Review of using gallium nitride for ionizing radiation detectionJinghui Wang, Padhraic Mulligan, Leonard Brillson, and Lei R. Cao Citation: Applied Physics Reviews 2, 031102 (2015); doi: 10.1063/1.4929913 View online: http://dx.doi.org/10.1063/1.4929913 View Table of Contents: http://scitation.aip.org/content/aip/journal/apr2/2/3?ver=pdfcov Published by the AIP Publishing Articles you may be interested in Radiation hardness of three-dimensional polycrystalline diamond detectors Appl. Phys. Lett. 106, 193509 (2015); 10.1063/1.4921116 Neutron detection using boron gallium nitride semiconductor material APL Mat. 2, 032106 (2014); 10.1063/1.4868176 Narrow-band radiation sensing in the terahertz and microwave bands using the radiation-inducedmagnetoresistance oscillations Appl. Phys. Lett. 92, 102107 (2008); 10.1063/1.2896614 Efficiency of composite boron nitride neutron detectors in comparison with helium-3 detectors Appl. Phys. Lett. 90, 124101 (2007); 10.1063/1.2713869 Fundamental limits to detection of low-energy ions using silicon solid-state detectors Appl. Phys. Lett. 84, 3552 (2004); 10.1063/1.1719272

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

50.4.102.168 On: Sat, 26 Sep 2015 15:55:25

APPLIED PHYSICS REVIEWS—FOCUSED REVIEW

Review of using gallium nitride for ionizing radiation detection

Jinghui Wang,1,2 Padhraic Mulligan,1 Leonard Brillson,3,4 and Lei R. Cao1,a)

1Nuclear Engineering Program, Department of Mechanical and Aerospace Engineering,The Ohio State University, Columbus, Ohio 43210, USA2Department of Radiology, Stanford University, Stanford, California 94305, USA3Department of Electrical and Computer Engineering, The Ohio State University, Columbus, Ohio 43210, USA4Department of Physics, The Ohio State University, Columbus, Ohio 43210, USA

(Received 12 July 2015; accepted 12 August 2015; published online 3 September 2015)

With the largest band gap energy of all commercial semiconductors, GaN has found wide

application in the making of optoelectronic devices. It has also been used for photodetection such as

solar blind imaging as well as ultraviolet and even X-ray detection. Unsurprisingly, the appreciable

advantages of GaN over Si, amorphous silicon (a-Si:H), SiC, amorphous SiC (a-SiC), and GaAs, par-

ticularly for its radiation hardness, have drawn prompt attention from the physics, astronomy, and nu-

clear science and engineering communities alike, where semiconductors have traditionally been used

for nuclear particle detection. Several investigations have established the usefulness of GaN for alpha

detection, suggesting that when properly doped or coated with neutron sensitive materials, GaN

could be turned into a neutron detection device. Work in this area is still early in its development,

but GaN-based devices have already been shown to detect alpha particles, ultraviolet light, X-rays,

electrons, and neutrons. Furthermore, the nuclear reaction presented by 14N(n,p)14C and various

other threshold reactions indicates that GaN is intrinsically sensitive to neutrons. This review

summarizes the state-of-the-art development of GaN detectors for detecting directly and indirectly

ionizing radiation. Particular emphasis is given to GaN’s radiation hardness under high-radiation

fields. VC 2015 Author(s). All article content, except where otherwise noted, is licensed under aCreative Commons Attribution 3.0 Unported License. [http://dx.doi.org/10.1063/1.4929913]

TABLE OF CONTENTS

I. INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

II. GAN MATERIAL PROPERTIES . . . . . . . . . . . . . . 2

A. Basic parameters . . . . . . . . . . . . . . . . . . . . . . . . 2

B. GaN growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

III. RADIATION DETECTION . . . . . . . . . . . . . . . . . . 4

A. Alpha particle response . . . . . . . . . . . . . . . . . . 4

1. Double-Schottky contact structure . . . . . . 4

2. Mesa structure . . . . . . . . . . . . . . . . . . . . . . . 4

3. Sandwich structure . . . . . . . . . . . . . . . . . . . 5

B. X-ray detection . . . . . . . . . . . . . . . . . . . . . . . . . 5

C. Betavoltaic application . . . . . . . . . . . . . . . . . . . 6

D. Neutron detection . . . . . . . . . . . . . . . . . . . . . . . 6

E. Intrinsic neutron sensitivity . . . . . . . . . . . . . . . 7

IV. HARSH-ENVIRONMENT PERFORMANCE . . . 8

A. Neutron irradiation damage . . . . . . . . . . . . . . . 8

B. High temperature performance . . . . . . . . . . . . 9

V. CONCLUSION. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

I. INTRODUCTION

Gallium nitride (GaN) semiconductors are now

commonly found in optoelectronic and high-power devices,

e.g., light-emitting diodes (LEDs),1,2 lasers,3 and high elec-

tron mobility transistors (HEMTs).4 GaN can also be used

for detecting ionizing radiation under extreme radiation

conditions due to its properties such as a wide band-gap

(3.39 eV), large displacement energy (theoretical values

averaging 109 6 2 eV for N and 45 eV for Ga),5 and high

thermal stability (melting point: 2500 �C).6 Compared to

narrower band-gap semiconductors such as silicon, GaN can

operate at higher temperatures; while a comparison with

other wide band-gap semiconductors, such as silicon carbide,

demonstrates GaN’s higher electron mobility7 and potential

for better carrier transport properties. In addition, the high

Z-value and density of GaN makes it a suitable material for

X-ray detection in medical imaging. The first group of

reports showing GaN as an alpha particle detector used devi-

ces in a double Schottky structure, fabricated from a

2–2.5 lm thick epitaxial GaN layer, grown via metal organic

chemical vapor deposition (MOCVD)8–10 on a sapphire

substrate. Based on these studies, a review article in 2006

compared the use of a few wide band-gap semiconductor

materials in very high radiation environments, to be used in

the next generation of high-energy physics experiments at

the Large Hadron Collider (LHC).11 The article concluded

that GaN is a very promising candidate for use in such

experiments, despite it still being a relatively immature

semiconductor material. Subsequent studies have furthera)Email: cao.152@osu.edu

1931-9401/2015/2(3)/031102/12 VC Author(s) 20152, 031102-1

APPLIED PHYSICS REVIEWS 2, 031102 (2015)

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

50.4.102.168 On: Sat, 26 Sep 2015 15:55:25

demonstrated that GaN-based sensors can detect a-par-

ticles12–14 and X-rays,15–17 indicating the growing potential

use of GaN for ionizing radiation detection. Furthermore, a

study investigating alternative materials for neutron detec-

tion, driven by the shortage of 3He gas, has considered the

use of GaN for neutron sensors in harsh environments.18

More recently, the growth methods for GaN are shifting

from foreign substrate epitaxial to free standing type, bulk

GaN with a thickness of several hundred micrometers. Based

on these different types of materials, various GaN sensor

structures have been fabricated for ionizing radiation detec-

tors. For a-particle detectors, the lateral double-Schottky

contact (DSC),8–10 mesa,6,12,13,19 sandwich,14,20 and p-i-n

structures21 have been tested. For X-ray detectors, Schottky

Metal-Semiconductor-Metal (MSM),15 Schottky diode,16,17

and p-i-n structures22 have all been reported. For electron

detection, the applications for making betavoltaic energy

converters,23,24 using pn,25 p-i-n,23,26–28 and Schottky type29

structures have now been tested. Thermal neutron detection

using a 6Li converter in a sandwich structure has been

recently reported.20 The nuclear reaction presented by14N(n,p)14C (1.8 b for neutron at 0.025 eV) and a various

other threshold reactions producing charged particles suggest

that GaN is intrinsically sensitive to neutrons, including fast

neutron, e.g., at 1 MeV or above. However, the commerciali-

zation of GaN detectors for radiation detection is still

impeded by the lack of high-quality materials. The various

defects present in GaN, including point defects, extended

defects, and surface defects form scattering centers, recombi-

nation and trapping centers limit the quality of the material.

In addition, the growth of p-type GaN is still under develop-

ment due to the lack of a suitable dopant. In this review, we

discuss the properties of various GaN device structures and

their constituent materials to understand the use of GaN for

detecting ionizing radiation at a fundamental level. In addi-

tion, device performance in a high radiation field and high

temperature environment is also summarized. This under-

standing will facilitate the effective application of GaN in

fabricating ionizing radiation detectors and the potential

applications in extreme radiation conditions such as those

found in nuclear power reactors, accelerators, and fusion

reactors, which require radiation-hard devices.

