ORIGINAL PAPER

www.pss-a.com

Influence of Undoped-AlGaN Final Barrier of MQWs on

the Performance of Lateral-Type UVB LEDs
Muhammad Ajmal Khan,* Eriko Matsuura, Yukio Kashima, and Hideki Hirayama

of high threading dislocation densities

Aluminium gallium nitride (AlGaN)-based ultraviolet-B light-emitting diodes (TDDs
1010 cm2) in the AlGaN buffer
(UVB LEDs) are expected to offer smart size, wider choice of UVB light emission layer (BL) grown on sapphire substrate,
in the wavelengths range of 280 nm > λ > 320 nm, and low cost as well as low which is caused by large lattice mismatch

16% at the interface of sapphire substrate
power consumption compared with other UV light sources including toxic
and BL, have presented a big challenge
mercury UV-lamps. The hole-tunneling from p-AlGaN side of UVB LED into the for many researchers.[7,8,11–14] In 1971,
multi-quantum-wells (MQWs) is strongly dependent on the thickness (TFB) and Manasevit et al.[15] successfully grown
Al-contents of undoped (ud)-AlGaN final barrier (FB). Herein, the photolumi- GaN and AlN by using metal-organic
nescence (PL) efficiency from MQWs of the UVB LED devices is investigated and chemical vapor deposition (MOCVD).
Initially, some other groups also attempted
compared with the electroluminescence (EL) spectra as a function of TFB.
to the crystal growth of bulk AlN single
Subsequently, the dependence of PL efficiency, external quantum efficiency crystal for the applications of AlGaN-based
(EQE), and light output power on the TFB of UVB LEDs is attempted, using the optoelectronic devices.[8,13,16] However,
same growth condition for all samples except variation in TFB. When TFB is set to due to the high cost as well as due to the
6–7 nm, improvements in the EQE and light output power, respectively, from issue of low transmittance of UV-light
4.3% and 7 mW to the high values of 5.6% and 17 mW in emission band of through the bulk AlN substrate, several
researchers preferred to grow AlGaN-based
295–300 nm under continuous-wave (cw) at room temperature (RT) are achieved.
UV LEDs either on AlN template on sap-
phire substrate or to grow it on high-
temperature annealing (HTA) of sputtered
AlN template.[7,8,14,17,18] The crystal quality
1. Introduction of AlN template grown on c-(0001) sapphire substrates was
reasonably improved (total-TDDs
5108 cm2) by our group,
Low cost, smart, safe, and low power consumption aluminium using a well-known technique of “ammonia (NH3) pulsed-flow
gallium nitride (AlGaN)-based ultraviolet-B light-emitting diodes multilayer (ML) growth.”[7,8,13,14] But still the epitaxial growth
(UVB LEDs), compared with the other UV light sources includ- of AlGaN on an AlN template can have a lattice mismatch as large
ing toxic mercury UV-Lamps, are needed for real-world applica- as
4%, depending on the molar ratio of the Al-contents in the
tions, such as those in immunotherapy (psoriasis, vitiligo), grown layer. However, the performances of AlGaN-based UVA
vulgaris treatment, plant growth under UVB lightning, the pro- and UVB LED devices grown on AlN template on sapphire
duction of vitamin D3 in the human body, and the production of substrates are still very low, due to the UVB emissions in the
phytochemicals in green leaves of vegetables.[1–10] Based on these challenging wavelength ranges under 340 nm.[19–22] Engineering
motivations, we began our research and development of AlGaN- of UVB light emission based on the Al composition around 40%
based UVB LED devices to be grown on aluminum nitride (AlN) in AlGaN is quite challenging due to the complex growth issues
template, with medical science (294–310 nm) as well as agricul- in MOCVD. Such complex growth of AlGaN might be caused
tural applications (310 nm) in mind.[1–10] However, the existence either by kinetic separation as well as 3D growth-mode, which
is further associated with the Ga-rich localized energy states
or due to the existence of vertically propagating TDDs in the
Dr. M. Ajmal Khan, E. Matsuura, Y. Kashima, Dr. H. Hirayama
RIKEN Cluster for Pioneering Research
n-AlGaN electron injection layer (EIL) underneath the multi-
2-1 Hirosawa, Wako, Saitama 351-0198, Japan quantum-wells (MQWs).[19–25]
E-mail: muhammad.khan@riken.jp Some research groups reported about the AlGaN-based UVB
Dr. M. Ajmal Khan, Dr. H. Hirayama LED devices, either grown on an AlN template or on HTA-
RIKEN Center for Advanced Photonics (RAP) sputtered AlN template, where the maximum light output power
2-1 Hirosawa, Wako, Saitama 351-0198, Japan remains restricted to the range of 2.9 mW at 250 mA and 18 mW
E. Matsuura, Y. Kashima at 500 mA, respectively, with a maximum external quantum effi-
Marubun Corporation ciency (EQE) of 1.8% at emission band of 290–310 nm (excluding
Chuo, Tokyo 109-8577, Japan
the data from company catalog, where we cannot access to their
The ORCID identification number(s) for the author(s) of this article device structure’s information as well as growth conditions).[26–28]
can be found under https://doi.org/10.1002/pssa.201900185. After the reduction of TDDs (total-TDDs
1  109 cm2) in the
DOI: 10.1002/pssa.201900185 n-AlGaN EIL underneath the MQWs region leads us to a

Phys. Status Solidi A 2019, 216, 1900185 1900185 (1 of 9) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.pss-a.com

significant increase in the internal quantum efficiency (IQE) up Therefore, in this work, the influence of ud-AlGaN FB both in
to 47% from the AlGaN MQWs of UVB LED, which is measured terms of potential barrier heights (as a function of Al-contents)
by the method given in previous study.[29] Subsequently, EQE of as well as TFB on the enhancement of photoluminescence (PL)
4.4% as well as maximum light output power of 13 mW at 100 mA spectral emission, electroluminescence (EL) spectral emission,
from 295 nm-band UVB LED was achieved.[30,31] But, in our pre- EQE, and light output power of the AlGaN-based lateral-type
vious work of UVB LED for the emission band of 295–310 nm, the (Lat)-UVB LEDs was investigated, as shown in the schematic view
influence of ud-AlGaN FB thickness (TFB) and its Al-contents on of Figure 1c. Especially the hole tunneling from MQB-EBL toward
the hole injection (tunneling) in to the MQWs as well as on the the MQWs via suitable TFB is very critical, and at the same time,
suppression of Mg-diffusion from multi-quantum-barrier electron the suppression of Mg-diffusion from the p-side of MQB-EBL
blocking layer (MQB-EBL) toward the MQWs were not stud- toward the MQWs of Lat-UVB LED is extremely inevitable.
ied,[30,31] as shown in the schematic view of Figure 1a–c.

