Magnetic and Structural Properties of Co, Cr, V ion-implanted GaN
J.S. Lee, D.J. Lim,
Z.G. Khim, Y.D. Park[a])
School of Physics
Seoul National University
Seoul 151-747 Korea
S.J. Pearton[b])
Department of Materials Science and Engineering
University of Florida
Gainesville, FL 32611
S.N.G. Chu
Agere Systems
Murray Hill, NJ 07974
ABSTRACT
We report on the magnetic and structural properties of epitaxial metal organic chemical vapor deposition (MOCVD) grown p-GaN:Mg/Al2O3 implanted with Co, Cr, and V ions at varying high dosages at 350°C. Magnetic and structural properties were investigated after a short anneal at 700°C to remove implantation damage. Magnetic properties determined from superconducting quantum interference device (SQUID) magnetometer measurements indicate ferromagnetic-like ordering for Co and Cr doped samples up to 320°C, while V doped samples show paramagnetic behavior for all temperatures considered. For all samples studied, structural characterization techniques such as x-ray diffraction (XRD), high-resolution cross-sectional transmission electron microscopy (HRXTEM), and selected area diffraction pattern (SADP), indicate no second phases that may contribute to the magnetic properties measured. Transport measurements (resistivity as a function of temperature) reveal all samples to show insulating-like behavior.
INTRODUCTION
It has been more than two decades since researchers have shown that ferromagnetic transition metals grow epitaxially on semiconductor substrates. Since then, much have been learned about the fundamental physics of ferromagnetic systems at low dimensions as well as discovery of such phenomena as oscillatory coupling and giant magnetoresistance (GMR), which have had immediate technological impact. Perhaps most importantly, the demonstration of epitaxial growth of Fe on GaAs has led to development of ideas of utilizing spin as well as charge of carriers to develop novel devices and add new functionalities.[1] Although much progress have been made in realization of spintronic devices, one of the most critical elements in such a device, the ability to robustly inject a net spin population into a semiconductor channel, have been elusive.[2] Realization of a new class of materials, dilute magnetic semiconductors (DMS), showing ferromagnetic ordering at near or above room temperature, have received much experimental and theoretical attention. Several groups have reported Tc’s near or above room temperature for semiconductors and oxides doped with high concentrations of transition metals.[3] Of particular, wide band gap III-V semiconductor GaN have received a considerable amount of attention since the prediction of a TC above room temperature[4] for p-type (p = 3 x 1020 cm-3) (Ga,Mn)N doped with Mn (5 at.%). Several groups have reported ferromagnetic-like behavior near or above room temperature for (Ga,Mn)N[5],[6], and as high as 940 K reported by Sonoda et al.[7] Recently, theoretical work by Sato and Katayama-Yoshida have predicted various magnetic properties ranging from spin-glass –like to ferromagnetic for GaN incorporating various concentrations of Cr, Co, Fe, Mn, Ni, and V based on a local spin-density approximation which assumed that Ga atoms were randomly substituted with the magnetic atoms and did not take into account any additional carrier doping effects.[8]
Recently, several groups have reported above room temperature TC for (Ga,Cr)N. S.E. Park et al. reported room temperature ferromagnetism (TC = 280 K) for bulk single crystal Cr-doped GaN prepared by sodium flux growth method.[9] Epitaxial (Ga,Cr)N prepared by ECR molecular beam epitaxy with TC > 400 K was reported by Hashimoto et al.[10] Ferromagnetic ordering in low temperature molecular beam epitaxy (LT-MBE) Cr doped GaAs DMS has also been reported.[11] Although Co doped TiO2 and ZnO have reported TC’s above 300 K, we are unaware of reports of Co incorporated GaN DMS thus far. For V-doped, we are not aware of any reports of DMS material systems. As in case of all DMS preparation, possible contributions from magnetic second phases are of concern. For example, in the well-studied (Ga,Mn)As DMS system, for high substrate growth conditions, ferromagnetic intermetallic (TC ~ 300 K) MnAs clusters form. Binary intermetallic phases for the transition metals Cr, Co, and V in GaN DMS preparation ranges from superconducting to anti-ferromagnetic in magnetic behavior. Of note, are CrN,[12] anti-ferromagnetic which Neél Temperature of 273 K (prepared by MBE and annealing at 800° C) – which is similar to magnetic transition reported by S.E. Park et al. (prepared by sodium flux method at 750° C); CoN,[13] Pauli paramagnetic for all phases (CoN, δ-Co2N, and γ-Co3N prepared by annealing reactive ion sputtering); and ferromagnetic phase - Cr2N.
EXPERIMENT
Ion-implantation of transition metals into various semiconductors has been effective in surveying possible DMS systems exhibiting ferromagnetic ordering.3 For the samples reported here, ion-implantation of Cr, Co, and V were conducted similarly to that reported previously elsewhere[14], [15] . The epitaxial p-GaN:Mg films(approximately 2 microns thick) were prepared by metal organic chemical vapor deposition (MOCVD) at a growth temperature of 1010 °C .The Mg acceptor level is known to be 170 meV above the valence band and only fractional amount of incorporated dopants are expected to contribute holes at room temperature, producing an effective hole concentration of ~3x1017 cm-3.
