Stutzmann, Mat. Sci. Soc. Symp. 622, T4.6.1 (2000). M. Stutzmann, O. Ambacher, M. Eickhoff, U. Karrer, A. Lima Pimenta, R. Neuberger, J. Schalwig, R. Dimitrov, P. J. Schuck, and R. D. Grober (submitted to Phys. Stat. Sol.).

I9.9.6