What does Pressure Tell Us about
Nitrogen in III-V Alloys?
Bernard A. Weinstein
State University of New York, Buffalo, NY USA
¾ The talk will explore the idea that applied pressure can filter
energetically overlapping localized and extended states.
Topic will be developed by way of three examples
¾ Yellow luminescence in GaN
¾ N-pair (cluster) states in GaAs1-xNx dilute alloys (work at
Buffalo: Phys. Rev. B68, 035336 (2003))
Consider a heuristic model from amorphous semicond.
¾ IR luminescence in InN
¾ Summary
Yellow luminescence & related deep levels in unintentionally doped GaN Films
I. Shalish, L. Kronik, G. Segal, Y. Rosenwaks, Y. Shapira, U. Tisch, J. Salzman
Phys. Rev. B 59, 9748 (1999) “The deep level energy distribution associated with the
well-known “yellow luminescence” in GaN is studied by means of … photoluminescence
and surface photovoltage spectroscopy. … results show that the yellow luminiescence is
due to capture of conduction band electrons, …by a deep acceptor level with a broad
energy distribution, centered at ~ 2.2eV below the conduction band edge. …the density of
yellow luminescence related states possesses a significant surface component.”
I have a better idea!
I. Shalish, et. al. Phys. Rev.
B61, 15573(2000)
T. Suski, P. Perlin, et. al. Appl. Phys. Lett. 67, 2189 (1995)
Mechanism of yellow luminescence in GaN
dE PL
meV dE g
= 3.0 ± 0.2
≈
dP
kbar
dP
These authors found:
From this it was suggested that the PL transition was from
CB (shallow donor)
But this depended on
⇒
Deep Acceptor
dE g
dP
≈
dECBE
dP
or
dEVBE
≈0
dP
Schematic of high-pressure low-temperature
photoluminescence apparatus
BACKROUND: PL spectra in GaAs doped with N at
impurity levels (< 1019cm-3).
GaAs:Si:N(unintend)
[N] ~ 5x1016 cm-3
[nd-na] = 2x1014 cm-3
Tmeas = 8K
N impurities give rise to a
complex spectrum of sharp
PL lines, which depends
strongly on pressure. See,
e.g., Hsu, et. al. PRB 16, 1597
(1977).
Nx + phonon
replica
1.6
1.7
1.8
1.9
2.0
Energy (eV)
2.1
2.2
2.3
M. Li, Ph.D. thesis, UB 1992
1.9
1.8
En
erg
y
(eV
)
1.7
x = 0.0025
1.6
x = 0.004
1.5
The PL is due to excitons
localized at isolated NAs and NNipair (i=1-5) centers, whose states
are either bound or CB-resonant
at 1 atm, but can be driven into
the gap by pressure.
[N]~5 x1017cm-3
1.4
0
10
20
Pressure (kbar)
30
Liu, et. al., APL 56, 1451(1990)
In GaAs1-xNx with N at dilute-alloy levels, x ~ 0.05%-5%
N-Impurity Limit -- Liu,
1 56,
et. al., Appl. Phys. Lett.
1451 (1990)
Impurity Bands,
as perhaps in GaP1-xNx --
Xin, et.al., APL 76,1267(2000)
There are three views of
how the CB-structure of
GaAs1-xNx dilute-alloys
evolves with increasing
N-content.
2
3
NAslevel-CBE Mixing
Walukiewicz, 2-level model
Shan et. al., J. Appl. Phys. 86,
2349 (1999)
Multiband hybridization
of NAs & N-pair levels
Kent & Zunger, PRL
86, 2613 (2001)
PL spectra at several pressures of the
GaAs1-xNx/GaAs sample with x=0.0025
Peak Assignments
Best choices are:
#1 NN1-2LO
#2 NN1-3LO
#3 NN1-LO
Sample #615
GaAs1-xNx/GaAs
x=0.0025,T=9K
#4
&
#5
NN3-(1,2)LO
&?
}
NN4-(2,3)LO
1 atm
• Numbers label observed PL features of N-N pairs.