(a)
2. Experimental Section
Conventional Ni/Au New Ni/Al
To grow AlGaN-based Lat-UVB LED devices on AlN template,
first an AlN template layer was grown on c-plane of sapphire
substrate, using well known, ammonia (NH3) pulsed-flow ML
growth method in low-pressure (LP) MOCVD reactor (SR-
290-300nm- 4000). Subsequently, a well baked and H-purifier-assisted LP-
UVB LED
(sample A-D) MOCVD reactor were used for the growth of Lat-UVB LEDs
devices, as shown in the schematic view of Figure 1a.
p-AlGaN twofold MQB-EB Valley~12nm
In-Dot Trimethylgallium (TMGa), trimethylaluminium (TMAl), tetrae-
ud-AlGaN FB ( TFB ): 4, 6, 8, 10nm thylsilane (TESi), bis (ethylcyclopentadienyl) magnesium
Al0.32Ga0.68N (QW) / Al0.47Ga0.53N (QWB) (Cp2Mg), and ammonia (NH3) were used as a gas precursor
MQWs in the reactor. The bubbler temperatures for the TMGa,
This Work TMAl, TESi, and Cp2Mg gas precursors were set to 4, 17,
30% Relaxed n-Al0.47Ga0.53N(EIL)~200 nm
10, and 40  C, respectively, under pressure of 760 Torr. The
24% Relaxed n-Al0.55Ga0.45N-BL1 ~ 1.7 µm
TMAl flows were set to 20.0 sccm for the growth of all samples
except the MQWs, ud-AlGaN FB, and MQB-EBL crystal structure
AlN~4µm of the samples. Following the high quality of 4 μm-thick AlN
C-Sapphire substrate template,[15,30] a 1.7 μm-thick Si-doped n-Al0.55Ga0.45N BL and
then relatively Ga-rich 200 nm-thick Si-doped n-Al0.47Ga0.53N
EIL were grown, using the growth condition of TMGa ¼
(b) 4.0 sccm (BL), 6.0 sccm (EIL), and TESi ¼ 0.1 sccm (for both
BL and EIL). The Si-concentration in the n-AlGaN BL and
n-AlGaN EIL layers, respectively, was kept at the same level
of
6  1018 cm3. Next, threefold MQWs structure with
2 nm-thick Al0.32Ga0.68N quantum-well (QW)/8 nm-thick
Al0.47Ga0.53N-quantum well barrier (QWB) was grown, using
the growth condition of TMGa ¼ 5.0 sccm (QWB) and 4.5 sccm
(QW), TMAl ¼ 8.0 sccm (QW) and 20.0 sccm (QWB). All the
QWBs of the MQWs in all samples (A–D) were lightly doped with
(c) TESi ¼ 0.02 sccm. Next, 6 nm-thick (sample A), 8 nm-thick
(sample B), 10 nm-thick (sample C), and 4 nm-thick (sample D)
ud-Al0.55Ga0.45N FB were grown, using the growth condition
of TMGa ¼ 6.0 sccm and TMAl ¼ 28.0 sccm. Subsequently,
twofold (EBL-I amd EBL-II) thin p-Al0.55Ga0.45N blocking
(20-30nm)/p-Al0.16Ga0.80N Valley (12 nm) structures were grown
as a MQB-EBL between ud-AlGaN FB and p-AlGaN HIL, using
the growth condition of TMGa ¼ 3.4 sccm (blocking) and
6.0 sccm (valley), Cp2Mg ¼ 30.0 sccm (for both blocking and val-
ley), shown in Figure 1. Similarly, TMAl ¼ 28.0 sccm (blocking)
and 16.0 sccm (valley) were supplied for the growth of MQB-EBL.
Figure 1. a) Schematic cross-sectional image of an AlGaN-based UVB LED
Finally, a 250 nm-thick Mg-doped p-Al0.47Ga0.53N HIL including
fabricated on a ML-AlN template on sapphire substrate. b) Schematic
view of the energy band diagram of the previous UVB LED at 294 nm emis-
p-AlGaN contact layer was grown, using the growth condition of
sion with un-optimized ud-AlGaN FB, and c) proposed energy TMGa ¼ 5.4 sccm (for both p-AlGaN HIL and p-AlGaN contact
band diagram model, to explain the hole migration (hole tunnelling) layer). Cp2Mg ¼ 70.0 sccm for p-AlGaN HIL and Cp2Mg ¼
from the p-AlGaN side via optimized ud-AlGaN FB toward the MQWs 80.0 sccm for p-AlGaN contact layer were supplied. The
(this work). Mg-concentration in the p-AlGaN HIL was kept around

Phys. Status Solidi A 2019, 216, 1900185 1900185 (2 of 9) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.pss-a.com

6  1019 cm3. Figure 1 shows the schematic diagram of the

grown Lat-UVB LED samples (A, B, C, and D), which is not
up to the scale, but only for the eye guidance.
The crystallinity of n-AlGaN EIL layer underneath the MQWs
of one typical sample B was investigated by X-ray rocking curves
(XRCs), using Malvern Panalytical Japan, and the same Si-doped
n-AlGaN EIL layer was used in all samples A, C, and D. The inte-
grated PL, spectra of samples A, B, C, and D, respectively, were
measured using a 20 mW Ar-SHG (244 nm) laser as an excitation
source both at room temperature (RT) as well as at low tempera-
ture (LT). Depth profiles of Mg-atoms in the Mg-doped AlGaN
regions of UVB LED (sample A) were characterized by secondary
ion mass spectroscopy (SIMS) using PHI ADEPT 1010 with pri-
mary ion species: Csþ and primary acceleration voltage: 2.0 kV.
Recently, highly reflective p-electrode (87%), which consists of

