Recent Progress of Electrically Pumped AlGaN Diode Lasers in the UV-B and -C Bands

Abstract

The development of electrically pumped semiconductor diode lasers emitting at the ultraviolet (UV)-B and -C spectral bands has been an active area of research over the past several years, motivated by a wide range of emerging applications. III-Nitride materials and their alloys, in particular AlGaN, are the material of choice for the development of this ultrashort-wavelength laser technology. Despite significant progress in AlGaN-based light-emitting diodes (LEDs), the technological advancement and innovation in diode lasers at these spectral bands is lagging due to several technical challenges. Here, the authors review the progress of AlGaN electrically-pumped lasers with respect to very recent achievements made by the scientific community. The devices based on both thin films and nanowires demonstrated to date will be discussed in this review. The state-of-the-art growth technologies, such as molecular beam epitaxy (MBE) and metalorganic chemical vapor deposition (MOCVD); and various foreign substrates/templates used for the laser demonstrations will be highlighted. We will also outline technical challenges associated with the laser development, which must be overcome in order to achieve a critical technological breakthrough and fully realize the potential of these lasers.

1. Introduction

The electrically pumped (EP) and continuous-wave (CW) operating AlGaN-based diode lasers in the ultraviolet (UV)-B (320–280 nm) and UV-C (280–100 nm) wavelengths have significant potential in the four major application areas: free space non-line-of-sight communications [1], sensing [2], disinfection [3,4,5], and biomedicine [6,7]. A compilation of all the relevant applications in each area is shown in Figure 1. More recently, light sources operating at these wavelengths are discovered to be useful for sterilization of surfaces or objects, a necessary step to fight the global spread of coronavirus disease 2019 (COVID-19) [8,9]. While arguably some of these applications could be enabled by light-emitting diodes (LEDs), a wide range of applications of these UV-LEDs are limited due to their large-size, high-cost, and energy-inefficiency. Their advanced counterparts, e.g., lasers, however, show the promise of achieving low size, weight, power, and cost (SWaP-C) enabling devices [10]. Most importantly, lasers alleviate the light extraction and efficiency-droop constraints commonly found in III-nitride LEDs and extend their applications toward disinfection to air and large-surface sterilization at standoff distances because of their high-power density and light directionality.

Owing to its extraordinary band tuning capability, ternary Al_xGa_1−xN alloys with high to mid-range aluminum (Al) compositions are one of the most promising candidates to cover the entire UV-B and up to 206 nm of the UV-C spectral region [11]. Despite successful demonstrations of UV electroluminescence (EL) back in 1970 [12], several attempts to achieve AlGaN EP lasers failed with only reported success limited to the UV-A spectral range until 2015 [13,14,15,16,17,18,19]. The recent encouraging results of AlGaN optoelectronic devices including optically-pumped (OP) lasers [20,21,22,23,24] suggest the promise of Al_xGa_1−xN alloys. However, optical pumping significantly increases the complexity and cost of the overall system due to the use of an additional power-hungry pump laser. The use of AlGaN as active layers is made possible mainly due to significant efforts made in the past decades for developing low-defect density templates and high-quality epi-layers through optimized growth processes. Hence, the AlGaN material system shows great potential for realizing EP lasers emitting in the range of 206–320 nm.

From the perspective of material types, both thin films and one-dimensional nanostructures are already studied, and fascinating laser results are achieved. Since obtaining high-quality substrates to grow low-defect AlGaN epitaxial thin films has been a bottleneck for many years, nanowires (NWs) are considered to be a viable alternative to access at these wavelengths [25,26,27,28,29]. In 2019, the first EP thin-film UV-C laser diode was demonstrated [30]. Since then, thin film-based devices continue to show progress, sporadically covering both the UV-B and UV-C wavelengths [30,31,32,33,34,35,36,37]. Due to materials challenges and difficulties in device processing, the cavity type is still restricted to only simple Fabry–Pérot edge-emitting lasers which need to be expanded to achieve complete utilization in multipurpose applications.

NWs offer fundamental advantages over thin film-based devices because of a number of intriguing properties including the ease of NWs growth without strain [38,39]; reduction of dislocations and other structural defects; enhanced dopant incorporation [40,41] and high-quality material development by effective strain management due to a large surface-to-volume ratio [25]. Recent studies show that AlGaN NWs with N-polarity and a tailored geometry can be utilized for designing low-threshold random lasers. Interestingly, the first EP sub-300 nm top-emitting lasers were demonstrated on the NW-based platform [28]. After the first demonstration of NW-based cryogenically-cooled EP lasers emitting at UV-C [28], their room temperature (RT) operation at the UV-B range was reported [27,29,42].

Given that the newly developed AlGaN laser technology for realizing UV-B and -C lasers coincides with the rapid increase in interest for the intended applications, a comprehensive review of recent progress and research activities is of paramount importance. A brief review on AlGaN UV-LEDs [43] explained the technologies involved in developing low threading dislocation density (TDD) AlGaN on sapphire, some of which have been implemented on laser structures. A shorter review covering AlGaN UV-lasers based on NWs with different UV wavelengths can be found in references [44,45]. Although there are book chapters and reviews on EP AlGaN lasers based on thin films [46,47,48] and NWs [45,49] at the broad UV band, focused reviews on such devices emitting, in particular, at the UV-B and -C spectral bands, however, are still missing. MBE-grown AlGaN thin film-based OP lasers and NW-based EP lasers emitting in the UV-C and part of UV-B (sub-300 nm) wavelengths were discussed in the very recent review [50].

