During the past few years, greatly impressive development of III-V nitride based light emitting devices have been witnessed for illuminated devices, laser diode light sources for high-density optical storage system, full-color display, and medical applications. High-efficiency GaN-based LEDs between the UV and amber spectral regions are now commercially available [1–6]. Especially, UV LEDs have recently attracted much attention to be strong candidate in biological agent detection, air cleaner, and transceivers for convert nonlinear-of-sight optical communications. As well, using UV LEDs as light emitters to excite phosphor can achieve white fluorescence for solid state lighting. The LEDs with an emission wavelength less than 370 nm are specifically significant due to most of the phosphors can absorb and provide a more effective down-conversion quantum efficiency. To date, the UV LEDs in a spectral region of 267–370 nm have been successfully fabricated [7–10]. However, as decreasing UV LED emission wavelength to a deeper level in the fabrication of UV LEDs, a rapid degradation of light output power is observed. The reason had been found to be attributed to the large threading dislocation, poor carrier confinement in the QW active region, and self absorption in the thick GaN contact layers [11].
As expected to be a replacement of conventional incandescent and fluorescent lamps, high-efficiency UV LEDs are strongly demanded. In particular, the UV LED with an emission wavelength between 365–370 nm is more favorable due to it can provide relatively higher output performance to pump phosphors [3].
3-1 Literature survey
To fabricate an UV LED with 365–370 nm emission, InGaN, AlGaN or quaternary AlGaInN are expected to be applied as QW active region in UV LEDs due to the
naturally direct high-bandgap transition. However, as compared with those of the visible LEDs, the illuminated power and efficiency of the UV LEDs are extremely worse.
While, the reasons had been observed due to some factors, including defects in the quantum well active region, device heating under continuous-wave operation, and especially carrier leakage from the active region.
Significant progress of UV LEDs had been made in emission wavelength region of 325–400 nm with mW output [12–15]. For UV LEDs with an emission wavelength of 267 nm, AlGaN-based LEDs with a pulse operation output power of 4.5 mW and 165 µW at 435 mA under CW operation for an array of four LEDs in parallel had been demonstrated by Yasan et al. [16]. For 365 nm UV LEDs, with the use of laser-induced liftoff and polishing technologies, Morita et al. had obtained an output power of 100 mW with an external quantum efficiency of 5.6% at 500 mA [17]. On the other hand, with an emission wavelength shorter than the characteristic wavelength corresponding to the GaN bandgap, Nishida et al. adopted a short period alloy superlattice as p-type cladding and p-type contact layers to have a transparent 348–351 nm LED with 7 mW output at 220 mA [18].
These efforts had made the UV LEDs to have an acceptable output characteristic;
however, it needs to be mentioned that substantial work is necessitated with an aim to improve the emitting power and efficiency. For 370 nm UV LED, the use of quaternary AlGaInN QW active region is beneficial for obtaining better material quality and higher internal quantum efficiency [19]. Zhang et al. had also pointed out that the use of quaternary AlGaInN as barriers could drastically improve the heterostructure quality [20]. In this chapter, we presented the growth and fabrication of UV LEDs operating at 370 nm based on quaternary AlGaInN QW active region. To create more efficient UV LED, we further theoretically investigated the relationships of the Al composition in AlGaN electron-block layer and the QW number to the LED output performance, with
an aim to reduce the electron leakage current.
3-2 Device fabrication and characteristics
The LEDs used in this study were grown on c-face sapphire substrate by low-pressure horizontal-flow MOCVD system using a 30-nm-thick low-temperature GaN nucleation layer at 550 °C, followed by a 2-µm-thick high-temperature undoped GaN buffer layer and a 1-µm-thick Si doped GaN to form the n-contact layer at 1050 °C.
Next, a 50-nm-thick graded n-type AlxGa1-xN (x=0.1 to 0.14) was deposited for cladding. Then, the growth temperature was linearly decreased to 850 °C to grow the active region of the UV LED. The active region was consisted of three quaternary Al0.06Ga0.85In0.09N multiple QWs sandwiched by four Al0.05Ga0.94In0.01N barriers. The temperature was afterwards increased to 1050 °C for growing a 25-nm-thick p-type Al0.19Ga0.81N electron-block layer, followed by a 125-nm-thick p-type Al0.09Ga0.91N and a 10-nm-thick p-GaN contact layer to complete the structure.
