Simulation and Experimental Results on the Forward J–V Characteristic of Al Implanted 4H–SiC p–i–n Diodes

Francesco G. Della Cortea, Fortunato Pezzimentia, , Roberta Nipotib
aDIMET—Faculty of Engineering, Mediterranea University of Reggio Calabria, Via Graziella—Feo di Vito, 89100 Reggio Calabria, ItalybCNR-IMM Via Gobetti 101, 40129 Bologna, Italy
Received 24 July 2007; accepted 30 September 2007Available online 5 November 2007

 

Corresponding author. Tel.: +39096 5875274; fax: +39096 5875463. E-mail addresses: francesco.dellacorte@unirc.it (F.G. Della Corte), fortunato.pezzimenti@unirc.it (F. Pezzimenti), nipoti@bo.imm.cnr.it (R. Nipoti).
0026-2692/$ - see front matter ©2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.mejo.2007.09.024


Abstract

In this work the forward J–V characteristics of 4H–SiC p–i–n diodes are analyzed by means of a physics based device simulator tuned by comparison to experimental results. The circular devices have a diameter of 350 mm. The implanted anode region showed a plateau aluminum concentration of 6 x 1019 cm-3 located at the surface with a profile edge located at 0.2 mm and a profile tail crossing the n-type epilayer doping at 1.35 mm. Al atom ionization efficiency was carefully taken into account during the simulations. The final devices showed good rectifying properties and at room temperature a diode current density close to 370 A/cm2 could be measured at 5V. The simulation results were in good agreement with the experimental data taken at temperatures up to about 523K in the whole explored current range extending over nine orders of magnitude. Simulations also allowed to estimate the effect of a different p+ doping electrically effective profile on the device current handling capabilities.

Keywords: 4H–SiC p–i–n diode; J–V characteristic; Ion implantation; Doping profile; Junction depth

 

1. Introduction

During the last decade, the efforts put in the development of sophisticated physical finite-element simulators of semiconductor devices has leaded these tools to gain a key role in the optimization of existing devices or in the design of devices based on new concepts. Even though this observation applies particularly well to the whole Si device area, also devices based on different semiconductors have taken advantages from these tools, and among them certainly silicon carbide (SiC), recognized as an optimal semiconductor for the next generation of power devices [1–4]. Compared to other more conventional semiconductors, like Si or GaAs, the most advantageous properties of SiC, which occurs in many different crystal structures, are its wide-bandgap energy, low intrinsic carrier concentration, high carrier saturation velocity, high thermal conductivity, high operational junction temperatures and high breakdown electric field [5–8]. Furthermore the 4H\ polytype of SiC is particularly suited for vertical devices because of its higher vertical mobility.

Ion implantation is the most feasible technique for the\ selective impurity doping of SiC, since the diffusion coefficients of dopants are generally so low that thermal diffusion is not possible at any reasonable temperature. Aluminum is the preferred dopant to produce p-type regions in SiC because of its stability during postimplantation annealing. In fact, the severe ion-induced lattice damage caused by the implantation process requires an appropriate post-implantation annealing for the electrical activation of the dopant and for recovering the crystalline structure, even when an ion implantation at high temperature is performed [9]. Unfortunately at the annealing temperatures and depending on the annealing conditions such as time, ambient, heating and cooling ramp, the surface morphology of the implanted SiC, as well as the not-implanted regions of the wafer, degrade basically due to an exodiffusion of Si. Therefore, after the thermal treatment, this results in an undesirable surface roughness and, however, in small 4H–SiC regions that no longer have the desirable crystal structure [10]. These play a key role on the devices electrical properties, independently if planar or mesa semiconductor technology is performed.

In this paper the ATLAS [11] commercial finite-element device simulator is used to study the DC characteristics of 4H–SiC aluminum implanted p–i–n diodes. The tool, that supports such particular SiC polytype and its distinct set of electronic properties, has been tuned and tested by comparison to experimental results. In particular the electron and hole mobility values were used as fitting parameters.