Our focused review will cover only EP AlGaN lasers based on both thin films and NWs grown by either MBE or MOCVD. All the experimental efforts for realizing EP AlGaN lasers in the wavelengths covering UV-B and -C bands are summarized here. Due to the multidisciplinary nature of material growth, device design, fabrication, and characterization, we have striven to make this review accessible to individuals with varying backgrounds by giving a comprehensive overview of important aspects of AlGaN laser technology and providing details related to laser demonstrations.

We begin with a brief review of the recent developments of EP UV-B and -C laser technology in Section 2. The technical challenges associated with this UV-B and -C laser technology are then described. We then focus on the four major technical areas including substrate material, growth technology, hole injection and laser gain medium—in which dedicated effort needs to be made on in order to make a critical technological breakthrough. We next present all the laser experimental demonstrations that are reported to date as well as their relevant design, fabrication, and device characterization details. Finally, we provide concluding remarks with a general outlook for technological advancement that will pave the way for the exciting prospects of UV laser-based photonics.

2. Overview of Recent Progress of AlGaN Ultraviolet (UV)-B and -C Lasers

UV-B and -C LEDs have already successfully ensured reasonably good performance metrics. Hence, these devices are commercially available from a number of suppliers [51,52]. However, the additional complexity in terms of epi-structures [53,54], thicker layers [23,55,56] and higher crystalline quality [57,58,59] requirements have enabled achievement of lasers limited to only the UV-A wavelength span; which requires low-Al containing AlGaN heterostructures. For a long time until 2015, the shortest wavelengths reported for EP AlGaN-based UV lasers were 336 nm [60] and 334 nm [26] for thin films and NWs, respectively.

The first-ever sub-300 nm EP AlGaN-based thin film-based lasers with an emission wavelength of 271.8 nm was demonstrated in 2019 [30]. Only four months after the first demonstration, pulse and RT operating UV-B lasers at 298 nm were reported [31]. Similarly, in reference [33], the authors presented laser devices emitting at ~279 nm by following the epi-layer design in reference [30] with lower threshold current density. Omori et al. reported the improvement in 298 nm UV-B laser performance [32] by employing a slightly different active region. The technological advances over time for EP UV-B and -C lasers are illustrated in Figure 2.

Due to the unique ability of lateral stress relaxation associated with large surface area, developing high-quality AlGaN NWs has been quite successful with a wide range of Al compositions. In 2015, the first EP AlGaN based NW laser emitting at 262.1 nm was achieved [28]. Later in the same year, the same group demonstrated the EP AlGaN-based NW lasers, for the first time, at UV-B [27]. With further exploration, Zhao et al. reported RT operating EP AlGaN [29] based lasers at 239 nm in 2016 which, to date, is the lowest EP laser wavelength among all the thin film and NW-based devices.

Table 1. A brief summary of experimental realizations of AlGaN electrically pumped (EP) lasers listed based on the material types.

Reference	Growth Method	Material Type	Lasing Wavelength (nm)	Threshold (kA/cm2)	Substrate	Operating Temp.	Operating Mode
Zhang et al., 2019 [30]	MOCVD	Thin film	271.8	25	AlN single crystal	RT	Pulse
Sato et al., 2020 [31]	MOCVD	Thin film	298	41	AlN/sapphire	RT	Pulse
Sakai et al., 2020 [33]	MOCVD	Thin film	278.9	19.6	AlN single crystal	RT	Pulse
Omori et al., 2020 [32]	MOCVD	Thin film	298	25	AlN/sapphire	RT	Pulse
Kushimoto et al., 2021 [34]	MOCVD	Thin film	271.2		AlN single crystal	RT	Pulse
Tanaka et al., 2021 [35]	MOCVD	Thin film	300	13.3	AlN/sapphire	RT	Pulse
Tanaka et al., 2021 [36]	MOCVD	Thin film	290	35	AlN/sapphire	RT	Pulse
Zhao et al., 2015 [28]	MBE	Nanowire	262.1	0.2	Si	77K	CW
Zhao et al., 2015 [27]	MBE	Nanowire	289	0.3	Si	RT	CW
Zhao et al., 2016 [29]	MBE	Nanowire	239		Si	RT	CW

MOCVD = metalorganic chemical vapor deposition, MBE = molecular beam epitaxy, RT = room temperature, CW = continuous-wave.

provides a summary of AlGaN lasers and their corresponding performance parameters demonstrated by different research teams in the world.

3. Challenges of AlGaN UV-B/C Lasers

Despite significant development in growing single-crystalline AlGaN materials, realizing high-efficiency AlGaN-based laser diodes in the range of UV-B and -C is still challenging owing to several technical issues which are briefly described in Section 3. Figure 3 illustrates the major four challenges associated with the development of AlGaN laser technology.

3.1. Substrates and Defects

AlN templates with low dislocation density and point defects are essential to enhance the emission efficiency of UV lasers. Nearly 80 times improvement in IQE was reported for AlGaN QW grown on AlN/sapphire templates with a TDD of 5 × 10⁸ cm⁻² compared to those grown on conventional templates with TDDs 2 × 10¹⁰ cm⁻² [61,62,63]. Due to the employment of different growth techniques such as HTA-sputtered AlN, epitaxial lateral overgrowth (ELO), patterned sapphire substrate (PSS) and thick AlN layer growth, significant advancement was made on AlN/sapphire over the years [64,65,66,67]. This led to TDDs in the range of 10⁷–10⁹ cm⁻². Using AlN bulk substrate which has a very low defect density in the range of 10³–10⁴ cm⁻² has been quite popular for obtaining high-quality epilayers [30,68].