Figure 3.1 A schematic plot of the UV LED device.
After MOCVD growth, the fabrication process began from partially etching by reactive ion etching from the surface of the p-type GaN contact layer until the n-type GaN exposed. Ni/Au metal was evaporated onto the p-type GaN, and Ti/Al metal was
evaporated onto the n-type GaN. The chip size was 300×300 µm2 and was formed by packing into 5 mm lamps with a standard process for output characteristic measurement.
A schematic plot of the UV LED device was shown in Figure 3.1.
The device characteristics were obtained with a probe station, Keithlyn 238 current source, and Newport 1835C power meter module. Electroluminescence (EL) spectrum was measured by Advantec optical spectrum analyzer (OSA) with a 0.1 nm spectrum resolution. Figure 3.2 showed the EL spectrum of the UV AlGaInN LED under continuous-wave operation when the input current was in a range of 10–100 mA. The main peak of the emission wavelength designed slightly longer than 365 nm was for the purpose of preventing strong internal absorption by the bulk GaN, and it shifted from 368 nm to 372 nm with increased input current from 10 mA to 100 mA.
0 500 1000 1500 2000 2500 3000
350 355 360 365 370 375 380 385 390 10 mA
20 mA 30 mA 40 mA 50 mA 60 mA 70 mA 80 mA 90 mA 100 mA
Intensity (a. u.)
Wavelength (nm)
Figure 3.2 EL spectrum of the UV AlGaInN LED under continuous-wave operation when the input current was in a range of 10–100 mA.
Figure 3.3 showed the output characteristics of the UV LED when the device temperature was varied in a range of 300–380 K. During the measurement of the temperature dependent output characteristics, the UV LED was mounted on a hot plate, and the device temperature was monitored with a thermal coupler. The room-temperature UV power of the LED was near 0.8 mW at 20 mA with 3.6 V
operation voltage and increased to 4 mW when the LED was driven at 125 mA under CW operation. The wall plug efficiency was approximately 1.1% and the external quantum efficiency was 1.2%. When the device temperature was operated at 380 K, the UV LED could still provide 0.5 mW output when the input current was 20 mA. A maximum output power of 2.1 mW could be achieved at 380 K when the input current was 100 mA.
0 1 2 3 4 5 6 7
0 0.5 1 1.5 2 2.5 3 3.5 4
0 25 50 75 100 125 150
Voltage (V) Power (mW)
Current (mA)
Figure 3.3 Output characteristics of the UV LED when the device temperature was varied in a range of 300–380 K.
For blue and green LEDs, which were made by III-nitride materials, the reports had shown that the main peak of the emission wavelength decreased first and increased afterward as the input current was increased. This phenomenon was found to be attributed to the localized states resulting from indium inhomogeneity. However, there was no short-wavelength shift in our UV AlGaInN LED, and it was supposed that the indium composition in the AlGaInN QWs was much less than that in the blue or green LEDs and the segregation of indium could not dominate under this situation.
3-3 Theoretical analysis
It is universally known that qualitatively theoretical analysis can advantageously provide a way to realize the output characteristic of the optoelectronic semiconductor device. Numerical simulation is always required to model and optimize the output characteristic of the device. In this study, the numerical simulation was executed with the use of an advanced physical model of semiconductor devices (APSYS) [21], which was utilized as a full two-dimensional simulator that solved the Poisson’s equation, current continuity equations, photon rate equation and scalar wave equation, and accounts for current spreading in this specific study.
For this specific simulation, the temperature dependent bandgap energies of binary InN, GaN, AlN alloys were governed by Varshni equation and the bowing factors of ternary GaInN, AlGaN and AlInN alloys were 2.4, 0.7, and 2.5 eV respectively [22, 23].
