author's personal copy - uic engineeringmazumder/sse_1.pdf · author's personal copy ......

This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

http://www.elsevier.com/copyright

Author's personal copy

Atomistic and electrical simulations of a GaN–AlN–(4H)SiC heterostructureoptically-triggered vertical power semiconductor device

Srikanta Bose, Sudip K. Mazumder ⇑Department of Electrical and Computer Engineering, Laboratory for Energy and Switching-Electronics Systems, University of Illinois, Chicago, USA

a r t i c l e i n f o

Article history:Received 20 July 2010Accepted 16 March 2011Available online 4 May 2011

The review of this paper was arranged byProf. S. Cristoloveanu

Keywords:Vertical power semiconductor deviceGaN4H–SiCAlNOptical triggeringSimulation

a b s t r a c t

In this paper, a comprehensive simulation study is conducted to investigate the switching characteristics,gain, and breakdown voltage of a GaN–AlN–(4H)SiC based optically-triggered (OT) heterostructure verti-cal power semiconductor device (PSD). It comprises a 1 nm AlN buffer layer between the GaN and SiC het-erointerface to achieve a reasonable compromise between lattice mismatch and lower forward drop. Theresults are compared with an all-(4H)SiC OT PSD. The all-(4H)SiC homostructure PSD is based completelyon SiC and has no buffer layer. Further, it has the same structure, dimensions, and doping densities as thatof the GaN–AlN–(4H)SiC based heterostructure PSD. While there have been studies on GaN–AlN–SiC lat-eral heterostructures, their primary focus has been on lateral conduction in the GaN structure with athick (typically >300 nm) AlN buffer layer residing on top of a SiC substrate. Such an approach will notbe useful for our vertical PSD because of the thick AlN layer. As such, first, a scaled molecular dynamicssimulation (MDS) is carried out in DMol3 emulating the GaN–AlN–(4H)SiC heterointerface pn junction ofthe vertical PSD (with 1 nm AlN buffer) to assess the possibility of vertical conduction and stability of theheterointerface by calculating the density of states (DOS) at the Fermi level and the potential energy,respectively. Subsequently, detailed electrical simulations of the GaN–AlN–(4H)SiC and all-(4H)SiC ver-tical PSDs are carried out in Silvaco to assess their switching performances, gain, on-state drop, andblocking capabilities. The overall results indicate that, the GaN–AlN–(4H)SiC vertical PSD provides supe-rior switching performance and optical absorption compared to the all-(4H)SiC vertical PSD, while thelatter provides better gain. The blocking capabilities and forward drops are found to be comparable forboth the PSDs from a practical standpoint.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Optically-triggered (OT) power semiconductor devices (PSDs)have great potential for high-voltage (HV) and high-power-elec-tronics application. By optically controlling the PSD of a HV powerstage, complete electrical isolation between the low-voltage con-troller and HV power stages can be ensured. Further, electromag-netic-interference (EMI) immunity of the control link betweenthe controller and power stage is realized as well. The earlier workon OT PSD has focused primarily on Si- and GaAs-based devices.The choice of direct bandgap GaAs has been motivated by its highoptical absorption coefficient (OAC), while the choice of indirectbandgap Si has been primarily motivated by cost reduction. How-ever, these GaAs- and Si-based OT PSDs have breakdown voltageand current–density limitations due to the low bandgap and

thermal conductivity of these materials. The semiconductor mate-rials, GaN and SiC, have great potentials for high-temperature andhigh-power-electronics application because of their attractivematerial properties such as large bandgap energies, high break-down fields and high thermal conductivities [1–3]. The materialGaN has very good optical absorption coefficient and short carrierlifetime [1–3]. Recently a SiC based OT thyristor has been pro-posed. The choice of thyristor as the PSD structure is motivatedby the high electrical gain of the PSD that somewhat compensatesfor the low OAC of SiC thereby precluding the need for a high-power short-wavelength laser, which is a difficult proposition fromcost and availability standpoints. Unfortunately, the inefficientturn-off characteristics of the SiC thyristor due to its internal latch-ing action limits the operation of the PSD for high-frequencypower-electronics application and, instead, more suited forpulsed-power application.