AlN bulk substrates developed by physical vapor transport (PVT) or hydride vapor phase epitaxy (HVPE) are now commercially available to be implemented in UV-optoelectronics [21,77,78,79]. However, while Al_xGa_1−xN with any Al mole fraction x was possible to grow on AlN/sapphire templates without degrading overgrown epilayers, pseudomorphic growth on top of AlN bulk substrates with no strain relaxation was reported up to x = 0.65 (Figure 3a) [80,81]. Large-scale growth of AlN bulk crystal is still under development, making the substrate quite expensive. Along with the high cost, other issues such as Mg-dopant defect-related absorption, and unavailability of conducting substrates impede the UV-optoelectronics from enabling completely [78,82,83,84]. While AlN bulk substrates or AlN/sapphire templates are the material of choice for growing AlGaN thin films, silicon is usually used in growing high-quality AlGaN NWs. The potential solutions to overcome the challenges regarding high-quality substrates and low-defect templates are presented in detail in Section 4.1 and Section 4.2.

3.2. P- and N-Doping for High-Al

As UV-B and -C lasers require high-quality, low-defect and thick AlGaN cladding layers for improved waveguiding [24,85], their electrical performance is equally important. To achieve the high current density required in the active regions for laser operation, cladding layers need to be conductive which is particularly challenging for p-type doping for all x of Al_xGa_1−xN, as shown in Figure 3b [72]. Behind ineffective p-doping, the most formidable challenges include large acceptor ionization energy for Mg dopants in AlGaN [86,87], formation of low energies by compensating defects (donors) like nitrogen vacancies [88], limitation of solubility of Mg in AlGaN [71,89], hydrogen passivation of the Mg dopants [70,90] and the existence of parasitic impurities, such as hydrogen (H), carbon (C) and oxygen (O) [69,91,92]. While significant efforts were made to develop low-defect, highly p-doped AlGaN layers using AlN bulk substrate and engineered sapphire templates, approaches such as distributed polarization doping (DPD), short-period superlattice (SPSL) and tunnel junction (TJ) will bypass the p-doping problem. A brief discussion on these three approaches can be found in Section 4.3.

Compared to p-type doping, n-type doping is easily attainable and Si-doped AlGaN up to x = 0.8 with a reasonable doping concentration was achieved [73,74,75]. However, a sharp increase in resistance was observed due to the high activation energy of Si which increases exponentially from 25 meV–250 meV once the x > 0.8 for n-Al_xGa_1−xN [73]. On top of this, the presence of compensating impurities carbon and oxygen in AlN substrates as well as MOCVD reactor impurities decrease donor concentrations in AlGaN with x > 0.85 by an order of magnitude [68,91,93].

3.3. Low-Resistive Ohmic Contact

High-doping concentrations in both n- and p-contact layers are usually required to form low-resistive ohmic contacts for current injection [94,95]. However, it becomes challenging for AlGaN materials due to the aforementioned reasons. At a nearly degenerately doping level, a “knee behavior” in conductivity due to self-compensation (vacancy-silicon complexes in n-type and nitrogen vacancies in p-type) appears [82]. A common practice is to grow a degenerately-doped GaN layer to make ohmic contacts, which requires linear Al compositional grading down to x = 0 [96,97]. Due to the use of insulating substrates, large-area backside n-contact formation is not possible. Figure 3c shows the representative AlGaN lasers with co-planar or intracavity contacts, which is the only possible way to form bottom n-side contacts. As the improvement in n- and p-type conductivity in Al-rich AlGaN continues, the low resistive ohmic contact realization emerges. As development continues in n- and p-type doping in AlGaN layers, as discussed in Section 3.2 and Section 4.3, the low resistive ohmic contact realization emerges.

3.4. Facet Formation

Unlike the other III–V compound semiconductor-based lasers, III-nitride materials provide a reflectivity of only ~19% from the naturally cleaved facet at the semiconductor–air interface. This may necessitate a high-reflection coating on one of the facets in order to overcome resonator losses and obtain lasing, as schematically shown in Figure 3d. Another issue is that both sapphire and AlN are extremely hard materials. Hence, backend processes including substrate lapping and laser scribing to obtain mirror-finish facets pose many technical challenges to complete laser fabrication [98,99]. One sometimes breaks processed lasers and polishes the cavity-ends to obtain high-quality facets, which, however, increases manufacturing cost.

4. Critical Technical Areas for AlGaN UV-B and -C Lasers

A great deal of research has been devoted to achieving technical knowledge on the way towards overcoming the challenges mentioned in Section 3. When it comes to EP laser technology development, good cavity design is one of the first steps. Good device performance can be obtained by an optimized laser heterostructure design, such as choosing layer thicknesses and compositions, as well as separate confinement heterostructures with strong transverse mode confinement to maximize the overlap with AlGaN QWs. This has to be combined with reducing the optical losses in EP lasers which is particularly challenging due to absorption tails in the highly Mg-doped p-AlGaN cladding and waveguide layers. A number of reviews, regular papers, and book chapters already provided a comprehensive discussion of this important aspect for both UV-B [100,101] and -C [24,46,54,102,103,104,105]. Therefore, we have considered this discussion in our review paper to be redundant. Instead, this review aims to capture some of the critical technical aspects that are not well-addressed in the literature. Figure 4 schematically shows the four major technical pillars that will determine the success of UV-B and -C laser technology and underpin the implementation of high-performance AlGaN devices.