 

2. Device Analysis

4H–SiC p–i–n diodes were fabricated by Al+ ion implantation at 673K into an n/n+ epitaxial wafer where, via a photolithographic process, circular areas with a diameter of 350 mm were selected. The starting material was a n+ 4H–SiC substrate with a doping density of 5x1019 cm-3 on which a 5 µm-thick epitaxial layer, with a donor doping of 3x1015 cm-3, was grown [12]. Aluminum implantation was performed to create an anode region with a depth profile peaking at 6x1019 cm-3 at the surface. The profile edge and the junction depth are located at 0.2 and 1.35 mm, respectively, as verified by Secondary Ion Mass Spectrometry (SIMS) measurements. The post-implantation annealing process was done at 1873K for 30 min with a heating rate of 40ºC/s by using an inductively heated furnace. During annealing samples are exposed to a high pure Ar atmosphere. Ohmic contacts were made by deposition of Ti/Al and Ni on p+ implanted regions and n+ back surface of the wafer, respectively. The sample surface was not passivated. Further details about the fabrication process were reported elsewhere [13].

After dicing, the chips were encapsulated into a TO18 metal case and contacted by aluminum wires in order to reduce the contact resistance during measurements. The HP4155 Semiconductor Parameter Analyzer was used to trace the device experimental characteristics. All the diodes showed good rectifying characteristics measured from room temperature up to about 523 K. The limit forward current density was set at about 400 A/cm2 in all measurements to avoid excessive heating of the aluminum wires used for bonding. The reverse breakdown voltage of several diodes was found close to –1000V with a leakage current of the order of 1 nA. The diode reverse recovery switching characteristics were also observed at 298K as shown in Fig. 1. The diodes were switched from an initial forward bias current of 330mA to a reverse bias voltage of –10 V. A storage time of about 20 ns was measured.

The device geometry and doping levels described above were subsequently imported into the simulator. The measured J–V characteristics were also imported and used for comparison to the ATLAS outputs.

 

3. Simulation Models

The fundamental geometrical and electrical parameters used in this work are summarized in Table 1 [5] and Table 2 [14–16].

 


4H-SiC regions p+ n- n+

Thickness (mm)
0.2 4.8 300
Doping (cm-3)

6 x 1019
(peak value)
3 x 1015 5 x 1019
Bandgap energy (eV)
3.2 3.2 3.2
Saturated velocity (cm2/s)
2 x 107 2 x 107 2 x 107
Intrinsic concentration (cm-3)
1.99 x 10-8 1.99 x 10-8 1.99 x 10-8
Dielectric constant
9.66 9.66 9.66
Table 1. Physical parameters considered for the 4H–SiC diode analysis at room temperature [5].

 

The effective carrier lifetime p(base) in the n- region was determined by the turn-off switching analysis as in Ref. [6]:

Here IR and IF are the reverse current during storage time ts and the forward on-state current, respectively. From experimental data a value of 15 ns was found for hole life time in the epilayer material. The same carrier lifetimes were assumed for electrons and holes throughout the device as reported in Table 2.


p,n (base) (ns)
15
Nn,p (cm-3)
7 x 1016
Cn,p (cm6/s)
5 x 10-31, 2 x 10-31
EA (meV)
230
ED (meV)
50
(cm2/V s)
200, 20
(cm2/V s)
200, 20
(cm-3/V s)
200, 20
n,p
-0.5
n,p
-0.2
n,p
0.7
n,p
-7, 5.5

Table 2. Simulation model parameters assumed at room temperature [14–16].

 

The lifetimes in the heavily doped anode and cathode regions were accordingly calculated from the semi-empirical formula proposed in [17]

where N is the doping density for the highly doped emitter layer (anode or cathode) and Nn,p is a characteristic parameter for electrons and holes, respectively, dependent on both the material and diode fabrication process [15]. An Nn,p value of 7x1016cm-3 for both electron and holes (see Table 2) has been used in the simulations.

These carrier lifetimes values, a function of impurity concentration, aid to define the Shockley–Read–Hall\ recombination model within the bulk of 4H–SiC using the expression [18]

where ni is the effective intrinsic carrier concentration and Etrap is the difference between the trap energy level and the intrinsic Fermi level. For sake of simplicity a difference Etrap 1/4 0 was assumed during this analysis.