II. GaN MATERIAL PROPERTIES

A. Basic parameters

Despite the limited number of devices reported for radi-

ation detection, GaN holds several advantages over other

semiconductors for high temperature and high radiation field

applications. Compared to the widely used Si and Ge detec-

tors, both of which are limited to either room or liquid nitro-

gen cooled temperatures, respectively, GaN is characterized

by a much wider band-gap, making it capable of working in

environments well above room temperature. Shortcomings

in other wide band-gap semiconductors such as short carrier

lifetimes (10 ns) in GaAs due to the dominant EL2 native

deep-level defect,30 the large number of deep-level defects31

in AlN, and the high cost of diamond32 limits their

implementation as radiation detectors. Compared to SiC that

has an in-direct band-gap, GaN has a higher mobility and

thus better electrical properties. For instance, GaN can form

a high mobility 2D electron gas by the polarity effect for

field effect transistor applications. GaN may also be more

radiation hard due to its higher ionic bond strength, large

crystal density, and fewer polytypes.33,34 GaN also has a

higher Z-value and should thus be more suitable for X- and

c-ray detection. Although there are other candidates with

high Z-value suitable for radiation detection, for example

HgI2, the issues of difficulty in growing large scale crystals

and controlling material quality have hindered their further

applications.35 It is indeed the progress in the photonics and

electronics areas that drives the advancement of GaN,

mainly from materials synthesize, which in turn benefits its

applications in the relatively small market represented by the

radiation detections. In addition, GaN is a superior material

for optoelectronic applications, since it has a direct band-gap

and can alloy with Al and In, representing a tunable band-

gap value of 1.9 (InN) to 6.2 eV (AlN).36 Note that the band

gap of InN is still under debate, while a new value of

�0.7 eV is recently more in consensus theoretically37 and

experimentally.38 A comparison of the properties of these

semiconductors can be found in Table I.

B. GaN growth

Growth methods and defects formed in GaN have been

previously reviewed in various publications.47–52 A brief

review is given here with the focus on GaN’s properties that

are most relevant to radiation detection, though these proper-

ties are also important for other applications. Due to the lack

of a native substrate, GaN is mostly grown on foreign

substrates, such as AlN,53 Si,54 sapphire,55 and SiC.56 The

mismatch between the two layers results in a high density of

threading dislocations in the GaN epilayer, which includes

pure edge, pure screw, and mixed dislocations. These dislo-

cations have a significant effect on the device behavior.

They behave as non-radiative recombination centers with

energy levels in the forbidden gap and thus form trapping

centers, act as charged scattering centers,57 and provide a

leakage current pathway.58,59 Recent research has found that

pure screw components, which are solely responsible for the

leakage paths, are uncharged, while edge dislocations behave

as negatively charged scatterers because the associated traps

are filled with electrons.60 The edge dislocation has a repul-

sive potential around its line, which will not deteriorate the

device’s performance in which electrons transport parallel to

the edge. The screw dislocations, however, are a major con-

cern in terms of device performance.57

For epitaxial GaN, the threading dislocation density can

be as high as 108–1010 cm�2 when GaN is grown directly

on a foreign substrate.47 By introducing buffer layers61 and

using the epitaxial laterally overgrown (ELOG) tech-

nique,50,62 the density can be lowered down to 106 cm�2

(Ref. 50) and 105 cm�2 (Ref. 63), respectively. When grow-

ing bulk material, such as in hydride vapor phase epitaxy

(HVPE), dislocations can be controlled by increasing the

thickness of the material, resulting in interactions between

031102-2 Wang et al. Appl. Phys. Rev. 2, 031102 (2015)

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

50.4.102.168 On: Sat, 26 Sep 2015 15:55:25

dislocations which leads to a decrease in the dislocation den-

sity near the top surface.64 Research shows that by removing

the top layer of the HVPE substrate, a high-quality bulk GaN

(several hundred micrometers thick) with a dislocation den-

sity of �106 cm�2 can be produced,65 such as those produced

by one of GaN wafer suppliers.66

In addition to dislocations, another key growth factor

that could dominate the device performance is the doping

level. Undoped GaN shows n-type properties due to the re-

sidual shallow donors such as oxygen in MOCVD-grown

GaN13 and silicon in HVPE-grown GaN67 (shallow donors

are defined as impurities that ionize at room temperature,

which corresponds to an activation energy of 100 meV).

Oxygen donors result in a background free-carrier concen-

tration between 1015 and 1017 cm�3 with an activation

energy of 30–33 meV (Ref. 68) that is below the conduction

band minimum (CBM). GaN can be intentionally doped

with Si to form an n-type material through either as-growth

or post-growth implantation. For Si-doped GaN, activation

energy levels of 12–17 meV,69 30–60 meV,70 18.1 meV,

and 273.9 meV (Ref. 71) have all been reported. Most of

these levels can be ionized at room temperature to donate

an electron to the material.33

For p-type doping, Mg provides a hole by occupying a

Ga site.72 However, the hole population is limited to within

1018 cm�3 in p-type GaN:Mg, and the resistivity is always

greater than 104 X cm (Ref. 73) due to (a) the large thermal

activation energy of Mg in GaN (120–250 meV) resulting

in low activation efficiencies of 1%–2%;74 (b) the con-

sumption of Mg by hydrogen passivation, i.e., the forma-

tion of the electrically inactive neutral complex (Mg-H)0

during the growth or high-temperature annealing pro-

cess;75 (c) the hole compensation by oxygen impurities

causing a high-resistivity, semi-insulating (SI) material;76

and (d) the consumption of Mg by self-compensation, i.e.,

the formation of a deep donor with a nitrogen vacancy,

MgGaVN.77,78

Besides n- and p-type doping, semi-insulating GaN has

also been successfully produced. Fe ions can be introduced

to compensate the residual donors to obtain SI GaN, result-

ing in SI substrates with a high resistivity of �1010 X/�.79

Fe forms the charge transfer deep levels FeGa3þ/2þ when it

occupies the Ga lattice sites,80 and the charged FeGa3þ state

can transform to FeGa2þ by capturing an electron and thus

compensate the residual donors. The energy level that repre-

sents the energy required to emit electrons captured from

the donor by the Fe acceptor is between 0.34 and 0.87 eV

below the conduction band edge, so the compensation

should be thermally stable at room temperature.81 SI

GaN:Fe shows good crystal quality and the strain-free incor-

poration of Fe.82 However, the Fe doping pins the Fermi

level to approximately 0.5–0.6 eV below the CBM.83 In

addition to Fe, Mg ions can also be used to compensate for

the residual donors, and moreover, Mg can improve the

crystal quality. Research has shown that crystal strain and

point defects are largely eliminated by the substitution of

Ga by Mg atoms.84 For example, SI GaN:Mg produced by

one of GaN wafer suppliers has a low dislocation density of

�104 cm�2.85 Although the compensated semi-insulating

GaN may exhibit a good crystal quality, it is not suitable for

making radiation detectors due to the short carrier lifetime

caused by the high-density of impurities.

TABLE I. Material properties (mechanical, electrical, thermal) of major semiconductors for radiation detection at 300 K.11,39 Note: lh: light hole; hh: heavy

hole; tr: transverse; l: longitude; hz: heavy hole at kz direction, hx: heavy hole at kx direction, lz: light hole at kz direction, lx: light hole at kx direction. GaN

has the largest breakdown voltage among all the commercial semiconductors, its electron mobility exceeds all but GaAs, Ge, and diamond, its thermal conduc-

tivity exceeds all but AlN and diamond.

Property AlN Diamond GaN 4H-SiC CdTe GaAs Si Ge

Crystal structure Wurtzite Diamond Wurtzite Wurtzite Zinc blende Zinc blende Diamond Diamond

Average atomic number 10 12 19 10 50 31.5 14 32

Bandgap (eV) 6.2 5.5 3.39 3.23 1.44 1.424 1.12 0.661

Density (g/cm3) 3.23 3.515 6.15 3.211 5.85 5.32 2.33 5.32

e-h pair creation energy (eV) 15.3 12 8.9 7.8 4.43 4.2 3.62 2.96

Effective electron masses (m0) 0.4 1.40 (l, at 85 K) 0.2 0.29 (l) 0.11 0.063 0.98 (l) 1.6 (l)

0.36 (tr, at 85 K) 0.42 (tr) 0.19 (tr) 0.08 (tr)

Effective hole mass (m0) 3.53 (hz) 0.8 0.4

10.42 (hx) 2.12 (hh, at 1.2 K) 1.75 (l) 0.51 (hh) 0.49 (hh) 0.33 (hh)

3.53 (lz) 0.70 (lh, at 1.2 K) 0.66 (tr) 0.082 (lh) 0.16 (lh) 0.043 (lh)

0.24 (lx)

Electron mobility (cm2/(V�s)) 300 1800–2200 1000 800–1000 1100 �8500 1450 �3900

Hole mobility (cm2/(V�s)) 14 1200–1600 30 50–150 100 �400 450 �1900

Breakdown field (106 V/cm) 1.2–1.8 1–10 �5 3–5 — 0.4 0.3 0.1

Saturated electron drift velocity

(107 cm/s)

1.4 2.7 3 (Ref. 40) 2.0 1.3 (Ref. 41) 1.2 1.0 6.5

Threshold displacement energy (eV) 43 35 (Ref. 42) Ga: 18, N: 22

(Ref. 43)

C: 20, Si: 35

(Ref. 44)

Te: 7.8, Cd: 8.9

(Refs. 45 and 46)

10 13–20 25

Melting point (�C) 3000 4373 (at 125 kbar) 2500 2857 (at 35 atm) 1092 1240 1412 937

Thermal conductivity (W/cm �C) 2.85 6–20 1.3 3.7 0.06 0.55 1.3 0.58

Thermal expansion coefficient

(10�6/ �C)

5.27 0.8 5.59 3.7 5.9 5.73 2.6 5.9

031102-3 Wang et al. Appl. Phys. Rev. 2, 031102 (2015)

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

50.4.102.168 On: Sat, 26 Sep 2015 15:55:25

III. RADIATION DETECTION

A. Alpha particle response

1. Double-Schottky contact structure

GaN-based a-particle detection was first realized by

Vaiktus et al.9 using a DSC structure shown in Fig. 1(a). The

active layer in this device was a 2-lm-thick epitaxial GaN

(epi-GaN) layer grown by MOCVD on a sapphire substrate.