1 nm-thick Ni film followed by 200 nm-thick Al film, was
deposited on the over layer of p-AlGaN contact layer instead
of conventional Ni/Au p-electrode, as shown in the inset of
Figure 1a. In-dot metal deposition on the n-AlGaN EIL was used
for the fabrication of n-electrodes on all samples A, B, C, and D of
Lat-UVB LEDs for quick check, as shown in the schematic view
of Figure 1a. The reported output power values in this work were
corresponding to the total radiant flux from the LEDs, and it was
measured on bare-wafer conditions using a Si-photodetector,
which was located just behind the test samples. This was cali-
brated to measure the luminous flux for accurate values of light
output power measurement.[32,33] The performances (I-Ls and
I-EQEs) of all Lat-UVB LED devices were evaluated at RT under
continuous-wave (cw)-operation using p-type electrodes with
chip size area of 0.35  0.35 mm2on bare-wafer level condition.
Figure 2. XRCs of the n-AlGaN EIL underneath the MQWs, with the
FWHM values along (0002) plane, and the (10–12) plane of sample B
(the same n-AlGaN EIL layer was used in all samples A, B, C, and D).
3. Results and Discussion
3.1. Crystallinity and PL Characterization LEDs (sample A, B, C and D) were observed both at RT
(300 K) and at LT (77 K), with band-emission wavelength of
The crystalline quality of AlN template layer was investigated and 290–300 nm, as shown in Figure 3a–d. Furthermore, a clear
confirmed using XRD and high-resolution of transmission elec- shoulder peak around 270 nm, respectively, was observed in
tron microscope (HR-TEM),[7,13,34] where the full width at half all samples A, B, C, and D, respectively, both at LT as well
maximum (FWHM) values of the XRCs along (0002) and as at RT, which were originated from the n-Al0.47 Ga0.53N
(10–12) planes, respectively, are 200 and 350 arcsec (total-TDDs EIL, shown in Figure 3a–d. In addition, some other interesting

5  108 cm2). High crystalline quality of n-AlGaN EIL under- results were observed when comparing the main PL emission
neath the MQWs and suitable design of ud-AlGaN FB along with peaks of these four samples A, B, C, and D at LT, where the
highly transparent p-AlGaN HIL are believed to be critical for real- emission from MQWs possessed an apparently smaller
izing high-efficiency Lat-UVB LED devices. A similar, 200 nm- FWHM value <12 nm (narrow-band, NB), shown in
thick n-AlGaN EIL was designed and fabricated on the over layer Figure 3a–d. The FWHM of the PL spectra at LT for samples
of 1.7 μm-thick n-AlGaN BL for all samples A, B, C, and D, where A, B, C, and D, respectively, was estimated to be 6, 11, 9,
the FWHM of XRCs along (0002) and (10–12) planes, respectively, and 11 nm. Furthermore, the NB-EL emission from AlGaN-
was reduced to
311 and
442 arcsec, shown in Figure 2a,b. The based UVB MQWs is strongly dependent on the NB-PL spectral
edge-, screw-, and mixed-type TDDs respectively of the same emission, with slight shift in the emission peak, as shown in
n-AlGaN EIL structure underneath the MQWs were previously Figure 3a–d and Figure 4a. The lowest FWHM value of 6 nm
confirmed to be 4.5  108, 5.5  107, and 4.7  107cm2 (total- in sample A can be associated to two possible reasons to explain
TDDs
5.5  108 cm2) as determined from a cross-sectional about band narrowing for expected medical applications. The
transmission electron microscope (TEM) image.[30] first reason could possibly be due to the reduction of quan-
Based on the relatively good crystalline quality of n-AlGaN EIL tum-confined Stark effect (QCSE) in the structure with opti-
underneath the MQWs,[30] four new samples A, B, C, and D were mized QWB (8–10 nm) and QW (1–2 nm) in the MQWs,[31]
prepared with different TFB. Therefore, the influence of TFB on and the second reason could possibly be due to the reduction
the relative PL spectral emission from MQWs of samples A, B, C, of TDDs in n-AlGaN EIL (total-TDDs
5.5  108 cm2) under-
and D was investigated. PL spectral emissions from Lat-UVB neath the MQWs.[7,8,30,31,35] The need of NB UVB emission is

Phys. Status Solidi A 2019, 216, 1900185 1900185 (3 of 9) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.pss-a.com

(a) (b)
Time: 0.1sec Time: 0.1sec sample B
Slit: 0.1mm sample A Slit: 0.1mm
3 Ar-SHG(244nm) Ar-SHG(244nm)
P=20mW P=20mW
TFB ~ 6nm, PLRT/PLLT= 53%
TFB ~ 8nm, PLRT/PLLT= 48%
77K 2
77 K

PL (RT)/PL (LT)
PL (RT)/ PL (LT)

300K
1
1 300K

0 0
270 280 290 300 310 320 270 280 290 300 310 320
Wavelength [nm] Wavelength [nm]
(c) (d)
Time: 0.1sec sample C Time: 0.1sec sample D
Slit: 0.1mm Slit: 0.1mm
Ar-SHG(244nm) Ar-SHG(244nm)
1.5 P=20mW 1.5 P=20mW

TFB ~10nm, PLRT/PLLT= 40% TFB ~ 4nm, PLRT/PLLT=31%

PL (RT)/ PL( LT)

77K
PL (RT)/ PL (LT)

77K
1 1

0.5 0.5
300K
300K

0 0
270 280 290 300 310 320 260 280 300 320
Wavelength [nm] Wavelength [nm]

Figure 3. Integrated PL spectra of a) sample A, b) sample B, c) sample C, and d) sample D, measured using a 20 mW Ar-SHG (244 nm) laser as an
excitation source both at RT (300 K) and at LT (77 K).