4.1. Substrate Materials

The most commonly used two substrate materials for the growth of AlGaN laser epilayers are bulk AlN crystal substrates, and MOCVD-grown AlN templates on sapphire. For NW-based lasers, silicon is one of the most popular substrate materials. Though AlN bulk substrates offer low TDD, their use is still limited to fabricating high-cost devices and rarely considered for commercial applications. Figure 5 schematically shows all the possible substrates that are used for implementing EP AlGaN lasers. Unlike GaAs or Si, bulk nitrides cannot be grown using the conventional Czochraslski or Bridgman techniques because of their high melting temperature and very high dissociation pressure [106,107,108]. To develop AlN substrate with low dislocation either PVT/sublimation technique [109,110] or HVPE [111,112] is employed. Despite the successful demonstration of growing low TDD AlGaN layers on top of AlN substrates [80,113], the use of AlN buffer templates grown on top of sapphire is still the most popular way of growing AlGaN based LEDs and lasers as well as other electronic devices.

4.1.1. Substrates for UV-C Lasers

Conventionally UV-C AlGaN based devices are realized on AlN bulk substrates without templates. Pseudomorphic growth of doped or undoped AlGaN layers with >60% Al content [82,114] on AlN substrates have been possible with dislocation density less than 10³ cm⁻², being over-qualified for making EP lasers. The AlN bulk crystal substrate cost largely varies with the amount of dislocation density. With a substrate with <10³ cm⁻² dislocation density, wafer costs exceeds over $5 K. A list of commercially available AlN substrates and AlN/sapphire template with their respective cost and selling companies is given in

Table 2. Different companies selling commercial AlN substrates and AlN/sapphire templates.

Company	Cost/cm2	Dislocation Density (cm−2)	Largest Available Wafer Dimension	Substrate
Hexatech [115]	$275	<1 × 103	5.08 cm dia	Bulk crystal
MTI Corporation [116]	$1995	<1 × 105	1 cm × 1 cm	Bulk crystal
Kymatech [117]	$18	Threading 5 × 108	5.08 cm dia	AlN/sapphire
MSE supply [118]	$22	Screw 5.5 × 107 Edge 1.8 ×109	5.08 cm dia	AlN/sapphire
MTI Corporation [119]	$5		10.16 cm dia	AlN/sapphire

.

In spite of the high cost, bulk AlN crystal is still the material for choice for realizing UV-C lasers. To our best knowledge, all the EP UV-C lasers experimentally realized so far used materials that were grown on AlN bulk substrates by leveraging its low dislocation density property. It should be noted that substrate contributes only 10–20% of the overall cost in UV lasers or LEDs [120], thus solving the issues regarding active region optimization, p-AlGaN contacts should receive much more attention.

4.1.2. Substrates for UV-B Lasers

To grow AlGaN lasers suitable for UV-B applications, it was found that strain relaxation is inevitable due to the large lattice mismatch between AlN and AlGaN. AlGaN layers with 55% Al composition on top of both AlN bulk crystal and AlN/sapphire were successfully grown and no clear dependence on substrates was found when the thickness of AlGaN layers was >5 µm [23,31]. A high-temperature annealing (HTA) process at any thickness of the initially deposited AlN by sputtering was found to result in an order of reduction in dislocation density compared to non-annealed AlN thin films [121]. The HTA process of thin AlN layers to reduce TDD is a relatively new process [122,123,124], and subsequent continuous growth of AlN at a relatively high temperature followed by AlGaN growth with appropriate Al composition by MOCVD for laser epilayers was undertaken.

The 3D schematic of the HTA-sputtered AlN/regrown AlN templates is shown in Figure 5c. Such templates help bend TDDs while annealing, which was one of the keys to obtaining high-quality AlGaN active epilayers for EP lasers [31,124]. Soon after, transferring from direct 2D growth on top of the HTA AlN layer to a two-step growth method was reported. In the two-step method, 3D growth was first initiated on top of the HTA AlN layer prior to proceeding to 2D growth, as shown in Figure 5d. The dislocation density was significantly reduced in this novel growth technique [23]. While both templates were successful in terms of EP laser demonstrations in the UV-B range [31,32], the templates shown in Figure 5d are preferred. Very recently, another innovative two-step growth method has been reported [36]. Instead of using self-nucleated 3D AlN, patterned growth was employed in this novel technique where AlN films were first turned into a 3D periodic plano cocavo-convex patterns (PCCPs) as schematically shown in Figure 5e. However, the early stage demonstration of the process produced templates with a higher defect density compared to the self-nucleated 3D growth templates (see Figure 5d) and the resulting UV-B lasers of poorer performance.

4.1.3. Silicon Substrates for Nanowire Lasers

The development of AlGaN-based NWs grown on Si was largely influenced due to the large demand of Si photonics. With three-fold atomic symmetry (triangular lattice), Si (111) surface can grow AlGaN wurtzite crystal structures with an epitaxial relation to it [26,125,126,127,128]. Growing high-quality NWs on top of highly-desired CMOS-compatible Si (100) still remains a challenge and this area experienced very limited success [129,130].

4.2. Growth Technology

The two most popular methods available for epitaxial growth AlGaN UV laser materials are molecular beam epitaxy (MBE) [128,131,132,133] and metalorganic chemical vapor deposition (MOCVD) [134,135,136]. When it comes to templated substrates, MOCVD is superior to MBE for growing high-quality AlN templates due to high growth-temperatures and -rates [137,138,139]. One can then grow laser materials by employing either of the growth techniques on top of such MOCVD-grown templates. MOCVD-grown AlGaN QW UV-C LEDs with 10–20% EQE [62,97,140] and IQE as high as 85% [141] were achieved. Considering OP lasers [102,142], and very recent demonstrations of EP lasers at both UV-B [31] and -C [30], MOCVD has been the most popular choice mostly due to its wide growth window in terms of various growth parameters.

