About the 1996 Electronic Materials Conference: Wednesday Afternoon Sessions (June 26)

Session I: SiC Processing

1:30PM, I1

"High-Voltage [[alpha]]-SiC pn Junction Diodes Formed by Hot Implantation of N⁺ into p-Type Epilayers:" T. KIMOTO, N. Inoue ⁽¹⁾, H. Matsunami, T. Nakata, M. Inoue, Department of Electronic Science and Engineering, Kyoto University, Sakyo, Kyoto 606-01, Japan, Ion Eng. Research Inst. Corp., Tsuda, Hirakata 573-01, Japan

Silicon carbide (SiC) has been recognized as a vital wide bandgap material for high-power and high-temperature devices, owing to its outstanding properties and the availability of "device-quality" epilayers. Ion implantation is an important technique of selective doping for SiC, in which the diffusion coefficients of impurities are extremely low. In particular, characteristics of pn junctions formed by implantation is a key factor which determines the device performance. In this paper, high-voltage -SiC pn junction diodes are fabricated using N⁺ implantation into p-type epilayers. The effects of implantation temperature are discussed.

Samples used in this study were 6H-SiC epilayers grown on off-oriented p-type 6H-SiC(0001) substrates by atmospheric-pressure chemical vapor deposition in a SiH4-C3H8-H2 system. The p-type epilayers with an acceptor concentration of 4x10¹⁶ cm^-3 were produced by B doping using B2H6 as a dopant source. N⁺ implantation was performed at room or elevated temperatures (500 or 800deg.C), followed by annealing in Ar at 1500deg.C. A box profile was obtained by triple implantation 140, 80, 30 keV), and the total dose was varied in the range of 3x10¹³~1x10¹⁶ cm^-2. Rutherford backscattering spectroscopy measurements revealed that the implantation-induced damage was significantly reduced by utilizing hot implantation.

The sheet resistance and electrical activation ratio of implanted layers were examined by the van der Pauw method. The reduction of sheet resistance could be achieved in the high-dose region (>10¹⁵ cm^-2) by utilizing hot implantation at 500~800deg.C. The lowest sheet resistance was 550 / obtained by implantation at 500 or 800deg.C with an implant dose of 4x10¹⁵ cm^-2.

Mesa pn junction diodes were fabricated by reactive ion etching using CF4+O2 gases. The surface was passivated with thermal oxides, and ohmic contacts were annealed Al/Ti for both n- and p-SiC. The diode exhibited a high breakdown voltage of 615 V, which is almost the same as the ideal value (~620 V) calculated from the breakdown field (3.0x10⁶ V/cm) and doping concentration. The forward current can be clearly divided into recombination and diffusion current components. The diffusion current becomes dominant with increasing temperature, due to the rapid increase of the intrinsic carrier concentration. The reverse leakage current was very low, 2.7x10^-8 A/cm² at -100 V (room temperature). The reverse current at room temperature was proportional to V^1/2, indicating that generation current in the depletion layer is dominant. Although the dependence of breakdown voltage on implantation temperature was very small, remarkable difference was observed in leakage current. We made the histograms of leakage current at -100 V for the diodes formed by implantation at room temperature and 800deg.C (implant dose = 4x10¹⁵ cm^-2). The histogram for 800deg.C-implanted diodes demonstrated a much lower median value of leakage current density with narrower distribution. The results indicated that hot implantation is a promising technique for SiC device fabrication.

