0.5 µm/hr. For p-doping, the Magnesium source
temperature was 280°C. The main GaN epilayers had a typical thickness
of 2 µm.
* Department of Materials Science and Mineral Engineering, UC
Berkeley, Berkeley,
CA 94720
** Lawrence Berkeley National Laboratory, Materials Science
Division, Berkeley,
CA 94720
1 joachim@socrates.berkeley.edu
The paper describes the influence of strain on the optical quality of GaN films grown by MBE on c-plane sapphire. The photoluminescence (PL) line width of the donor-bound exciton can be designed to be as narrow as 1.2 meV by actively utilizing hydrostatic and biaxial stress components. Unstrained p-type Mg-doped GaN films exhibit comparably narrow near band edge transitions. A sharp PL line at 3.261 eV in some of our films is identified as the donor bound exciton of the cubic phase. The formation of these cubic inclusions can be stimulated by a high III/V flux ratio at the growth temperature of T = 725°C. The PL spectrum of an InGaN multi quantum well structure is significantly broadened compared with the spectra of single quantum well structures. Combination of PL and TEM indicates that this effect relates to a progressive increase of the quantum well widths and their spacing along the growth direction. It is argued that strain affects the growth rate and the incorporation of Indium into the quantum well structures.
In recent years, the growth of GaN has been developed within an unprecedented short time [1]. Metal-organic-chemical-vapor-deposition (MOCVD) and related methods were established to be the major crystal growth technologies for the large scale production of optoelectronic devices on GaN, such as LED's and lasers. Drawbacks of these rapid developments relate to the poor understanding of physical processes that determine and limit the material quality. Strain was recently found to be a key issue that influences many of the materials properties, such as the film morphology, the bandgap, electrical or optical properties.
It is the purpose of this paper to focus on selected optical properties of GaN that can be altered by the presence of strain. In case of GaN thin film growth, the strain originates from the growth on lattice mismatched substrates (15%) with a thermal expansion coefficient that differs from GaN. It is pointed out that the strain can be engineered by the growth of suitable buffer layers.
In case of InN/GaN quantum well structures, an additional strain component comes from the 11% lattice mismatch between GaN and InN. Experimental results indicate that this strain affects the growth of multi quantum well structures.
The GaN films were grown using a rebuilt Riber 1000 chamber. Gallium was
evaporated from a Knudsen effusion cell. The activated nitrogen species were
provided by a Constricted Glow Discharge Plasma Source (for details see [2]).
Prior to the growth, the sapphire substrate was nitridated for 10 minutes.
Subsequently, a buffer layer of typically 25 nm thickness was grown. Typical
growth conditions are: Ga source temperature: 1210 K, nitrogen flow rate: 35
sccm, buffer layer growth temperature: 775K, main layer growth temperature:
1000K, growth rate 0.5 µm/hr. For p-doping, the Magnesium source
temperature was 280°C. The main GaN epilayers had a typical thickness
of 2 µm.
Photoluminescence was excited using the 325nm line of a 50mW HeCd laser. The luminescence light was then dispersed by a 0.85m double monochromator and detected with a photomultiplier using a lock-in technique. The sample temperature was varied between 4 K and 300 K. Hall effect measurements were carried out at room temperature using the van der Pauw configuration.
Thin films of hetero-epitaxially grown GaN are known to incorporate strain of
up to 1 GPa, due to lattice mismatch and different thermal expansion
coefficient of substrate and GaN epilayer. The amount and sign of strain are
determined by the thickness and growth temperature of the buffer layers as well
as by the film stoichiometry [3]. In particular, it is be noted that films
grown on sapphire were found to be either under compressive or tensile stress
depending on the growth conditions.
From the inspection of a variety of samples grown under different conditions strain is concluded to be the main broadening mechanism for PL lines at helium temperatures. Consequently, crystals being strain free at cryogenic temperatures exhibit PL line widths of the donor bound exciton (DX) as low as 1.2 meV (FWHM), figure 1. To the best of our knowledge, this represents the smallest line width detected so far for a hetero-epitaxially grown GaN film. This value has only been surpassed by MOCVD-films grown on bulk GaN (line width of donor bound exciton = 0.55 meV), [8]. Further indications of high optical quality of our films are the low yellow-to-exciton intensity ratio of about 10-5 as well as the fact that phonon replica of the excitonic transitions up to the third order are visible.
To date, a free p-type carrier density of [p] = (1-2) * 1017 cm-3 along with a carrier mobility of 10 <= µ <= 41 cm2/Vs could be achieved for Mg-doped p-type MBE-grown materials. The PL spectrum is dominated by either the donor-acceptor-transition at about 3.26 eV (ZPL, see figure 2) or the broad luminescence centered at 3.0 eV which has been assigned to the Mg acceptor. The barely visible excitonic transitions (line width = 3 meV) consist of the DX at 3.4687 eV and the acceptor bound exciton (AX) at 3.4637 eV. In agreement with [8] we assign this line to the Mg acceptor.
Figure 3 shows a PL spectrum which was frequently detected for our MBE-grown samples. The donor bound exciton (DX) of the hexagonal phase at 3.468 eV (FWHM=2.2meV) and a 9.0 meV broad line at 3.261 eV are clearly visible. In the following paragraph we present argu-ments assigning this transition to the DX of the cubic phase. Also, we find a structure consisting of three lines (ZPL = 3.157 eV). Temperature varia-tion proved this to be the donor-acceptor-pair (plus two phonon replica) in the cubic phase, in agreement with cathodo-luminescence results taken from cubic GaN grown on GaAs [7].
The following signals of different origin are commonly observed in the
energetic region between 3.25 eV and 3.28 eV and can overlap with the DX
signal of the cubic phase: the second phonon replica of the acceptor bound
exciton of the hexagonal phase, the zero-phonon-transition of the
donor-acceptor of the hexagonal phase [4], and the 6. replica of the resonant
Raman line [5]. By temperature variation, one can clearly distinguish these
signals from the excitonic transition of the cubic phase.
TEM confirmed the presence of cubic inclusions in the MBE-grown materials [10]. They account for about 1% of the hexagonal matrix and have been found through-out the whole epilayer volume. This indicates a spontaneous nuclea-tion, with no pre-ference for the substrate interface. In a series grown with different nitro-gen flux we found the existence of cubic inclusions to be stoichiometry rela-ted, crystals grown under gallium-rich conditions show the strongest signal.
In order to identify the exact nature of the luminescent transition of the cubic phase the temperature dependence of the line positions of the hexagonal and cubic phase transitions was taken, figure 4. The DX of the cubic phase is visible up to room temperature, exhibiting a line width of 45.0 meV (Lorentzian). The line shape and width is identical with the free excitonic transition (FX) of the hexagonal phase. This finding additionally underlines the excitonic character of this line.
Generally, the bandgap temperature dependence of a semiconductor can be described by Cody's formula [6]:
 |
with 
being a constant and 
the Einstein temperature.
FX (T=4K)
[eV]
|
DX (T=4K)
[eV]
|
[meV]
|
[K]
| |
| hexagonal
phase
|
3.4741
|
3.4683
|
140 *
± 11
|
340 *
± 25
|
| cubic
phase
|
3.272 *
± 0.002
|
3.2605
|
135 *
± 14
|
340
|
(Parameters derived by fitting Cody's equation (1) are marked with * )
We have fitted this equation to both the hexagonal and the cubic
excitonic PL signals. Due to the higher line width of 9 meV, the transition
from the DX to the FX signal is not resolved for the cubic phase. Apparently,
the excitonic line shifts to higher energies for temperatures up to 50K. We
believe this to be caused by the relative intensity increase of the free
exciton. For the same reason, we were only able to fit Cody's relation
for temperatures above 150 K, figure 4. This restriction doesn't allow
to derive the Einstein temperature from the experimental data, so that we
assumed it to be identical with the value of the hexagonal phase which has been
determined to be = 340 K. Such a value was also reported by other
authors ([9] and references therein). Consequently, we get the energetic
position of the free exciton at 4K as (3.272 ± 0.002) eV, table I. This
is also in agreement with CL investigations of cubic GaN [7].
Figure 5 shows the low temperature PL spectra of a two single quantum well (SQW) and one multi quantum well structure (MQW). The MQW structure consists of 10 wells grown under identical conditions.
Two effects are apparent: The PL line width increases with increasing Indium content for the InGaN single quantum well structure. More strikingly, for the multi quantum well structure we observe an apparent line width of about 400 meV. Closer inspection reveals the line as being composed of several transitions. This finding can not be explained by quantum effects within the well. Therefore, a structural analy-sis of the well structures was conducted using high reso-lution TEM. The Indium distri-bution was spatially resolved probed [11, 12]. One realizes that the Indium content varies by up to 20% within the quantum wells. In fact, it was found for the MQW structure that the well width as well as the well spacing continuously increase with progressive growth. However, the PL results also indicate that the Indium concentration varies from well to well. Since strain increases with the growth of each well, we feel that it is strain that effect both the growth rate and the Indium incorporation into the wells.
In this paper we show that strain engineering plus the usage of a constricted glow discharge plasma source providing reactive species with low kinetic energy results in GaN films of high optical quality. We present PL spectra of p-type GaN with clearly resolved Mg-related acceptor bound exciton transition. Also, we identify the donor bound exciton of the cubic phase and analyze the PL line width broadening mechanism of multi quantum well structures. The selective examples depict how strain impacts optical properties of GaN thin films.
We would like to gratefully acknowledge the supply of the MOCVD-grown quantum well structures by Nichia Chemical Industries, APA Optics, and Hewlett Packard Company. We acknowledge discussions with Zuzanna Liliental-Weber. This work was supported by the Office of Computational and Technology Research, Advanced Energy Projects and the Laboratory Technology Research Program (ERLTR) of the U.S. Department of Energy under Contract No. DE-AC03-76SF00098.
[1] S.Nakamura; SPIE, Proc. 2693, 43 (1996)
[2] A.Anders and S.Anders; Plasma Sources Sci. Technol. 4, 571 (1995)
[3] C.Kisielowski, J.Krüger, M.S.H.Leung, R.Klockenbrink, H.Fujii, T.Suski, Sudhir G.S., J.W.Ager III, M.Rubin, and E.R.Weber; Proceedings of the 23th ICPS, Berlin 1996 (World Scientific, Singapore) 1996, p. 513
C.Kisielowski, J.Krüger, S.Ruvimov, T.Suski, J.W.Ager III, E.Jones, Z.Liliental- Weber, H.Fujii, M.Rubin, E.R.Weber, M.D.Bremser, and R.F.Davis; Phys.Rev.B II 54, 17745 (1996)
[4] S.Fischer, C.Wetzel, E.E.Haller, and B.K.Meyer; Appl.Phys.Lett. 67, 1298 (1995)
[5] D.J.Dewsnip, AVAndrianov, I.Harrison, D.E.Lacklison, J.W.Orton, J.Morgan, G.B.
Renz, T.S.Cheng, S.E.Hooperz, and C.T.Foxon; Semicond.Sci.Technol.12, 55 (1997)
[6] G.D.Cody, in Semiconductors and Semimetals; Vol. 21B,
edited by J.I.Pankove (Academic Press, New York, 1984), chapt. 2, 11 - 79
[7] J.Menniger, U.Jahn, O.Brandt, H.Yang, and K.Ploog; Phys.Rev.B 53, 1881 (1996)
[8] J.M.Baranowski and S.Porowski; Proceedings of the ICPS 23, Berlin 1996, (World Scientific Publishing, Singapore), p. 497 (1996)
[9] M.O.Manasreh; Phys.Rev.B 53, 16425 (1996)
[10] M.S.H.Leung, R.Klockenbrink, C.Kisielowski, H.Fujii, J.Krüger, Sudhir G.S.,
A.Anders, Z.Liliental-Weber, M.Rubin, and E.R.Weber, Mat.Res.Soc.Symp.
Vol. 449, p. 221 (1997)
[11] Christian Kisielowski, Zuzanna Liliental-Weber, and Shuji Nakamura; submitted
to Jap.J.Appl.Phys.
[12] Christian Kisielowski, Joachim Krüger, Zuzanna Liliental-Weber, E.R.Weber,
Jinwei Yang, Asif Khan, and Chihping Kuo; to be publ.
© 1997 The Materials Research Society
