The Si-doped AlN films were grown on sapphire (0001) by pulsed sputtering deposition (PSD), and their structural and electrical properties were investigated. A combination of PSD and high-temperature annealing process enabled the growth of high-quality AlN (0001) epitaxial films on sapphire (0001) substrates with atomically flat stepped and terraced surfaces. The transmission electron microscopy observations revealed that the majority of the threading dislocations in AlN belonged to the mixed- or edge-type, with densities of 2.8 × 108 and 4.4 × 109 cm−2, respectively. The Si-doping of AlN by PSD yielded a clear n-type conductivity with a maximum electron mobility of 44 cm2 V−1 s−1, which was the highest value reported for AlN that was grown on sapphire. These results clearly demonstrated the strong potential of the PSD technique for growing high-quality conductive n-type AlN on sapphire.

AlN and its alloy with Ga are promising materials for ultraviolet (UV) light-emitting diodes (LEDs) because of their tunable wide bandgap that ranges from 3.4 to 6.2 eV. For such device applications, an efficient doping technique for AlN-based wide-bandgap materials is urgently required to enable the control of the n- or p-type conductivity. The feasibility of both n- and p-type doping in AlN has been experimentally and theoretically investigated; furthermore, the operation of the AlN p–n junction diodes with an emission wavelength of 210 nm has been demonstrated in 2006.1 For typical AlGaN-based UV LEDs, sapphire is usually selected as the starting material because of its transparency in the deep-UV region. However, few studies have reported on the electrical properties of intentionally doped AlN on sapphire because the large lattice mismatch between AlN and sapphire makes the epitaxial growth of high-quality epilayers difficult. To achieve n-type conductivity in AlN, Si atoms are typically used as dopants. The Si-doping of AlN on sapphire via metal-organic chemical vapor deposition (MOCVD) yielded a poor n-type conductivity of 2 mS cm at an electron concentration of 4 × 1015 cm−3.2 This poor n-type conductivity was attributed to the large activation energy of the dopant, the low crystalline quality, and the formation of intrinsic and extrinsic point defects, including the DX-like behavior of Si.3 

Recently, we developed a new growth technique that can be referred to as pulsed sputtering deposition (PSD) for the nitride-related epitaxial films.4 This sputtering process has enabled us to prepare device-quality group-III nitride films at growth temperatures that are considerably lower than those used in conventional MOCVD.5–7 PSD also provides an extensive doping range as compared to that provided by conventional growth techniques because of its highly nonequilibrium nature.8 For n-type doping in GaN by PSD, the electron concentration can be controlled between 1 × 1016 and 5 × 1020 cm−3, while maintaining a high electron mobility. For heavily doped n-GaN, PSD has yielded a record-low resistivity of 0.16 mΩ cm with a mobility of 100 cm2 V−1 s−1 at an electron concentration of 3.9 × 1020 cm3.9 The high electron mobility at such a high doping level can be explained by the incorporation of a negligible amount of compensating centers. These advantages indicate a strong potential for growing conductive AlN and AlGaN by PSD.

In this study, we describe the growth of high-quality AlN epitaxial films on the sapphire (0001) substrates and investigation of the feasibility of Si-doping in AlN on sapphire by PSD.

One of the most promising approaches to reduce the dislocation density in AlN on sapphire substrates is a high-temperature (HT) annealing process. Although several researchers have investigated the annealing of AlN films on sapphire substrates,9–11 recent studies have revealed that HT annealing under an N2 atmosphere at 1700 °C results in a drastic improvement in the crystalline quality of AlN on sapphire.12 In particular, we expect that PSD will be a suitable low-cost epitaxial technique for mass producing AlN on the sapphire templates because the sputtering process enables the deposition of AlN on a large-area substrate with a high throughput and a high growth rate.

