The terahertz (THz) spectral range offers promising applications in science, industry, and military. The THz range is also important for astronomical research as 98% of the photons emitted since Big Bang are in sub-millimeter and far-infrared regime. The difference in THz absorption for different materials could also be used for three dimensional mapping. Absorption of THz frequencies by water can be used to distinguish cells with different water ratios (fat vs. lean cells). The body parts with low water contents (like teeth) could be fully mapped for any sign of decay within. On the order of millimeter penetration inside the body helps identifying cancer such as basal cell carcinoma (BCC); a most common form of cancer worldwide with over 1 million annual reported incidence in the USA. The tumor usually contains more water and less fat, which is resolved in terahertz imaging as darker areas.

Based on THz frequency range excitation of intermolecular interactions, NASA used a 2.5 THz laser to measure the concentration and distribution of the hydroxyl radical (OH^–) in the stratosphere, a critical component in the ozone cycle. THz spectroscopy enables characterization of solid materials as well as analysis of different forms of active pharmaceutical ingredients (for example, in a drug). Penetration through nonconductors (fabrics, wood, plastic) enables a more efficient way of performing security checks (for example at airports), as illegal drugs and explosives could be detected via this technique. Being a non-ionizing radiation, THz radiation is environment-friendly enabling a safer analysis environment than conventional X-ray based techniques. The compact solution to THz application will lead to a continuous monitoring of an environment ensuring a better security than conventional security check-points without effecting privacy. As an industrial impact, quality control of the packed goods could be managed by THz based systems.

A gyrotron is a vacuum tube based high power (kW-MW) emitter based on cyclotron resonance maser. Cost, maintenance, and space allocations limit the usage to applications such as heating of nuclear fusion plasmas and industrial thermal treatment of materials. A backward wave oscillator is an electron tube based emitter requiring highly homogeneous magnetic field of ~ 10 kG. The power is ~10 mW and wavelength is limited between 0.1 and 1.5 THz. A more convenient source is the optically pumped terahertz laser (OPTL) which consists of a grating-tuned carbon dioxide pump laser and a far-infrared gas cell mounted in the laser resonator. The stability of an OPTL is affected by slight changes in pumping wavelength, changes in the cavity length, and feedback interaction between the pump laser and the terahertz laser. The usable wavelength is 0.3-10 THz. Other sources such as direct multiplied sources (directly multiplies millimeter-wave sources up to terahertz frequency (~ 1THz)), and frequency mixing have low powers (<~10 μW) with limited upper frequencies (<10 THz).

Table 1: Comparison of THz Sources

 

GaAs/Al_xGa_(1-x)As based conventional Quantum Cascade Laser (QCLs) emitting in the THz regime have recently been demonstrated. New QCL designs have been proposed for THz emission using LO phonon depopulation. Even with novel ideas (such as usage of usage of photonic crystals, or dual-wavelength quantum cascade lasers) the GaAs system, which has ~36 meV LO-phonon energy, requires low operating temperatures for long wavelengths. With QCLs, no emission is reported around the LO phonon energy for such narrow gap materials. As a result, the GaN system, with a LO-phonon energy ħωLO~90meV, is a promising candidate for room temperature operation of far-infrared QCLs. First, the large LO-phonon energy can increase the lifetime of the upper laser state by decreasing the scattering rate of hot electrons in the upper subband. And on the other hand in order to establish the population inversion, ultrafast LO-phonon scattering in GaN/AlGaN quantum wells can be used for a rapid depopulation of the lower laser state.

Achieving high quality AlN/GaN SLs requires high quality AlN and GaN layers, and a fast transition between subsequent layers with well ordered interfaces. The optimum MOCVD growth conditions such as growth temperature/pressure and V/III ratio are different for AlN and GaN. Thus, realizing optimum SL growth conditions requires a trade-off between AlN and GaN quality.

The AlN and GaN have different optimum growth temperatures; this is primarily due to the different inherit atomic bond strengths of aluminum and gallium. The optimum MOCVD growth temperature for AlN is ~1200ºC whereas that of GaN is ~1050ºC. As parasitic pre-reactions between the TMAl and NH₃ precursurs occurs, lower pressures are generally desired for AlN growth whereas GaN typically obtains higher quality when grown under higher V/III ratios. Aluminum also has a smaller diffusion length due to strong bonding and small atomic radii. High growth temperatures and low pressures have been employed to enhance surface migration of aluminum adatom species and to suppress parasitic gas-phase pre-reaction between TMAl and NH₃. Due to these facts, the few reports of AlN/GaN SLs grown by MOCVD employ low growth temperatures (~885-935ºC). Although MOCVD growth temperature are typically at least 200ºC higher than those used in Plasma-assisted MBE leading to higher diffusion lengths and less surface segregation, it is still below that conventionally employed in MOCVD for GaN (1050ºC) and AlN (1200ºC).

Figure 1: Left) The pulsing sequence of AlN/GaN SLs. One period consists of two phases: (1) AlN deposition, which is realized via pulsing TMAl and NH3 to enhance aluminum adatom mobility, and (2) GaN deposition, which is realized by conventionally supplying TMGa and NH3 simultaneously. Right) A cross-sectional schematic of the AlN/GaN SLs grown on PALE AlN/LT-AlN/Sapphire.

One way to decrease the parasitic reactions while depositing AlN is via temporal separation of TMAl and NH3. This enhances the surface adatom migration and maximizing the growth efficiency can be achieved via this modulation, which employs a growth scheme of the separate introduction of the group III and group V precursors into the growth chamber in an alternating sequence. This pulsing enhances the diffusion length of aluminum adatoms leading to higher quality material.

Our pulsed deposition technique separates the SL growth into four parts as shown in Fig. 1: Steps (I) and (II) result in AlN deposition. In order to have a high quality AlN layer, the pulsed growth technique is used for AlN growth. This leads to lower effective NH₃ during the deposition of AlN, and thus, minimizes the parasitic pre-reactions. This resulted in 50% increase in growth rate of AlN. Step (III) deposits GaN. Using a bulk deposition of GaN ensures a high NH₃ partial pressure and thus high V/III ratio for the GaN. Step (IV) nitridizes the surface pre- aluminum deposition in preparation for the next SL period. This step prepares the surface for AlN growth via eliminating the excess Ga on the interface that may otherwise form an AlGaN interlayer degrading the interface sharpness. By this technique, V/III ratio for AlN is decreased whereas GaN is not affected.

Figure 2: (a) AlN thickness vs number of AlN pulses for TMAl duration of 2 s. Inset shows a cross-sectional diagram of a SL. (b) GaN thickness vs number of GaN pulses for TMGa duration of 2 s. Inset shows the (1x1 µm²) AFM SL surface without TMIn (height scale=1.5 nm). (c) GaN thickness vs TMGa duration for a single GaN pulse. Inset shows the (1x1 µm²) AFM SL surface with TMIn (height scale=1.5 nm).

In summary, a pulsed MOCVD deposition technique for high optical and structural quality AlN/GaN SL is introduced. Indium is shown to improve the surface and structural quality. Tunability of AlN and GaN thicknesses is demonstrated. ISB absorption at 1.53 µm is achieved at room temperature via MOCVD grown material. Lower (redshift) and higher ISB transition energy (blueshift) is observed in strainrelaxed and well- or barrier-doped SLs, respectively.

Figure 3: Relative (p-polarization) transmission of undoped, well- or barrier-doped, and uncapped, 30 or 100 nm capped 50 period {1.9-nm-thick GaN/3.2-nm-thick AlN} SL.

In summary, SLs with thinner wells (grown at the same TS) or grown at lower TS (employing the same well width) demonstrate higher strain effects leading to lower PL energy and ISB absorption energy. The higher strain effects in SLs with thinner wells and SLs grown under lower TS are attributed to higher well quality (due to growing thinner than critical thickness) and less thinning of the GaN well (due to less etching effect of Al adatoms). Well width and the growth temperature are shown to be key parameters for AlN/GaN SLs absorbing in optical communications wavelengths. Theoretical analysis and simulations are realized to explain experimental observations.

Figure 4: Transmission is plotted as a function of wavelength for four well widths of 1.5, 2.7, 4.3 and 7 nm. For all cases, AlN barrier thickness and growth temperature is 3.0 nm and 1035ºC, respectively. Inset shows simulated and observed ISB absorption energies as a function of well width (for the transition from the first state to the second state where other transitions are possible).

In summary, we have studied Al_0.2Ga_0.8/GaN superlattices grown by MOCVD with various well thicknesses via XRD, PL, and polarization dependant ISB absorption measurements. A theoretical model was developed taking into account strain in binary/ternary superlattices to correlate with both experimental interband emission and ISB absorption results. Ultimately we demonstrate tunability of ISB absorption from 4.5 to 5.3 µm in a GaN/Al_0.2Ga_0.8N superlattice. This demonstration is an important step towards employing GaN/AlGaN SLs in longer wavelength ISB devices.

Figure 5: (a) Solid line is the p-polarization transmission for 2.6 nm well width and 2.9 nm barrier with, wells are Si doped to 1×1018 cm-3, the dashed line corresponds to s-polarization transmission for the same sample. Inset: ISB absorption wavelength as a function of well width – experimental results (squares) and simulation (circles) for a fully strained structure. (b) From left to right, ISB absorbance for 1.2 nm (black), 2.6 nm (red), 3.7 nm (blue), and 5.1 nm (magenta) well widths. Lorentzian fits are shown (dashed lines) for the 3.7 and 5.1 nm well widths.