See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/293642859

Effect of GaN buffer polarization on electron

distribution of AlGaN/GaN heterostructure

Article in Journal of Alloys and Compounds · February 2016

DOI: 10.1016/j.jallcom.2016.02.047

CITATIONS READS

0 41

6 authors, including:

Xiaoguang He Jing Yang

Chinese Academy of Sciences CAS
34 PUBLICATIONS 42 CITATIONS 20 PUBLICATIONS 15 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Xiaoguang He on 25 May 2016.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
 Journal of Alloys and Compounds 670 (2016) 258e261

Contents lists available at ScienceDirect

Journal of Alloys and Compounds

journal homepage: http://www.elsevier.com/locate/jalcom

Effect of GaN buffer polarization on electron distribution of AlGaN/

GaN heterostructure
Xiaoguang He, Degang Zhao*, Wei Liu, Jing Yang, Xiaojing Li, Xiang Li
State Key Laboratory on Integrated Optoelectronics, Institute of Semiconductors, Chinese Academy of Sciences, P.O. Box 912, Beijing 100083, China

a r t i c l e i n f o a b s t r a c t

Article history: The formation of 2DEG in AlGaN/GaN heterostructure is discussed in detail. A misunderstanding about
Received 29 December 2015 the 2DEG sheet density expression is clarified. It is predicted by theoretical analysis and validated by self-
Accepted 6 February 2016 consistent SchrodingerePoisson numerical simulation that under the force of GaN polarization, large
Available online 8 February 2016
amounts of electrons will accumulate at the GaN/substrate interface in AlGaN/GaN/substrate HEMT
structure.
Keywords:
© 2016 Elsevier B.V. All rights reserved.
GaN
HEMT
Polarization
2DEG

1. Introduction s ε

ns ¼  2 q4b þ Ef  DEc ; (1)
q q d
High electron mobility transistors (HEMTs) based on AlGaN/GaN
heterostructures have shown great potential in high-power and where ε is the dielectric constant of AlGaN, d is the AlGaN layer
high-frequency applications [1]. Owing to improvements made thickness, q is the electron charge, 4b is the surface barrier height, Ef
during the last decades in both material quality and device pro- is the Fermi level, DEc is the conduction band offset between GaN
cessing, AlGaN/GaN HEMTs have been improved significantly in and AlGaN, as is shown in Fig. 1. By transposing, Eq. (1) can be
both DC and RF performance. An accurate expression of the 2DEG changed into:
sheet carrier concentration ns is of great importance to the further
understanding of AlGaN/GaN HEMTs. There is an important and 1

s qns
very much widely cited equation reported for calculating the 2DEG q4b þ Ef  DEc ¼ d  : (2)
q ε ε
concentration ns (see Eq. (1) in this article, which will be discussed
later in detail). However, the meaning of “s” in this formula is According to Fig. 1, the potential drop of AlGaN barrier layer can
confusing and has some logical contradictions. In this work, we be expressed as:
carefully analyze the process of the 2DEG formation and provide a
clear physical image of the AlGaN/GaN heterostructure. The qDVAlGaN ¼ qfb þ Ef  DEC : (3)
meaning of “s” in the formula should be reinterpreted.
Putting Eq. (3) into Eq. (2), we can get:

DVAlGaN s qns
¼  : (4)
d ε ε

2. Aboudt the “s” Eq. (4) indicates that the electric field in AlGaN barrier layer is
determined by two parallel plate capacitors, of which sheet charge
The widely cited 2DEG sheet density expression is given as density is ±s and ±qns respectively.
[2e4]: The structural profile and charge distribution of a AlGaN/GaN
heterojunction is shown in Fig. 2. We can see that there are indeed
two pairs of coupled sheet charges at the upper surface and the
* Corresponding author. lower interface of AlGaN: ① ss and qns, ② ±sAlGaN. ss is the sheet
E-mail address: dgzhao@red.semi.ac.cn (D. Zhao). charge density of ionized surface donors. It equals to the product of