Most Luttinger-like valence band parameters we used in this study such as A1, A2…A6
for nitrides and the deformation potentials, elastic constants, etc. were also obtained from Ref. 13, as listed in Table 1.1, excepted that the electron and hole mobilities were taken from the default database values given in the APSYS material macro file [21].
For the treatment of device heating, the thermoelectric power and thermal current induced by temperature gradient were solved utilizing the methods provided by Wachutka et al. [24–26]. Various heat sources, including Joule heat, generation/recombination heat, Thomson heat and Peltier heat, were taken into account in this specific study. The boundary temperature between the LED contacts and the ambiance was solved assuming that the contacts were connected to a thermal conductor with a fixed temperature set by the simulator. The calculation of the interface charge density including spontaneous and piezoelectric polarization in the ternary III-nitride material as a function of composition and microscopic structure was by the use of ab
initio density-functional techniques and Berry phase method [27], while we assumed the charges at multiple QWs were with partial 85% screening. For the interface charge of quaternary AlGaInN QW, it was predicted by ternary interpolation formulas [28]:
),
while the calculated interface charge densities in the fabricated UV AlGaInN/GaN LED structure were summarized in Table 3.1.
Table 3.1 Net surface charge density at each interface of the UV LED.
Interface Surface charge density
GaN/Al0.1Ga0.9N +5.50×1015 m-2
The numerical spontaneous emission rate spectrum of the UV AlGaInN LED as a function of the input current was shown in Figure 3.4. The inset in Figure 3.4 depicted the main peaks of the numerical spontaneous emission rate spectra and the experimental EL spectra. It was clearly seen that the main peaks of the numerical spontaneous emission rate spectra and the experimental EL spectra were of great agreement. The spontaneous emission rate calculated in this study was given by [29]:
,
and fi, fj represented the Fermi functions for the ith and jth levels. D(E) was the optical
350 355 360 365 370 375 380 385 390 10 mA
Figure 3.4 Numerical spontaneous emission rate spectrum of the UV AlGaInN LED as a function of the input current. The inset shows the main peaks of the numerical
spontaneous emission rate spectra and the experimental EL spectra.
The spontaneous emission power (Pspon) was calculated by the integral of all modes spontaneity:
and the extracted output power from the LED top surface was
,
where A was the active region cross section in the LED and we regarded the LED as a special case of Fabry-Perot laser with cavity length of L. n and ng were the indices, and gl was a integral constant. We wished to underscore that the package loss assumed in this study was zero, and therefore the ratio of PL/Pspon gave the LED external efficiency.
Besides, a current efficiency could also be obtained from the ratio of the spontaneous recombination current (Ispon) to the total current, which was the sum of the spontaneous recombination current, nonradiative recombination current (Inr), and leakage current (I ):
.
To fit the experimental LED output characteristic, some parameters were used such as the radiative recombination coefficient for the bulk AlGaInN, AlGaN, GaN was 2.9×10−15 m3/s, the auger coefficients for n and p carrier were both set to 4×10−40 m6/s, and the carrier lifetime was 1 ns [30]. A large internal loss value of 1000 m−1 was assumed due to the large defect density in III-V nitride material. It was clearly seen that the temperature dependent light output versus current (L-I) characteristic obtained numerically was fit in with the experiment despite that the differential resistance was slightly inconsistent. Practically, the characteristics of the UV AlGaInN LED could be quantitatively analyzed by the simulator with the parameters shown above, which remained unchanged throughout the study. To enhance the output power of the UV AlGaInN LED, we further investigated the effects of the aluminum composition in AlGaN electron-block layer and the QW number on the UV AlGaInN LED. The purpose was with an aim to reduce the electron leakage current and therefore improved the output performance.
Figure 3.5 Numerical temperature dependent output characteristics of the UV AlGaInN LED.