Co, Cr, and V ions were implanted at doses of 3 x 1016 cm-2 and 5 x 1016 cm-2 with energy of 250 keV with the film held at ~350° C to minimize amorphization during the process.[16],[17] Although depth profiling of Co, Cr, and V implanted GaN have not been performed, from previous secondary ion mass spectrometry (SIMS) measurements of implanted Mn ions in GaN at the same energy conditions, the projected range of these ions is roughly 1500 Å and follows a Gaussian profile that is slightly skewed towards the surface due to associated sputtering that occurs at such high doses. The peak concentration of implanted ion is found to be approximately 3 and 5 at. % respectively from the SIMS measurements, using a low dose (1015 cm-2) as a calibration standard. After ion-implantation, samples were annealed for 5 minutes at 700° C to remove as much implantation damage as possible without formation of magnetic secondary phases.
EXPERIMENTAL RESULTS
The structural properties of Co, Cr, and V ion-implanted p-GaN films have been examined by x-ray diffraction (XRD), selective area diffraction pattern (SADP), and high resolution cross-sectional transmission electron microscopy (HRXTEM). Figure 1 depicts a representative XRD θ-2θ scan following implantation and annealing processes, GaN:Co (5 x 1015 cm-2). All peaks correspond to expected GaN epilayer and sapphire substrate, and possible unidentified peaks corresponding to second phases were not observed. Although such systems as (Ga,Mn)As and (In,Mn)As show a distinct change in lattice parameter linear to percent Mn incorporated,[18] for the two differing doses, such a relationship in Co, Cr, V ion-implanted GaN was not clearly found. As ion-implantation and subsequent short anneal would incorporate transitional metals only in a fraction of the original GaN epilayer, the resulting XRD scan is dominated by undoped epilayer and substrate, although the peak corresponding to GaN (0002) show a slight shouldered peak to its right which may be attributed to the DMS phases.
The absence of second phases is also supported by selected area diffraction patterns (SADP) and cross-sectional transmission electron microscopy (TEM). SADP of the sample before and after implantation do not show much difference except for expected broadening of the spots due to implantation damage. There are no unaccountable diffraction patterns compared to before and after implantation. Figure 2 depicts a representative Cr ion-implanted GaN before and after implantation. Cross-sectional TEM micrographs reveal no observable second phases other than the expected Cr-implantation and subsequent anneal processes leaving residual damage in the form of dislocation loops.
Although within the detectable limits of the characterization techniques applied no second phases were observed, we cannot completely rule out existence of clustering, especially those phases that may contribute to the magnetic properties. Such undetected clusters if ferromagnetic would be most likely super-paramagnetic, meaning that due to the small size of the clusters, thermal energy would randomize the magnetization of individual clusters. Such behavior can be discerned by such measurement techniques as zero-field cooled (ZFC)/field cooled (FC) measurements and/or AC susceptibility measurements. In all metallic super-paramagnetic systems such as Co and Ag where two elements are immiscible, much of the structural and magnetic properties have been studied. Xiong et al. reports of co-sputtered Co/Ag granular systems to form Co clusters as small as 2 nm which were discerned from XRD and TEM measurements.[19] It is also worthy of note that for the smallest Co clusters that Xiong et al. reports observation of the extraordinary hall effect, which effects are difficult to observed in low carrier systems such as wide band gap semiconductors.
Magnetic measurements on ion-implanted samples were conducted using Quantum Design MPMS SQUID magnetometer. As DMS systems in general have low saturation magnetization (MS) values, sample preparation and sample area were carefully maintained. Magnetization as function of applied field at 5 K is plotted in figure 3 (diamagnetic contributions from the sapphire substrate has been carefully subtracted in all magnetic data). For Co and Cr ion-implanted samples, ferromagnetic-like loops were observed. For two dosages, magnetic saturation was not observed for applied fields up to 5 Tesla for both Co and Cr implanted samples. In other DMS systems with ferromagnetic ordering, for samples which are on the insulating side of the metal-insulator transition, technical saturation of magnetization was found to be difficult.18 The larger paramagnetic background for higher dose samples is consistent with a carrier mediated model of ferromagnetism in DMS systems at low carrier regimes.[20],[21] Such a material system as wide band gap GaN DMS, at low temperatures, carriers will be localized to the magnetic impurities. If these bound carriers do not overlap with other polarized bound carriers to form a percolation network, ferromagnetic ordering is not maintained. Thus, samples with higher impurity concentrations show greater paramagnetic background as there are more bound carriers to magnetic impurities. If the magnetic properties originate from an undetected ferromagnetic, ferrimagnetic, or anti-ferromagnetic phase, we would expect the higher dose samples to show magnetization approach saturation faster than lower dose samples. For all V implanted samples, magnetization as a function of applied field measurements indicate the resulting samples to be paramagnetic as magnetic susceptibility was found to be constant.