http://dx.doi.org/10.1016/j.jallcom.2016.02.047
0925-8388/© 2016 Elsevier B.V. All rights reserved.
 X. He et al. / Journal of Alloys and Compounds 670 (2016) 258e261
Fig. 1. Conduction band diagram of AlGaN/GaN heterostructure.
Fig. 2. Structure profile and charge distribution of AlGaN/GaN/substrate
heterostructure.
electron charge and the 2DEG sheet density qns, because the elec-
trons in 2DEG is considered coming from the surface states of
AlGaN [5,6]. sAlGaN is the polarization charge sheet density of
AlGaN. Therefore, according to the above mentioned discussion, the
“s” in Eq. (4) refers to the AlGaN polarization charge density, i.e.,
s ¼ sAlGaN.
However, according to previous works [2e4,7e10] which have
mentioned the 2DEG sheet density expression (Eq. (1)), the term
“s” refers to the difference value of AlGaN polarization charge and
GaN polarization charge, i.e., s ¼ sAlGaN  sGaN. This argument
makes the formula conforms to the result of experiment well, but
has some logical contradictions. Admittedly, the net polarization
charge sheet density at the AlGaN/GaN interface is
snet ¼ sAlGaN  sGaN. However, if s ¼ sAlGaN  sGaN, according to Eq.
(4), the absolute value of net polarization charge density at the
AlGaN upper surface should be equal to sAlGaN  sGaN, too. But this is
obviously contradictory to the fact that the only polarization charge
at the AlGaN surface is sAlGaN. There is not a “þsGaN” at the AlGaN
surface. Moreover, it should be noticed that there is a positive po-
larization charge þsGaN at the GaN/substrate interface, as is shown
in Fig. 2. ±sGaN at the upper and lower interface of GaN layer form a
parallel plate capacitor, which will only have nonzero electric field
outside the GaN buffer layer. This means that polarization charge of
GaN buffer will have no affect on the potential drop of AlGaN bar-
rier layer.
259
Actually, in previous works, polarization effect of GaN buffer is
neglected when calculating the band diagram. Polarization sheet
charge density of AlGaN is considered as the relative value
comparing with GaN's: sAlGaN  sGaN, and polarization charge of
GaN is regarded as zero. But why?
3. New calculation
Taking polarization effect of GaN buffer into consideration, there
will be a polarization field FGaN in GaN layer pointing from bottom
to the top. Adding the potential drop caused by this electric field,
the conduction band diagram should be like the one shown in
Fig. 3, but not the one in Fig. 1. Moreover, we should notice that all
the above-mentioned discussion is based on an a priori assumption
that the 2DEG locates at the AlGaN/GaN interface. In fact, charge
distribution shown in Fig. 2 is not stable, because that electrons will
move towards the GaN/substrate interface under the force of FGaN,
rather than fix at the AlGaN/GaN interface.
The conduction band diagram and the corresponding electron
distribution of a 20 nm Al0.25GaN0.75/20 nm GaN/substrate HEMT
structure is calculated by self-consistent SchrodingerePoisson nu-
merical simulation, as is shown in Fig. 4a. The lattice parameter a,
polarization intensity of Psp, Ppz, Ptotal of GaN and AlGaN that we
used in the calculation are shown in Table 1. Here we only discuss
under an assumption that the substrate material has a higher
conduction band minimum than GaN's (e.g., sapphire). Since the
specific band structure of substrate does not affect the electron
distribution much, the conduction band edge of substrate is arti-
ficially set as 3 eV higher than GaN's. According to our calculation,
electrons not only accumulate at the AlGaN/GaN interface, there is
also another electron distribution peak at the GaN/buffer interface.
It is found that the width of GaN buffer affect neither the shape of
the two electron peaks nor the shape of conduction band near the
two interfaces. As long as the polarization of GaN is taken into
consideration, electrons will move toward the substrate, and as
long as the substrate has a higher conduction band edge than GaN,
electrons will accumulate at the GaN/substrate interface. The band
diagram and electron distribution of a HEMT structure with 40 nm
GaN buffer is shown in Fig. 4b as an example. Therefore, for defi-
niteness and without loss of generality, we will only discuss the
characteristics of the HEMT structure with 20 nm GaN buffer in the
following discussion.
Comparing with conventional works, the only difference of our
calculation is that polarization of AlGaN and GaN is set as the
original value, i.e., sAlGaN and sGaN respectively. For comparison,
band diagram and electron distribution of a 20 nm AlGaN/20 nm
GaN/20 nm substrate HEMT structure is also calculated under
conventional conditions, i.e., polarization sheet charge density of
AlGaN and GaN is considered as sAlGaN  sGaN and 0 respectively, as
is shown in Fig. 4c. It can be found that Fig. 4a and c are almost the
same at the region of 0e30 nm. They have very similar band dia-
gram and electron distribution shape. The electron sheet density of
20 nm AlGaN/20 nm GaN/substrate structure from 0 to 30 nm, i.e.
the integral intensity of the first electron distribution peak, is
Fig. 3. Conduction band profile of AlGaN/GaN heterojunction when taking polarization
effect of GaN buffer into consideration.
260 X. He et al. / Journal of Alloys and Compounds 670 (2016) 258e261

Fig. 4. Conduction band diagram and electron distribution of AlGaN/GaN/substrate heterostructure (a) with 20 nm GaN buffer, (b) with 40 nm GaN buffer, (c) with 20 nm GaN
buffer and under conventional conditions.