• Arrows mark calculated QW-bandgap using BAC model.
• Insert: QW and GaAs-substrate absorption spectrum at
1 atm for this sample
PL spectra at several pressures of the
GaAs1-xNx/GaAs sample with x=0.004
Peak Assignments
#1 NN1-2LO
#2 NN1-3LO
#3 NN1-LO
#4
&
#5
•
•
} Missing!
Numbers label observed PL features of N-N pairs.
Arrows mark calculated QW-bandgap using
the BAC model.
Pressure-dependence of the PL features
observed in the x=0.0025 sample
Slope of NN1-2LO
from Liu et. al.
• Data points obtained from multi-oscillator fits to spectra.
• Solid curve is calculated QW-bandgap using the BAC model.
• Stars (and dotted fit) are measured for bulk GaAs bandgap in
sample's substrate.
Pressure-dependence of the PL features
observed in the x=0.004 sample
•
•
•
Data points obtained from multi-oscillator fits to spectra.
Solid curve is calculated QW-bandgap using the BAC
model.
Stars (and dotted fit) are measured for bulk GaAs
bandgap in substrate of x=0.0025 sample.
TO SUMMARIZE
A key result believed to arise from delocalization:
Features 1, 2, & 3, the NN1-pair harmonics are present at
equivalent pressures in GaAs1-xNx for x= 0.25% & x=0.4%.
Features 4, 5, & 6, related to NN3 & NN4 harmonics are
present for x=0.25% but not for x=0.4% at the equivalent
pressure.
x=0.25%
60 kbar
62.2 kbar
x=0.4%
16 kbar
x=0.25%
17.5 kbar
x=0.4%
Heuristic picture based on a 3-channel PL mechanism in
dilute-N GaAs 1-x Nx/GaAs QWs
ε
εa
εb
CBE
Energy
0
VBE
n(ε)
Density of States
Ia
= (QE )ar N a β as τ sr
Is
r
I b (QE )b N bβ bs
=
+ (QE )rb N bβ ba τ ar
I a (QE )r N a βas
a
PL line-width σ ~ 0.2µ3/2(δEg), where δEg = x (1 − x ) ∂E g ∂x
Question: Why do some of the localized N-N
pair (N-cluster) states vanish, but not others??
0.44% - N
62.2 kbar
60 kbar
NN3-pairs
(NN2-pairs)
0.25%-N
Two possibilities:
Features 4 & 5 become X-like and forbidden as in GaP1-xNx.
NO -- the Γ-X crossover should be at ~ 80kbar.
Features 4 &5 no longer emerge below the CB-edge. Their
states stay resonant due to full hybridization into the band.
More likely. Implies: Localized
De-localized transition.
Kent & Zunger, PRB 64, 115208 (2001)
Recently, the pressure-shift of the IR luminescence peak
in MBE grown InN and In-rich InGaN alloys was measured
and found to be very small, dEPL ~ +0.6 meV .
dP
kbar
S. X. Li, et. al., Appl. Phys. Lett. 83, 4963 (2003)
Strong localization of the exciton state was invoked and
the full shift of 0.6meV/kbar was attributed to the VBE.
This is pushing the accuracy. Localized states often have
dE
P-shifts < 1meV/kbar, but it gives a useful estimate for VBE .
dP
Summary
¾ Use of high pressure as a filter to distinguish overlapping localized & extended states was discussed.
¾
Examples were illustrated for the
1) “Yellow luminescence” in GaN
2) NN-pair (cluster) states in dilute GaAs1-xNx
3) Recent studies of the IR PL in InN
The method is very useful when combined with:
a) corroborating results, as for the surface
studies of the yelow PLdeep acceptor by
Shalish et. al.
and
b) careful analysis and calculations that go
beyond simple estimates.
Much thanks is due to my collaborators:
S. Stambach,T. Ritter, J. Maclean, D. Wallis
A special thanks for inviting me to Tel Aviv to help
celebrate this happy time with you and Yoram.