Under the epi-GaN is a 2-lm-thick, highly doped, n-type

GaN buffer layer that is employed to minimize the disloca-

tion density. Two Au Schottky contacts are deposited on the

top of the epi-GaN to complete the detector. Alpha particle

detection was performed using 5.48 MeV a-particles emitted

from an 241Am source, and the spectra under different bias

voltages before breakdown (29 V) is shown in Fig. 2.9 One

can see that as the bias voltage increases to 27 V, any further

increase in the voltage does not change the peak channel

number, which indicates that the device is fully depleted. A

fit of the pulse peak with a Gaussian distribution gives a

Charge collection efficiency (CCE) of �92%. A better fit

can be obtained with two Gaussian functions, and Vaitkus

et al.9 suggested that the double-peak structure is related to

the complicated drift of the charge carriers in a layer contain-

ing different drift and trapping barriers. This two Gaussian

fit to alpha spectrum has also been reported in devices

utilizing a p-i-n structures.21 Subsequent research10 verified

the electric path for this double-Schottky structure: electric

field lines through the epilayer are perpendicular to each

Schottky contact, connected by the highly doped buffer

layer.

The double-Schottky contact structure has three main

advantages:86 (a) The fabrication process is simple in that it

requires only one masking step; (b) the structure can be used

without any optimized Ohmic contact and is thus usually

employed to study and optimize Schottky contacts; and (c)

since the thickness of epi-layer is usually less than the range

of the a-particles, both electrons and holes contribute to the

detection signal and significant trapping of charge carriers is

not expected.10 On the other hand, the structure suffers from

several drawbacks: (a) The thin epilayer (<10 lm) results in

only partial energy deposition; (b) multiple Gaussian func-

tions are needed to fit the a spectrum; and (c) parasitic capac-

itance and resistance may exist due to the long distance

between the two metal contacts.

2. Mesa structure

The mesa structure (mesa-1) is based on thin-film GaN

as shown in Fig. 1(b).6,12 An Ohmic contact on the highly

doped buffer layer is realized by etching through the

epi-GaN layer. Polyakov et al.12 deposited 1 mm diameter

Ni Schottky contacts on mesa structures to study three types

of undoped GaN with carrier concentrations of more than

1015 cm�3: 3-lm GaN grown by MOCVD, 3-lm GaN grown

by molecular-beam epitaxy (MBE), and 12-lm GaN grown

by MOCVD using the ELOG technique. For all three sam-

ples, deep-level transient spectroscopy (DLTS) analysis has

revealed two electron traps with activation energies of 0.25

and 0.6 eV, respectively, and that the MBE sample had the

highest concentration. Due to these traps, the CCE of the

MBE device is lower than that of the MOCVD and ELOG

devices, both of which are close to 100%. Additionally, all

three CCEs are higher than those reported based on semi-

insulating materials, which again is a result of the low

trapping impurity concentration.

The other mesa structure (mesa-2) shown in Fig. 1(c) is

that reported by Lu et al.13 In this structure, the etching does

not reach the buffer layer, and the Ohmic contact is realized

on the same epilayer. The epi-GaN is a 3-lm layer of

undoped GaN (carrier concentration of �4� 1016 cm�3)

FIG. 1. Structures of a-particle GaN

detectors: (a) Double-Schottky contact

structure, (b) and (c) mesa structure,

and (d) sandwich structure.

FIG. 2. Alpha particle pulse height spectra from the double-Schottky contact

structure a-particle detector. Reproduced with permission from Vaitkus

et al., Nucl. Instrum. Methods Phys. Res., Sect. A 509, 60–64 (2003).

Copyright 2003 Elsevier.

031102-4 Wang et al. Appl. Phys. Rev. 2, 031102 (2015)

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

50.4.102.168 On: Sat, 26 Sep 2015 15:55:25

grown by MOCVD on a sapphire substrate. The Schottky

contact is realized by depositing a circular Ni/Au contact

with a diameter of 940 lm, and the Ohmic contact employs

the Ti/Al/Ni/Au metal scheme, achieved by etching away

2-lm of the epi-GaN. Using this mesa structure, Lu et al.13

found that the a-particle pulse height spectra have one dis-

tinct peak that can be fitted precisely by a Gaussian.

Compared with lateral devices, i.e., the double-Schottky

contact structure, mesa geometry detectors exhibit superior

performance, e.g., a lower noise level and fewer dislocation

mobility fluctuations,87 which are related to the better trans-

port properties of carriers flowing vertically. For these two

mesa structures, mesa-1 may suffer from an over-etching

problem, since the applied buffer layer is usually thin (less

than 2 lm), while mesa-2 needs a high temperature annealing

process to form the Ohmic contact. Moreover, for both struc-

tures, the distance between the Schottky and Ohmic contacts

should be kept small (less than several tens of micrometers)

in order to minimize the parasitic capacitance and resistance.

If this is overlooked, the device will have a high on-

resistance (as seen in the current–voltage curve), and the

capacitance–voltage data may no longer be reliable.

3. Sandwich structure

The sandwich structure a-particle detector (Fig. 1(d))

fabricated on bulk GaN was reported recently by Lee et al.14

It consists of a �500–lm-thick n-type free-standing GaN

purchased from Kyma (grown by HVPE), which is charac-

terized by a very low carrier concentration of

�1013–1014 cm�3 in the upper �30-lm (the concentration in

the rest of the GaN is �1016 cm�3). The Schottky contact is

made on the Ga side by a circular Ni contact with a diameter

of 1 mm. DLTS and ODLTS (DLTS measurements with op-

tical injection) measurements show that the low carrier con-

centration is due to compensation by unidentified acceptor

centers with an activation energy of 0.2 eV and by the major

H5 hole traps with an activation energy of 1.2 eV. However,

owing to these H5 traps, the CCE of the device is limited,

and a voltage of 120 V is necessary to overcome the trapping

and obtain 100% charge collection.

Sandwich structure has also been fabricated on normally

grown bulk GaN (Kyma) by Mulligan et al.20 The 450 lm

wafer had an unintentionally n-type doping (impurity Si) of

1.68� 1016 cm�3. Due to the relatively high doping concen-

tration, the depletion region is only around 2 lm under a

reverse bias of less than 20 V. Experiment results showed

that the charge collection efficiency within this thin depletion

region is almost 100%. Further simulation88 indicated that

the carrier gain due to impact ionization in the depletion

region and the diffusion components from the undepleted

region is significantly small compared with the drift carriers

in the depletion region, and the carrier loss due to hole trap-

ping is negligible.

Compared with thin-film-based structures, the bulk-

sandwich structure has a potential advantage of offering a

large depletion region and thus a full a-particle energy depo-

sition. Furthermore, it has superior carrier transport proper-

ties and, in particular, suffers less of a current crowding

problem.88,89 At the same time, the structure maintains fabri-

cation simplicity. We may therefore conclude that the sand-

wich structure is a preferred choice for a-particle detection.

B. X-ray detection

X-ray detection with thin-film GaN is difficult due to the

low absorption coefficient shown in Fig. 3. Both theoretical

calculations and experimental measurements indicate that

the absorption is large only for photon energies between 10

and 20 keV.15 At higher energies such as 40 keV, the coeffi-

cient is about 50 cm�1, which corresponds to a 1/e attenua-

tion length of 200 lm. Thus for high-energy X-rays, a

multilayer GaN film or bulk structure is needed to absorb a

significant number of incoming photons.

Nevertheless, GaN-based X-ray detection has been dem-

onstrated by Duboz et al. using both the Schottky MSM15

and Schottky diode structures.16,17 The MSM structure was

fabricated on 10-lm-thick undoped GaN grown by MOCVD

and employed a Pt/Au Schottky contact with the finger width

varying from 2 to 20 lm and the spacing varying from 2

to 10 lm. When the X-ray source was switched off, the

signal showed a fast decrease in less than 1 s followed by a

long exponential transient with a time constant of 40 s.

The Schottky diode detector was fabricated on 20-lm

homogeneous-growth, undoped GaN with a Pt/Au Schottky

contact area that varied between 1 and 2 mm2. Fig. 4 shows

the transient characteristics after the X-ray source was

switched on and off at two different power levels.16

X-ray detection has also been performed with a p-i-n

structure,22 but despite the different structures employed, a

fast transient increase or decrease (of less than 1 s) followed

by a slow exponential increase or decay is seen in all the

measured photocurrents. According to Duboz et al.,15 this

photocurrent characteristic can be explained in terms of two

currents: photovoltaic and photoconductive. The photovoltaic

current generated in the depletion region has a fast response,

while the photoconductive current created by the X-ray

activated carrier traps has a slow response. Recently, Lu’s

group decoupled the photoconductive current into several

FIG. 3. Experimental (dots) and theoretical (line) absorption coefficient of

GaN as a function of the photon energy. Reproduced with permission from

Appl. Phys. Lett. 92, 263501 (2008). Copyright 2008 AIP Publishing LLC.

031102-5 Wang et al. Appl. Phys. Rev. 2, 031102 (2015)

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

50.4.102.168 On: Sat, 26 Sep 2015 15:55:25

components related to electrons trapped in different energy

levels characterized by their own lifetime constants.22,90 To

decrease or eliminate the slow transient component, the only

solution is to fabricate high-quality materials.

C. Betavoltaic application

A biased semiconductor detector could easily be per-

ceived as a voltaic device when no bias is applied. Whether

it is alpha-voltaic, beta-voltaic, gamma-voltaic, or even

neutron-voltaic, the only difference is which source causes

ionization. Due to its superior properties already mentioned,

GaN is suitable to make radiation-voltaic batteries for aero-

nautic, marine, and medical applications that require a long

life time and high reliability power supply.23 Currently, GaN

has been investigated for beta particle response in its applica-

tion of making beta-voltaic micro-batteries.

Theoretical calculations from a conceptual design24,91

utilized two GaN pn junctions sandwiched with a 63Ni radio-

isotope source (with a half-life of 100 yr). The results indi-

cated a maximum short circuit current of 1.1 lA/cm2,

maximum open circuit voltage of 2.7 V, and ideal efficiency

of 25%. The efficiency obtained is higher than those pres-

ently available from thermoelectric converters made of

Silicon, which is around 15%. Similar results were recently

obtained by another calculation using the same structure and

a 147 Pm source.92

Two groups have tested their GaN-based beta-voltaic

devices simultaneously. Chen’s group studied both pn and

p-i-n structures.25–27 In one of their p-i-n devices, using a63Ni source with activity of 2 mCi, an open-circuit voltage of

1.62 V, short-circuit current density of 16 nA/cm2, filling

factor of 55%, and energy conversion efficiency of 1.13%

were obtained.26 Another group lead by Lu studied both

Schottky and p-i-n structures,23,28,29 reporting a p-i-n device

with an open circuit voltage of 1.07 V, short circuit current

of 0.554 nA, and a filling factor of 24.7% using a 147 Pm

source.23

The performance of GaN-based beta-voltaic devices is

still far from its theoretical calculated values, and further

improvement largely relies on the availability of high quality,

thick GaN. High purity GaN would result in a wide active

region with low recombination and trapping effects,25,29

increasing the energy conversion efficiency. In addition, the

structure of the device should be optimized by using a thin

electrode layer to minimize the dead layer and backscattering

for electrons,28 and a thin passivation layer to decrease the

leakage current.93 Admitting that the development of GaN

beta-voltaic is still at its early infancy, GaN is highly promis-

ing as a potential candidate for long-life nuclear micro-

batteries used as power supplies for microelectrochemical

system devices.

D. Neutron detection

For neutron detection using a semiconductor device,

neutron sensitive converters are typically employed to pro-

duce charged particles for subsequent electron-hole produc-

tion. Commonly used neutron converters are 10B, 6Li, and157Gd,94 and the converter can either be coated directly onto

the surface of the GaN as a thin-film or be incorporated by

doping,95,96 as illustrated in Fig. 5. While these commonly

applied neutron convertors have high cross-sections in the

thermal energy region (�2200 m/s), the detection of fast

neutron will involve bulky moderation materials or ineffi-

cient reaction process (e.g., proton recoil). For thin-film-

coated detectors, the fabrication process is relatively simple,

but the intrinsic neutron detection efficiency is limited by the

fact that only one of the two charged particles generated

from the neutron capture reaction can enter the active region

FIG. 4. Photocurrent transient measured for two incident X-ray powers of

P¼ 10 and P¼ 40 (corresponding to currents of 10 and 40 mA on the X-ray

tube, respectively). The applied reverse voltage is �10 V. Reproduced with

permission from J. Appl. Phys. 105, 114512 (2009). Copyright 2009 AIP

Publishing LLC.

FIG. 5. (a) Thin-film-coated and (b)

solid-form semiconductor thermal

neutron detectors.

031102-6 Wang et al. Appl. Phys. Rev. 2, 031102 (2015)

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

50.4.102.168 On: Sat, 26 Sep 2015 15:55:25

of the device. Another fundamental limitation on detection

efficiency is the conflict between the relatively long neutron

mean free path and the short range of the charged particles in

these convertor materials. In solid-state detectors, the

conversion atoms may be introduced by high-temperature

diffusion, ion implantation, or in-grow processes, during

which defects will likely be introduced and degrade the over-

all performance of the detector. Both configurations shall be

explored to study the behavior of GaN-based neutron detec-

tors. In addition, c-ray discrimination is always a valid con-

cern with solid-state neutron detection devices.97 Although

the thin-film device is relatively gamma blind, the use of

high-Z elements as the convertor, such as Gd, may generate

a considerable number of X-rays via c-ray activation. These

phenomena, however, boost the c-ray detection efficiency

when Gd is combined with GaN or other semiconducting

materials.98,99

A rare-earth-doped GaN thin film might lead to signifi-

cant improvements in device performance in sensor applica-

tions; GaN is potentially neutron sensitive when doped with

Gd.100,101 Melton et al.102 found that the carrier concentra-

tion in Gd-doped GaN increases as a result of the Gd doping.

Cao and Myers employed a 500-nm superlattice doping

structure grown by MBE to maintain the quality of the

material.18 Although the device showed a large leakage

current and low breakdown voltage, which indicated a large

dislocation density, increasing the gap between Gd clusters

may improve the charge collection efficiency, possibly lead-

ing to a working superlattice device. Melton et al.103 also

fabricated a GaN-based neutron scintillator by coating

Si-doped GaN (n-type doping of �5� 1018 cm�3) with 6LiF

and Gd converters. When electron–hole pairs are created in

GaN by secondary charged particles, they recombine to

produce scintillation photons, which are then detected by

photo-sensors. The Si-doped sample was chosen because its

near-band-edge recombination intensity is an order of mag-

nitude higher than undoped samples.104

A thermal neutron induced spectrum in GaN was first

realized by Cao.105 A sandwich structure Schottky diode with

a diameter of 1 mm was fabricated on special growth n-type

GaN grown by Kyma. A 6LiF:ZnS thin film (0.3 mm) was

used as the neutron-conversion material placed in front of the

detector. The set up was then placed directly in the 3-cm neu-

tron beam with a nuclear reactor operating at a power of

250 kW. The spectrum of the neutron response is given in Fig.

6. The two charged particles emitted from 6Li capture, i.e., 3H

at 2727 keV and 4He at 2055 keV were clearly discernible in

the spectrum. However, because the depletion depth of the de-

tector was 1.6 lm, much less than the range of these particles

in GaN, only a fraction of the initial energy of the particles

was deposited into the depletion region. Because the stopping

power of 4He is greater than that of 3H, 4He deposits more

energy in the depletion region than 3H.

E. Intrinsic neutron sensitivity

GaN is uniquely qualified for intrinsic neutron detection

due to a 584 keV monoenergetic proton emission following

neutron capture in 14N. Detection of this proton serves as an

indication of a neutron interaction with the detector.

Although the cross section of 14N is significantly smaller

than other isotopes typically employed as converter material

in neutron detectors, this apparent shortcoming could be

quite advantageous in high flux environments. Detection effi-

ciency can be sacrificed in neutron rich environments by

choosing conversion materials with a smaller cross section.

Materials with high neutron cross sections can be depleted

quickly in such environments, while materials with a smaller

cross section result in less material depletion during a given

time interval and thus, a longer operational lifetime. The 1.8

b cross section of 14N at thermal energy could be sufficient

for high flux neutron detection, while avoiding the acute

depletion seen with other conversion materials (Fig. 7). On

the other hand, 14N’s sensitivity to fast neutron is in the

same order of magnitude with that of 10B because of its

reduced difference in neutron capture cross-section at, for

example, 1 MeV (0.03 b for 14N versus 0.22 b for 10B), and

also taking into account 14N’s much larger natural abundance

ratio (99.6% for 14N versus 20% for 10B). Transmutation of

the semiconductor material, leading to unwanted doping and

changes in the detector’s characteristics, should also be con-

sidered in high neutron flux environments. Although the

cross sections of 69Ga and 71Ga (60% and 40% atomic abun-

dance, respectively) are somewhat larger than other semicon-

ductor materials suited for high flux environments (namely,

SiC), Ga holds a cross section similar to N, and will therefore

FIG. 6. Observed spectrum due to charged particle emissions following 6Li

neutron capture and 14N neutron capture (a), and experimental setup (b).

Reproduced with permission from L. Cao, Battelle Energy Alliance, LLC

Project No. 11–3004, 2015, pp. 1–44. Copyright U.S. Department of

Energy, Nuclear Engineering University Program 2015.

031102-7 Wang et al. Appl. Phys. Rev. 2, 031102 (2015)

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

50.4.102.168 On: Sat, 26 Sep 2015 15:55:25

deplete concurrently with N. A high flux neutron detector of

GaN would therefore suitably detect neutrons, while delay-

ing the effects of converter material depletion and neutron

transmutation doping in the semiconductor. Work has been

completed to test the efficacy of a GaN based neutron detec-

tion by placing a sample of bulk GaN in a high flux neutron

beam and measuring the proton emission spectrum in vac-

uum. Results shown in Fig. 8 indicated a sufficient number

of pulses were detected for GaN to be used as a neutron con-

version material, even though a very small solid angle

(�0.001 Sr.) was used in the experiment.106

The energy spectrum of proton emissions from GaN

was measured with a Si detector, detecting those protons

produced inside GaN by nuclear reactions and escaping the

materials to reach a Si detector a few cm away. The spectrum

indicates that proton particles induced by thermal neutrons

can effectively produce electron-hole pairs within GaN if it

were to be fabricated into a rectifying junction. The difference

in spectrum width can be explained by the different thick-

nesses of two samples measured. The detectable depth limit of

the 14N in GaN is 4.13 lm from the surface, based on the

Stopping and Range of Ions in Matter (SRIM) calculation of

the projected range of the 584 keV proton in the GaN material.

In Fig. 8, the epi-layer of GaN in epi-layer GaN sample is

�2 lm, and Ammono is much thicker than proton range, thus

an extended peak into the background region.

IV. HARSH-ENVIRONMENT PERFORMANCE

A. Neutron irradiation damage

The radiation damage to GaN materials and devices

under various radiation species, such as proton, neutron,

gamma-ray, and electrons has been reviewed in other refer-

ences;107,108 here we only provide a concise discussion in

neutron irradiation effects. Both fast and thermal neutrons

damage the GaN lattice. Fast neutrons cause damage by

elastic collisions and the thermal neutrons by nucleus recoil

during neutron activation of Ga and N atoms.109 For fast

neutron irradiation, Nordlund et al.5 did a simulation study

by employing molecular dynamics method, they found that

although GaN has low threshold displacement energies

(22 6 1 eV for Ga and 25 6 1 eV for N), the average values

are relatively high (45 6 1 eV for Ga and 109 6 2 eV for N).

The average values for displacement are higher than the

threshold is due to the fact that in some collisions, energies

are transferred via other mechanisms instead of displace-

ment, such as generation of phonons. A 1 MeV fast neutron

can transfer up to 55 keV to a primary knock-on atom (PKA)

of Ga or 230 keV to a PKA of N; these energies exceed the

displacement energies and will thus lead to cascade colli-

sions.109 A study in 2006 (Ref. 110) showed that fast neutron

irradiation with a fluence of 6.7� 1018 cm2 produces approx-

imately 1900 and 7200 displaced atoms per PKA of Ga and

N, respectively.

Thermal neutrons can produce indirect atomic displace-

ments as a result of radioactive capture (n,c) reactions. The

main thermal reactions in GaN are listed in Table II.111,112

Among these reactions, the energy of recoil atoms generated

by the first two reactions is sufficient to cause approximately

2� 105 displacements according to a SRIM simulation. The

third reaction releases most of its energy to c-rays, and the

effect of the last reaction is negligible due to the extremely

small cross section (2.42� 10�5 b). Note that isotopes of

FIG. 7. Neutron capture cross section plot of primary constituent isotopes in

GaN and SiC (a) and charged particle emission cross section plot (b) in com-

monly used neutron converter materials. 14N has a sufficiently small cross

section to delay the effects of material depletion in high flux environments.

FIG. 8. Energy spectrum of protons emitted following 14N neutron capture

in two types of GaN wafers.106

031102-8 Wang et al. Appl. Phys. Rev. 2, 031102 (2015)

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

50.4.102.168 On: Sat, 26 Sep 2015 15:55:25

70Ga, 72Ga, and 16N will also undergo beta decay; however,

these beta-particles and c-rays will not cause serious damage

to the GaN lattice, since most of their energy will be lost to

interactions with outer orbital electrons.

Neutron irradiation may change the lattice constant and

induce strain inside GaN. Research by Marques et al.109 and

Lorenz et al.113 indicates that under neutron irradiation, Ga

atoms are preferentially displaced along the c-axis, but the in-

plane lattice parameter does not change. For instance, for a

neutron fluence of 8� 1019 cm�2, the lattice constant c

increases by 0.38% and the lattice constant a nearly

unchanged.114 Neutron irradiation can also alter the resistivity

or even cause type-reversion of the material. Compensation

schemes have been identified for both p-type115 and n-type116

GaN when the neutron fluence exceeds 1016 cm�2. Wang

et al.116 suggested that the neutron-irradiation-induced struc-

ture defects GeGa give rise to carrier trap centers that are

responsible for the observed reduction in the carrier concen-

tration in irradiated n-type GaN. Due to these induced

defects, trapping centers are created, and the Fermi level is

pinned within the energy levels of these defects. For example,

Polyakov et al.117 found that under a high neutron fluence of

�1018 cm�2, a deep electron trap with an activation energy of

0.75 eV was introduced in both n- and p-type GaN, and the

Fermi level was pinned to near Ec-0.85 eV. For Mg-doped

p-type GaN, the Fermi level pinning near Ec-(0.8–0.9) eV

has also been reported.115 For Fe-compensated semi-insulat-

ing GaN, both N vacancies and deep level defects were

observed by DRCLS (depth-resolved cathodoluminescence

spectroscopy) analysis after fast and thermal combined

neutrons irradiation with fluences from 1014 to 1016 n/cm2

(Ref. 118). For low neutron fluences on the order of

�1011 cm�2, the damage can be repaired by a self-annealing

process at room temperature for several days.119,120 For me-

dium (�1014 cm�2) and high fluences (�1019 cm�2), most of

the lattice damage can be repaired under a high temperature

annealing process at around 1000 �C; however, a certain

amount of optical and electrical damage remains.113

In a general sense, neutron irradiation will degrade the

device performance. For instance, Grant et al.6,19 found that

neutron irradiation of Schottky diode detectors fabricated on

2- and 12-lm thin-film semi-insulating GaN produced a

non-linear increase in the leakage current as the neutron

fluence was increased from 1014 to1016 cm�2, and the CCE

of the 12-lm device under a fluence of 1016 cm�2 decreased

from 53% to 20%. They suggested that the degradation is

mainly due to an increased number of recombination/trap-

ping centers. Similar results were reported by Mulligan

et al.121 after studying the degradation of the electrical

properties of a Schottky diode device, a turning point at a

total neutron fluence of 1015 n/cm2 was discovered.

However, there may be an optimal fluence for which the de-

vice performance is enhanced. One study by Wang122,123

shows that after neutron irradiation with a fluence of

1� 1013 cm�2, the Au/GaN Schottky barrier photodetectors

showed superior current�voltage characteristics, which was

attributed mainly to the effective repression of the deep elec-

tron traps Et.

The existence of the threshold neutron fluence on the

performance of GaN Schottky diodes was further validated

by Lin,124 which found that fast neutrons with fluences

�1015 n/cm2 increases deep level defects densities, while

thermal neutron with fluences �1015 n/cm2 anneal these

defects and improve the GaN crystalline quality. Their fur-

ther studies125 revealed that both fast and thermal neutrons

introduce recombination centers in GaN, forming an insulat-

ing phase between GaN and metal contact, raising the

Schottky contact ideality factor, increasing the Ohmic con-

tact sheet resistance, and lowering the current transport for

both contacts. Above the threshold fluence, thermal neutrons

can introduce thermal spikes and elevated temperatures that

drive GaN out diffusion and reaction with metal contacts.

Fig. 9 shows the current-voltage characteristics of the

Schottky diodes under different neutron fluences, from

which the Schottky barrier height and ideality factor were

derived and analyzed.

B. High temperature performance

Both Ohmic and Schottky contacts require a thermal

annealing process to improve the quality of the contact.

After annealing at a certain temperature, the resistance of the

Ohmic contact can be reduced, and the leakage of the

Schottky contact can be decreased. However, annealing at

temperatures higher than the optimum value degrades the

contacts.

In terms of Ohmic contacts, Dobos et al.126 showed that

at an annealing temperature of 700 �C, there was a lateral

TABLE II. Thermal-neutron-induced reactions in GaN.111,112

Reactions

Main c-ray energy

(main transition probability) (MeV)

Recoil atom

energy (MeV)

69Gaðn; cÞ70Ga 1.039 (65%) 6.62271Gaðn; cÞ72Ga 0.834 (96%) 5.69014Nðn; cÞ15N 10.82 (50%) 0.01415Nðn; cÞ16N Negligible due to low cross section (2.42� 10�5 b)

FIG. 9. The log(jcurrentj) vs. voltage characteristics for the neutron-

irradiated GaN Schottky diodes for voltage from �5 V to 3 V. Reproduced

with permission from J. Appl. Phys. 115, 123705 (2014). Copyright 2014

AIP Publishing LLC.