1:50PM, I2

"Free Electron Laser Annealing of Silicon Carbide:" HIDEAKI OHYAMA, Toshiji Suzuki, Kazuhisa Nishi, Tuneo Mituyu, Takio Tomimasu, Free Electron Laser Research Institute, Inc., 4547-44 Tsuda, Hirakata, Osaka, 573-01 Japan

2:10PM, I3+

"Annealing of Ion-Implanted 6H-SiC by Microwave Processing:" JASON GARDNER, Mulpuri V. Rao, Y.L. Tian, Department of Electrical and Computer Engineering, George Mason University, Fairfax, VA 22030

In the fabrication technology for silicon carbide (SiC) devices, the annealing process after ion implantation is important in order to remove the crystalline damage and activate the implanted atoms by moving them onto substitutional sites. However, extremely high temperatures (> 1300deg.C) are required for activating dopants using a conventional annealing technique. It is desirable to lower the annealing temperature in view of the actual device fabrication. The direct excitation of the lattice vibration with intense ultrashort-pulsed laser may induce the reconstruction of disordered atoms and activating dopants even at room temperature. The free electron laser (FEL) has two main characteristics such as wavelength tunability and ultrashort pulse operation (ps) with intense peak power (MW). Therefore, FEL annealing of silicon carbide was investigated using it tuned to the strong absorption peak at 12.6 um which corresponds to the Si-C stretch mode at room temperature.

As the first step of the present study, we have attempted the FEL annealing of amorphous silicon carbide (a-SiC) thin films at room temperature. An a-SiC film with 1.6 um in thickness was deposited on semi-insulating Si substrates by conventional rf-sputtering technique. The amorphous phase was obtained at the substrate temperature below 500deg.C. The present experiments were performed using linac pumped FEL. The FEL delivered macropulses with a length of 20 us and at a repetition rate of 10 Hz. Each of these macropulses consisted of ultrashort (10 ps) micropulses with a 45 ns spacing and maximum power of 1 MW. The original laser beam passed the mechanical aperture and was focused down to about 500 um ø through the ZnSe optical lens. The average power on the sample was about 16 mW.

FT-IR measurements of the a-SiC film showed that the broad peak of absorption spectrum became sharp after the irradiation of FEL. X-ray diffraction measurements revealed that a peak related to the SiC crystal for the irradiated sample. These measurements suggested that the crystallization of the a-SiC occurred at room temperature. Recrystallization of damaged layer will be discussed with the electronic properties of ion-implanted SiC crystal.

2:30PM, I4

"Doping of 3C-SiC by Implantation of Nitrogen at High Temperatures:" R. LOSSY, W. Reichert, E. Obermeier, W. Skorupa*, Technical University Berlin, TIB 3.1, Gustav-Meyer-Allee 25, 13355 Berlin, Germany; *F.Z. Rossendorf, POB 510119, 01314 Dresden, Germany

Heteroepitaxially grown 3C-SiC is particularly useful for the design of microsensors operated at high temperatures. Structural parts are fabricated in the substrate using silicon micromachining whereas transducer elements are made from 3C-SiC on top. Temperatures during processing of SiC should be well below the melting point of silicon in order to maintain the quality of the silicon for the micromachining process. Annealing temperatures after room temperature implantation, depending on the activation required, will exceed these values. Therefore, implantation at elevated temperatures without post-annealing needs to be employed.

The crystal damage induced during hot implantation was deduced from Channelling Rutherford Backscattering Spectroscopy (RBS/C) and results reveal that at temperatures as high as 800deg.C the damage is already at a very low level (4%). Thus, the investigation concentrated on the temperature range from 800deg.C to 1200deg.C.

Implantation profiles were measured by SIMS, using a Cameca IMS-4f instrument. Single energy implants were used to enable unambiguous comparison of the evolving doping profiles. The profiles were characterized using Pearson distributions. For implantation at room temperature SIMS measurements are in agreement with results obtained from Monte Carlo-Simulation using the TRIM code. High temperature implantation at T (&) gt; 800 deg.C reveals only moderate spreading of the profiles during implantation.

Resistivity and Hall measurements were made using van der Pauw test structures. Ohmic character of the sputtered and annealed contacts was verified by I-V measurements. The carrier concentration and mobility in the implanted layers was determined in the temperature range between 80 K and 723 K. The electrical activation at room temperature varied from 15% to 45% for the implantation temperatures investigated (800 - 1200deg.C). The mobility of the samples increased with measuring temperature and are situated around 20 cm²/Vs.