One way of addressing the limitations of SiC is to synthesize anOT PSD based on GaN, which has higher OAC compared to SiC. Cur-rently, however, this technological possibility is limited by size andquality of GaN substrates even though progress is being made. An

0038-1101/$ - see front matter � 2011 Elsevier Ltd. All rights reserved.doi:10.1016/j.sse.2011.03.008

⇑ Corresponding author. Tel.: +1 312 355 1315; fax: +1 312 996 6465.E-mail addresses: [email protected] (S. Bose), [email protected] (S.K.

Mazumder).

Solid-State Electronics 62 (2011) 5–13

Contents lists available at ScienceDirect

Solid-State Electronics

journal homepage: www.elsevier .com/locate /sse


alternate option, which is pursued in this paper, is to synthesize aGaN–(4H)SiC heterostructure OT PSD with GaN addressing the

optical-triggering aspect of the device due to its higher OAC whileSiC addressing the high-temperature sustenance of the device. The

Emitter metal contact

Collector metal contact

GaN – N+ emitter(1e19/cm3)

GaN – P base (2e16/cm3) 15 micron

AlN-buffer (1nm)

4H-SiC - N- drift (3e15/cm3)

4H-SiC - N+ Collector substrate (1e19/cm3)

Optical base injection window

(350 nm)

20 micron

20 m

icro

n

0.2

0.5

1.3

6

5.5

5

4.5

4

3.5

3

Ban

dgap

ene

rgy

(eV

)

GaN AlN 4H-SiC

3.43 eV 3.27 eV

6.129 eV

Ban

d en

ergy

(eV

)

12

10

8

6

4

2

0

-2

-4

n+-GaN p-GaN AlN n--4H-SiC n+-4H-SiC

Conduction band energy (eV)

Valence band energy (eV)

Conduction band offset

Valence band offset

(a)

(b)

(c)

Fig. 1. (a) GaN–AlN–(4H)SiC based OT heterostructure PSD. (b) Bandgap profile of GaN–AlN–(4H)SiC heteromaterial system. (c) Conduction- and valence-band energiescorresponding to the device structure shown in (a).

6 S. Bose, S.K. Mazumder / Solid-State Electronics 62 (2011) 5–13


HV aspect of the OT PSD is addressed by both GaN and SiC due totheir wide bandgap, which yields an overall smaller PSD as com-pared to a Si- and GaAs-based PSD.

The main problem with such a heteroepitaxial GaN–SiC deviceis the lattice mismatch of around 3.4% between GaN and SiC mate-rials. To address this, conventional approach typically introduces arelatively thick (typically >300 nm) AlN buffer layer between SiCand GaN growth structures [4–6]. This approach has had successwith lateral low-power and low-voltage devices. Because such de-vices conduct laterally, introduction of a thick high-bandgap AlNbuffer layer over SiC substrate does not affect the device operation.Unfortunately, for HV (and high-power) PSDs, vertical structure iswidely used to address high electric fields and high current densi-ties. In such a vertical GaN–SiC heterostructure PSD, use of a thickAlN buffer layer will affect the conductivity of the PSD. As such, inthis paper, we focus on the vertical conduction of a GaN–(4H)SiCvertical OT PSD structure comprising a 1 nm thick AlN buffer layerto provide balance between vertical conduction and mitigating theeffect of lattice mismatch. It is noted that, the literature in the areaof vertical GaN–(4H)SiC heterostructure PSD is rare [7–14].