Although not as extensive as MOCVD, MBE-grown UV materials, i.e., both thin film and NWs, have also been studied over the years. As a matter of fact, MBE offers some distinct benefits over MOCVD in terms of interface diffusion, defect control, memory effect, high hole concentration in p-AlGaN, Mg passivation and significant control over defect incorporation [143]. Due to its relatively low growth temperature MBE grown samples do not suffer from surface damage due to chemical reaction at high temperature often associated with MOCVD. This suggests the superiority of MBE over MOCVD in terms of growing high-quality active regions. Noticeable success in terms of implementing EP UV-C lasers using, in fact, NWs was achieved [144,145,146]. In MBE, AlGaN NWs are typically grown spontaneously in N-rich conditions on Si substrate with GaN NW template grown over it.

Lasers with buried TJ may require a hybrid MOCVD/MBE growth approach [147,148]. In such structures, a base structure, containing all the layers up to an active region and the p⁺-side of the TJ, is grown by MOCVD, and the remainder of the device including the n⁺-side of TJs, n-cladding and n⁺-contact layers can be overgrown during the second epitaxial growth by MBE. Hence, the hybrid approach, comprising a MOCVD-grown base structure with the active region and MBE-grown TJs, is another viable alternative towards obtaining high-performance UV-B and -C lasers. As a matter of fact, the MOCVD-grown TJs are found to be more resistive compared to the MBE-grown equivalent structures [149,150,151]. Thus, growing the samples in two growth steps involves a tradeoff between the process complexity/device cost and attainable device performance.

4.3. Three Major Techniques for Hole Injection

In addition to the substrates issue, poor hole injection into active regions and the resulting reduced optical efficiency is another critical area of research. Low achievable doping for p-AlGaN cladding layers has been the fundamental roadblock to this problem. Extensive efforts are underway to improve p-type doping in AlGaN alloys. Using heavily-doped low-bandgap and non-transparent p-GaN as contact layers on top of thick p-AlGaN cladding circumvent the problem of the formation of low-resistive ohmic contacts. However, it comes at the expense of growing complex graded compositional profiles. Note that low light extraction efficiency in LEDs due to the high absorption nature of p-GaN at the UV spectral range is not a concern for resonant device like lasers [152]. In this section, we review the three state-of-the-art approaches to alleviate the p-AlGaN doping issue for implementing all-AlGaN EP lasers. We also emphasize their suitability to be adopted in different structures that translates improved device performance.

4.3.1. Distributed Polarization Doping

Introducing dopants in AlGaN by compositionally grading, commonly addressed as DPD takes advantage of both spontaneous and piezoelectric polarization. Since its first demonstration, DPD has been widely popular as an effective means to enhance p-AlGaN doping [153]. In all the experimentally demonstrated EP lasers at UV-B and -C, DPD was utilized within the laser epitaxial structures. The compositionally-graded AlGaN layers have their polarization charge distributed throughout the grading demonstrating high sheet carrier densities and immune to carrier freeze-out at low temperature. A hole carrier concentration of 4.2 × 10¹⁷ cm⁻³ was achieved for an undoped AlGaN layer with an average of x = 0.86 [154].

Figure 6a shows how compositionally graded AlGaN creates polarization-induced 3D charges that result in free carriers. The nature of the free charges, acceptors or donors is controlled by changing the direction of grading. Although hole conductivity has been achieved with DPD without doping, most p-type DPD studies introduced Mg doping as well despite the optical loss due to Mg-induced absorption [155,156,157]. To minimize Mg absorption at the DPD layer, both UV-C lasers [30,33] employed the undoped-graded AlGaN structure for DPD, while the UV-B structures introduced Mg doping to a certain degree. In the DPD structure, a two-step process with undoped AlGaN adjacent to the waveguide and p-doped graded AlGaN layers was implemented on the p-side contact of the UV-B lasers [31].

Because of very little difference in refractive index for Al-rich AlGaN layers used in thin film based UV devices, developing proper waveguide layers has remained a major challenge. The DPD structure integration on the p-AlGaN side has had major success in developing the UV-lasers. Along with its intended purpose of increasing hole injection, this high Al-containing DPD layer also improves the confinement factors due to a high refractive index contrast.

4.3.2. Short-Period Superlattice

A short-period superlattice, comprising of alternate layers of Al_xGa_1−xN with high and low x doped with Mg (Al_xGa_1−xN/Al_yGa_1−yN with x > y) was used to improve hole activation with the help of band offset and strong built-in spontaneous and piezoelectric polarization fields instead of thermal energy [158,159,160,161]. By improving vertical conductivity, the resistance of an SPSL reduced nearly 12 times compared to a bulk layer with the same composition, making the SPSL layer suitable to operate at a high voltage [56]. A hole concentration of 3.4 × 10¹⁸ cm⁻³ and resistivity of 6.4 Ω-cm for a 200 nm-thick Al_0.8Ga_0.2N layer was achieved by controlling the SL period and well thickness [160]. A schematic diagram along with band structure is shown in Figure 6b. The strong electric field built at the barrier and well enhances the ionization of Mg acceptors and reduces the activation energy from above 400 meV to 40–67 meV for Al_xGa_1−xN layers with x = 0.8. Besides hole-conductivity, the high Al-containing cladding layer [56] is also required for good optical confinement to reduce optical loss. Martens et al. experimented on different SPSL cladding layers with an average x of 0.37, 0.57 and 0.81 to investigate the optical loss and developed low-loss, highly-conductive p-AlGaN, being suitable for EP UV-C lasers [56]. The work demonstrated a positive slope in series resistance as x increases, while nearly an order of magnitude reduction in series resistance is obtained for lasers with p-SPSL compared to the devices with uniform composition [56].