Comparison of the hot implanted samples and conventional annealing after cold implantation reveal that an electrical activation of approx. 15% is already achieved at 800deg.C for hot implantation whereas the conventional method requires 1100deg.C for the same amount of activation.

2:50PM, I5

"DLTS Studies on Bulk and Interfaces Traps in 6H & 4H-SiC:" M. KOTHANDARAMAN, B.J. Baliga, Power Semiconductor Research Center, North Carolina State University, Raleigh, NC 27695

There is very limited data available on the nature of levels found in SiC as-grown epitaxial layers or those induced by processing [1]. In this work, we report for the first time, the observation of a bulk deep level in as-grown 4H-SiC epitaxial layers and of surface defects produced due to residual damage from etching 6H-SiC using two methods.

Bulk traps in as-grown 6H-SiC and 4H-SiC epitaxial layers: DLTS measurements were performed using Titanium Schottky barrier diodes fabricated on as-grown 6H and 4H-SiC N-type epitaxial layers with doping concentration 2x10¹⁶ and 8x10¹⁵ cm^-3, respectively. The bias conditions for the DLTS experiment were chosen such that the surface traps were eliminated. The DLTS spectra obtained on 4H-SiC sample revealed one discrete peak with NT = 1x10¹³ cm^-3, EC-ET = 0.52 eV and = 4.36x10^-19 cm² whereas no such peak was obtained in the case of the 6H-SiC sample.

Residual damage from etching 6H-SiC: Reactive Ion Etching (RIE) is a popular technique for etching trenches in SiC [2]. Another technique that has been reported is the amorphization method [3] in which Si-C bonds are broken by high dose ion implantation to produce an amorphous layer and this layer is subsequently removed using a wet etch. In this work, residual damage from etching 6H-SiC using RIE and the amorphization method was compared using DLTS. Unmasked (blanket) RIE was done using SF6 and O2 on a 6H-SiC n-type epitaxial layer with a doping concentration of 2x10¹⁶ cm^-3 at 200 W, 50 mTorr, 22.5deg.C, % O2 = 50% with a total flow rate of 10 sccm till about 0.4 um of the epitaxial layer is removed. The amorphization implant was done at 50, 130, 200, 300 & 400 KeV successively at a dose of 1x10¹⁵ cm^-3. The amorphized region was then etched using a wet etch (1 HF:1 HNO3) at 50deg.C to remove 0.4 um. Titanium Schottky barrier diodes were then fabricated on these surfaces. A standard (unetched) sample was also included for comparison.

DLTS measurements on the amorphized/etched sample showed a deep level located at EC-ET = 0.7 eV with a concentration of 6 x 10¹⁵ cm^-3 plus a number of peaks closer to the conduction band with lower trap concentrations. The spectrum of the RIE and standard sample showed a low concentration of broadened peaks suggesting the presence of a band of states. The magnitude of the DLTS peaks seen in the RIE sample was lower than that of the standard sample indicating a superior interface. A depth profiling of these levels showed that the traps in the RIE and standard samples were surface traps whereas those in the amorphized/etched sample extended deeper into the bulk as expected from the tails in the implant profiles.

In conclusion, the presence of a deep level in as-grown 4H-SiC epitaxial layer has been detected for the first time. The study on process induced defects indicates that RIE is superior to the amorphization method for making trenches in 6H-SiC. This study also suggests that RIE can be used as a surface cleaning technique for SiC prior to fabricating Schottky barrier diodes.

____________________________________

References:

[1] M.G. Spencer, "Properties of Silicon Carbide," edited by G. L. Harris, INSPEC, the Institution of Electrical Engineers, 1995.

[2] P.H. Yih, A.J. Steckl, "Residue free Reactive Ion etching of Silicon Carbide in fluorinated plasmas. II.6H-SiC," J. Electrochem. Soc., v. 142, pp. 312-9, 1995.