In this study, we used 2-in. sapphire (0001) substrates as starting materials. After the wafer surface was cleaned, it was introduced into a PSD-growth chamber; furthermore, a 300-nm-thick AlN layer was epitaxially grown using a pulsed sputtering source. To improve the crystalline quality of the deposited AlN on sapphire, we transferred the wafer to a furnace and annealed it under an N2 atmosphere at 1700 °C.12 A 1.2-μm-thick Si-doped AlN layer was further grown on HT annealed AlN/sapphire by PSD. We confirmed that the wafer surface remained specular and crack-free after the growth despite the high doping concentrations above 1 × 1019 cm−3. Figures 1 and 2 depict the atomic force microscopy (AFM) surface images and X-ray rocking curves (XRCs) for the as-deposited 300-nm-thick AlN on sapphire, the HT-annealed AlN/sapphire, and the 1.2-μm-thick Si-doped AlN on the HT-annealed AlN/sapphire. In Fig. 1(a), the surface morphology of the as-deposited sample is characterized by c-axis-oriented small grains with a diameter of several hundred nanometers. The FWHM values of XRCs for 0002 and 101¯2 diffractions were 113 and 1843 arc sec, respectively. Such a sharp XRC for 0002 and a relatively broad XRC for 101¯2 were typically observed for highly c-axis-oriented columnar AlN. After HT annealing, disordered step and terrace structures were observed with several pits, as depicted in Fig. 1(b). The high density of surface pits was likely to be caused because of the surface thermal decomposition during the annealing process. Despite the relatively rough surface morphology, the annealing process drastically improved the crystalline quality of AlN on sapphire. In Fig. 2, the XRCs for the 0002 and 101¯2 diffractions became sharp after HT annealing; furthermore, their FWHM values were as narrow as 58 and 409 arc sec, respectively. According to Xiao et al., the c-axis oriented AlN columnar structures coalesce into submicron-sized grains at the initial stage of HT annealing and the solid phase reaction between such grains improves the crystalline quality by reducing the number of grain boundaries at the elevated temperature.13 After the growth of 1.2-μm-thick Si-doped AlN on the HT annealed AlN/sapphire, the surface morphology became smooth, with a root mean square (RMS) surface roughness of 0.49 nm. A clear stepped and terraced structure with a step height of two monolayers (0.5 nm) was observed. For the Si-doped AlN on the annealed template, the FWHM values of XRCs were approximately similar to those of the annealed template, as depicted in Fig. 2, which indicated that formation of the dislocations at the interface between Si-doped AlN and the annealed template was not vigorous.

FIG. 1.

The AFM surface images of 300-nm-thick AlN (a) before and (b) after HT annealing process and (c) 1.2-μm-thick Si-doped AlN.

FIG. 1.

The AFM surface images of 300-nm-thick AlN (a) before and (b) after HT annealing process and (c) 1.2-μm-thick Si-doped AlN.

Close modal
FIG. 2.

The X-ray rocking curves for AlN (a) 0002 and (b) 101¯2 diffractions.

FIG. 2.

The X-ray rocking curves for AlN (a) 0002 and (b) 101¯2 diffractions.

Close modal

To further clarify the dislocation propagation in Si-doped PSD-AlN grown on the annealed template, the samples were observed by transmission electron microscopy (TEM). Figures 3(a) and 3(b) display the cross-sectional dark-field TEM images of g = [0002] and [112¯0]. As depicted in Figs. 3(a) and 3(b), the nucleation of dislocations at the interface between Si-doped AlN and the annealed template was not serious. According to the invisible criterion of g·b = 0, two dislocations that were indicated by the white arrows in Figs. 3(a) and 3(b) were assigned as mixed dislocations. The others were determined to be edge dislocations; furthermore, no screw dislocation was observed in these images. The dislocation density was calculated by dividing the number of dislocations by the measured area. The mixed and edge dislocation densities were determined to be 2.8 × 108 cm−2 and 4.4 × 109 cm−2, respectively. These values were as low as or even lower than those of the commercially available AlN/sapphire wafers. We also investigated the surface polarity of the PSD-grown AlN from the high-resolution TEM Z-contrast image in Fig. 3(c). In these images, bright spots were observed to correspond to the position of Al atoms; furthermore, careful observation determined the atomic arrangement of Al and nitrogen to be as depicted in this schematic. From this interpretation, we can conclude that PSD-AlN on sapphire has Al-polarity, which is used for fabricating light-emitting devices. These results indicate that the combination of a HT annealing process and the PSD technique is a promising approach for the low-cost fabrication of UV LEDs.

FIG. 3.

The cross-sectional TEM images of PSD-grown Si-doped AlN on sapphire with (a) g = [0002] and (b) g = [112¯0]. (c) The high-resolution Z-contrast TEM image of an Si-doped AlN layer.

FIG. 3.

The cross-sectional TEM images of PSD-grown Si-doped AlN on sapphire with (a) g = [0002] and (b) g = [112¯0]. (c) The high-resolution Z-contrast TEM image of an Si-doped AlN layer.

Close modal

Finally, we investigated the electrical properties of the PSD-grown Si-doped AlN using the Hall-effect measurements. As ohmic electrodes, Ti/Al/Ti/Au stacks were deposited by e-beam evaporation; the samples were further annealed under an N2 atmosphere at 900 °C with a rapid thermal annealing apparatus. The Hall-effect measurements were performed using a ResiTest 8400 (Toyo Corp.) based on the assumption that the Hall scattering factor was unity. Figure 4 depicts the relation between the room-temperature electron mobility and the electron concentration of PSD-grown Si-doped AlN on the HT-annealed templates. The electron concentration varied from 1.4 × 1014 to 3.7 × 1015 cm−3 depending on the Si-dopant concentration. The maximum n-type conductivity was 8.3 mS cm−1 at an electron concentration of 3.7 × 1015 cm−3. As the electron concentration decreased, the electron mobility gradually increased and became 44 cm2 V−1 s−1 at an electron concentration of 2.1 × 1014 cm−3. Although little work has been conducted to investigate the electron mobility of Si-doped AlN on sapphire, the authors of an MOCVD study reported an electron mobility of 30 cm2 V−1 s−1 at an electron concentration of 4 × 1014 cm−3.2 The Si-doped AlN grown by molecular beam epitaxy (MBE) on sapphire also yielded a low electron mobility of 4–6 cm2 V−1 s−1 at an electron concentration of 1.1 × 1015–1.7 × 1015 cm−3.14 For higher electron mobility in Si-doped AlN on sapphire, one of the most important factors is a reduction in a dislocation density in AlN. In fact, the MBE work reported the electron mobility shown in Fig. 4 was mainly limited by a dislocation scattering due to low crystalline quality. The lower electron mobility of the MOCVD sample can also be attributed to the lower crystalline quality than that of our samples.2 To the best of our knowledge, our material exhibits the highest electron mobility among the reported values for Si-doped AlN on sapphire to date. The improvement in the crystalline quality of AlN by the HT-annealing process is responsible for our high electron mobility.

FIG. 4.

The relation between the electron concentration and the electron mobility in an Si-doped AlN grown by PSD.

FIG. 4.

The relation between the electron concentration and the electron mobility in an Si-doped AlN grown by PSD.

Close modal

Figure 5(a) depicts the temperature-dependent electron concentration of Si-doped AlN with a room-temperature electron mobility of 44 cm2 V−1 s−1. From the temperature dependence of the electron concentration, we obtained the donor concentration, ND, the acceptor concentration, NA, and the donor activation energy, ED, under the assumption of a simple charge-neutrality equation for an n-type semiconductor with a shallow donor level and a deep acceptor level. The fitting curve matched well with the experimental data with ND = 4.6 × 1019 cm−3, NA = 4.8 × 1018 cm−3, and ED = 305 meV. The obtained ND is similar to the Si concentration estimated by secondary-ion mass spectrometry. The compensation ratio is as low as ND/NA = 0.10 even at such high Si doping concentrations. Heavy Si-doping at concentrations of greater than 3 × 1019 cm−3 via MOCVD leads to the formation of highly resistive AlN films because of the self-compensation effect.15 A recent theoretical study has revealed that the formation of VAl–Si complexes is a main obstacle for increasing the free-electron concentration in MOCVD-grown AlN.16 On the other hand, the PSD growth of AlN proceeded in the high-density nitrogen based plasma at lower growth temperatures compared to MOCVD. We have speculated that such a nonequilibrium growth condition is likely to cause different defect distributions in Si-doped AlN. Our observed n-type conductivity in the higher doping range indicates that the use of PSD allows the growth of highly conductive n-type AlN.

FIG. 5.

(a) Arrhenius plot of the electron concentration and (b) the temperature dependence of the electron mobility in Si-doped AlN grown by PSD, the calculated electron mobilities limited by the individual scattering mechanism, and the total electron mobility that can be obtained on the basis of Matthiessen’s rule.

FIG. 5.

(a) Arrhenius plot of the electron concentration and (b) the temperature dependence of the electron mobility in Si-doped AlN grown by PSD, the calculated electron mobilities limited by the individual scattering mechanism, and the total electron mobility that can be obtained on the basis of Matthiessen’s rule.

Close modal

Figure 5(b) exhibits the detailed temperature dependence of electron mobility and the calculated electron mobility limited by the interactions of electrons with ionized and neutral impurities, dislocations, and lattice vibrations, as calculated using the material parameters reported in the literature.17 In Fig. 5(b), the experimental electron mobility agrees well with the calculated total electron mobility, μtotal. At approximately 300 K, the electron mobility was mainly limited by the ionized impurity scattering.

Thus, the combined usage of PSD and an HT-annealing process enabled the growth of high-quality AlN (0001) epitaxial films on sapphire (0001) with atomically flat stepped and terraced surfaces. The TEM observations revealed that almost all of the threading dislocations in AlN were mixed- and edge-type and that their densities were 2.8 × 108 cm−2 and 4.4 × 109 cm−2, respectively. The Si-doping in AlN that was prepared by PSD on sapphire yielded a clear n-type conductivity, and the maximum electron mobility was observed to be as high as 44 cm2 V−1 s−1. These results clearly demonstrated the strong potential of the PSD technique for preparing UV LED materials on sapphire substrates.

This work was partially supported by the JST ACCEL Grant No. JPMJAC1405 and the JSPS KAKENHI Grant Nos. JP16H06414 and JP16H06415.

1.
Y.
Taniyasu
,
M.
Kasu
, and
T.
Makimoto
,
Nature
441
,
325
(
2006
).
2.
S. B.
Thapa
,
J.
Hertkorn
,
F.
Scholz
,
G. M.
Prinz
,
R. A. R.
Leute
,
M.
Feneberg
,
K.
Thonke
,
R.
Sauer
,
O.
Klein
,
J.
Biskupek
, and
U.
Kaiser
,
J. Cryst. Growth
310
,
4939
(
2008
).
3.
R.
Zeisel
,
M. W.
Bayerl
,
S. T. B.
Goennenwein
,
R.
Dimitrov
,
O.
Ambacher
,
M. S.
Brandt
, and
M.
Stutzmann
,
Phys. Rev. B
61
,
R16283
(
2000
).
4.
K.
Sato
,
J.
Ohta
,
S.
Inoue
,
A.
Kobayashi
, and
H.
Fujioka
,
Appl. Phys. Express
2
,
011003
(
2009
).
5.
E.
Nakamura
,
K.
Ueno
,
J.
Ohta
,
H.
Fujioka
, and
M.
Oshima
,
Appl. Phys. Lett.
104
,
051121
(
2014
).
6.
K.
Ueno
,
E.
Kishikawa
,
S.
Inoue
,
J.
Ohta
,
H.
Fujioka
,
M.
Oshima
, and
H.
Fukuyama
,
Phys. Status Solidi RRL
8
,
256
(
2014
).
7.
J. W.
Shon
,
J.
Ohta
,
K.
Ueno
,
A.
Kobayashi
, and
H.
Fujioka
,
Sci. Rep.
4
,
5325
(
2014
).
8.
Y.
Arakawa
,
K.
Ueno
,
H.
Imabeppu
,
A.
Kobayashi
,
J.
Ohta
, and
H.
Fujioka
,
Appl. Phys. Lett.
110
,
042103
(
2017
).
9.
K.
Ueno
,
T.
Fudetani
,
Y.
Arakawa
,
A.
Kobayashi
,
J.
Ohta
, and
H.
Fuijoka
,
APL Mater.
5
,
126102
(
2017
).
10.
H.
Miyake
,
G.
Nishio
,
S.
Suzuki
,
K.
Hiramatsu
,
H.
Fukuyama
,
J.
Kaur
, and
N.
Kuwano
,
Appl. Phys. Express
9
,
025501
(
2016
).
11.
M.
Nemoz
,
R.
Dagher
,
S.
Matta
,
A.
Michon
,
P.
Vennegues
, and
J.
Brault
,
J. Cryst. Growth
461
,
10
(
2017
).
12.
H.
Miyake
,
C. H.
Lin
,
K.
Tokoro
, and
K.
Hiramatsu
,
J. Cryst. Growth
456
,
155
(
2016
).
13.
S.
Xiao
,
R.
Suzuki
,
H.
Miyake
,
S.
Harada
, and
T.
Ujihara
,
J. Cryst. Growth
502
,
41
(
2018
).
14.
S.
Contreras
,
L.
Konczewicz
,
J. B.
Messaoud
,
H.
Peyre
,
M. A.
Khalfioui
,
S.
Matta
,
M.
Leroux
,
B.
Damilano
, and
J.
Brault
,
Superlattices Microstruct.
98
,
253
(
2016
).
15.
Y.
Taniyasu
,
M.
Kasu
, and
N.
Kobayashi
,
Appl. Phys. Lett.
81
,
1255
(
2002
).
16.
J. S.
Harris
,
J. N.
Baker
,
B. E.
Gaddy
,
I.
Bryan
,
Z.
Bryan
,
K. J.
Mirrielees
,
P.
Reddy
,
R.
Collazo
,
Z.
Sitar
, and
D. L.
Irving
,
Appl. Phys. Lett.
112
,
152101
(
2018
).
17.
Y.
Taniyasu
,
M.
Kasu
, and
T.
Makimoto
,
Appl. Phys. Lett.
85
,
4672
(
2004
).