critical and the NB-UVB light sources release a smaller range of emission peak from sample C remains stationary. Previously,
UV light with FWHM of 4 nm. NB-310 nm UVB phototherapy the grown n-type Al0.47Ga0.53N EIL was in a relaxed state (with
can be used to cure cancer and skin diseases while minimizing a relaxation ratio of 30%) with respect to the fully relaxed AlN
the adverse side effects to the neighboring non-affected substrate, and such highly strained (70%) n-AlGaN EIL might
tissues.[8,36,37] cause the piezoelectric (PZ) polarization effect in the MQWs
The PLRT/PLLT ratio of the AlGaN-based MQWs of samples A, of the UVB LEDs.[30,31] Because the EQE was limited up to
B, C, and D was evaluated, respectively, as 53%, 48%, 40%, and 4.4% with maximum light outpower of 12 mW at 293 nm emis-
31% as a function of TFB, under the same measurement condi- sion and high IQE around 47% measured by temperature depen-
tion as shown in Figure 3a–d. The PL spectral peak position of dent and power density dependent PL measurement.[29,30] The
AlGaN/AlGaN MQWs at RT in sample A is expected to be relative PLRT/PLLT ratio of sample A (this work) was estimated
slightly suffered by red-shift relative to that of LT due to the tem- to be 53%, indicating that the 295 nm-MQWs structure with rel-
perature-induced band-gap shrinkage, shown in Figure 3a. atively thin TFB
6 nm can effectively enhance the emission effi-
However, significant blue-shifts were also observed in the PL ciency from the MQWs, which is speculated to be due to the hole
spectral emission from samples B and D, whereas the PL tunneling enhancement from p-AlGaN HIL side including

Phys. Status Solidi A 2019, 216, 1900185 1900185 (4 of 9) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.pss-a.com

(a) (b)
sample C 50 sample C
sample D sample D
(290-300nm)-Band sample A sample A
sample B 40 sample B

EL Intensity [a. u]
FB (approx: 6 nm)

Current [mA]
FB (approx: 8 nm)
30
FB (approx: 4 nm)
I-V Characteristics
FB (approx: 10 nm)
20
AlGaN-based UVB LEDs on AlN
template/sapphire substrate 10
measured at pulsed (RT)

0
300 350 400 450 500 0 10 20 30 40 50 60 70
Wavelength [nm] Voltage [V]
(c) (d)
6
Output power [mW]

20 2 sample A (approx: 6 nm)

sample A (power) 0.35×0.35 mm
12 15 under 82 mA
sample B (approx: 8 nm)
Output power [mW]

5
10
5 4 sample C (approx: 10 nm)

EQE [%]
8 0
0 20 40 60 80 3 sample D (approx: 4 nm)
Current [mA]
EQE measured under CW operation at RT
2 sample A, B, C and D
4

Power measured under 52 mA 1

at RT sample A, B, C and D
0 0
0 10 20 30 40 50 10 20 30 40 50
Current [mA] Current [mA]

Figure 4. a) Dependence of EL spectral emission intensity on the thickness (TFB) of ud-AlGaN FB in Lat-UVB LEDs of samples A, B, C, and D measured
under DC drive operation at RT. Device performances of 295–300 nm-UVB LEDs (samples A, B, C, and D): b) I–V characteristics (the QCSE effect has
been illustrated in the inset); c) current versus output power (I–L); d) current versus EQEs (I–EQE) characteristics, just on bare-wafer level conditions
under cw-operation at RT.

MQB-EBL toward the MQWs as well as due to the band align- 100
ment at the interface of ud-AlGaN FB and MQB-EBL structure, sample A
when compared with our previous studies[30,31], as shown in
Relative transmittance [%]

80
Figure 1b,c.

60
3.2. Optical and Electrical Properties (Performances Evaluation)
of Lat-UVB LED Devices UVB Range
40
Mg-doped p-Al0.55Ga0.45N contact layer of UVB LEDs (previous
work),[30,31] where highly relative transmittance ≥90% in the
20
UV-emission band of 293–340 nm, was successfully achieved.
Similarly, the Mg-doped p-Al0.47Ga0.53N contact layer including
p-Al0.47Ga0.53N HIL of the Lat-UVB LED (sample A; this work), 0
where highly relative transmittance ≥90% for UVB range, was 220 240 260 280 300 320 340 360 380 400
also confirmed, shown in Figure 5, and the same p-AlGaN struc- Wavelength [nm]
ture was used in all samples A, B, C, and D. Such highly trans-
Figure 5. Relative transmittance through p-AlGaN-based UVB LED struc-
parent p-AlGaN layers are very essential for the enhancement of
ture (sample A).
light extraction efficiency (LEE) in the Lat-UVB LED devices of
samples A, B, C, and D to replace the p-GaN.
For the development of Lat-UVB LED (n-AlGaN EIL/AlGaN-
MQWs/ud-AlGaN FB/p-AlGaN MQB-EBL/p-AlGaN HIL) struc- and uncontrolled diffusion of Mg-atoms into MQWs may decrease
tures, we require optimization of the ud-AlGaN FB and p-type the performance of UVB LED devices. Figure 6a presents the
AlGaN material growth process in a way to avoid Mg-diffusion SIMS depth profiles of Mg-atoms (NMg) in a newly designed
into MQWs. Especially, the non-uniform distribution of Mg- Lat-UVB LED epitaxial heterostructure (sample A). The Mg-depth
dopants (NMg) in the p-AlGaN HIL as well as in the MQB-EBL profile is found to be uniform in the p-AlGaN HIL, but relatively

Phys. Status Solidi A 2019, 216, 1900185 1900185 (5 of 9) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.pss-a.com

(a) MQB-EBL Figure 1c. Al-composition of
55% in the ud-AlGaN FB of

FB
MQWs

MQWs was chosen for the dual purposes of Mg-blocking from

ud-AlGaN
EBL-II

EBL-I
the MQWs as well as for the blocking of low-energy electron from

Valley
22
10 p-AlGaN-HIL n-AlGaN EIL
overshooting toward the p-AlGaN HIL, as shown in Figure 1c. It is
21
10 clearly visible that the Al-compositions of 55% in the ud-
sample A Mg
Mg [atoms-cm-3]