Table 1 the GaN conduction band. Finally, they keep moving to the GaN/
Lattice parameter a, polarization intensity of Psp, Ppz, Ptotal of GaN and AlGaN. substrate interface, blocked by the barrier of substrate and accu-
GaN Al0.25Ga0.75N mulate there. Therefore, the stable location for electrons is at the
lattice parameter a (nm) 0.3189 0.3170
GaN/substrate interface but not at the AlGaN/GaN interface. As is
Psp (C/cm2) 0.0290 0.0384 shown in Fig. 6a, the electrons at the GaN/substrate interface and
Ppz (C/cm2) 0 0.0067 their corresponding ionized positive charge at AlGaN surface will
Ptotal (C/cm2) 0.0290 0.0452 form a parallel plate capacitor and its electric field Fns2 will weaken
Ptotal  P(GaN) (C/cm2) 0 0.0162
the polarization field in GaN and make the potential drop smaller.
As the 2DEG sheet density increases, the slope of GaN conduction
band will become smaller and smaller. When ns ¼ sGaN, the polar-
ization field in GaN will be totally counteracted, the conduction
Z nm
30
band will be flat, as is shown in Fig. 5b. Under this circumstance, if
ns1 ¼ nðzÞdz ¼ 0:75  1013 cm2 additional electrons are ionized from surface states into GaN layer,
0 nm no force will drive them to the GaN/substrate interface, on the
contrary, they will be attracted by their corresponding positive
which is almost the same value as the total sheet density of con- charge of ionized surface states and accumulate at the AlGaN/GaN
ventional result. While the sheet density of the second peak is: interface. These electrons form the 2DEG that we usually refer to.
The charge distribution diagram and corresponding electron field
Z
80nm
distribution under the circumstance of ns > sGaN is shown in Fig. 6b,
ns2 ¼ nðzÞdz ¼ 1:81  1013 cm2 we can see that the potential drop of AlGaN can be given as:
30nm

which is just the same value of the sheet number density of GaN DVAlGaN sAlGaN qns1 qns2
polarization charge: ¼  
d ε ε ε
sGaN According to the above-mentioned discussion, we know that
nGaN
s ¼ ¼ 1:81  1013 cm2
q ns2 ¼ sGaN, so the equation can be given as:
Is this a coincidence?

DVAlGaN sAlGaN  sGaN qns1

4. Process of 2DEG formation ¼ 
d ε ε

The process of 2DEG formation is modulated by self-consistent Thus we can get an expression of 2DEG sheet density exactly the
SchrodingerePoisson numerical simulation. According to the same as Eq. (4). But the electric field in AlGaN barrier layer is
above-mentioned calculation, the total electron sheet density of determined by three parallel plate capacitors, but not two. The
the AlGaN/GaN/substrate structure is reason why the s term should be given as sAlGaN  sGaN rather than
Z 80nm sAlGaN is that the electrons accumulation at the GaN/substrate
13 2 GaN
ns ¼ nðzÞdz ¼ ns1 þ ns2 ¼ 2:56  10 cm ¼ 1:41ns . We interface and their corresponding positive charge of ionized surface
0nm
states form a third parallel plate capacitor, partially counteracting
artificially set the total electron sheet density as 0:9nGaN , nGaN , and
s s the field of AlGaN polarization.
1:1nGaN
s respectively to calculate the electron distribution and band However, the high density electrons at the GaN/substrate
diagram under different circumstances, as is shown in Fig. 5aec. It interface maybe hard to observe since usually the GaN epilayer
is found that if ns < nGaN
s , electrons will all accumulate at the GaN/ contains a bad-quality nucleation layer [11] near the GaN/substrate
substrate interface, and only when ns exceeds the value of nGaN
s , the interface. It is reported that there is a conductive GaN layer near the
2DEG at the AlGaN/GaN interface begins to form. GaN/sapphire interface due to the thermal diffusion of the oxygen
Let's discuss the process of the formation of 2DEG. First, the in sapphire substrate [12e14]. According to our investigation, the
electrons at the AlGaN surface states are stimulated to the con- formation of conductive layer near the substrate may also have
duction band [5,6]. Then, driven by the force of polarization field in something to do with the electron accumulation caused by the
AlGaN, electrons slide toward AlGaN/GaN interface and drop into force of GaN. Further experiments are needed to test the theory.
 X. He et al. / Journal of Alloys and Compounds 670 (2016) 258e261 261

Fig. 5. Conduction band diagram and electron distribution of AlGaN/GaN/substrate heterostructure with the total electron sheet density artificially set as (a) 0:9nGaN
s , (b) nGaN
s , and
(c) 1:1nGaN
s respectively.