Several reports had shown that the electron leakage current played an important role in the III-V nitride material due to the large discrepancy of electron and hole effective masses and the low p-type doping concentration [31–36]. For UV LEDs with emission wavelength of 305–365 nm, the Al composition in AlGaN electron-block layer was typically in a range of 23–70% [37–40]. When the emission wavelength was shorter, the Al composition in AlGaN electron-block layer should be increased accordingly because the conduction band offset was decreased and more electrons overflowed to the p-type layers. Figure 3.6 showed the L-I characteristics of the UV AlGaInN LED with variant Al compositions in AlGaN electron-block layer when the device temperatures were 300 K and 380 K. It was observed that the output power was enhanced when the Al composition in AlGaN electron-block layer was increased, and the output power at 300 K was limited when the Al composition in AlGaN electron-block layer was higher than 19%. For the UV LED operated at 380 K, increasing the Al composition in AlGaN electron-block layer to be higher than 19%
could also effectively prevent the electrons overflow, and the output power remained unchanged when the input current was lower than 110 mA.
0 0.5 1 1.5 2 2.5 3 3.5 4
0 50 100 150
Al=0.13 Al=0.15 Al=0.17 Al=0.19 Al=0.21 Al=0.23 Al=0.25
Power (mW)
Current (mA)
300 K
380 K
Figure 3.6 L-I characteristics of the UV AlGaInN LED with variant Al compositions in AlGaN electron-block layer when the device temperatures were 300 K and 380 K.
Figure 3.7 showed the current efficiency of the UV AlGaInN LED as a function of the input current for variant Al compositions in AlGaN electron-block layer when the device temperature were 300 K and 380 K. Increasing device temperature with decreased internal efficiency was found due to the increase of recombination loss and carrier leakage from the active region. With higher input current and higher device temperature, the electron leakage current and the nonradiative recombination increased which in turn resulted in reduction in current efficiency. The simulated results suggested that the low internal efficiency might be limited by the electron leakage current and the large nonradiative recombination we assumed for the purpose of fitting the experimental output characteristics.
0 0.05 0.1 0.15 0.2 0.25
0 50 100 150
Al=0.13 Al=0.15 Al=0.17 Al=0.19 Al=0.21 Al=0.23 Al=0.25
Current efficiency (%)
Current (mA)
Figure 3.7 Current efficiency of the UV AlGaInN LED as a function of the input current for variant Al compositions in AlGaN electron-block layer when the device temperature
were 300 K and 380 K.
To further reduce the electron leakage current and improve the output characteristics, we subsequently investigated the QW number effect on the output characteristics of the UV AlGaInN LED. In this specific study, the Al composition in AlGaN electron-block layer was 19% since our numerical analysis indicated that the relatively better output characteristics were indicated when the Al composition in
AlGaN electron-block layer was 19%. The internal loss value was set and remained unchanged at 1000 m−1 when varying the QW number in a range of 1–11, without contemplating the increased defects by increasing QW number during crystal growth.
Figure 3.8 showed the L-I characteristic of the UV AlGaInN LED with variant QW numbers when the device temperatures were 300 K and 380 K. Figure 3.9 showed the percentage of electron leakage current as a function of the device temperature when the QW number was in a range of 1–11. It could be observed that the lowest output power was obtained when the QW number was one. As the QW number increased, the output power was enhanced respectively; however, we wished to underscore that the output power at lower injection current was decreased as the QW number was more than five, and the hole leakage was much small in the numerical analysis. Nevertheless, more QWs in the active region could undoubtedly reduce the electron leakage current and provide higher output power for higher injection current operation. Thus, numerical results suggested that the optimized QW number in the UV AlGaInN LED was in a range of 5–7.
0 0.8 1.6 2.4 3.2 4 4.8
0 50 100 150
1 well 3 wells 5 wells 7 wells 9 wells 11 wells
Output power (mW)
Current (mA)
380 K 300 K
Figure 3.8 L-I characteristic of the UV AlGaInN LED with variant QW numbers when the device temperatures were 300 K and 380 K.
0 2 4 6 8 10 12 14 16
300 310 320 330 340 350 360 370 380 1 well 3 wells 5 wells 7 wells 9 wells 11 wells
Percentage of electronic leakage current (%)
Temperature (K)
Figure 3.9 Percentage of electron leakage current as a function of the device temperature when the QW number was in a range of 1–11.