4.3.3. Tunnel Junctions (TJs)

While efficient hole injection is feasible for UV lasers using DPD and SPSL structures, absorption and Joule heating at p-AlGaN leads to high internal loss and large series resistance, resulting in a high threshold. With the help of polarization doping, successful demonstration of GaN-based TJ was reported since 2001 [162] and in 2016 the first AlGaN based TJ was reported [163]. While p-AlGaN is necessary for p-side cladding, implementing a TJ eliminates the thick p-cladding layer requirement. The use of TJs as an intracavity contact for hole injection through interband tunneling is reported in a number of experimental studies [55,150,151,164,165]. With the non-equilibrium hole injection and ultra-low absorption at the TJ interface, the introduction of TJ simultaneously improves hole injection. So far, TJ has successfully been implemented only in UV-C LEDs based on both thin films [165] and NWs [166]. Hence, the TJ is another potential candidate for solving the problem of hole injection for making conductive p-AlGaN even with high x.

TJ resistance as low as 2 × 10⁻³ Ω-cm² was achieved for structures with x = 0.7 [165]. The resistance however increased significantly once the x is further increased due to the increase of tunnel barrier, raising the voltage penalty subsequently. To reduce the voltage penalty, the use of an ultra-thin InGaN layer in between the p⁺- and n⁺-AlGaN are described [163]. The insertion of a lower-bandgap InGaN, however, could dramatically increase the optical loss and subsequent threshold current increase that will very likely prevent devices from lasing. FP edge-emitting sub-300 nm lasers with “offset quantum-well” design to reduce optical loss on the p-cladding side with p⁺-AlGaN/i-GaN/n⁺-AlGaN TJs were recently proposed (see Figure 6c) [100,101].

4.4. Active Region

The low differential efficiency of UV-B and UV-C lasers may not be completely related to insufficient hole generation and injection mechanism, as explained in Section 4.3. For the successful design and demonstrations of UV-B and -C lasers, one primarily requires a good AlGaN-based active region to obtain high material gain [22,53,154]. This includes not only a good material quality with low-defects and sharp interfaces but also an optimal number of QWs, and the right thicknesses of QWs and barrier with optimal band offsets and adjusting the waveguide thickness [16,167,168]. These considerations will be added by the polarization switching phenomena of emitted light for the unique AlGaN material system [169,170]. Using a higher number of QWs appears to be non-conventional in the wavelengths of interest due to the constraint of uniform pumping of all the QWs with carriers to obtain material gain. If one of the quantum wells cannot be pumped enough, they will operate as a band-edge absorbing layer and then it fails to lase.

As far as the UV-A regime goes, a different trend holds. References [18,60], reporting UV-A emission, used three 3 nm thick QWs [171]. As opposed to conventional III-nitride laser structures, EP UV-B and -C lasers used relatively thicker QWs in active regions. One is often interested in a modal gain which is material gain times transverse confinement factor. Therefore, a thicker QW provides higher modal gain, however, at the expense of poor overlap between electron and hole wave functions in III-nitrides. Zhang et al. [30] in their first EP UV-C laser used a 9 nm thick compositionally-optimized AlGaN layer to emit at 271.8 nm [30]. This provided a transverse confinement factor of 2.9%. Compared to UV-C lasers, UV-B lasers used a slightly lower Al composition, i.e., x = 0.35 in the active region [31] with SQW of similar thickness or MQW of lower thickness. While all previously reported structures [30,31,33] used identical thickness, a double QW-based active region with a well thickness of 4 nm and barrier 8 nm was used to demonstrate nearly 40% improvement in threshold current with no change in emission wavelength [32].

Contrary to thin multi-QW based active regions used in conventional diode lasers, NW laser structures implemented “thick layer-based” unique active regions [27,28,29]. Interestingly, it was found that Al-rich AlGaN NWs feature extensive Ga-rich nanoclusters. The atomic scale Ga-rich AlGaN striation laterally exhibit discontinuity in the direction perpendicular to NW growth [27]. This indicates that atomic-scale compositional fluctuations possess quantum dot/dash-like structural characteristics within the Al-rich NWs. Such Ga-rich i-AlGaN nanoclusters have thicknesses varying from 0.25–2 nm along the growth direction, and lateral sizes in the range of 2–10 nm. Hence, the spontaneously formed quantum dot/dashes embedded within the i-AlGaN layer serves as an active region, providing strong 3D confinement and high material gain.

5. Demonstration of AlGaN Lasers

Significant efforts have already been undertaken to develop EP UV-B and -C lasers using thin film based epi-structures. With the recent advancement in high-quality AlN bulk- or AlN/sapphire templated substrates, low-defect active epilayers and improved hole injection, EP AlGaN lasers have turned into reality. To the best of our knowledge, only two research teams worldwide have recently experimentally demonstrated EP AlGaN lasers which will now be discussed. The structural descriptions and the achieved results of each device will also be briefly provided. The discussion is classified based on the material types used in the devices.

5.1. Thin Film Lasers

In 2019, the first thin film-based EP AlGaN laser at UV-C was achieved by researchers at Nagoya University, Japan, in cooperation with Asahi Kasei Corporation, Chiyoda City, Japan and Crystal IS, Inc., Green Island, NY, USA [30]. The laser materials were pseudomorphically grown on (0001) bulk AlN substrates by MOCVD. The devices used a single AlGaN 9-nm-thick QW to emit at 271.8 nm. For hole injection, pseudomorphic DPD layers were used on top of p-waveguide. Figure 7a schematically shows the fully processed FP lasers.