Al0.55Ga0.45N FB and TBF
6 nm can effectively block the Mg-

20
10 diffusion into the MQWs, as shown in Figure 6a.
19 To confirm the single peak EL spectral emission from the LED
10 devices, a typical sample A was chosen for log scale data evalua-
18 tion, as shown in Figure 6b, and the same results were also con-
10
firmed in the other samples B, C, and D. The EL spectral
10
17 emissions under the variation of dc drive (1–8 mA) show that
all other emissions (including the deep-level emissions, with
16
10 wavelengths emission between 350 and 550 nm) were found
0 50 100 150 200 250 300 to be smaller than the main emissions peaks around 300 nm
Depth [nm] from MQWs, shown in Figure 6b.
(b) Finally, the performances of Lat-UVB LED devices based on
sample A
sample A, B, C, and D were evaluated, at RT under cw-operation,
4 I = 1 mA as shown in Figure 4a,b. Figure 4a shows the EL spectral emis-
10 I = 3 mA
I = 8 mA sions from samples A, B, C, and D at injection currents of 5 mA
under the pulsed operation at RT, where single peak emissions
3
EL Intensity [a. u]

10 were confirmed in the emission band of 290–300 nm. A marked

increase in the EL intensity was observed with TBF
6 nm, which
2 is attributed to the efficient hole injection (tunneling) into the
10 MQWs as well as due to the band alignment between ud-
AlGaN FB and MQB-EBL, as illustrated in the schematic view
1 of Figure 1b,c and also confirmed by the enhancement of PL
10
emission intensity both at RT as well as LT, shown in
Figure 3a–d. However, the polarization field within the ud-
0
10 AlGaN FB can prevent the holes from moving into the MQW.
300 350 400 450 500
The polarization in the III-Ns materials system consists of both
Wavelength[nm] spontaneous (SP) and PZ components. When looking along the
Figure 6. a) The SIMS depth profiles of Mg-atoms in the newly designed c-direction of AlGaN crystal growth, pure Ga and pure N sub-
AlGaN-based UVB LED structure (sample A), where five distinct regions layers can be identified in the UVB LED crystal structure, as
can be identified as: i) p-AlGaN HIL, ii) MQB-EBL, iii) ud-AlGaN FB, shown in Figure 1b,c. Wurtzite III-Ns materials exhibit sponta-
iv) MQWs, and v) n- AlGaN EIL, respectively (as marked on top). neous polarization because of the highly ionic bond result from
b) EL spectra of UVB LED (sample A) with peak emission wavelength the large difference in electronegativity between Ga, Al, and N,
at 300 nm on a log scale under DC drive operation at RT. and therefore, the wurtzite structure is lacking inversion symme-
try, i.e., same atoms on the same plane of crystal. The polarity of
the AlGaN crystal decides the direction of SP polarization in the
high peak of Mg-atoms was observed in EBL-II of MQB-EBL, devices. In addition, the III-Ns materials exhibit piezoelectricity
shown in Figure 6a. It was also confirmed that the implemented too, i.e., the characteristic of generating internal electric field due
ud-AlGaN FB between MQWs and p-AlGaN MQB-EBL success- to the existence of mechanical stress, in the grown layer of
fully separated them, enough to prevent the acceptor dopant, n-AlGaN EIL underneath the MQWs on AlN template. The lattice
NMg from diffusing into the MQWs, as shown in Figure 6a. mismatch between AlN and n-AlGaN EIL underneath the MQWs
From the macroscopic viewpoint, the NMg was uniformly distrib- gives rise to compressive strain in the MQWs, which might fur-
uted within the p-AlGaN HIL layer. The average NMg of p-AlGaN ther induce an internal PZ field. The polarity of the crystal and
HIL remained constant around 6  1019 cm3 but then slightly the type of elastic strain, i.e., compressive or tensile strain, may
increased up to 1  1020 in the EBL-II of MQB-EBL, as shown decide the direction of SP and PZ polarization all together in the
in Figure 6a. The Mg-depth profile shows that no segregation UVB LED devices. Therefore, the orientation of the SP and PZ
and diffusion phenomena were observed from MQB-EBL toward polarization is parallel in the case of tensile strain and antiparallel
the MQWs via newly optimized ud-AlGaN FB (sample A), as in the case of compressive strain. In the present work, n-AlGaN
shown in Figure 6a. It is important to mention here that previously EIL is 30% relaxed layers (70% strained) underneath the MQWs
in our old UVB LED, Al-composition of
47% in the ud-AlGaN with respect to the fully relaxed AlN template, measured by the
FB and Al-composition of
47% in the normal QWBs were used, XRD measurement of reciprocal space mapping (RSM).[30,31]
as shown in the schematic view of Figure 1b. However, in the Therefore, the total internal field in the AlGaN/AlGaN MQWs
newly designed UVB LED, Al-composition of
47% (20.0 sccm) might be existed, which comprised of both SP and PZ polariza-
in the normal QWBs and Al-composition of
55% (28.0 sccm) tion in UVB LED devices. In this work, we did not measure the
in the ud-AlGaN FB of MQWs were chosen, as shown in internal field due the polarization effects in the AlGaN MQWs,

Phys. Status Solidi A 2019, 216, 1900185 1900185 (6 of 9) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.pss-a.com