The outline of the paper is as follows. Section 2 outlines theoverall structure and operating principle of the GaN–AlN–(4H)SiCbased OT heterostructure PSD comprising a GaN–AlN–(4H)SiC pnjunction epitaxial interface. Subsequently, in Section 3, a first-prin-ciple molecular dynamic [15] analysis of the scaled buffered het-erojunction is carried out on a TeraGrid Supercluster [16] (ofNational Center for Supercomputing Applications) using DMol3

[17] first-principle simulation module of Materials Studio 5.0 soft-ware [18]. To determine whether the high-bandgap AlN buffer af-fects the vertical conduction, the atomistic simulation is carriedout for the heavily doped p-(Ga-face)GaN/n-doped (Si-face)(4H)SiCheteromaterial system with one-layer of (Al-face)AlN as the inter-face or buffer material. The result shows that, the total density ofstates (DOS) at the Fermi level is relatively high, clearly indicatingthe possibility of good vertical conduction across the GaN–AlN–(4H)SiC heterostructure. Based on this favorable observation, inSection 4, an electrical-simulation study is carried out in Silvacosimulator [19] to investigate the turn-on and turn-off characteris-tics, gain, and breakdown voltage of the GaN–AlN–(4H)SiC PSD.The results are then compared with those obtained using an all-(4H)SiC OT vertical homostructure PSD. The all-(4H)SiC PSD isbased completely on 4H–SiC and has no buffer layer. Further, ithas the same structure, dimensions, and doping densities as thatof the GaN–AlN–(4H)SiC based heterostructure PSD. Section 5 out-lines the summary and the conclusions.

2. Device structure and principle of operation

Fig. 1a shows the device structure of the GaN–AlN–(4H)SiC OTPSD. The bipolar PSD comprises emitter and collector metal con-tacts with the collector connected to a positive bias. Right belowthe metal contacts, are the GaN N+ emitter and SiC N+ collector re-gions that are heavily doped to ensure ohmic contacts. The GaN P-base region is excited using a short-wavelength optical pulse toturn the PSD on. The GaN–AlN–(4H)SiC based heterostructure pnregion blocks the applied PSD voltage when the device is off, withthe space-charge layer expanding primarily in the 4H–SiC n� driftregion. The thickness of the GaN based p-base region has to matchor exceed the critical optical absorption depth. Fig. 1b and c showsthe bandgap profile of the heterointerface and the conduction- andvalence-band energies corresponding to the PSD structure shownin Fig. 1a. Because of high-bandgap differences among AlN, GaN,and 4H–SiC, the band offsets are very high at the heterointerfaces.In view of this, a 1 nm AlN buffer layer is incorporated to achieve areasonable compromise between lattice mismatch and lower for-

ward drop. Overall, the doping densities and thicknesses of theemitter, base, and drift regions have to be so optimized (via simu-lation) such that satisfactory compromise among low leakage cur-rent (under blocking condition), low forward drop, and high opticalgain is achieved. The OT PSD is turned on when the GaN p-base re-gion is excited with an optical pulse leading to carrier generationthrough band-to-band transitions resulting in the generation ofadditional electrons and holes [20]. While the photoconductive-ly-generated electrons in the p-base region move towards collectorregion, the holes recombine with the electrons that are injectedfrom the emitter region. The remaining (and majority of the) emit-ter electrons are collected at the collector electrode. After the opti-cal pulse is turned off, and following rapid recombination in thebase region, the PSD enters a dark state with the heterostructurepn junction blocking the high collector-to-emitter voltage.

3. Atomistic analysis for heterointerface density-of-state (DOS)calculation

To analyze whether the high-bandgap 1 nm AlN buffer layer af-fects the vertical conduction of the PSD shown in Fig. 1, an atom-istic MDS is carried out for the heteromaterial system comprisinga moderately p-doped (Ga-face)GaN and a lightly n-doped (Si-fa-ce)(4H)SiC with monolayer of (Al-face)AlN as the interface mate-rial. For performing the MDS, the supercell approach is adopted[21] and a typical setup is shown in Fig. 2. The total number ofatoms in the cell is kept at sixty and the atoms in (Si-face)(4H)SiCare constrained whereas the Ga, Al, and N atoms are relaxed. Theatoms are placed in their respective positions with the coordinatesfollowing [22,23]. To pose the initial condition for the MDS, it is as-sumed that, the heteroepitaxial GaN–AlN structure follows thecrystallographic growth axis of the (Si-face)(4H)SiC substrate.The molecular dynamics ensemble is set to NVT (where N is thenumber of atoms in the cell, V is the volume of the cell, and T isthe cell temperature) so that the temperature of the cell can becontrolled (using Nose–Hoover algorithm [24,25]) and its effecton the atoms can be monitored as the simulation progresses intime. The exchange correlation energy functional is set to LDA/

4H-SiC

N-dopant

Mg-dopant

AlN-layer

GaN

C-atomSi-atom

Al-atom

Ga-atom

Fig. 2. A typical supercell approach for performing atomistic simulation of theGaN–AlN–(4H)SiC heteromaterial system.

S. Bose, S.K. Mazumder / Solid-State Electronics 62 (2011) 5–13 7


PWC [26,27] (i.e. Local Density Approximation/Perdew–Wang–Ceperley) to provide an increased accuracy of the predicted energyand structure with lesser computational time. The temperature of

the ensemble is set at 800 K to emulate the GaN experimentalgrowth condition [13,14]. For energy minimization in the MDSwith computational overhead, a Harris-approximation [28] is used.

16

14

12

10

8

6

4

2

0

0 2e-06 4e-06 6e-06 8e-06 1e-05 1.2e-05

Inpu

t op

tica

l pul

se (

W/c

m2 )

Time (s)

Fig. 4. Optical pulse of 15 W/cm2 used to turn the GaN–AlN–(4H)SiC and the all-(4H)SiC PSDs on or off.

DOS = 9.01 electrons/eV

Fermi level

2000 1900 1800 1700 1600 1500 1400 1300 1200 1100 1000 900 800 700 600 500 400 300 200 100

0

Den

sity

of

stat

es (

elec

tron

s/H

a)

Energy (Ha)

-0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3

0 0.3 0.4 0.5 0.2 0.1

Time (ps)

-19651.66

-19651.67

-19651.68

-19651.69

-19651.70

-19651.71

-19651.72

-19651.73

-19651.74

-19651.75

Pot

enti

al e

nerg

y (H

a)

(a)

(b)

Fig. 3. (a) The DOS of heavily p-doped GaN over n-doped (Si-face)4H–SiC heteroepitaxial material system, with one-layer of (Al-face)AlN as the interface material. (b) Thepotential energy of the GaN–AlN–(4H)SiC heteromaterial system.



Using this supercell setup, the DOS of moderately p-doped (Ga-face) GaN over lightly n-doped (Si-face)(4H)SiC heteroepitaxialsystem with one-layer of (Al-face)AlN, emulating the PSD structurein Fig. 1a, is investigated. This is shown in Fig. 3a. To achieve the p-doping of GaN, the Ga-site is replaced with two Mg atoms. Simi-larly, C-site of 4H–SiC is replaced with one N atom for n-doped4H–SiC. The unit of energy in Fig. 3a has been converted fromhartree to electron volt (1 Ha = 27.2 eV) while reporting the DOS

value. Fig. 3a shows a high DOS at the Fermi level for the GaN–AlN–(4H)SiC heteroepitaxial system with one-layer of (Al-face)AlNas the interface material. This suggests that, a macroscopic electri-cal-device structure, comprising the GaN–AlN–(4H)SiC hetero-structure with 1 nm AlN interface layer, can potentially supportvertical conduction. The MDS also revealed that, the potential en-ergy of the GaN–AlN–(4H)SiC heteromaterial system reduces to aminimum steady-state value (as shown in Fig. 3b), indicating a sta-ble heterostructure configuration.

4. Electrical simulation for steady-state and switchingperformances of the OT PSD

Based on the MDS analysis in Section 3, a set of electrical sim-ulation is carried out on the OT PSD. To assess the performanceof the GaN–AlN–(4H)SiC, a matching set of simulation is also car-

Device dimension in µµm (X-direction)0 2 4 6 8 10 12 14 16 18 20

0

4

8

12

16

20

Device d

imen

sion in

µm

(Y-d

irection)

Photogeneration rate (/s.cm3)

Photogeneration is confined to the p-base region of GaN/4H-SiC NPN device

Photogeneration rate (/s.cm3)

0 2 4 6 8 10 12 14 16 18 20

Device dimension in µm (X-direction)

Device dim

ension in µm (Y

-direction)

0

4

8

12

16

20

Photogeneration occurs across the depth of all-4H-SiC NPN device

(a)

(b)

Fig. 5. Photogeneration rates in (a) the GaN–AlN–(4H)SiC and (b) the all-(4H)SiC PSDs.

Table 1Summary of electrical simulation results of the OT PSD.

OT PSDstructure

Breakdownvoltage (V)

Forwarddrop (V)

Leakagecurrent(A/lm)

Risetime(ls)

Falltime(ls)

Gain

GaN–AlN–(4H)SiC 1950 4.0 1 � 10�10 0.04 0.2 190.0All-(4H)SiC 2000 3.8 4 � 10�11 0.552 1.569 345.661



ried out on an all-(4H)SiC homostructure PSD, which is basedcompletely on SiC and has no buffer layer. Additionally, the all-SiC PSD has the same structure, dimensions, and doping densitiesas that of the GaN–AlN–(4H)SiC based PSD. For transient analysis,to turn the OT PSD on or off, an optical pulse (as shown in Fig. 4)with a wavelength of 350 nm and an intensity of 0 or 15 W/cm2

is incident on the 15 lm p-base optical window of the PSD. Theperformance of the PSD under steady-state and transient condi-tions is provided in Table 1 with the following device specifica-tions: emitter doping = 1 � 1019/cm3, base doping = 2 � 1016/cm3,drift doping = 3 � 1015/cm3, collector doping = 1 � 1019/cm3,emitter thickness = 0.2 lm, base thickness = 0.5 lm, drift thick-ness = 1.3 lm, optical window = 15 lm, Z-dimension = 1 � 104 lm,and optical intensity = 15 W/cm2.

Fig. 5a and b shows the photogeneration in the GaN–AlN–(4H)SiC and all-(4H)SiC PSDs, respectively. From Fig. 5a, it is seenthat, the photogeneration rate is maximum within the GaN regionand almost negligible in other parts. This is achieved by setting thethickness of the p-base region to be compatible with the OAC ofGaN. In contrast, in Fig. 5b, the photogeneration rate is diminished

because of the lower OAC of indirect bandgap 4H–SiC. Most of theincident light penetrates the SiC base region since the energy of thephoton is comparatively higher than the bandgap of SiC. Due to theabove reasons, the available photocurrent current for GaN–AlN–(4H)SiC PSD is higher than the all-(4H)SiC PSD. This is shown inFig. 6a and b, respectively.

Fig. 7 shows that, the breakdown voltage and leakage current ofthe all-(4H)SiC PSD is slightly better than the GaN–AlN–(4H)SiCPSD, indicating a relatively higher breakdown-field strength ofthe former PSD.

Fig. 8 shows how the collector currents for the GaN–AlN–(4H)SiC and all-(4H)SiC PSDs vary with time. If one compares Figs.6 and 8, the gain of GaN–AlN–(4H)SiC PSD (1 9 0) is found to beless than that of all-(4H)SiC PSD (3 4 5). While the collector currentis higher in the GaN–AlN–(4H)SiC PSD, its photogeneration is alsohigh as well which reduces the overall device gain.

However, Fig. 8 also shows that the rise and fall times for theheterostructure GaN–AlN–(4H)SiC PSD are lower than that of theall-(4H)SiC PSD. GaN has higher OAC and hence it absorbs the lightin the base region much more effectively resulting in higher

0.007

0.006

0.005

0.004

0.003

0.002

0.001

0

0 2e-06 4e-06 6e-06 8e-06 1e-05 1.2e-05

Cur

rent

(A

)

Time (s)

0.007

0.006

0.005

0.004

0.003

0.002

0.001

0

0 2e-06 4e-06 6e-06 8e-06 1e-05 1.2e-05

Cur

rent

(A

)

Time (s)

(a)

(b)

Fig. 6. Photogenerated current in (a) the GaN–AlN–(4H)SiC and (b) the all-(4H)SiC PSDs.



Col

lect

or c

urre

nt (

A/µµ

m)

Collector voltage (V) 0 400 800 1200 1600 2000 0 400 800 1200 1600 2000 2400

1.4e-06

1.2e-06

1e-06

8e-07

6e-07

4e-07

2e-07

0

1950 V with leakage of 1 x 10-10 (A/µm)

1e-06

8e-07

6e-07

4e-07

2e-07

0

Col

lect

or c

urre

nt (

A/ µ

m)

Collector voltage (V)

2000 V with leakage of

4 x 10-11 (A/µm)

(a) (b)

Fig. 7. Breakdown voltage characteristic of (a) the GaN–AlN–(4H)–SiC and (b) the all-(4H)SiC PSDs.

1

0.8

0.6

0.4

0.2

0

Col

lect

or c

urre

nt (

A)

0 4e-06 6e-06 8e-06 1e-05 1.2e-05

Time (s)

0.9464 A

2e-06

1

0.8

0.6

0.4

0.2

0

Col

lect

or c

urr

ent

(A)

0 4e-06 6e-06 8e-06 1e-05 1.2e-05

Time (s)

0.972 A

late turn-off

2e-06

(a) (b)

Fig. 8. Collector-current variation with time for (a) the GaN–AlN–(4H)SiC and (b) the all-(4H)SiC PSDs.

5000

4000

3000

2000

1000 Cur

rent

den

sity

(A

/cm

2 )

Cur

rent

den

sity

(A

/cm

2 )

X Y

3000

2500

2000

1500

1000

500

YX

00

2 446

8810

121214

161618

20

02

46

810

1214

1618

20 20

04

812

1620

(a) (b)

Fig. 9. Current density of (a) the GaN–AlN–(4H)SiC and (b) the all-(4H)SiC PSDs captured at 0.1 ls (with reference to the optical pulse shown in Fig. 4).



current density (see Fig. 9a) and lower rise time. On the other hand,for the all-(4H)SiC device, part of the light penetrates the baseregion because of poorer OAC of SiC resulting in lower currentdensity (see Fig. 9b) and slower rise time. The fall time of theGaN–AlN–(4H)SiC is lower than that of the all-(4H)SiC PSD primar-ily because of the shorter carrier lifetime of GaN. Thus, once thelight is shut off, the carriers in the base region of the heterostruc-ture PSD recombine much faster. A further reason for the higherfall time of the all-(4H)SiC is due to the additional photogeneratedcarriers in the drift region since light is not effectively captured inthe base region due to the poorer OAC of 4H–SiC.

Fig. 10 shows the recombination rate for GaN–AlN–(4H)SiC andall-(4H)SiC NPN device during turn-off at 2 ls with reference to in-put pulse shown in Fig. 4. Due to shorter carrier lifetime and highrecombination rate of GaN material, the GaN–AlN–(4H)SiC PSD isturning-off faster in comparison to all-(4H)SiC PSD which we seefrom Figs. 8 and 10. Another reason for late turn-off for all-(4H)SiC PSD is that incident light is getting penetrated from baseto drift to collector region, (as shown in Fig. 5b) since, the energyof the photon is comparatively higher than its bandgap. Hence,even though it has poor absorption coefficient, photo carriers aregenerated not only in base but also in the drift and collector re-gions though the amount is small whereas in case of GaN–AlN-(4H)SiC PSD, the photogeneration is confined to base only (shownin Fig. 5a). Thus, during the turn-off, due to lower recombinationrate, there still exists some extra carriers due to photogeneration,contributing to the total current density and hence, the delay.