The light-current-voltage (L-I-V) characteristics of the devices are shown in Figure 7b. The devices operated at RT under pulse conditions. The measured threshold voltage and current density of the devices were 13.8 V and 25 kA/cm², respectively. The spectral characteristics of the devices are shown in the inset of Figure 7b. Despite the low operating voltage (voltage drop across the active region ~4.6 V), the large threshold current density observed hindered the CW operation of such devices. The lasers operated only under extremely short current-pulses of 50 ns and a repetition frequency of 2 kHz. This breakthrough was ascribed to a low defect density AlN substrate and incorporation of a polarization doped Mg-free p-AlGaN DPD-based cladding layer.

The same research team engineered the same laser epilayers by combining dry and wet etching [172,173,174] to form smooth-vertical sidewalls on mirror facets [33]. The facets were then coated with a distributed Bragg reflector (DBR) composed of HfO₂ and Al₂O₃, yielded 49.6% reflectivity. This significantly reduced the threshold current density of the device [175]. A cross-sectional scanning electron microscopy image with DBR coating is shown in Figure 8a. The L-I-V characteristics are shown in Figure 8b with the lasing spectrum in the inset. Despite the same epilayers in reference [30], the devices emitted at a peak reflectivity of DBR, i.e., 278.9 nm. The measured threshold current density of the device was 19.6 kA/cm². The authors demonstrated the process steps used in fabricating these devices over a 2″ wafer, showing a promise of large-area production in the future.

In the next year, the achievement of demonstrating the first UV-B diode laser was reported by the industry researchers from Asahi Kasei Corporation, Chiyoda City, Japan, in collaboration with Meijo University, Mie University and Nagoya University, Japan [31]. Unlike the UV-C devices, the UV-B devices used the epilayers grown on top of AlN/sapphire templates. The laser structures remain nearly the same as reference [30] except for the use of Al_0.35Ga_0.65N-QW as a gain medium.

Figure 9a presents the L-I-V characteristics for the UV-B lasers. The threshold current density and voltage were 41 kA/cm² and 27 V, respectively, which were much higher than the UV-C lasers [30,33]. A much higher threshold voltage was attributed to a large voltage drop at n- and p-contacts of the devices. The threshold current was experimentally confirmed to be highly sensitive to the p-contact width of the devices. To be specific, more than 400% reduction in threshold current was obtained by increasing the contact width from 2 µm to 11.5 µm. Figure 9b presents the current-dependent spectra, showing multimode emission around 298 nm.

A few months after the previous demonstration, Omori et al. from the same group reported another UV-B lasers emitting at 298 nm [32]. Compared to the previous structure, this laser structure used a two QW-based design and a thicker waveguide layer, yielding a larger modal gain due to a higher confinement factor. Figure 10a presents L-J-V characteristics with different cavity length. The pulsed threshold current density was 25 kA/cm². In addition to the highly-resistive n- and p-contacts, high absorption loss at the highly graded p-layer was attributed to the high threshold current. The high threshold voltage also forced to operate this device under pulse conditions. It was experimentally verified that the AlN molar grading at the DPD cladding layer had no significant influence on the internal loss owing to the polarization doping instead of impurity doping. Hence, the device epilayer design was considered as a suitable optical resonator after determining the internal loss to a few cm⁻¹. Very recently, a follow-up experimental study, investigating the carrier injection efficiency in the same laser structure, has been conducted by Sato et al. [37]. An imbalance in carrier injection leading to electron overflow was found responsible for the high threshold current density for the UV-B lasers and can be overcome by improving injection efficiency and obtaining highly-reflective mirror facets.

To further investigate the high threshold current for UV-lasers and possible ways to reduce it, very recently Kushimoto et al. reported the impact of the heat treatment on lasers [34]. Experiments revealed that the dark spots at the active region near the mesa sidewalls appeared after heat treatment and were directly related to point defects created during dry etch. The point defects diffused from the sidewalls to the device gain medium, introducing non-uniformity in active area composition [176,177]. To avoid these dark spots, depositing the p-contact away from the sidewall was experimentally confirmed to be better. The threshold current density as a function of the p/mesa gap is presented in Figure 10c. The wider mesa area allowed metal contacts to keep a higher gap from the sidewalls. Hence, a reduction in threshold current density was observed with increasing the p/mesa gap.

As an extension of the early UV-B laser effort of reference [32] for further improvement in device performance, the same set of Japanese academic and industry researchers very recently published new results. Using the same epi-layer structure of reference [32], the role of waveguide layer and p-type cladding layer thickness on the threshold current density of a device [35] was carefully analyzed. By reducing waveguide layer thickness from 150 nm to 50 nm and changing the p-type waveguide composition as well as doping, UV-B lasers emitting 300 nm with a threshold current density of 13.3 kA/cm² were achieved. Figure 11a,b represents the L-J-V characteristics for the improved lases along with the laser diode reported in reference [32] and the emission spectra, respectively. The p-type doping optimization placed a significant effect on the reduction of the turn-on voltage and differential resistance. The waveguide layer thickness reduction on the other hand improved the injection efficiency and optical confinement factor resulting in the record-low threshold current density at the UV-B spectral band.

At nearly the same time as their earlier publication [35], the same researcher also demonstrated a UV-B laser grown on the novel patterned AlN template (see Figure 5e) [36]. The objective behind employing a patterned AlN template was to achieve a controlled 3D growth of AlGaN film to reduce dislocation formation. Compared to the epi-layer structure [32] grown on a self-nucleated AlN sample, the PCCP-ALN-based lasers showed 10 kA/cm² higher threshold current density as well as 8 nm red shift in the lasing wavelength as shown in Figure 11c. The higher threshold current density observed in the study was mainly attributed to higher defect density in AlGaN films on PCCP-AlN templates.