3-4 Summary
We had fabricated high-performance 370-nm AlGaInN UV LED. The AlGaInN LED could provide an output power of 0.8 mW at 20 mA with 3.6 V operation voltage and 4 mW at 125 mA under continuous-wave operation. With the help of numerical analysis, we further investigated the effects of the Al composition in AlGaN electron-block layer and the QW number on the 370-nm AlGaInN LED. The results obtained numerically suggested that the 370-nm AlGaInN LED could provide better output characteristics when the Al composition in AlGaN electron-block layer was in a range of 19%–21% and the AlGaInN QW number was in a range of 5–7. The qualitative analysis was significant for improving the output characteristics of the UV LED.
References
[1] M. Yamada, T. Mitani, Y. Nalukawa, S. Shioji, I. Niki, S. Sonobe, K. Deguchi, M.
Sano, and T. Mukai, “InGaN-based bear-ultraviolet and blue-light-emitting diodes with high external quantum efficiency using a pattern sapphire substrate and a mesh electrode,” Jpn. J. Appl. Phys., vol. 41, pp. L1431–L1433, 2002.
[2] S. Nakamura, M. Senoh, N. Iwasa, S. Nagahama, T. Yamada, and T. Mukai,
“Superbright green InGaN single-quantum-well-structure light-emitting diodes,”
Jpn. J. Appl. Phys., vol. 34, pp. L1332–1335, 1995.
[3] T. Mukai, D. Morita, and S. Nakamura, “High-power UV GaInN/AlGaN double-heterostructure LEDs,” J. Cryst. Growth, vol. 189, pp. 778–, 1998.
[4] T. Mukai, M. Yamada, and S. Nakamura, “Current and temperature dependences of electroluminescence of InGaN-based UV/blue/green light-emitting diodes,” Jpn.
J. Appl. Phys., vol. 37, pp. L1358–L1361, 1998.
[5] T. Mukai, H. Narimatsu, and S. Nakamura, “Amber InGaN-based light-emitting diodes operable at high ambient temperatures,” Jpn. J. Appl. Phys., vol. 37, pp.
L479–L481, 1998.
[6] M. Yamada, Y. Narukawa, and T. Mukai, “Phosphor free high-luminous-efficiency white light-emitting diodes composed of InGaN multi-quantum well,” Jpn. J. Appl. Phys., vol. 41, pp. L246–L248, 2002.
[7] V. Adivarahan, W. H. Sun, A. Chitnis, M. Shatalov, S. Wu, H. P. Maruska, and M.
Asif Khan, “250 nm AlGaN light-emitting diodes,” Appl. Phys. Lett., vol. 85, pp.
2175–2177, 2004.
[8] J. P. Zhang, S. Wu, S. Rai, V. Mandavilli, V. Adivarahan, A. Chitnis, M. Shatalov, and M. Asif. Khan, “AlGaN multiple-quantum-well-based, deep ultraviolet light-emitting diodes with significantly reduced long-wave emission,” Appl. Phys.
Lett., vol. 83, pp. 3456–3458, 2003.
[9] J. P. Zhang, A. Chitnis, V. Adivarahan, S. Wu, V. Mandavilli, R. Pachipulusu, M.
Shatalov, G. Simin, J. W. Yang, and M. Asif Khan, “Milliwatt power deep ultraviolet light-emitting diodes over sapphire with emission at 278 nm,” Appl.
Phys. Lett., vol. 81, pp. 4910–4912, 2002.
[10] H. Hirayama, A. Kinoshita, T. Yamabi, Y. Enomoto, A. Hirata, T. Araki, Y.
Nanishi, and Y. Aoyagi, “Marked enhancement of 320–360 nm ultraviolet emission in quaternary InxAlyGa1-x-yN with In-segregation effect,” Appl. Phys. Lett., vol. 80, pp. 207–209, 2002.
[11] V. Adivarahan, S. Wu, J. P. Zhang, A. Chitnis, M. Shatalov, V. Mandavilli, R.
Gaska, and M. Asif Khan, “High-efficiency 269-nm emission deep ultraviolet
Gaska, and M. Asif Khan, “High-efficiency 269-nm emission deep ultraviolet