Magnetization as function of temperature is plotted in figure 4. Co and Cr implanted samples magnetization as a function of temperature indicates many multiple exchange interactions in that its decay cannot be easily fit to classical description of ferromagnetism, again in agreement with current theories concerning DMS systems with low carrier concentrations. Although Co implanted samples show the highest magnetic moment at the lowest temperatures considered, Cr implanted samples retain magnetization up to the measured temperature of 320 K. This observation is consistent with epitaxially prepared (Ga,Cr)N magnetic properties observed by Hashimoto et al. who have reported TC’s higher than 400K.10 In our measurements, the reported absolute magnetization per volume (emu/cc) are lower bounds as DMS layer thickness was conservatively estimated as the projected range of the implantation and annealing process. In comparing Co implanted samples to Cr implanted (with magnetization above room temperature) and V implanted (with paramagnetic properties for all temperatures considered), Co magnetization drops below Cr implanted samples around 110 K and approached the values for V implanted samples around 280 K, indicating possible presence of multiple complex exchange interactions (including spin-glass like mechanisms) for Co implanted samples; thus, quoting a TC or even θC (from Curie-Weiss relationship incorporating multiple exchange mechanisms) is difficult for these samples. Along with magnetization as a function of temperature measurements, ZFC/FC measurements were made on representative samples and the data did not indicate any blocking temperatures that can be associated with super-paramagnetic behavior arising from undetected magnetic second phase clusters.
Magnetic properties- large paramagnetic background at low temperatures and complex magnetization as a function of temperature relationship- indicate the samples to be most likely non-metallic based on theories of carrier mediated magnetic ordering in DMS systems and DMS systems with low carrier densities. Transport measurements as a function of temperature were conducted with conventional four point probe geometry with In soldered contacts. Each contact was checked to be ohmic by an I-V measurement. Figure 5 shows normalized resistivity as a function of temperature. We did not observe any drastic changes in slope corresponding to magnetic transition as reported by S.E. Park et al. for bulk single crystal (Ga,Cr)N. All samples show semiconducting behavior consistent with wide band gap material with low carrier concentrations. Carrier concentration could not be easily and accurately determined by transport techniques (Hall measurements) due to the fact that implantation creates a Gaussian profile leaving complex parallel channels of differing doping concentrations.
SUMMARY
Structural and magnetic properties of Co, Cr, and V ion implanted p-GaN epitaxial films were studied. Structural characterization revealed that the ion-implantation and subsequent annealing process induced expected damage in forms of dislocation loops and defects, but no formations of second phases were observed. Magnetic properties showed ferromagnetic-like ordering at low temperatures for Co and Cr implanted samples, while V implanted samples show paramagnetic behavior. Our results partly agree with predictions made by Sato and Katayama-Yoshida for GaN DMS systems for various transition metals. Cr implanted GaN show magnetization above 300 K from magnetization as a function of temperature measurements in agreement with Hashimoto et al. and S.E. Park et al, and as predicted by Sato and Katayama-Yoshida. Co implanted GaN show complex relationship between magnetization and temperature which partly agree with spin-glass like state predicted for (Ga1-xCox)N (0.05 ≤ x ≤ 0.25) by Sato and Katayama-Yoshida.
ACKNOWLEDGEMENTS
This work at SNU is partly supported by Samsung Electronics Endowment and KOSEF through the Center for Strongly Correlated Materials Research. We will also like to thank ReCOE at SNU in assistance with XRD measurements. The work at UF was partially supported by NSF and ARO.
FIGURE CAPTIONS
Figure 1 - A representative XRD θ-2θ measurement of GaN/sapphire (0001) ion implanted with 5 x 1015 cm-2 Co (corresponding to ~ 5 at.%) followed by short aneal is plotted. The inset is an expanded view near the peak corresponding to GaN (0002).
Figure 2 - A representative Cross-sectional TEM and SADP of GaN/Sapphire before (a) and after (b) Cr-implantation and subsequent anneal – ion implantation process leaves residual damage in the form of dislocation loops (as expected), but no detectable secondary phases.
Figure 3 - SQUID magnetometer measurements of magnetization (emu/cc) as a function of applied field (kOe) at T = 5K for a) Co ion-implanted p-GaN/Sapphire (0001) with 3 x 1015 cm-2 (~ 3%) and 5 x 1015 cm-2 (~ 5%) (inset is ± 10 kOe for 3%); b) Cr ion-implanted; c) V ion-implanted. Only V ion-implanted samples show a linear relationship between magnetization and applied field.
Figure 4 - SQUID magnetometer measurements of magnetization (emu/cc) as a function of temperature (H = 1 kOe) for Co, Cr, and V ion-implanted GaN with 5 x 1015 cm-2 dose.
Figure 5 – Normalized dc (I = 1 μA) resistance measured by
standard four point probe geometry using In soldered contacts as a function of
temperature for Co, Cr, and V ion-implanted GaN with 3 x 1015 cm-2
dose.
Figure 1 – J.S. Lee et al.
Figure 2 – J.S. Lee et al.
Figure 3 – J.S. Lee et al.
Figure 4 – J.S. Lee et al.
Figure 5 – J.S. Lee et al.