Fig. 6. Structure profile and charge distribution of AlGaN/GaN/substrate heterostructure (a) ns  sGaN (b) ns > sGaN.

5. Conclusions
It is theoretically analyzed and validated by self-consistent
calculation that under the force of GaN polarization, large
amounts of electrons will accumulate at the GaN/substrate inter-
face in AlGaN/GaN/substrate HEMT structure. Only when the
amount of electrons at the GaN/substrate interface is sufficient to
totally counteract the polarization field in GaN buffer, the 2DEG at
the AlGaN/GaN is able to form. The existence of the GaN/substrate
interface electrons remains to be proved by experiments.
Acknowledgments
The authors acknowledge the support from the National Natural
Science Foundation of China (Grant Nos. 61474110, 61377020,
61376089, 61223005 and 61176126), the National Science Fund for
Distinguished Young Scholars (Grant No. 60925017), One Hundred
Person Project of the Chinese Academy of Sciences, and Basic
Research Project of Jiangsu Province (Grant No. BK20130362).
References
[1] A. Soltani, J.C. Gerbedoen, Y. Cordier, D. Ducatteau, M.
M. Chmielowska, M. Ramdani, J.C. De Jaeger, IEEE Electron Device Lett. 34 (4)
(2013) 490e492.
[2] E.T. Yu, X.Z. Dang, L.S. Yu, D. Qiao, P.M. Asbeck, S.S. Lau, G.J. Sullivan,
K.S. Boutros, J.M. Redwing, Appl. Phys. Lett. 73 (13) (1998) 1880.
[3] P.M. Asbeck, E.T. Yu, S.S. Lau, G.J. Sullivan, J. Van Hove, J. Redwing, Electron.
Lett. 33 (14) (1997) 1230e1231.
[4] E.T. Yu, G.J. Sullivan, P.M. Asbeck, C.D. Wang, D. Qiao, S.S. Lau, Appl. Phys. Lett.
71 (19) (1997) 2794.
[5] I.P. Smorchkova, C.R. Elsass, J.P. Ibbetson, R. Vetury, B. Heying, P. Fini, E. Haus,
S.P. DenBaars, J.S. Speck, U.K. Mishra, J. Appl. Phys. 86 (8) (1999) 4520.
[6] J.P. Ibbetson, P.T. Fini, K.D. Ness, S.P. DenBaars, J.S. Speck, U.K. Mishra, Appl.
Phys. Lett. 77 (2) (2000) 250.
[7] O. Ambacher, B. Foutz, J. Smart, J.R. Shealy, N.G. Weimann, K. Chu, M. Murphy,
A.J. Sierakowski, W.J. Schaff, L.F. Eastman, R. Dimitrov, A. Mitchell,
M. Stutzmann, J. Appl. Phys. 87 (1) (2000) 334.
[8] B.K. Ridley, O. Ambacher, L.F. Eastman, Semicond. Sci. Technol. 15 (3) (2000)
270e271.
[9] M. Gonschorek, J.F. Carlin, E. Feltin, M.A. Py, N. Grandjean, V. Darakchieva,
B. Monemar, M. Lorenz, G. Ramm, J. Appl. Phys. 103 (9) (2008) 093714.
[10] O. Ambacher, J. Smart, J.R. Shealy, N.G. Weimann, K. Chu, M. Murphy,
W.J. Schaff, L.F. Eastman, R. Dimitrov, L. Wittmer, M. Stutzmann, W. Rieger,
J. Hilsenbeck, J. Appl. Phys. 85 (6) (1999) 3222.
[11] S. Nakamura, Jpn. J. Appl. Physics. pt Lett. 30 (1991).
[12] K.S. Kim, M.G. Cheong, H.K. Cho, E.K. Suh, H.J. Lee, Appl. Phys. Lett. 80 (5)
(2002) 799.
[13] D.C. Look, R.L. Jones, X.L. Sun, L.J. Brillson, J.W. Ager, S.S. Park, J.H. Han,
R.M. Molnar, J.E. Maslar, J. Phys. -Condes. Matter 14 (48) (2002)
13337e13344.
[14] D.C. Look, R. Molnar, Appl. Phys. Lett. 70 (25) (1997) 3377e3379.
Rousseau,

View publication stats