but the PL emission from samples A, B, C, and D and highly effective mass of electron and therefore low probability of hole
strained (70%) n-AlGaN EIL underneath the MQWs gives tunneling toward MQWs is speculated. Similarly, too much
extensive clues of the existence of polarization field in the thinner TFB
4 nm is also not useful because of the possibility
MQWs, as shown in Figure 3a–d. of Mg-atoms diffusion into the MQWs might be higher and there-
The current versus voltage (I–V) characteristic of the devices fore TFB
6 nm thickness is speculated to be thinner enough to
was also measured, where driving voltages in the range of allow the hole injection into the MQWs (sample A) and at the
22–30 V at 5 mA, respectively, were identified using Ni same time to suppress the Mg-diffusion from the MQWs, as
(1 nm)/Al (200 nm) p-electrodes for samples A, B, C, and D, shown in the SIMS profile of Figure 6a. However, the whole
respectively, shown in Figure 4b. However, the turn on voltages scenario can be explained more easily, by considering the quan-
in the LEDs (sample B and C) were found to be higher than the tum mechanical phenomena at the interface of p-AlGaN-based
LED (sample A and D), as shown in Figure 4b. These relatively MQB-EBL and ud-AlGaN FB, as depicted schematically in the
low voltages as depicted in I–V curve for samples A and B as energy band diagram of Figure 1b,c. When compared with the
compared with the samples C and D at low current of 10 mA p-AlGaN MQB-EBL, such energy band bending in the thin
are attributed to the hole tunneling enhancements as a function ud-AlGaN FB layer is decided by the polarization-induced electric
of TFB. However, the decrease in ud-AlGaN FB thickness, TFB field. Especially, the surface depletion region width is precisely
would also cause a reduction in the series resistance and hence determined by the ud-AlGaN FB layer thickness. Figure 4a–d
a drop in voltage, as shown in Figure 4b. In the case of AlGaN- also illustrates that if the TFB is further thickened, the surface
based MQWs structure, a spatial separation of the electron and depletion region width will be larger and the intra-band tunn-
hole wave-functions can be induced by the large electric field in eling efficiency for the hole carriers becomes low, and such
the QW due to the polarization effect along the c-axis of the AlN hypothesis is supported by the results of EL spectral intensities
crystal,[7,8,35,38] as shown in the inset of Figure 4b. However, elec- (Figure 4a), shows an even more degraded emission efficiency
tron–hole (e–h) separation is suppressed as the QW thickness in the Lat-UVB LED with the TFB
10 nm at 20 mA (sample C),
decreases, which is the collective suppression of PZ polarization when compared with TFB
6 nm (sample A) under the DC
electric fields in the QW of MQWs (QCSE), as shown in the inset drive mode.
of Figure 4b. Therefore, the use of a very thin QW and suitable One potential reason of the improvement in the device
thickness of ud-AlGaN FB are expected to enhance the radiative performance was achieved by shifting from conventional
recombination in the MQWs.[30,31,35,38] Generally, the uniform design[30,31] of ud-AlGaN FB (Al-contents
47%) to the new
distribution of e-h pairs across the whole region of MQWs is also design of ud-AlGaN FB (Al-contents
55%). Conventionally,
desired for higher radiative recombination, which is possible due the conduction band offset at ud-AlGaN FB/MQB-EBL interface
to the collective contribution of each QW and QWBs of active was utilized for the suppression of electron-leakage; however,
region (MQWs). Liu et al. investigated the influence of QWB for III-Ns, the polarization field-induced band tilt has an impor-
on the hole transport and carrier distribution in InGaN/GaN tant impact on the effective potential barrier heights.[7,8,13]
MQWs of visible LEDs and found that the thicker QWB and Previously, the polarization effects in the MQWs region were
QW layers were causing problem to the uniform distribution reasonably reduced, by keeping the QWB thickness around
of e-h pairs in the active region for efficient radiative recombina- 8–10 nm and QW thickness around 1.5–2 nm.[30,31] However,
tion.[39] We also found the same issue in AlGaN-based UVB LED the EQE and light output power, respectively, from Lat-UVB
at 310 nm emission, and such issue was resolved later after the LEDs were limited up to 4.4% and 12 mW at 293 nm emission
reduction of QWB thickness from 20 to 8 nm.[31,40] by using the same Al-composition of 47% for both in the final
During the performances evaluation of the newly fabricated barrier (FB) of MQWs (ud-Al0.48Ga0.52N FB) as well as in all
UVB LED devices, light output powers were varied linearly from other QWBs of the reference sample.[30,31] One major issue
7.5 to 17 mW with varying TFB at the maximum current range of of electron-leakage current was realized (nonlinear behavior
52–85 mA at RT, under cw-operation, as shown in Figure 4c. of I–L curve), which might be caused by intrinsically low hole
Also, the output power in all Lat-UVB LED devices (sample A, concentration in Mg-doped p-type AlGaN due to high acceptor
B, C and D) shows a reasonably linear increase with increasing activation energy as well as due to the injection mismatch
the drive current, as shown in Figure 4c. When, TFB was reduced between electrons (n-AlGaN EIL) and holes (p-AlGaN HIL)
from
10 nm (sample C) to
6 nm (sample A) respectively, the of Lat-UVB LED devices.[31,40] In our previous work, both the
maximum light output power was enhanced from 7 to 17 mW on conduction and valance band nearby ud-AlGaN FB/MQB-EBL
bare-wafer level measurement under cw-operation at RT, as interface were drastically dropped, due to the positive polariza-
shown in the inset of Figure 4c. Similarly, the maximum EQE tion charges, resulting in a band dip, which is caused by the
was also improved from 3.7% to 5.6%, with relatively thin TFB design of high Al-composition (55%) in the MQB-EBL as well

6 nm under cw-operation at RT, using Ni (1 nm)/Al as low Al-composition (47%) in the ud-AlGaN FB (conduction
(200 nm) p-electrode, shown in Figure 4d. Furthermore, some and valance band offset). In this scenario, plenty of electrons
interesting phenomenon was observed, when TFB was either might be gathered, and these parasitic electrons may not con-
increased up to
10 nm (sample C) or decreased up to tribute to the desired light-emission (non-radiative recombina-

4 nm (sample D), the maximum output power and EQE, tion) from the MQWs of UVB LEDs, as shown in Figure 1b.
respectively, remained the same as 7.5 mW and 3.7% with emis- Furthermore, the effective potential barrier height (PFBh) for
sion band of 295–300 nm at 30 mA, shown in Figure 4c,d. These the hole transport was found to be unsuitable in the previous
results also confirm that the thicker TFB is not useful for hole work, as shown in Figure 1b. In this work, we shifted to slightly
tunneling because the effective mass of hole is larger than the a new design of the ud-Al0.55Ga0.45 FB in the LED, which has