5. Summary and conclusions

The detailed simulation studies based on the GaN–AlN–(4H)SiCbased heterostructure vertical optically-triggered (OT) PSD revealsthe feasibility of the device. First a scaled molecular dynamicssimulation (MDS) is carried out that suggests possibility of verticalconduction (with the high-bandgap AlN buffer layer) and a stableheterointerface. This is assessed, respectively, by determining thedensity of states at the Fermi level and by determining the poten-tial energy. Subsequently, electrical simulations are carried out toanalyze the switching performance, gain, and blocking capabilityof the GaN–AlN–(4H)SiC based OT PSD. These results are also com-pared with an all-(4H)SiC OT PSD. The comparison clearly shows

the superiority of the switching performance (which is crucial forhigh-frequency power-electronics application) and optical absorp-tion (which important to reduce laser cost) of the heterostructuredue to the shorter lifetime and optical absorption coefficient ofGaN. The device electrical gain of the all-(4H)SiC PSD is found tobe better. The forward drop of the GaN–AlN–(4H)SiC is found tobe practically the same as that of the all-(4H)SiC. A slight differenceis due to the high-bandgap AlN buffer layer. Finally, the breakdowncharacteristics of the all-(4H)SiC PSD shows practically comparableresults to the GaN–AlN–(4H)SiC PSD because of higher critical fieldstrength of 4H–SiC material. These results clearly indicate that theheterostructure device has potential for high-frequency and high-voltage power-electronics.

Acknowledgements

This work is supported in part by the award of US National Sci-ence Foundation (under Award Number 0823983), received byProf. S.K. Mazumder. However, any opinions, findings, conclusions,or recommendations expressed herein are those of the authors anddo not necessarily reflect the views of the NSF.

References

[1] http://cst-www.nrl.navy.mil/.[2] http://www.ioffe.ru/SVA/NSM/Semicond.[3] Group IV elements, IV–IV and III–V compounds. Part a – lattice properties.

Springer-Verlag; 2006. <http://www.springermaterials.com/navigation/navigation.do?m=l_2_132697_Group+IV+Elements%2C+IV-IV+and+III-V+Compounds.+Part+a+-+Lattice+Properties>.

[4] Yu H, Ozturk M, Ozcelik S, Ozbay E. A study of semi-insulating GaN grown onAlN buffer/sapphire substrate by metalorganic chemical vapor deposition. JCryst Growth 2006;293:273–7.

[5] Nam O, Bremser M, Zheleva T, Davis R. Lateral epitaxy of low defect densityGaN layers via organometallic vapor phase epitaxy. Appl Phys Lett1997;71:2638–40.

[6] Honda Y, Okano M, Yamaguchi M, Sawaki N. Uniform growth of GaN on AlNtemplated (1 1 1)Si substrate by HVPE. Phys Status Solidi c 2005;2:2125–8.

[7] Kuznetsov N, Gubenco A, Nikolaev A, Melnik Y, Blashenkov M, Nikitina I, et al.Electrical characteristics of GaN/6H–SiC n–p heterojunctions. Mater Sci Eng B1997;46:74–8.

[8] Torvik J, Leksono M, Pankove J, Zeghbroeck B, Ng H, Moustakas T. Electricalcharacterization of GaN/SiC n–p heterojunction diodes. Appl Phys Lett1998;72:1371–3.

[9] Torvik J, Qiu C, Leksono M, Pankove J. Optical characterization of GaN/SiC n–pheterojunctions and p-SiC. Appl Phys Lett 1998;72:945–7.