031102-9 Wang et al. Appl. Phys. Rev. 2, 031102 (2015)

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

50.4.102.168 On: Sat, 26 Sep 2015 15:55:25

diffusion of Al for the Al-only-based contact, and the conti-

nuity of the Ti layer was broken for the Ti/Al contact. After

annealing at a temperature of 900 �C, the two contacts no

longer exhibited linear behavior. For the Ti/Al/Ni/Au

contact, Fan et al.127 reported an optimum annealing temper-

ature of 750 �C, and the multi-metal layer scheme has been

widely used to form Ohmic contact on GaN.128–132

In Schottky contacts, the degradation varies according

to the metal employed. For Pd, Pt, Au, and Ni, degradation

starts after exposure to temperatures as low as 300, 400, 575,

and 600 �C, respectively.33 Kim et al.133 showed that W con-

tacts degrade at annealing temperatures greater than 500 �C,

and the metal silicide WSix shows a stable Schottky barrier

height of �0.5 eV at temperatures above 600 �C. Thus, the

W-based Schottky contact is currently the most promising

contact that can sustain temperatures of up to 600 �C. The

thermal stability of the Schottky contact is also related to the

metal scheme employed. Monroy et al.134 studied Pt- and

Ni-based Schottky contacts and demonstrated the smooth

decay of the barrier height in both Ni/Au and Pt/Ti/Au

diodes as the temperature increased, and a decrease of 20%

after annealing at 500 �C. This is in contrast to the Pt/Au

contact, which suffers from barrier height degradation at

300 �C. They attributed this difference to the thin Ti layer,

which may behave either as a diffusion barrier or as a

wetting layer to prevent the Pt from deforming during

the annealing process. High-temperature resistant Schottky

contacts are essential for harsh-environment applications.

Fig. 10 summarizes the behavior of GaN and related devices

at different temperatures.70,71,120,126,135

The long-term stability of Schottky diodes has also been

studied. Research by Luther et al.120 showed that Pt

Schottky contacts can maintain at 400 �C for 500 h without

degradation. O’Mahony et al.119 studied Ni-based Schottky

contacts and found that under a forward current of 1.3 A/cm2

or a reverse bias of �3.5 V while stored at 300 �C in an N2

environment for 466 h, the diodes exhibit a drift of less than

10% in both the ideality factor and barrier height. Despite

these cases, more research is still needed to understand the

long-term operating performance.

V. CONCLUSION

GaN is one promising semiconducting material for ion-

izing radiation detection, particularly in hash environments.

The defects present in GaN, such as the dislocations and

unintentional doping, still presents a main challenge in terms

of improving device-level performance. In spite of the

defects, prototype detectors fabricated on GaN have shown

response to alpha particles, electrons, X-rays, and neutrons.

Their special capability in operating in high radiation fields

and elevated temperature conditions are promising for many

applications. The nitrogen neutron caption reaction also

makes GaN an interesting material for intrinsic neutron

detection in high flux environments. Future progress in crys-

tal growth to produce a low carrier concentration region

comparable to the ranges of various radiations in GaN, and

fabrication techniques to minimize the leakage current and

parasitic capacitance are still required to expand GaN’s

application as a mature radiation detection material.

ACKNOWLEDGMENTS

We acknowledge the support of this work by U.S.

Department of Energy Office of Nuclear Energy’s Nuclear

Energy University Programs and Department of Defense,

Defense Threat Reduction Agency [Grant No. HDTRA1-11-

1-0013]. The authors also gratefully acknowledge National

Science Foundation, Grant No. DMR-1305193 (Charles

Ying and Haiyan Wang) for support of this work.

1M. Meneghini, L. R. Trevisanello, G. Meneghesso, and E. Zanoni, IEEE

Trans. Device Mater. Reliab. 8(2), 323–331 (2008).2S. Nakamura, T. Mukai, and M. Senoh, Appl. Phys. Lett. 64(13),

1687–1689 (1994).3N. Trivellin, M. Meneghini, E. Zanoni, K. Orita, M. Yuri, and G.

Meneghesso, “A review on the reliability of GaN-based laser diodes,” in

2010 IEEE International Reliability Physics Symposium (IRPS), 2–6 May

2010, pp.1–6.4J. A. del Alamo and J. Joh, Microelectron. Reliab. 49(9), 1200–1206

(2009).5J. Nord, K. Nordlund, J. Keinonen, and K. Albe, Nucl. Instrum. Methods

Phys. Res., Sect. B 202, 93–99 (2003).6J. Grant, R. Bates, W. Cunningham, A. Blue, J. Melone, F. McEwan, J.

Vaitkus, E. Gaubas, and V. O’Shea, Nucl. Instrum. Methods Phys. Res.,

Sect. A 576(1), 60–65 (2007).7H. Arabshahi, Braz. J. Phys. 39(1), 35–38 (2009).8J. Vaitkus, E. Gaubas, T. Shirahama, S. Sakai, T. Wang, K. M. Smith,

and W. Cunningham, Nucl. Instrum. Methods Phys. Res., Sect. A

514(1–3), 141–145 (2003).9J. Vaitkus, W. Cunningham, E. Gaubas, M. Rahman, S. Sakai, K. M.

Smith, and T. Wang, Nucl. Instrum. Methods Phys. Res., Sect. A

509(1–3), 60–64 (2003).10P. J. Sellin, D. Hoxley, A. Lohstroh, A. Simon, W. Cunningham, M.

Rahman, J. Vaitkus, and E. Gaubas, Nucl. Instrum. Methods Phys. Res.,

Sect. A 531(1–2), 82–86 (2004).11P. J. Sellin and J. Vaitkus, Nucl. Instrum. Methods Phys. Res., Sect. A

557(2), 479–489 (2006).12A. Y. Polyakov, N. B. Smirnov, A. V. Govorkov, A. V. Markov, E. A.

Kozhukhova, I. M. Gazizov, N. G. Kolin, D. I. Merkurisov, V. M. Boiko,

FIG. 10. High temperature behavior of

GaN and related devices (triangle

arrow indicates the commonly used

values).

031102-10 Wang et al. Appl. Phys. Rev. 2, 031102 (2015)

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

50.4.102.168 On: Sat, 26 Sep 2015 15:55:25

A. V. Korulin, V. M. Zalyetin, S. J. Pearton, I. H. Lee, A. M. Dabiran,

and P. P. Chow, J. Appl. Phys. 106(10), 103708 (2009).13M. Lu, G. G. Zhang, K. Fu, and G. H. Yu, Chin. Phys. Lett. 27(5),

052901 (2010).14I. H. Lee, A. Y. Polyakov, N. B. Smirnov, A. V. Govorkov, E. A.

Kozhukhova, V. M. Zaletin, I. M. Gazizov, N. G. Kolin, and S. J.

Pearton, J. Vac. Sci. Technol., B 30(2), 021205 (2012).15J. Y. Duboz, M. Lauegt, D. Schenk, B. Beaumont, J. L. Reverchon, A. D.

Wieck, and T. Zimmerling, Appl. Phys. Lett. 92(26), 263501 (2008).16J. Y. Duboz, B. Beaumont, J. L. Reverchon, and A. D. Wieck, J. Appl.

Phys. 105(11), 114512 (2009).17J. Y. Duboz, E. Frayssinet, S. Chenot, J. L. Reverchon, and M. Idir, Appl.

Phys. Lett. 97(16), 163504 (2010).18J. Wang, P. Kandlakunta, T. F. Kent, J. Carlin, D. R. Hoy, R. C. Myers,

and L. Cao, Trans. Am. Nucl. Soc. 14, 209–210 (2011).19J. Grant, W. Cunningham, A. Blue, V. O’Shea, J. Vaitkus, E. Gaubas, and

M. Rahman, Nucl. Instrum. Methods Phys. Res., Sect. A 546(1–2),

213–217 (2005).20P. Mulligan, J. H. Wang, and L. Cao, Nucl. Instrum. Methods Phys. Res.,

Sect. A 719, 13–16 (2013).21G. Wang, K. Fu, C. S. Yao, D. Su, G. G. Zhang, J. Y. Wang, and M. Lu,

Nucl. Instrum. Methods Phys. Res., Sect. A 663(1), 10–13 (2012).22C. S. Yao, K. Fu, G. Wang, G. H. Yu, and M. Lu, Phys. Status Solidi A

209(1), 204–206 (2012).23F. H. Li, X. Gao, Y. L. Yuan, J. S. Yuan, and M. Lu, Sci. China Technol.

Sci. 57(1), 25–28 (2014).24C. Honsberg, W. A. Doolittle, M. Allen, and C. Wang, in IEEE

Photovoltaic Specialists Conference, 2005, pp. 102–105.25Z. J. Cheng, H. S. San, Y. F. Li, and X. Y. Chen, in Proceedings of the

5th IEEE International Conference on Nano/Micro Engineered andMolecular Systems, 2010, pp. 582–586.

26Z. J. Cheng, H. S. San, X. Y. Chen, B. Liu, and Z. H. Feng, Chin. Phys.

Lett. 28(7), 078401 (2011).27Z. J. Cheng, X. Y. Chen, H. S. San, Z. H. Feng, and B. Liu,

J. Micromech. Microeng. 22(7), 074011 (2012).28M. Lu, G. Wang, and C. S. Yao, Adv. Mater. Res. (Switz.) 343–344,

56–61 (2012).29M. Lu, G. G. Zhang, K. Fu, G. H. Yu, D. Su, and J. F. Hu, Energy

Convers. Manage. 52(4), 1955–1958 (2011).30M. Rogalla, T. Eich, N. Evans, R. Geppert, R. Goppert, R. Irsigler, J.