Auger recombination was also considered and calculated introducing the coefficients Cn and Cp given in [14] by means [18]:

ATLAS accounts for the incomplete ionization of dopant impurities using the Fermi–Dirac statistics assuming a single donor or acceptor level. In this case the ionized acceptor and donor concentrations N-A and N+D were expressed as [19]


where NA and ND are the substitutional p-type and n-type doping atoms concentrations, EA and ED are the acceptor and donor energy levels, EFn and EFp are the quasi-Fermi energy levels for electrons and holes, whereas gD and gA are the appropriate degeneracy factors for the conduction and valence band [20].

Finally, the carrier mobilities were modeled with the Caughey and Thomas analytic model [21] and used as variable parameters for the optimal matching of experimental and simulated diode J–V characteristics throughout the whole temperature range investigated:

Here N is the total doping density and is the doping concentration at which the mobility is halfway between the min and max values, namely the mobility in highly doped and undoped material. Both and are temperature- dependent parameters according to

with and mobility values estimated at room temperature [22]. Terms , ,and are fitting coefficients.

 

4. Result and Discussion

A schematic cross-sectional view of the manufactured 4H–SiC p–i–n diode used in the simulations is shown in Fig. 2. The fundamental current components are also shown. Here JnA is the electron injection current into the anode, while JpC and Jr represent the two components of the hole current coming from the anode, namely the hole injection current into the substrate and the recombination current inside the p+/n– depletion layer and inside the neutral region of the base. The top epitaxial layer thickness Wl is 5 mm whereas the n+ substrate thickness Ws is 300 µm. The assumed device area is 1µm2.

During the simulations the incomplete ionization of Al atoms in the anode region was considered as described by Eq. (5) and (6). Fig. 3 reports the difference between the SIMS measured doping profile (solid line) and the electrical acceptor concentration profile calculated by the simulator (dotted line). In fact, in 4H–SiC aluminum produces relatively deep acceptor levels and only a fraction of the Al atoms that occupy a substitutional position in the crystal actually gives origin to free holes [23]. Moreover, the postimplantation annealing process brings only a fraction of the implanted Al atoms in substitutional position. As is well known, the complete ionization of Al atoms is expected only at low doping levels, i.e. for an acceptor concentration lower than 1016 cm-3[24].

The simulated (solid line) and experimental (symbols) forward J–V characteristics at six distinct temperatures of the 4H–SiC p–i–n diodes investigated in this work are shown in Fig. 4. It is worthwhile noting the good agreement over almost nine orders of magnitude for current. At room temperature an ideality factor lower than 2 can be observed at the medium biases of the characteristic where the current transport mechanism is to be related to the carrier recombination effects. For example = 1.9 at 2.2V and = 1.6 at 2.4 V. In this voltage range an ideality factor close to 2 can be observed only at the highest temperatures. For increasing temperatures, the J–V curves shift left and up, an explicit effect of the increased material intrinsic carrier concentration ni on diffusion and recombination current contributions that result essentially to be proportional to ni2 and ni, respectively [6].

The absence of a J–V region with >2 at room temperature before high carrier injection is achieved should be a proof of an efficient damage recovery of ion implanted 4H–SiC crystal. Nevertheless the device conduction characteristics are strongly influenced by the real SiC material quality at the end of the fabrication process [13,25]. It should be noted that the effective carrier lifetimes assumed and the average carrier mobilities estimated in this study suggest that the electrically active defects could trap out electrons and holes. Additionally, the rough topography of the surface at the border of the anode region would be more prone to scattering the carriers.

Interesting information on the device physics can be extracted if one plots the current components depicted in Fig. 2. These are illustrated in Fig. 5 at T 1/4 298K between 2.0 and 5.0 V. In particular, the electron injection into the anode JnA was calculated at the p+/n– depletion region edge inside the anode, while the hole injection JpC was calculated at n–/n+ interface. The recombination component Jr was instead simply evaluated as the difference Jr 1/4 JtotJnAJpC. The total current Jtot is also reported. It is worth noting that, at low and medium current densities Jr dominates over JnA and JpC whereas at larger biases the dominant current component is the electron current density JnA, while the hole current density JpC results even 102 times lower.