[3] D. Alok, B.J. Baliga, "A novel method of etching trenches in silicon carbide," J. Electron. Mater., v. 24, pp. 311-4, 1995.

3:30PM, I6

"EBIC Measurements of Diffusion Lengths in Silicon Carbide:" R. RAGHUNATHAN, B.J. Baliga, Power Semiconductor Research Center, Box 7924, NCSU, Raleigh, NC 27695

Introduction: Diffusion lengths (L) and lifetimes ( ) of charge carriers in a semiconductor determine many important characteristics of the devices under operation. Although silicon carbide has emerged as a highly successful candidate for high temperature, high power and high frequency applications [1], there has been very little literature available on diffusion lengths and lifetimes in SiC and particular on variation of these parameters with temperature. A. M. Strel'chuk [2] reported ^[[tau]] and L values for 6H-SiC and their variation with temperature. However, diffusion lengths in this work were calculated using the photocurrent method. The disadvantage of using light source for generating carriers is the poor resolution of such a technique, specially while dealing with very small diffusion lengths as is the case in SiC. In this work, we report the use of the Electron Beam Induced Current (EBIC) technique for determining the minority carrier lifetimes in two polytypes of SiC (4H- and 6H-SiC) for both n and p-type material.

Advantages of EBIC: One of the primary advantages of EBIC is its high resolution which is determined by the radius of the generation volume [3] in the semiconductor created by the incident electron beam (~0.05 um). Secondly, this technique is simple and requires no high temperature processing as sample preparation consists only of Schottky barrier metal deposition thereby allowing us to characterize the material without altering its properties.

Methodology: Carriers generated by the electron beam of the SEM in the vicinity of the junction diffuse to the depletion region where they are collected and consequently induce a current in an external circuit. EBIC variation with distance should follow the diffusion equation including the effect of surface recombination velocity ( ):

I = Io x^-aexp(-x/L); where = 0 for vs = 0 and =3/2 for vs =*

Hence the inverse of the slope of a plot of ln(I/Io) versus x should give us the diffusion length after accounting for .

Experimental Procedure: The setup was designed and interfaced with the PC in order to allow us to digitize the data. The unique capability of this setup is its ability to measure diffusion length at different temperatures with a heating stage that has been specially configured for our SEM. In this work, Schottky barrier diodes were fabricated using a shadow mask with sequential evaporation of Ti (1000 Å) and Al (1000 Å). Blanket evaporation of Ti/Al layer was done on the heavily doped substrate for backside ohmic contact. The material used for the fabrication of these diodes were nitrogen doped 6H-SiC (1.1x10¹⁶) and 4H-SiC (1.4x10¹⁶); and Al doped 6H-SiC (6.0x10¹⁵) and 4H-SiC (9x10¹⁵) epitaxial layers obtained from CREE Research Inc. Measurements were made on all four structures and a variation of the minority carrier diffusion length with temperature was studied.

Experimental Result: Diffusion lengths in 4H- and 6H-SiC were extracted using various values of . It is important to use low energies for the electron beam so that the generation volume is small in order to obtain high resolution for the measurement. Secondly, since SiC is anisotropic it is essential to generate carriers close to the surface so that we measure only the lateral diffusion length. Measurements made in order to study the variation of Lp in 6H- and 4H-SiC for accelerating voltage of 4, 5 and 6 kV show that it is not sensitive to variations in beam energy. Hence for our measurement purposes was assumed to be 0 and an accelerating voltage of 5 kV was used. The extracted value of Lp for n-type 6H-SiC was 0.39 um and 1.4 um for n-type 4H-SiC. Similar measurements were made for p-type 6H- and 4H-SiC. Ln for 6H-SiC was calculated to be 1.19 um and for 4H-SiC was found to be 1.05 um. The variation of Ln and Lp with temperature was studied for the two polytypes and was found to be non-monotonic as also observed by Strel'chuk [2] for n-type 6H-SiC grown in Russia.