Ludwig, K. Runge, T. Schmid, and D. G. Marder, Nucl. Instrum.

Methods Phys. Res., Sect. A 395(1), 49–53 (1997).31V. A. Soltamov, I. V. Ilyin, A. A. Soltamova, E. N. Mokhov, and P. G.

Baranov, J. Appl. Phys. 107(11), 113515 (2010).32D. R. Kania, M. I. Landstrass, M. A. Plano, L. S. Pan, and S. Han,

Diamond Relat. Mater. 2(5–7), 1012–1019 (1993).33X. A. Cao, S. J. Pearton, and F. Ren, Crit. Rev. Solid State 25(4),

279–390 (2000).34F. Nava, G. Bertuccio, A. Cavallini, and E. Vittone, Meas. Sci. Technol.

19(10), 102001 (2008).35A. Burger, D. Nason, and L. Franks, J. Cryst. Growth 379, 3–6 (2013).36J. I. Pankove, Mater. Sci. Eng., B 61–62, 305–309 (1999).37P. Carrier and S. H. Wei, J. Appl. Phys. 97(3), 033707 (2005).38K. S. A. Butcher and T. L. Tansley, Superlattices Microstruct. 38(1),

1–37 (2005).39See http://www.ioffe.ru/SVA/NSM/Semicond/GaN/ for the detailed me-

chanical, thermal and electrical properties of GaN.40Y. Zhou, D. Wang, C. Ahyi, C. C. Tin, J. Williams, M. Park, N. M.

Williams, and A. Hanser, Solid-State Electron. 50(11–12), 1744–1747

(2006).41C. Canali, M. Martini, G. Ottavian, and K. R. Zanio, Phys. Rev. B: Solid

State 4(2), 422–431 (1971).42J. C. Bourgoin and B. Massarani, Phys. Rev. B 14(8), 3690–3694 (1976).43J. Nord, K. Nordlund, and J. Keinonen, Phys. Rev. B 68(18), 184104

(2003).44R. Devanathan and W. J. Weber, J. Nucl. Mater. 278(2–3), 258–265

(2000).45F. J. Bryant and E. Webster, Phys. Status Solidi 21(1), 315–321 (1967).46F. J. Bryant and A. F. J. Cox, Proc. R. Soc. London, Ser. A 310(1502),

319–339 (1969).47L. Liu and J. H. Edgar, Mater. Sci. Eng., R 37(3), 61–127 (2002).48A. Denis, G. Goglio, and G. Demazeau, Mater. Sci. Eng., R 50(6),

167–194 (2006).49J. A. Freitas, J. Phys. D: Appl. Phys. 43(7), 073001 (2010).

50B. A. Haskell, S. Nakamura, S. P. DenBaars, and J. S. Speck, Phys. Status

Solidi B 244(8), 2847–2858 (2007).51F. Scholz, Semicond. Sci. Technol. 27(2), 024002 (2012).52S. E. Bennett, Mater. Sci. Technol. 26(9), 1017–1028 (2010).53V. Avrutin, D. J. Silversmith, Y. Mori, F. Kawamura, Y. Kitaoka, and H.

Morkoc, Proc. IEEE 98(7), 1302–1315 (2010).54A. Dadgar, M. Poschenrieder, A. Reiher, J. Blasing, J. Christen, A.

Krtschil, T. Finger, T. Hempel, A. Diez, and A. Krost, Appl. Phys. Lett.

82(1), 28–30 (2003).55D. S. Peng, Y. C. Feng, W. X. Wang, X. F. Liu, W. Shi, and H. B. Niu,

J. Phys. D: Appl. Phys. 40(4), 1108–1112 (2007).56V. Ramachandran, R. M. Feenstra, W. L. Sarney, L. Salamanca-

Riba, and D. W. Greve, J. Vac. Sci. Technol., A 18(4), 1915–1918

(2000).57N. G. Weimann, L. F. Eastman, D. Doppalapudi, H. M. Ng, and T. D.

Moustakas, J. Appl. Phys. 83(7), 3656–3659 (1998).58J. S. Speck, Mater. Sci. Forum 353–356, 769–778 (2001).59D. S. Jiang, D. G. Zhao, and H. Yang, Phys. Status Solidi B 244(8),

2878–2891 (2007).60K. Leung, A. F. Wright, and E. B. Stechel, Appl. Phys. Lett. 74(17),

2495–2497 (1999).61S. Nakamura, Jpn. J. Appl. Phys., Part 2 30(10A), L1705–L1707 (1991).62S. Sakai, T. Wang, Y. Morishima, and Y. Naoi, J. Cryst. Growth 221,

334–337 (2000).63N. N. Morgan, Z. Z. Ye, and Y. B. Xu, Mater. Sci. Eng., B 90(1–2),

201–205 (2002).64Z. Liliental–Weber, J. Jasinski, and D. N. Zakharov, Opto-Electron. Rev.

12(4), 339–346 (2004).65J. Jasinski and Z. Liliental-Weber, J. Electron. Mater. 31(5), 429–436

(2002).66See http://www.kymatech.com/ for the specifications of the HVPE grown

bulk GaN templates.67H. Morkoc, Mater. Sci. Eng., R 33(5–6), 135–207 (2001).68T. A. G. Eberlein, R. Jones, S. Oberg, and P. R. Briddon, Appl. Phys.

Lett. 91(13), 132105 (2007).69W. Gotz, N. M. Johnson, C. Chen, H. Liu, C. Kuo, and W. Imler, Appl.

Phys. Lett. 68(22), 3144–3146 (1996).70S. O. Kucheyev, J. S. Williams, and S. J. Pearton, Mater. Sci. Eng., R

33(2–3), 51–107 (2001).71C. Ronning, E. P. Carlson, and R. F. Davis, Phys. Rep. 351(5), 349–385

(2001).72M. H. Zaldivar, P. Fernandez, J. Piqueras, and J. Solis, J. Appl. Phys.

85(2), 1120–1123 (1999).73D. H. Youn, M. Lachab, M. S. Hao, T. Sugahara, H. Takenaka, Y. Naoi,

and S. Sakai, Jpn. J. Appl. Phys., Part 1 38(2A), 631–634 (1999).74O. S. Elsherif, K. D. Vernon–Parry, I. M. Dharmadasa, J. H. Evans-

Freeman, R. J. Airey, M. J. Kappers, and C. J. Humphreys, Thin Solid

Films 520(7), 3064–3070 (2012).75M. Lachab, D. H. Youn, R. S. Q. Fareed, T. Wang, and S. Sakai, Solid-

State Electron. 44(9), 1669–1677 (2000).76K. Saarinen, J. Nissila, J. Oila, V. Ranki, M. Hakala, M. J. Puska, P.

Hautojarvi, J. Likonen, T. Suski, I. Grzegory, B. Lucznik, and S.

Porowski, Physica B 273–274, 33–38 (1999).77A. F. Wright and T. R. Mattsson, J. Appl. Phys. 96(4), 2015–2022 (2004).78Q. M. Yan, A. Janotti, M. Scheffler, and C. G. Van de Walle, Appl. Phys.

Lett. 100(14), 142110 (2012).79Z. H. Feng, B. Liu, F. P. Yuan, J. Y. Yin, D. Liang, X. B. Li, Z. Feng, K.

W. Yang, and S. J. Cai, J. Cryst. Growth 309(1), 8–11 (2007).80S. Heikman, S. Keller, S. P. DenBaars, and U. K. Mishra, Appl. Phys.

Lett. 81(3), 439–441 (2002).81J. Dashdorj, M. E. Zvanut, J. G. Harrison, K. Udwary, and T. Paskova,

J. Appl. Phys. 112(1), 013712 (2012).82D. O. Dumcenco, S. Levcenco, Y. S. Huang, C. L. Reynolds, J. G.

Reynolds, K. K. Tiong, T. Paskova, and K. R. Evans, J. Appl. Phys.

109(12), 123508 (2011).83A. Y. Polyakov, N. B. Smirnov, A. V. Govorkov, V. I. Vdovin, A. V.

Markov, A. A. Shlensky, E. Prebble, D. Hanser, J. M. Zavada, and S. J.

Pearton, J. Vac. Sci. Technol., B 25(3), 686–690 (2007).84A. Sedhain, J. Li, J. Y. Lin, and H. X. Jiang, Appl. Phys. Lett. 96(15),

151902 (2010).85See http://ammono.com/ for the specifications of the Mg-doped semi-

insulating GaN wafer.86D. Alquier, F. Cayrel, O. Menard, A. E. Bazin, A. Yvon, and E. Collard,

Jpn. J. Appl. Phys., Part 1 51(1), 01AG08 (2012).

031102-11 Wang et al. Appl. Phys. Rev. 2, 031102 (2015)

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

50.4.102.168 On: Sat, 26 Sep 2015 15:55:25

87O. Katz, V. Garber, B. Meyler, G. Bahir, and J. Salzman, Appl. Phys.