These results can be also evidenced by studying the electron/hole concentration and the recombination rate depth profiles, calculated at several biases at T 1/4 298 K.

These are reported in Figs. 6 and 7 up to a forward bias of 4.7 V. In Fig. 6 it is seen that up to 3.5V the device does not reach the high injection regime in the n– region and therefore in fact it behaves as a p+–n long diode. By comparison with Fig. 5, it is also clear that the change of slope of the Jtot curve that takes place around Vd 1/4 2.9V marks the set up of a significant electron and hole injection into p+ and n– layers, respectively. On the other hand, the calculation of the carrier recombination depth profile shown in Fig. 7, reveals that, for biases below 2.6 V, the recombination current, which is the dominant component of Jtot, is largely concentrated inside the depletion layer.

Moving from these considerations, in order to fix the impact of the anode doping electrically effective profile and the role of the anode volume on the device characteristics, alternative versions of the starting diode were considered.

At first, considering the published experimental data on the ionization energy level of Al in 4H–SiC, approximately 210 meV 710–15% from the valence band edge [16,26], three different values of EA, namely 190, 210 and 230 meV were assumed. Fig. 8 reports the simulated J–V plots at room temperature and at the highest current regime. In comparison with Fig. 3, where EA = 230 meV, the effective peak aluminum concentration at the surface for EA = 210 and 190 meV results about 4x1017 and 7x1019 cm-3, respectively. However, it is evident that this variation has only a limited effect on the maximum current handling.

Afterwards the impact of different profile edges, We, was analyzed considering We = 200, 100, 50 and 25 nm and then a junction depth placed at 1.35, 1.25, 1.2 and 1.175 µm. The computations were referred to room Diode bias, Vd (V) temperature. In order to decouple the role of the series resistance of the n– base, the (WlWe) thickness was held constant at 4.8 mm for all devices through the parallel decrease of Wl. Also in this case the diodes on-state current, density increase at the larger biases was only in the range of 5–10%, a further proof that in this voltage range the diode forward J–V characteristics are not controlled by an anode effect.

 

5. Conclusion

The experimental J–V curves of 4H–SiC aluminum implanted p–i–n diodes under forward bias conditions and various temperatures have been analyzed by means of a numerical device simulator. The minority carrier lifetimes were determined investigating the turn-off switching waveform of diodes. The carrier mobilities were used as fitting parameters. The study reveals which current components are dominant at the different operating conditions. At higher current regimes, the diodes exhibit a J–V characteristic largely dominated by electron injection into the p+ layer and so the maximum current handling capability appears slightly dependent on the anode doping electrically effective profile shape.

 

Acknowledgements

The staffs of the clean rooms and of the ion beams Departments of CNR-IMM in Bologna Italy are acknowledged for the manufacture of the 4H–SiC diodes. Prof. A. Carnera of INFM and Department of Physics of the University of Padova Italy is thanked for the SIMS analyses of the implanted samples. The help provided by Dr. G. Foti in setting the simulator parameters is also gratefully acknowledged.

 