5e+23

4e+23

3e+23

2e+23

1e+23

0

XR

ecom

bina

tion

rat

e (/

s.cm

3 )Y

X Y

2.4e+23

2e+23

1.6e+23

1.2e+23

8e+22

4e+22

0

Rec

ombi

nati

on r

ate

(/s.

cm3 )

02 4

6810

1214

1618

20

02 4

6810

1214

1618

20

04

812

1620

04

812

1620

(a) (b)

Fig. 10. Recombination rate of GaN–AlN–(4H)SiC and all-(4H)SiC PSDs captured at 2 ls (with reference to the optical pulse shown in Fig. 4).



[10] Nikolaev A, Rendakova S, Nikitina I, Vassilevski K, Dmitriev V. GaN grown byhydride vapor phase epitaxy on p-type 6H–SiC layers. J Electron Mater1998;27:288–91.

[11] Torvik J, Leksono M, Pankove J, Heinlein C, Grepstad J, Magee C. Interfacialeffects during GaN growth on 6H–SiC. J Electron Mater 1999;28:234–9.

[12] Danielsson E, Zetterling C, Ostling M, Nikolaev A, Nikitina I, Dmitriev V.Fabrication and characterization of heterojunction diodes with HVPE-grownGaN on 4H–SiC. IEEE Trans Electron Dev 2001;48:444–9.

[13] Nakano Y, Suda J, Kimoto T. Direct growth of GaN on off-oriented SiC (0 0 0 1)by molecularbeam epitaxy for GaN/SiC heterojunction bipolar transistor. PhysStatus Solidi c 2005;2:2208–11.

[14] Brown A, Lusurdo M, Kim T, Giangregorio M, Choi S, Morse M, et al. The impactof SiC substrate treatment on the heteroepitaxial growth of GaN by plasmaassisted MBE. Cryst Res Technol 2005;40:997–1002.

[15] http://en.wikipedia.org/wiki/Molecular_dynamics.[16] Delley B. An all-electron numerical method for solving the local density

functional for polyatomic molecules. J Chem Phys 1990;92:508–17.[17] http://accelrys.com/.[18] http://www.ncsa.illinois.edu/UserInfo/Resources/Hardware/Intel64Cluster/

TechSummary.

[19] http://www.silvaco.com.[20] Sze S. Physics of semiconductor devices. New York: John Wiley and Sons;

1969. p. 654–83.[21] Jarvis M, White I, Godby R, Payne M. Supercell technique for total-energy

calculations of finite charged and polar systems. Phys Rev B 1997;56:14972–8.[22] Zoroddu A, Bernardini F, Ruggerone P, Fiorentini V. First-principles prediction

of structure, energetics, formation enthalpy, elastic constants, polarization,and piezoelectric constants of AlN, GaN, and InN: comparison of local andgradient-corrected density-functional theory. Phys Rev B 2001;64:1–6.

[23] Iuga M, Neumann GS, Meinhardt J. Ab-initio simulation of elastic constants forsome ceramic materials. Eur Phys J B 2007;58:127–33.

[24] Nose S. A molecular dynamics method for simulations in the canonicalensemble. Mol Phys 1984;52:255–68.

[25] Hoover WG. Canonical dynamics: equilibrium phase-space distributions. PhysRev A 1985;31:1695–7.

[26] Kohn W, Sham LJ. Self-consistent equations including exchange andcorrelation effects. Phys Rev 1965;140:A1133–8.

[27] Perdew JP, Wang Y. Accurate and simple analytic representation of theelectron–gas correlation energy. Phys Rev B 1992;45:13244–9.

[28] Harris J. Simplified method for calculating the energy of weakly interactingfragments. Phys Rev B 1985;31:1770–9.


author's personal copy - uic engineeringmazumder/sse_1.pdf · author's personal copy ......

Documents