5.2. Nanowire Lasers

While the thin film-based UV-optoelectronics was battling against high-quality substrates along with p-doping, nearly defect-free AlGaN NWs is possible on Si using MBE [42,44,178]. Spontaneously grown NWs show excellent lateral and transverse optical confinement and bring inherent advantage of the ease of p-dopant incorporation. Since the first demonstration in 2015, the evolution of NW-based EP UV-B and -C lasers will now be reviewed.

Three experimental demonstrations of NW-based EP AlGaN lasers in the wavelengths of interest are reported so far [27,28,29]. All of these three reports were made by the same research group from McGill University, Canada. The type of cavity adopted in all these studies used random NW arrays, yielding random lasing [179,180]. Figure 12a shows the schematic of AlGaN NW array lasers. The NW structure consisted of GaN:Si (∼250 nm), AlGaN:Si (∼100 nm), AlGaN (∼100 nm), AlGaN:Mg (∼100 nm), and GaN:Mg (∼10 nm) segments. The devices were designed to act as surface-emitting lasers although there was ~20 nm-thick metal p-contact on the top with negligible absorption.

In 2015, the first development was related to CW-operating EP lasers emitting at 262.1 nm [28]. The fully-processed device schematic is shown in Figure 12a. The first-ever EP UV-C lasers also reported the lowest threshold current density till date. The EL characteristics is shown in Figure 12b. The extracted L-I characteristics of the random lasers is presented in Figure 12c. The inset shows the L-I curve of the 289 nm lasing peak in a logarithmic scale. A low threshold current density of 0.2 kA/cm² for a lasing area of 10 µm² was observed for the devices. However, the device operation was limited to a cryogenic temperature, i.e., 77 K [28]. The inability to operate at RT was attributed to reduced gain due to the very large inhomogeneous broadening of quantum dots embedded in NWs and unoptimized cavity design.

Again a few months later, UV-B lasing from single-crystalline AlGaN NWs was demonstrated by the same research team from McGill University [27]. Figure 13a shows the current dependent EL spectra with a peak at 289 nm. The extracted L-I characteristics of the random lasers are presented in Figure 13b. Owing to the optimized 3D optical confinement structure, the threshold current density reduced to 0.3 kA/cm² under CW operation. The lasing area was 10 µm².

Despite the advantage of obtaining strong optical confinement, the quantum dot generation due to compositional variation at the i-AlGaN active region limits the shortest achievable lasing wavelength [28,181]. In 2016, the same research group for the third time addressed this issue and successfully achieved record short 239 nm lasing emission with improved uniformity and higher Al incorporation in the laser devices [29]. The tapered structure and the epilayers’ design and the resonator type all were kept unchanged except slight reduction in i-AlGaN thickness. Figure 14a shows the applied current-dependent EL spectra with much shorter linewidth compared to the previously reported results. RT and CW operation was observed from the devices. This is the shortest wavelength reported to date for any EP AlGaN laser. Figure 14b shows the extracted L-I characteristics for the random lasers at 239 nm.

6. Conclusions

The ultrawide bandgap AlGaN materials system offers the greatest promise for realizing compact and efficient diode lasers in the UV-B and -C spectral bands. This review provides the current status and issues in the development of EP AlGaN-based UV-B and -C diode lasers which serve as a critical component of future photonic devices and subsystems. Materials challenges are discussed in detail in four critical technical areas that require close attention for the successful implementation of emitters at these spectral bands. Compared to UV-B lasers, the reported UV-C lasers showed relatively better device performance due to the use of low-defect density bulk AlN substrates. Finding the right substrate materials for UV-B is not trivial. AlGaN templates with a target Al composition on AlN substrates have not yet shown their full potential in terms of working devices. Only sapphire template substrates with controlled strain engineering generated the working UV-B lasers so far.

Although EP thin film-based UV-B and UV-C lasers have been reported only very recently, they operated at a very high threshold current density and only under pulsed conditions. Compared to better-performing UV-A lasers, the devices under interest exhibited threshold current density ×3–4 higher. Hence, reducing threshold current density for UV-B and UV-C lasers to <10 kAcm⁻² is important to achieve CW operation. High optical losses and poor injection efficiency are the two potential issues of high threshold current density, which should be improved. An optimized heterostructure design, good active region by preventing electron overflow over QWs, balanced carrier injection, excellent thermal management and high-quality AlGaN layers are essential to obtain a reasonable optical loss and injection efficiency, enabling CW- and room temperature operating lasers with a low-threshold current.

All the reported EP AlGaN NW UV-B and -C lasers used a random cavity, where the resonance cannot be controlled precisely. For real-world applications, a controlled cavity with a defined resonance is necessary. Using AlGaN NWs grown on patterned substrates and exploiting photonic crystal effects could be a few possible solutions towards making application-suited devices. In addition to patterned growth, controlling NW geometry and spacing amongst the NWs, material bandgap engineering, excellent sidewall smoothness and surface quality are also important for the development of high-efficiency AlGaN NW-based UV-B and -C lasers.

The device physics of UV-B and -C lasers is still under investigation and a significant amount of work still has to be done to improve the laser performances. The state-of-the-art laser results consolidated in this review suggest the feasibility of EP lasing from AlGaN-based thin-film and NW material platforms. Further development of metal contacts especially on the p-side, p-SPSL cladding layer or low-resistive AlGaN TJ for improved hole injection are required in order to reach high carrier densities and obtain lasing. All in all, with the progress that has already been made and the potential solutions for some of the technical challenges, together with the technological advancement, this highly-demanding laser technology is expected to see rapid development over time. Continued and extended efforts will pave the way to achieve application-suitable lasers at these spectral bands and more critical breakthroughs are anticipated within the next few years.