References

  1. B.J. Baliga, Power Semicond. Devices, MA-PWS, Boston, 1996.
  2. C.E. Weitzel, J.W. Palmour, C.H. Carter Jr., K. Moore, K.J. Nordquist, S. Allen, et al., IEEE Trans. Electron Devices 43 (1996) 1732–1740.
  3. J.W. Palmour, R. Singh, R.C. Glass, O. Kordina, C.H. Carter Jr., Silicon carbide for power devices, in: Proceedings of the Ninth International Symposium on Power Semiconductor Devices and IC’s, vol. 1, 1997, pp. 25–35.
  4. J.A. Cooper, A. Agarwal, SiC power-switching devices—the second electronics revolution?, Proc. IEEE 90 (2002) 956–968.
  5. G.L. Harris, Properties of SiC, Institute of Electrical Engineers, London, 1995.
  6. S.M. Sze, Physics of Semiconductor Devices, Wiley, New York, 1981.
  7. W.J. Choyke, H. Matsunami, G. Pensl, Silicon Carbide—A Review of Fundamental Questions and Applications to Current Device Technology, Wiley, Berlin, 1997.
  8. A.R. Powell, L.B. Rowland, SiC materials—progress, status and potential roadblocks, Proc. IEEE 90 (2002) 942–955.
  9. M.V. Rao, Maturing ion-implantation technology and its device applications in SiC, Solid State Electron 47 (2003) 213–222.
  10. M.A. Capano, S. Ryu, J.A. Cooper Jr., M.R. Melloch, K. Rottner, S. Karlsson, et al., Surface roughening in ion implanted 4H–silicon carbide, J. Electron. Mater. 28 (1999) 214–218.
  11. Silvaco International, ATLAS User Manual, Santa Clara, CA, 2006.
  12. Purchased from CREE Research Inc., Durham, NC, USA.
  13. A. Poggi, F. Bergamini, R. Nipoti, S. Solmi, A. Carnera, Effects of heating ramp rates on the characteristics of Al implanted 4H–SiC junctions, Appl. Phys. Lett. 88 (2006) 162106–162109.
  14. A. Galeckas, J. Linnros, V. Grivickas, U. Lindefelf, C. Hallin, Auger recombination in 4H–SiC: unusual temperature behaviour, Appl. Phys. Lett. 71 (1997) 3269–3271.
  15. M.E. Levinshtein, T. Mnatsakanov, P. Ivanov, J.W. Palmour,S.L. Rumyantsev, R. Singh, et al., ‘‘Paradoxes’’ of carrier lifetime measurements in high-voltage SiC diodes, IEEE Trans. Electron Devices 48 (2001) 1703–1710.
  16. T. Ayalew, T. Grasser, H. Kosinam, S. Selberherr, Modeling of lattice site-dependent incomplete ionization in a-SiC devices, Mater. Sci. Forum 483 (2005) 845–848.
  17. P.T. Landsberg, G.S. Kousik, The connection between carrier lifetime and doping density in nondegenerate semiconductors, J. Appl. Phys. 56 (1984) 1696–1700.
  18. S. Selberherr, Analysis and Simulation of Semiconductor Devices, Springer, Wien, 1984.
  19. R.C. Jaeger, F.H. Gaensslen, Simulation of impurity freezeout through numerical solution of Poisson’s equations and application to MOS device behavior, IEEE Trans. Electron Devices 27 (1980) 914–920.
  20. D.C. Cole, J.B. Johnson, Accounting for incomplete ionization in modeling silicon based semiconductor devices, in: IEEE Proceedings of the Workshop on Low Temperature Semiconductor Electronics, 1989, pp. 73–77.
  21. D.M. Caughey, R.E. Thomas, Carrier mobilities in silicon empirically related to doping and field, Proc. IEEE 55 (1967) 2192–2193.
  22. M. Roschke, F. Schwierz, Electron mobility models for 4H, 6H, and 3C SiC, IEEE Trans. Electron Devices 48 (2001) 1442–1447.
  23. M. Lades, W. Kaindl, N. Kaminski, E. Niemann, G. Wachutka, Dynamics of incomplete ionized dopants and their impact on 4H/6H–SiC devices, IEEE Trans. Electron Devices 46 (1999) 598–604.
  24. M. Ruff, H. Mitlenher, R. Helbig, SiC devices: physics and numerical simulation, IEEE Trans. Electron Devices 41 (1994) 1040–1054.
  25. F. Moscatelli, A. Scorzoni, A. Poggi, M. Bruzzi, S. Sciortino, S. Lagomarsino, et al., Radiation hardness after very high neutron irradiation of minimum ionizing particle detectors based on 4H–SiC p+n junctions, IEEE Trans. Nucl. Sci. 53 (2006) 1557–1563.
  26. S.R. Smith, A.O. Eevwareye, W.C. Mitchel, M.A. Capano, Shallow acceptor levels in 4H– and 6H–SiC, J. Electron. Mater. 28 (1999) 190–195.

Download pdf version of this article