2012 International Conference on Solid-State and Integrated Circuit (ICSIC 2012)
IPCSIT vol. 32 (2011) © (2011) IACSIT Press, Singapore
Design and Fabrication of Nitride Heterostructures for Intersubband
Applications
Tommy Ive*, Kristian Berland, Martin Stattin and Thorvald G. Andersson
Department of Microtechnology and Nanoscience, Chalmers University of Technology,
SE-41296 Gothenburg, Sweden
Abstract. Models based on the effective-mass approximation and Schrödinger-Poisson calculations were
used to design AlN/GaN multiple quantum well structures for intersubband transitions. The structures were
grown by molecular beam epitaxy and the surface morphology and the structural quality were investigated.
GaN waveguides were fabricated using standard cleanroom techniques. This report is concerned with the
various challenges encountered when trying to fabricate quantum cascade lasers based on group III-nitrides.
The challenges are discussed in the light of our results.
Keywords: QCL, GaN, short wavelength
1. Introduction
Photonic devices with intersubband transitions have several advantages over the common interband ones.
The main application of intersubband devices is for quantum cascade lasers (QCL). Such lasers constitute a
large research field and about 20 companies worldwide are involved in commercial enterprises concerning
QCLs. The common use is for lasers within the mid-IR spectral range, 5-25 µm, for spectroscopy and
chemical detection of trace elements in liquids and gases. Large research efforts are made to extend the
applications to shorter or longer wavelengths. These spectral ranges depend on the applications. The mid-IR
is for detection, and spectroscopy, of chemical compounds. Far-IR, λ > 50 µm is for THz-applications. There
is a new spectral area of interest for λ < 5 µm (short wavelengths). Commercial QCLs are available from
3.25 µm and up. This shortest wavelength represents the vibration frequency of one of the lightest organic
molecules: C-H. In a range below this, 1.55 – 3.25 µm, we have telecommunication either in traditionally
fibres, in plastic fibres or in free space. A new rapidly expanding application area for the short wavelengths
is countermeasures for airplanes.
Existing QCLs are made of AlGaAs/GaAs, InGaAs/InAlAs/InP, InGaAs/AlAsSb/InP or InAs/AlSb/InAs.
The parameters that determine the properties of the device are the material combination giving the
conduction band edge offset, the layer compositions and thicknesses and the strain due to lattice constant
mismatch. The theoretical limit for λ in the InAs/AlSb material is ~2.7 µm. Additionally the intensity in
QCLs is reduced with reduced wavelength. For shorter wavelengths we need a material heterostructure with
larger band offset, which is fulfilled in AlN/GaN. In the nitrides, however, a new parameter is incorporated;
the polarization, which provide an electric field across the individual layers, cf. Fig. 1.
Fabricating intersubband devices based on group III-nitrides, eg. GaN and related alloys, is challenging
with respect to epitaxy, design and device processing [1-6]. Substrates, such as sapphire and SiC, are
commonly employed in nitride epitaxy due to the lack of commercially available nitride substrates that have
* Corresponding author:
e-mail tommy.ive@chalmers.se
62
both a sufficiently high structural quality and a large area. The consequence of heteroepitaxy on dissimilar
substrates, is a high defect density in the (Al,Ga)N/GaN structures due to the large lattice mismatch between
substrate and layer. This high defect density has a negative impact on the device performance. The intrinsic
internal electric fields across heterostructures grown along the c-direction represent a challenge since these
electric fields may effect both electron confinement and efficient current injection. The target emission
wavelength for nitride intersubband structures is between 1.55 and 3 µm [7-9]. This target wavelength
requires 0.5-1.5 nm wide barriers and homogeneous and abrupt AlN/GaN heterointerfaces. This is
challenging from an epitaxial point of view since a very precise control over the layer thickness must be
maintained throughout the full quantum cascade laser (QCL) structure [10-13] which can consist of 15-75
(Al,Ga)N/GaN periods [14,15]. To achieve efficient n-type doping of AlxGa1-xN alloys with a high Alcontent is yet another challenge [16].
In this report we outline the various challenges related to the fabrication of nitride-based QCL structures,
which we have investigated. Based on the effective-mass approximation and self-consistent SchrödingerPoisson calculations we have modelled an AlN/GaN QCL structure
2. Modelling
The intrinsic electric fields across heterostructure layers, shown along the c-direction in Fig. 1, influence
the electron confinement due to the QW distorsion. The short 1.55 - 3 µm wavelengths require 0.5-1.5 nm
wide homogeneous barriers with abrupt AlN/GaN heterointerfaces. Intersubband transitions were calculated
in systems with two or three energy levels [17-18]. This design was used as the basis for our experiments to
cast light over the challenges related to device simulation, epitaxy, characterization and device processing
[19]. In Fig. 1 the principal function of a simple 2-level QCL-structure is shown when under bias. The
electron is injected in the ground state of the first deep QW and relaxes to a state in the neighbouring deep
QW before this is emptied. The energy difference in the diagonal transition represents the emission
wavelength. Details of the band structure calculations are shown in [17].
Figure 1. QCL-design based on a diagonal transition for high quantum efficiency. The initial well on the right hand side
enhances the injection and lowers the internal voltage drop over the active structure.
3. Experiment
Our structures were grown in a Gen-II plasma-assisted molecular beam epitaxy (MBE) system using Aland Ga-solid sources. Active nitrogen was obtained from an SVT radio frequency N-plasma source. The
heterostructures were deposited on c-sapphire substrates as well as 3-4 µm thick GaN(0001)/c-sapphire
templates. The GaN template layers had a threading dislocation density of ~108 cm-2 or more. All structures
were grown at 800°C under Ga-stable conditions. The growth rate was 300 nm/h for both AlN and GaN.
The crystal quality was investigated with X-ray diffraction (XRD) by recording ω-2Θ-scans across the
(0002) reflection. The surface morphology was characterized by a Veeco Dimension 3000 atomic force
63
microscope (AFM) and a JEOL JSM-6301F scanning electron microscope (SEM). Cross-sectional
transmission electron microscopy (TEM) was used to investigate the microstructure. Temperature dependent
absorption measurements were previously performed to investigate the temperature stability of the
absorption peak [18].
4. Results and discussion
Figure 2 shows an ω-2Θ XRD scan across the GaN(0002) reflection from an intersubband superlattice
structure consisting of 20 periods of 3.2 nm AlN and 2.1 nm GaN. The superlattice was grown on a 100 nm
buffer on a GaN template. A 100 nm thick cap layer concluded the structure. The lower (red) curve denotes
the simulated XRD profile whereas the upper (blue) curve denotes the experimental result. A relatively good
match between experiment and simulation can be observed evidencing that the desired period thickness has
been achieved. The inset in Fig. 2 displays a cross-sectional TEM micrograph obtained from a section of the
superlattice. The bright layers correspond to the AlN barriers and the dark layers are the GaN wells. Both
layers are sharp and homogeneous throughout the depicted section indicating a precise control over the
epitaxial growth conditions and homogeneous atomic planes in the thin individual layers.
Figure 2. XRD ω-2Θ scan across the GaN(0002) reflection for a 20 period AlN/GaN superlattice. A good fit between
the measurement and simulation can be observed. Sharp satellites indicate abrupt and homogeneous interfaces. The
inset shows a cross-sectional TEM micrograph obtained from a superlattice. The bright fields correspond to 3 nm thick
AlN barriers whereas the dark fields correspond to 2 nm thick GaN wells
Crack formation is observed on sample surfaces as revealed by AFM. This is expected on structures
grown on GaN buffer layers, which exert a tensile strain on the AlN barrier layers [20]. Temperaturedependent absorption measurements show that the redshift of the absorption peak is vey small, only 15
µeV/K [18].
A waveguide for AlN/GaN based QCLs was designed and grown on c-sapphire. This structure consists
of a 500 nm thick gain region sandwiched between 500 nm thick n-doped AlGaN injection and extraction
layers but does not include an active region. The width of the waveguide is 8 µm and is located next to the 5
64
µm wide top metal contact. Injection occurs laterally and requires a second side contact. The white elliptical
area indicates the well confined, guided low loss TM polarized mode (see the inset in Fig. 3). It has been
shown that a single mode QCLs can be obtained using proper design parameters [21].
Figure 3. SEM micrograph of a GaN waveguide designed according to the QCL design shown in the inset. This
structure was used to optimize the dry etching processing parameters and does not include any active region. Nearly
vertical sidewalls and smooth surfaces can be observed. The white elliptical area in the inset indicates the guided low
loss TM polarized mode. Proper design parameters are expected to yield a single mode QCL.
5. Conclusions
In this work we have focused on the challenges related to the fabrication of a nitride-based QCL emitting
between 1.55 and 3 µm. To summarize, these challenges include the lack of large area native nitride
substrates having a sufficiently low defect density, non-flat-band conditions due to the internal electric fields,
crack-formation due to large lattice mismatch between structure and to substrate, weak intersubband
luminescence and inefficient current transport through the heterostructures. We have highlighted and
investigated the feasibility of several of the critical fabrication steps to obtain MBE-grown intersubband
heterostructures. We have shown that a very precise control of the epitaxial process can be achieved. This is
necessary for keeping the uniformity and homogeneity throughout the structure, which might consist ~150
periods. The obtained layers are sharp and homogeneous but the structures exhibit cracks. The intersubband
transitions are insensitive to temperature as only a weak dependence of the absorption peak on temperature
was observed. This speaks for the assumption that efficient and stable nitride QCLs can be made to operate
without external cooling systems. Further, we have fabricated GaN waveguides suitable for a QCL.
6. Acknowledgements
F. Fälth is acknowleged for his help with the XRD measurements.
7. References
[1] X. Y. Liu, P. Holmström, P. Jänes, L. Thylén, and T. G. A dersson, Phys. Status Solidi B 244 (2007) 2892.
[2] T. Aggerstam, T. G. Andersson, P. Holmström, P. Jänes, X. Y. Liu, S. Lourdudoss and L. Thylén, Proc. SPIE
6479 (2007) 64791E.
[3] C. Gmachl and Hock M. Ng, Electron. Lett. 39 (2003) 567.
65
[4] T. G. Andersson, X. Y. Liu, T. Aggerstam, P. Holmström, S. Lourdudoss, L. Thylén, Y. L. Chen, C. H. Hsieh, and
I. Lo, Microelectron. J. 40 (2009) 360.
[5] H. Sodabanlu, J.-S. Yang, M. Sugiyama, Y. Shimogaki, and Y. Nakano, Appl. Phys. Lett. 95 (2009) 161908.
[6] H. Machhadani, M. Tchernycheva, S. Sakr, L. Rigutti, R. Colombelli, E.Warde, C. Mietze, D. J. As, and F. H.
Julien, Phys. Rev. B 83 (2011) 075313.
[7] P. Holmström, IEEE J. Quantum Electron. 42 (2006) 810.
[8] C. Gmachl, H. M. Ng, S.-N. G. Chu, and A. Y. Cho, Appl. Phys. Lett. 77 (2000) 3722.
[9] P. K. Kandaswamy, F. Guillot, E. Bellet-Amalric, E. Monroy, L. Nevou, M. Tchernycheva, A. Michon, F. H.
Julien, E. Baumann, F. R. Giorgetta, D. Hofstetter, T. Remmele, M. Albrecht, S. Birner, and Le Si Dang, J. Appl.
Phys. 104 (2008) 093501.
[10] J. Faist, F. Capasso, D. L. Sivco, C. Sirtori, A. L. Hutchinson, and A. Y. Cho, Science 264 (1994) 553.
[11] J. Faist, C. Gmachl, F. Capasso, C. Sirtori, D. L. Sivco, J. N. Baillargeon, and A. Y. Cho, Appl. Phys. Lett. 70
(1997) 2670.
[12] G. Dehlinger, L. Diehl, U. Gennser, H. Sigg, J. Faist, K. Ensslin, D. Grützmacher, and E. Müller, Science 290
(2000) 2277.
[13] M. Beck, D. Hofstetter, T. Aellen, J. Faist, U. Oesterle, M. Ilegems, E. Gini, and H. Melchior, Science 295 (2002)
301.
[14] G. Scamarcio, F. Capasso, C. Sirtori, J. Faist, A. L. Hutchinson, D. L. Sivco, and A. Y. Cho, Science 276 (1997)
773.
[15] A. Ishida, Y. Inoue, H. Nagasawa, N. Sone, K. Ishino, J.J. Kim, H. Makino, T. Yao, H. Kan, and H. Fujiyasu,
Physica E 21 (2004) 765.
[16] T. Ive, O. Brandt, H. Kostial, K. J. Friedland, L. Däweritz, and K. H. Ploog, Appl. Phys. Lett. 86 (2005) 024106.
[17] K. Berland, T. G. Andersson, and P. Hyldgaard, Phys. Rev. B 84 (2011) 245313.
[18] K. Berland, M. Stattin, R. Farivar, D. M. S. Sultan, P. Hyldgaard, A. Larsson, S. M.Wang, and T. G. Andersson,
Appl. Phys. Lett. 97 (2010) 043507.
[19] F. Capasso: private communication.
[20] T. Ive, O. Brandt, H. Kostial, T. Hesjedal, M. Ramsteiner, and K. H. Ploog, Appl. Phys. Lett. 85 (2004) 1970.
[21] M. Stattin, K. Berland, P. Hyldgaard, A. Larsson, and T. G. Andersson, Phys. Status Solidi C 8 (2011) 2357.
66