Phys. Status Solidi A 2019, 216, 1900185 1900185 (7 of 9) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.pss-a.com

relatively high Al-composition (55%), when compared with the 4. Conclusions

normal QWBs (n-Al0.47Ga0.53N) of the MQWs. By comparison,
the optimized ud-AlGaN FB structure (sample A: this work) is In summary, the influence of ud-AlGaN FB thickness (TFB) on
much better and the band dip at ud-AlGaN FB/MQB-EBL inter- the performance of AlGaN-based UVB LEDs with peak emis-
face was minimized by keeping relatively high Al-composition sions band of 290–300 nm was investigated. A reasonable
(
55%) in the ud-AlGaN FB, which is equivalent to the Al- improvement in the UVB LED output power of 17 mW at
composition (
55%) in the MQB-EBL. The newly optimized 82 mA was successfully achieved by using a thin ud-AlGaN
structure may avoid the ud-AlGaN FB/MQB-EBL interface polar- FB (Al:55%) of
6 nm, along with highly transparent p-AlGaN
ization charges and therefore remove the band dip and the layers as well as highly reflective Ni/Al p-electrode. When the
parasitic electrons at this interface, as shown in Figure 1b,c. thickness of ud-AlGaN FB (TFB) was set to
6 nm and also
More simply, the conduction band offset at the interface of the band dip at ud-AlGaN FB/MQB-EBL interface was mini-
ud-AlGaN FB/MQB-EBL was reduced to a smaller value, due mized in the UVB LED, a value of improvement in EQE of
to the Al-composition enhancement from 47% to 55% in the 5.6% under 30 mA, at 300 nm emission was successfully demon-
ud-AlGaN FB. Subsequently, the small dips may lie above the strated on bare-wafer condition, under the cw-operation at RT.
electron quasi Fermi level for electron, and the downward bend- This improvement in the device performance is attributed to
ing at the interface of ud-AlGaN/ MQB-EBL (Figure 1c) is slightly the optimally thinner TFB as well as having suitable potential bar-
lower than the previous UVB LED,[30,31] and such arrangement rier height of ud-AlGaN FB, which is not only promoting the hole
may lead to the improved hole injection probability from tunneling toward the MQWs but at the same time also suppress-
p-AlGaN side toward the MQWs of UVB LEDs (this work), as ing the Mg-diffusion from MQB-EBL side toward the MQWs of
shown in Figure 1c. Finally, when the conventional MQWs struc- Lat-UVB LED.
ture was replaced with a new MQWs in the Lat-UVB LED, the
maximum light power was improved from 7.5 to 17 mW.
These results can further be improved, by promoting the Mg-
atomic activation in the p-AlGaN HIL and then enhancing to
Acknowledgements
the tunneling probability of hole from p-AlGaN HIL side includ- The authors would like to acknowledge the support by Special Postdoctoral
ing p-AlGaN MQB-EBL toward the MQWs of Lat-UVB LED. The Researcher (SPDR) Program office of Riken to make this research work
hole concentration up to 2  1015 cm3 in the p-AlGaN HIL via possible. This work was partially supported in part by the New Energy
and Industrial Technology Development Organization (NEDO), Japan.
special approach of 3D-hole generation, using polarization dop-
ing,[41] is desired. The measurement and detailed investigation of
the internal field due to both the SP and PZ polarization in the
AlGaN-based UVB LED devices grown on c-plan of AlN are also Conflict of Interest
inevitable in the future.[8,35]
The authors declare no conflict of interest.
The improved EQE of 5.6% at 20 mA and maximum light
output power of 17 mW at 82 mA dc drive measured on bare-
wafer level conditions in Lat-UVB LED are relatively small,
when compared with those obtained from flip-chip-based deep
Keywords
ultraviolet light-emitting diodes (DUV LEDs), using photonic
crystal (PhC) in p-AlGaN (p-GaN) as well as using lens-like pack- external quantum efficiency, hole tunneling, light output power, narrow-band
aging.[7,8,42] Therefore, further improvement in the Lat-UVB UVB, polarization effect, ultraviolet-B light-emitting diodes, undoped-AlGaN
final barrier
LEDs is expected, by using flip-chip technology, PhC in
p-AlGaN (p-GaN), polarization doping, and lens-like packag- Received: March 12, 2019
ing.[32,33,41–43] Revised: June 29, 2019
Last but not the least, the ultimate goal is to develop NB-UVB Published online: August 5, 2019
LED light sources, which are strongly needed for the specific real-
world applications, such as those in immunotherapy, for vulgaris
treatment, and plant lightning.[1–6,8] A small part of the NB-UVB [1] M. D. Daniel, J. S. Hill, ANZ. J. Surg. 1991, 61, 340.
spectrum can be used to treat skin conditions such as psoriasis [2] J. A. Parrish, K. F. Jaenicke, J. Invest. Dermatol. 1981, 76, 359.
and eczema.[8,36,37] NB-310 nm UVB phototherapy can be used to [3] P. J. Hargis Jr, T. J. Sobering, G. C. Tisone, J. S. Wagner, S. A. Young,
cure cancer and skin diseases while minimizing the adverse side R. J. Radloff, Proc. SPIE 1995, 2366, 147.
[4] P. M. Insel, R. E. Turner, D. Ross, Nutrition. 2nd Pkg edition, Jones &
effects to the neighboring nonaffected tissues.[8,36,37] Based on
Bartlett Publishing, Burlington, MA 2004.
the best clinical record, the UVB light sources with NB emission
[5] T. A. Kalajian, A. Aldoukhi, A. J. Veronikis, K. Persons, M. F. Holick,
were highly recommended for the treatment of psoriasis in the
Sci. Rep. 2017, 7, 11489.
national guidelines of the USA and in some other countries [6] C. M. Lerche, P. A. Philipsen H. C. Wulf, Photochem. Photobiol. Sci.,
too.[36,37] The FWHM of the PL spectra at LT for sample A 2017, 16, 291.
was estimated to be 6 nm, and FWHM of the EL spectra at [7] H. Hirayama, N. Maeda, S. Fujikawa, S. Toyoda, N. Kamata, J. Appl.
RT from the same sample was estimated to be 10 nm. Phys. 2014, 53, 100209.
However, we need to control the NB-UVB emission peak with [8] M. Kneissl, J. Rass, in III-Nitride Ultraviolet Emitters, Springer Series in
FWHM about 4 nm in the future, to improve the quality of life Material Science, Vol. 227, Chap. 1 (Eds: M. Kneissl, J. Rass),
(QOL). Springer, Cham, 2016.