Lett. 80(3), 347–349 (2002).88J. Wang, P. L. Mulligan, and L. R. Cao, Nucl. Instrum. Meth. A 761,

7–12 (2014).89I. Eliashevich, Y. X. Li, A. Osinsky, C. A. Tran, M. G. Brown, and R. F.

Karlicek, Proc. Soc. Photo-Opt. Instrum. Eng. 3621, 28–36 (1999).90K. Fu, G. H. Yu, C. S. Yao, G. Wang, M. Lu, and G. G. Zhang, Phys.

Status Solidi RRL 5(5–6), 187–189 (2011).91M. Mohamadian, S. A. H. Feghhi, and H. Afarideh, in Proceedings of the

13th International Conference on Emerging of Nuclear Energy System,

2007.92X. B. Tang, Y. P. Liu, D. Ding, and D. Chen, Sci. China Technol. Sci.

55(3), 659–664 (2012).93Y. L. Yuan, C. S. Yao, G. Wang, and M. Lu, Res. Process SSE 32(2), 1–5

(2012).94D. S. McGregor, M. D. Hammig, Y. H. Yang, H. K. Gersch, and R. T.

Klann, Nucl. Instrum. Methods Phys. Res., Sect. A 500(1–3), 272–308

(2003).95D. S. McGregor and J. K. Shultis, Nucl. Instrum. Methods Phys. Res.,

Sect. A 517(1–3), 180–188 (2004).96A. A. Bickley, C. Young, B. Thomas, J. W. McClory, P. A. Dowben, and

J. C. Petrosky, MRS Proc. 1341, 75–80 (2011).97A. N. Caruso, J. Phys.: Condens. Matter 22(44), 1–32 (2010).98P. Kandlakunta and L. R. Cao, J. Radioanal. Nucl. Chem. 300(3),

953–961 (2014).99P. Kandlakunta and L. Cao, Radiat. Prot. Dosim. 151(3), 586–590

(2012).100S. R. McHale, J. W. McClory, J. C. Petrosky, J. Wu, A. Rivera, R. Palai,

Y. B. Losovyj, and P. A. Dowben, Eur. Phys. J.: Appl. Phys. 55(3), 31301

(2011).101L. Wang, W. N. Mei, S. R. McHale, J. W. McClory, J. C. Petrosky, J.

Wu, R. Palai, Y. B. Losovyj, and P. A. Dowben, Semicond. Sci. Technol.

27(11), 115017 (2012).102A. Melton, E. Burgett, M. Jamil, T. Zaidi, N. Hertel, and I. Ferguson, in

IEEE SoutheastCon, 2010, pp. 402–403.103A. G. Melton, E. Burgett, T. M. Xu, N. Hertel, and I. T. Ferguson, Phys.

Status Solidi C 9(3–4), 957–959 (2012).104P. Ramvall, Y. Aoyagi, A. Kuramata, P. Hacke, K. Domen, and K.

Horino, Appl. Phys. Lett. 76(21), 2994–2996 (2000).105L. Cao, Battelle Energy Alliance, LLC Project No. 11–3004, 2015, pp.

1–44.106P. Mulligan, J. Qiu, C. H. Lin, L. J. Brillson, R. G. Downing, and L. Cao,

“Intrinsic neutron sensitivity of GaN and radiation effects on forward-

biased devices,” J. Nucl. Mater. Manage. (to be published).107S. J. Pearton, R. Deist, F. Ren, L. Liu, A. Y. Polyakov, and J. Kim,

J. Vac. Sci. Technol., A 31(5), 050801 (2013).108A. Y. Polyakov, S. J. Pearton, P. Frenzer, F. Ren, L. Liu, and J. Kim,

J. Mater. Chem. C 1(5), 877–887 (2013).109J. G. Marques, K. Lorenz, N. Franco, and E. Alves, Nucl. Instrum.

Methods Phys. Res., Sect. B 249, 358–361 (2006).110K. Kuriyama, Y. Mizuki, H. Sano, A. Onoue, K. Kushida, M. Okada, M.

Hasegawa, I. Sakamoto, and A. Kinomura, Nucl. Instrum. Methods Phys.

Res., Sect. B 249, 132–135 (2006).111P. N. Son, T. T. Anh, C. D. Vu, and V. H. Tan, J. Korean Phys. Soc.

59(2), 1761–1764 (2011).

112E. T. Jurney, J. W. Starner, and J. E. Lynn, Phys. Rev. C 56(1), 118–134

(1997).113K. Lorenz, J. G. Marques, N. Franco, E. Alves, M. Peres, M. R. Correia,

and T. Monteiro, Nucl. Instrum. Methods Phys. Res., Sect. B 266(12–13),

2780–2783 (2008).114V. M. Boyko, S. S. Verevkin, N. G. Kolin, A. V. Korulin, D. I.

Merkurisov, A. Y. Polyakov, and V. A. Chevychelov, Semiconductors

45(1), 134–140 (2011).115A. Y. Polyakov, N. B. Smirnov, A. V. Govorkov, A. V. Markov, N. G.

Kolin, D. I. Merkurisov, V. M. Boiko, K. D. Shcherbatchev, V. T. Bublik,

M. I. Voronova, S. J. Pearton, A. Dabiran, and A. V. Osinsky, J. Vac. Sci.

Technol., B 24(5), 2256–2261 (2006).116R. X. Wang, S. J. Xu, S. Fung, C. D. Beling, K. Wang, S. Li, Z. F. Wei,

T. J. Zhou, J. D. Zhang, Y. Huang, and M. Gong, Appl. Phys. Lett. 87(3),

031906 (2005).117A. Y. Polyakov, N. B. Smirnov, A. V. Govorkov, A. V. Markov, S. J.

Pearton, N. G. Kolin, D. I. Merkurisov, V. M. Boiko, C. R. Lee, and I. H.

Lee, J. Vac. Sci. Technol., B 25(2), 436–442 (2007).118J. Qiu, E. Katz, C. H. Lin, L. Cao, and L. J. Brillson, Radiat. Eff. Defects

Solids 168(11–12), 924–932 (2013).119D. O’Mahony, W. Zimmerman, S. Steffen, J. Hilgarth, P. Maaskant, R.

Ginige, L. Lewis, B. Lambert, and B. Corbett, Semicond. Sci. Technol.

24(12), 125008 (2009).120B. P. Luther, S. D. Wolter, and S. E. Mohney, Sens. Actuators, B

56(1–2), 164–168 (1999).121P. Mulligan, J. Qiu, J. H. Wang, and L. R. Cao, IEEE Trans. Nucl. Sci.

61(4), 2040–2044 (2014).122C. W. Wang, J. Vac. Sci. Technol., B 20(5), 1821–1826 (2002).123C. W. Wang, Appl. Phys. Lett. 80(9), 1568–1570 (2002).124C. H. Lin, E. J. Katz, J. Qiu, Z. C. Zhang, U. K. Mishra, L. Cao, and L. J.

Brillson, Appl. Phys. Lett. 103(16), 162106 (2013).125E. J. Katz, C. H. Lin, J. Qiu, Z. C. Zhang, U. K. Mishra, L. Cao, and L. J.

Brillson, J. Appl. Phys. 115(12), 123705 (2014).126L. Dobos, B. Pecz, L. Toth, Z. J. Horvath, Z. E. Horvath, A. Toth, E.

Horvath, B. Beaumont, and Z. Bougrioua, Appl. Surf. Sci. 253(2),

655–661 (2006).127Z. F. Fan, S. N. Mohammad, W. Kim, O. Aktas, A. E. Botchkarev, and H.

Morkoc, Appl. Phys. Lett. 68(12), 1672–1674 (1996).128Q. Feng, L. M. Li, Y. Hao, J. Y. Ni, and J. C. Zhang, Solid State Electron.

53(9), 955–958 (2009).129S. Ruvimov, Z. Liliental-Weber, J. Washburn, K. J. Duxstad, E. E.

Haller, Z. F. Fan, S. N. Mohammad, W. Kim, A. E. Botchkarev, and H.

Morkoc, Appl. Phys. Lett. 69(11), 1556–1558 (1996).130N. A. Papanicolaou, M. V. Rao, J. Mittereder, and W. T. Anderson,

J. Vac. Sci. Technol., B 19(1), 261–267 (2001).131Z. X. Qin, Z. Z. Chen, Y. Z. Tong, X. M. Ding, X. D. Hu, T. J. Yu, and

G. Y. Zhang, Appl. Phys. A: Mater. 78(5), 729–731 (2004).132Q. Z. Liu and S. S. Lau, Solid-State Electron. 42(5), 677–691 (1998).133J. Kim, F. Ren, A. G. Baca, and S. J. Pearton, Appl. Phys. Lett. 82(19),

3263–3265 (2003).134E. Monroy, F. Calle, R. Ranchal, T. Palacios, M. Verdu, F. J. Sanchez,

M. T. Montojo, M. Eickhoff, F. Omnes, Z. Bougrioua, and I. Moerman,

Semicond. Sci. Technol. 17(9), L47–L54 (2002).135N. Yildirim, K. Ejderha, and A. Turut, J. Appl. Phys. 108(11), 114506

(2010).

031102-12 Wang et al. Appl. Phys. Rev. 2, 031102 (2015)

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

50.4.102.168 On: Sat, 26 Sep 2015 15:55:25