_____________________________________

References:

[1] `Properties of Silicon Carbide' ed. by Gary L. Harris, an INSPEC publication.

[2] A. M. Strel''chuk, Semiconductors, 29, (7), 1995, pp. 614.

[3] H. J.Leamy, J. App. ., 53 (6), 1982, pp. R51.

4:10PM, I7

"Impurity Conduction and Dopant Activation Energy in Heavily Doped n-Type Bulk SiC:" A.O. EVWARAYE, S.R. Smith, W.C. Mitchel, M.D. Roth, Wright Laboratory, WL/MLPO, W-PAFB, OH 45433-7707, Department of Physics, University of Dayton, Dayton, OH 45469-2314, University of Dayton Research Institute, 300 College Park, Dayton, OH 45469

The electronic properties of heavily doped n-type 6H, 4H and 15R SiC samples have been studied by temperature dependent Hall effect and thermal admittance spectroscopy. Nitrogen was the n-type dopant in all cases. Hopping conduction was observed in all samples with electron concentrations above 1x10¹⁷ cm^-3 by both experiments. This conduction mechanism is characterized by the disappearance of the Hall voltage and a thermally activated resistivity with an activation energy significantly lower than the thermal activation energy of the donor level. In the 4H and 15R samples the hopping conduction activation energy, , was found to be in the range 3-5 meV while that for the 6H samples was in the range 9-11 meV. This conduction mechanism prevented the measurement of dopant activation energy by thermal admittance spectroscopy experiments on samples with carrier concentrations above 1x10¹⁷ cm^-3, but two energies, Ec - 0.053 and Ec - 0.1 eV were observed for the nitrogen donor in less heavily doped samples. Both hopping and band conduction were observed by Hall effect measurements. Dopant activation energies determined by Hall effect agreed with those measured by thermal admittance spectroscopy for the lightly doped samples. In the more heavily doped samples the nitrogen activation energies were significantly lower than the commonly accepted values for all three polytypes studied. Activation energies as low as Ec - 0.025 eV were observed. A dependence of activation energy on doping concentration has been observed in other semiconductors and is attributed to a broadening of the impurity level and a merging of the excited states with the conduction band.

4:30PM, I8+

"Electrical Properties of Metal-Diamond-Like Nanocomposite (Me-DLN) Contacts to N-Type 6H SiC:" K.J. SCHOEN¹, J.M. Woodall, A. Goel, C. Venkatraman, School of Electrical and Computer Engineering and the Engineering Research Center for Collaborative Manufacturing, Purdue University, West Lafayette, IN 47907, Advanced Refractory Technologies Inc., Buffalo, NY 14207

Reliable high temperature devices are of interest for high voltage (devices experience self-heating) and high temperature applications. SiC is a promising high temperature semiconductor because of its wide bandgap (3.0 eV for 6H-SiC) and thermal stability in air. Me-DLN is of interest as an electronic material because it has variable resistivity (10^-4 - 10¹⁴ [[Omega]]cm) and excellent thermal stability in air (Pt-DLN is stable in air up to 800deg.C). The basic amorphous structure of DLN consists of two random interpenetrating networks of C and Si which are stabilized, respectively, by H and O (a:C-H, a:Si-O). The two networks also mutually stabilize each other. DLN may be codeposited with a metal to form Me-DLN material with variable electrical properties. Therefore, a microelectronics system combining SiC and Me-DLN has potential advantages based on the capability to withstand high temperatures in air.

In this work, Tungsten-DLN has been deposited onto a 6H-SiC n-type epitaxial layer grown on a Si face heavily doped n-type SiC substrate. The epitaxial layer doping is 1.5x10¹⁶ cm^-3 and the layer thickness is 3.0 um. The W-DLN was deposited onto the SiC sample through a shadow mask to form circular contacts of 250 um diameter. Prior to W-DLN deposition, the SiC sample was degreased, dipped in HF, and rinsed in DI. Silver paste was used to mount the sample onto a Si wafer and served as the large-area backside contact.

The as-deposited W-DLN circular contacts' I-V and C-V characteristics were measured at room temperature. The I-V measurements indicated diode rectification with a non-abrupt breakdown at about 100 V reverse bias. The ideality factor of the diodes was 1.4. The C-V measurements showed the typical one-sided junction capacitance, and a Vbi of 1 V or a [[phi]]Bn of 1.2 eV. The sample was then annealed in a rapid thermal annealer for 5 minutes at 500deg.C with a N2 ambient. After annealing, the I-V and C-V room temperature measurements were repeated. The I-V measurements showed approximately a factor of 8 increase in current, however, the 1/C²-V plot showed some non-linearity that was not present prior to annealing. Based on these measurements, W-DLN contacts appear to be a candidate contact technology for high temperature rectifying and ohmic contacts to SiC with a potential capability of surviving in air.

4:50PM, I9

MOCVD Grown AlN on 6H-SiC for MIS Device Application: RONGXIANG HU, C.C. Tin, M.E. Zvanut², V. Madangarli, T.S. Sudarshan, Department of Physics, Auburn University, Auburn, AL 36849; Department of Physics, University of Alabama at Birmingham, Birmingham, AL 35294; Department of Electrical and Computer Engineering, University of South Carolina, Columbia, South Carolina 29208

SiO2 insulator on SiC is known to contain large number of interface states. While a lot of research is being carried out to improve the quality of SiO2 on SiC, work is also being performed on alternate dielectric materials. One alternative way of realizing good quality MIS devices is to use AlN as an insulating material instead of SiO2. AlN is thought to be a good candidate because of its relatively small lattice mismatch (~1%) with SiC. Low interface state densities in Al/AlN/SiC MIS structures have been demonstrated by using MBE grown AlN on 6H-SiC [1]. In this paper, we are going to present the results of MOCVD grown AlN on 6H-SiC with improved characteristics. A comparison of various approaches in MOCVD growth of AlN on 6H-SiC, and the resultant C-V, DC and fast ramped pulse breakdown characteristics of MOCVD AlN films on 6H-SiC will be discussed.

AlN films were grown epitaxially by both atmospheric-pressure-MOCVD (AP-MOCVD) and low-pressure-MOCVD (LP-MOCVD) on 6H-SiC (0001) Si face polished 3.5deg. off toward (1120). High purity NH3 gas and trimethylaluminum were used as precursors. The 6H-SiC substrates were n⁺ type with an n-type epitaxial layer 5-10 um thick and n-type doping concentration of 10¹⁶-10¹⁷ cm^-3. Prior to AlN epitaxial growth, the substrates were cleaned by organic solvents and DI water, etched by buffer oxide etchant to remove the native oxide, and then baked in H2 atmosphere at 1200deg.C for 20 minutes. The MOCVD reactor was modified to reduce the parasitic reaction of NH3 and TMAl at low temperature. The AlN growth temperature was 1200~1300 deg.C.

The high frequency C-V characteristics of AlN on 6H-SiC showed greatly reduced flat band voltage and hysteresis compared to SiO2 on 6H-SiC. AP-MOCVD grown AlN films on 6H-SiC showed a large leakage current, but LP-MOCVD grown AlN showed a greatly reduced leakage current. The leakage current was about 1 nA for a film thickness of about 1200 Å and a contact area of 5x10^-3 cm². In some samples, the forward leakage current had a sudden increase as forward biased voltage increased to about 3 volts, which indicated the AlN film was susceptible to breakdown at this voltage. Films with various thicknesses were also studied with the breakdown field estimated around 2~2.5x10⁵ V/cm.

____________________________________

[1] C. J. Harris, M. O. Aboelfotoh, R. S. Kern, S. Tanaka and R. F. Davis, presented at the International Conference on Silicon Carbide and Related Material, 1995, Japan.