Phys. Status Solidi A 2019, 216, 1900185 1900185 (8 of 9) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.pss-a.com

[9] C. Zhou, A. Ghods, V. Saravade, P. Patel, K. Yunghans, C. Ferguson, [27] J. Enslin, F. Mehnke, A. Mogilatenko, K. Bellmann, M. Guttmann,
Y. Feng, B. Kucukgok, N. Lu, I. Ferguson, ECS J. Solid State Sci. C. Kuhn, J. Rass, N. Lobo-Ploch, T. Wernicke, M. Weyers,
Technol. 2017, 6, Q149. M. Kneissl, J. Cryst. Growth 2016, 464, 185.
[10] A. Ramnemark, M. Norberg, U. Pettersson-Kymmer, M. Eliasson, Int J [28] N. Susilo, S. Hagedorn, D. Jaeger, H. Miyake, U. Zeimer, C. Reich,
Circumpolar Health 2015, 74, 27963. B. Neuschulz, L. Sulmoni, M. Guttmann, F. Mehnke, C. Kuhn,
[11] S. Krishnankutty, R. M. Kolbas, M. Ajmal Khan, J. N. Kuznia, J. M. Van T. Wernicke, M. Weyers, M. Kneissl, Appl. Phys. Lett. 2018, 112, 041110.
Hove, D. T. Olson, J. Electron. Mater. 1992, 21, 437. [29] T. Kohno, Y. Sudo, M. Yamauchi, K. Mitsui, H. Kudo, H. Okagawa,
[12] H. Hirayama, J. Appl. Phys. 2005, 97, 091101. Y. Yamada, Jpn. J. Appl. Phys. 2012, 51, 072102.
[13] H. Hirayama, in III-Nitride Ultraviolet Emitters, Springer Series in [30] M. Ajmal Khan, N. Maeda, M. Jo, Y. Akamatsu, R. Tanabe, Y. Yamada,
Material Science, Vol. 227, Chap. 4 (Eds: M. Kneissl, J. Rass) H. Hirayama, J. Mater. Chem. C 2019, 7, 143.
Springer, Cham 2016. [31] M. Ajmal Khan, T. Matsumoto, N. Maeda, N. Kamata, H. Hirayama,
[14] H. Hirayama, S. Fujikawa, J. Norimatsu, T. Takano, K. Tsubaki, Jpn. J. Appl. Phys. 2019, 58, SAAF0120.
N. Kamata, Phys. Status Solidi C 2009, 6, S356. [32] N. Maeda, J. Yun, M. Jo, H. Hirayama, Jpn. J. Appl. Phys. 2018, 57,
[15] H. M. Manasevit, F. M. Erdmann, W. I. Simpson, J. Electrochem. Soc. 04FH08.
1971, 118, 1864. [33] N. Maeda, M. Jo, Hideki Hirayama, Phys. Status Solidi A 2018, 215,
[16] M. Bickermann, B. Epelbaum, O. Filip, P. Heimann, S. Nagata, 1700436.
A. Winnacker, Phys. Status Solidi. C 2010, 7, 21. [34] H. Hirayama, Appl. Phys. Lett. 2007, 91, 071901.
[17] H. Miyake, G. Nishio, S. Suzuki, K. Hiramatsu, H. Fukuyama, J. Kaur, [35] F. Scholz, in Compound Semiconductors: Physics, Technology, and Device
N. Kuwano, Appl. Phys. Express 2016, 9, 025501. Concepts (Ed: F. Scholz). Pan Stanford, Singapore 2018, Ch. 11.
[18] H. Miyake, C.-H. Lin, K. Tokoro, K. Hiramatsu, J. Crystal Growth 2016, [36] J. C. Beani, M. Jeanmougin. Ann. Dermatol. Venereol. 2010,
456, 155. 137, 21.
[19] T. Matsumoto, M. Ajmal Khan, N. Maeda, S. Fujikawa, N. Kamata, [37] S. H. Ibbotson, D. Bilsland, N.H. Cox, R.S. Dawe, B. Diffey, C. Edwards,
H. Hirayama, J. Phys. D Appl. Phys. 2019, 52, 115102. P.M. Farr, J. Ferguson, G. Hart, J. Hawk, J. Lloyd, C. Martin, H. Moseley,
[20] I. Bryan, Z. Bryan, S. Mita, A. Rice, L. Hussey, C. Shelton, J. Tweedie, K. McKenna, L.E. Rhodes, D.K. Taylor, Br. J. Dermatol. 2004, 151, 283.
J.-P. Maria, R. Collazo, Z. Sitar, J. Cryst. Growth 2016, 451, 65. [38] H. Hirayama, N. Noguchi, T. Yatabe, N. Kamata, Appl. Phys. Exp.
[21] U. Zeimer, V. Kueller, A. Knauer, M. Weyers, M. Kneissl, J. Cryst. 2008, 1, 051101.
Growth 2013, 377, 32. [39] J. P. Liu, J. H. Ryou, R. D. Dupuis, J. Han, G. D. Shen, H. B. Wang,
[22] M. Marques, L. K. Teles, L. G. Ferreira, Phys. Rev. B 2007, 75, Appl. Phys. Lett. 2008, 93, 021102.
033201. [40] M. Ajmal Khan, T. Matsumoto, N. Maeda, M. Jo, N. Kamata,
[23] P. Cantu, F. Wu, P. Waltereit, S. Keller, A. Romanov, S. DenBaars, H. Hirayama, Ext. Abstr. Int. Symp. on Growth of III-Nitrides
J. S. Speck, J. Appl. Phys. 2005, 97, 103534. (ISGN-7, Poland), p-Mo4.4, Warsaw, Poland August 2018.
[24] D. M. Follstaedt, S. R. Lee, A. A. Allerman, J. A. Floro, J. Appl. Phys. [41] R. Dalmau, B. Moody, ECS Trans., 2018, 86, 31.
2009, 105, 083507. [42] T. Takano, T. Mino, J. Sakai, N. Noguchi, K. Tsubaki, H. Hirayama,
[25] J. Bai, T. Wang, P. J. Parbrook, Q. Wang, K. B. Lee, A. G. Cullis, Appl. Appl. Phys. Express 2017, 10, 031002.
Phys. Lett. 2007, 91, 131903. [43] Y. Kashima, N. Maeda, E. Matsuura, M. Jo, T. Iwai, T. Morita,
[26] K. H. Kim, Z. Y. Fan, M. Khizar, M. L. Nakarmi, J. Y. Lin, H. X. Jiang, M. Kokubo, T. Tashiro, R. Kamimura, Y. Osada, H. Takagi,
Appl. Phys. Lett. 2004, 85, 4777. H. Hirayama, Appl. Phys. Express 2018, 11, 012101.

Phys. Status Solidi A 2019, 216, 1900185 1900185 (9 of 9) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim