We designed and fabricated a semiconductor optical amplifier-integrated dual-mode laser (SOA-DML) as a compact and widely tunable continuous-wave terahertz (CW THz) beat source, and a pin-photodiode (pin-PD) integrated with a log-periodic planar antenna as a CW THz emitter. The SOA-DML chip consists of two distributed feedback lasers, a phase section for a tunable beat source, an amplifier, and a tapered spot-size converter for high output power and fiber-coupling efficiency. The SOA-DML module exhibits an output power of more than 15 dBm and clear four-wave mixing throughout the entire tuning range. Using integrated micro-heaters, we were able to tune the optical beat frequency from 380 GHz to 1,120 GHz. In addition, the effect of benzocyclobutene polymer in the antenna design of a pin-PD was considered.
Furthermore, a dual active photodiode (PD) for high output power was designed, resulting in a 1.7-fold increase in efficiency compared with a single active PD at 220 GHz.
Finally, herein we successfully show the feasibility of the CW THz system by demonstrating THz frequency-domain spectroscopy of an α-lactose pellet using the modularized SOA-DML and a PD emitter.
Keywords: CW terahertz, CW terahertz system, terahertz spectroscopy, photomixer, pin-photodiode, dual- mode laser, tunable lasers.
Manuscript received Oct. 8, 2015; revised Feb. 25, 2016; accepted Mar. 4, 2016.
Eui Su Lee (euisu@etri.re.kr), Namje Kim (namjekim@etri.re.kr), Sang-Pil Han (sphan
@etri.re.kr), Won-Hui Lee (whlee07@etri.re.kr), Kiwon Moon (kwmoon@etri.re.kr), Il-Min Lee (ilminlee@etri.re.kr), Jun-Hwan Shin (jhshin7@etri.re.kr), and Kyung Hyun Park (corresponding author, khp@etri.re.kr) are with the Broadcasting & Media Research Laboratory, ETRI, Daejeon, Rep. of Korea.
Donghun Lee (dhlee@etri.re.kr) is with the ICT Material & Components Research Laboratory, ETRI, Daejeon, Rep. of Korea.
I. Introduction
Owing to the unique properties of terahertz (THz) radiation, THz technologies have potential applications in spectroscopy, imaging, non-destructive testing, and wireless communications [1], [2]. Over the past decade, THz time-domain spectroscopy (THz-TDS) systems have been commercialized and are now commonly used in spectroscopy and non-destructive testing.
However, the applications of THz-TDS systems are limited because they require expensive and bulky femtosecond lasers.
Continuous-wave terahertz (CW THz) spectroscopy systems, which use an alternative method for generating THz frequencies, have several advantages over pulsed systems.
Their spectral resolution, which is less than 1 GHz, is superior to that of typical THz-TDS systems, which have a limited spectral resolution of a few GHz because of a mechanical delay and noise [3]. Additionally, CW THz systems have a lower cost and portability owing to the elimination of an expensive and bulky femtosecond laser. For the realization of a compact and hand-held CW THz system, a simple system configuration, small size, and low power consumption can be achieved by removing expensive optical components, such as the external power amplifier and polarization-maintaining fiber.
Current technological trends clearly indicate that the key components of THz systems are selected according to the specific applications for which they are used. THz-TDS systems, which typically employ wide-band photoconductive antennas based on low-temperature-grown (LTG) III-V materials, are mainly applicable to spectroscopic applications that require a bandwidth of at least 3 THz [4], [5]. High-power RF devices, such as resonant-tunneling diodes, Gunn diodes, and vacuum devices, such as gyrotrons, are suitable for
SOA-Integrated Dual-Mode Laser and
PIN-Photodiode for Compact CW Terahertz System
Eui Su Lee, Namje Kim, Sang-Pil Han, Donghun Lee, Won-Hui Lee, Kiwon Moon, Il-Min Lee, Jun-Hwan Shin, and Kyung Hyun Park
imaging applications [6], [7]. Photomixing methods have potential applications in thickness measurements, sensors, and hand-held systems requiring low power consumption and for which a moderate tuning range of ~1 THz is sufficient [8], [9].
Photomixing typically suffers from a low power-conversion efficiency of the photomixers, which are fabricated using LTG III-V materials [10]. However, systems that employ high-speed photodiodes (PDs) have recently shown an enhanced output power of CW THz radiation [11], [12]. Therefore, optical beat sources have also attracted a great deal of interest in recent years.
Optical beat sources for photomixing should have certain characteristics, including dual-mode operation, a tuning range of ~1 THz without mode-hopping, and an output power of greater than 30 mW when considering the responsivity of the photomixer. Several groups have reported the realization of optical beat sources for CW THz generation through the use of photomixing. Monolithically integrated dual-wavelength lasers with two independent laser cavities have been reported [13].
Although these devices exhibit a wide tuning range of the optical mode beat frequency using a distributed Bragg reflector laser diode (LD) structure or integrated micro-heaters (µ- heaters), a single cavity scheme is preferable because of the improved spectral purity of the resulting CW THz radiation.
The longitudinal single-mode operation of the reported dual- wavelength fiber lasers [14], [15] is a critical problem because it prevents fiber lasers from being used as optical beat sources.
In addition, optical-comb based optical beat sources [16]
appear to be particularly suitable as beat sources for CW THz generation from a physical point of view; however, they present problems in optical tunable filters, and their complex configurations have yet to be resolved. Recently, monolithically integrated dual-mode lasers (DMLs) with a single-cavity geometry have been reported [17], [18]. Although the output power of DMLs has increased by changing the grating structure from complex-coupled gratings to λ/4 phase- shifted gratings, this remains limited owing to the compound cavity modes, which limit the operating current and dual-mode operation. Consequently, integrating semiconductor optical amplifiers with DMLs can offer a promising solution for the development of a beat source with sufficient power to drive a photomixer or high-speed PD.
In this paper, we report a semiconductor optical amplifier- Integrated dual-mode laser (SOA-DML) module with an output power of greater than 15 dBm (31.6 mW), and a wide optical beat frequency tuning range of 380 GHz to 1,120 GHz.
We show that the SOA-DML can achieve a high side-mode suppression ratio (SMSR) of more than 35 dB throughout this tuning range. The use of an integrated spot-size converter (SSC) ensures stable mode behavior by suppressing reflections from the output facet, and provides a high coupling efficiency.
Furthermore, we describe the design a dual active PD for a high output power, resulting in a 1.7-fold increase in efficiency compared with the single active PD at 220 GHz. Finally, by demonstrating the THz frequency-domain spectroscopy of an α-lactose pellet, we confirm that the SOA-DML and packaged PD can be implemented as a compact beat source and an emitter for CW THz systems.
II. Device Fabrication of the SOA-DML
Figure 1 shows a schematic of the SOA-DML, and the inset shows the structure of the SSC region. The DML consists of two DFB LDs and one phase section, which controls the cavity losses by applying a reverse bias to suppress the compound cavity modes. Compound cavity modes are lasing modes that obtain the optical gain from the entire device structure. The integrated µ-heaters on top of each DFB LD section control the operating wavelength of each DFB LD independently. Because the µ-heaters adjust the temperature of the small active layer of the DFB LDs, the lasing wavelength tuning can be accomplished independently at a high speed (30 ms/THz) [19].
The lengths of the DFB LD, phase section, and SOA are 400, 50, and 700, respectively. Each active region is electrically isolated using deep-trench etching through a highly doped p- type cladding layer. The base waveguide for the SSC consists of a 150-nm thick 1.1Q-InGaAsP quaternary layer, a spacer, and a grating layer, and is fabricated using metal-organic chemical vapor deposition. The λ/4 phase-shifted gratings are then defined using electron-beam lithography. The Bragg wavelengths of the gratings were 1,300 nm and 1,304 nm, which correspond to an initial optical beat frequency of ~700 GHz. The grating layers are etched using reactive ion etching followed by a wet treatment to remove the polymer and damaged layer. The active layer for the DML and SOA regions consist of seven strained InGaAsP quantum wells sandwiched by a two-step separate-confinement heterostructure [18].
Because the SSC is composed of a passive layer and a tapered region, regrowth of the passive layer has been conducted
DFB LD2 Phase section DFB LD1 Spot-size
converter
SOA
µ-heater 1 µ-heater 2
p-clad Base waveguide Taper structure
(passive)
Fig. 1. Schematic diagram of the SOA-DML. Inset shows details of the SSC structure.
AR
Fig. 2. Optical propagation simulated using the BPM for the SSC region: (a) top view of the tapered region and active waveguide region, (b) top view of the base waveguide layer, (c) side view showing both waveguides, and (d) measurement output beam profile of the SOA-DML.
Pseudo-color plot of the field profile and FFA measurement.
–6 –4 –2 0 2 4 6 –6 –4 –2 0 2 4 6 –1 0 1 2 3 4 5 6 7 500
400
300
200
100
Z (μm)
x (μm) x (μm) y (μm)
(a) (b) (c)
–4,000
0 3,000 4,000 0
–4,000 0 1.2
y (μm) x (μm)
1.0
0.5
0.01.0
0.5
0.0
Intensity (normalized)
–4,000 –2,000 0 2,000 4,000 Z0
Z0 +1 mm Vertical
FFA = 19
Horizontal FFA = 16
(d) 0
through a butt-coupling process [20]–[22]. To reduce the internal loss and reflection at the butt-joint interface, the angle of the interface and the regrowth conditions were carefully optimized. A waveguide, including a tapered structure, was defined, and the regrowth process was applied to form a p-n-p current-blocking layer and a p-type cladding layer. The SSC is composed of a tapered region with a passive and base waveguide for the shallow-ridge structure. The width of the tapered region gradually decreases from 1.2 μm to 0.1 μm over a length of 300 µm, and the tip was fabricated using a two-step lithography process. The length of the shallow-ridge region is 50 μm. In addition, the height and width of the p-cladding layer are 3 μm and 5 μm, respectively.
The basic operating principle of an SSC is evanescent coupling between the tapered region and base waveguide, which has a large optical field. The output waveguide is tilted by 7° to reduce the facet reflections, which can degrade the spectral purity of the DFB LD and may even inhibit the single- mode operation of the DFB sections. Figure 2 shows the simulated optical propagation for the SSC region, created using the beam propagation method (BPM). Figures 2(a) and 2(b) show top views of the tapered region and base waveguide region, respectively, and Fig. 2(c) shows a side view of the beam propagation. It is clear that the optical beam propagating through the tapered region is gradually transferred into the base waveguide. An SSC with a short taper typically perturbs the output mode profile and increases the coupling losses to the single-mode optical fiber (SMF). Figure 2(d) shows the three- dimensional beam profile and far-field output beam profile of the SOA-DML in the vertical and horizontal directions. The beam profiles are clearly similar to a Gaussian curve. The far- field angles in the vertical and horizontal directions are 19° and 16°, respectively.
III. Optical Characteristics of the SOA-DML
Figure 3 shows the initial operating state of the SOA-DML.
The operating currents of the two DFB LDs are 50 mA (DFB1) and 80 mA (DFB2), the reverse bias of the phase section is –0.6 V, and the current in the SOA is 250 mA. The SOA-DML chip is mounted on a Cu plate, with the temperature set to 20 °C. The coupled power of the lasing modes is greater than 10 dBm, and strong four-wave mixing (FWM) signals can be observed. Although the FWM signals are generated principally in the SOA section because of the large input power and strong nonlinear effects of the semiconductor active layer, a DML without the SOA has been reported to exhibit strong FWM signals [23]. This clearly indicates the strong phase correlation of the two lasing modes and high spectral purity of the SOA-DML. In our experience, strong FWM signals are accompanied by clear optical beat signals, which are apparent in the autocorrelation measurements. Additionally, because FWM signals with a frequency detuning above 100 GHz are mainly caused by carrier density pulsations, we can expect a low phase noise and stable mode beating [23].
Compound cavity modes do not result from the all-active structure. Even if we use a passive waveguide for the phase section, compound cavity modes may be excited, and are difficult to suppress. However, the phase section with the active material provides an effective control mechanism for cavity losses in the compound cavity modes. In the SOA-DML, we used an SSC to extend the spatial extent of the optical modes
Fig. 3. Initial operating state of the SOA-DML. Operating currents of the two DFB LDs are 50 mA and 80 mA, reverse bias of the phase section is –0.6 V, and current in the SOA is 250 mA.
1,292 1,296 1,300 1,304 1,308 1,312 Wavelength (nm)
20 10 0 –10 –20 –30 –40 –50 –60
Output power (dBm)
Initial state
FWM DFB1
FWM DFB2
and hence reduce the reflections at the output facet. This approach also increases the mirror losses of the compound cavity modes and consequently reduces the reverse bias in the phase section, which requires a dual-mode operation. Without the SSC section, DMLs typically require –1.0 V to be applied to the phase section to ensure clear dual-mode operation throughout the entire tuning range [19]. However, the reverse bias in the phase section is only –0.6 V, and the SOA-DML exhibits dual-mode states at the specific beat frequency even with a forward bias of less than 0.5 V and with a negative current.
Figure 4 shows the output power characteristics of the SOA- DML as a function of the operating current of the SOA section.
The operating conditions of the DML are as described above.
The output power of the SOA-DML chip, measured using an integrating sphere, exceeds 60 mW (18 dBm) and is gradually saturated at 65 mW for an operating current of 300 mA. It must be noted that the large gain of the SOA section amplifies the reflected light from the output facet and degrades the spectral purity of the DFB LDs. The reflection coefficient at the output facet must be smaller than 10–6 [20]; in the SOA-DML, this is accomplished using the SSC with a 7° tilted output waveguide and an anti-reflection coating. The SSC expands the output beam profile, which helps reduce the coupling losses to the SMF and decreases the back-coupling (that is, the reflections) into the waveguide. The mode profile and spectral purity of the SOA-DML are maintained as the operating current of the SOA section increases. The spectral linewidths of the lasing modes are good figures of merit to measure the effect of the feedback and spectral purity. Herein, we measured the spectral linewidths using a delayed self-homodyne method and a time delay of 25 μs [24]. Each measured mode was selected using a narrow band-pass filter. The spectral linewidths in the dual-
Fig. 4. Output power of the SOA-DML chip as a function of the operating current of the SOA section. Here, DML operating conditions are fixed. Inset shows the spectral linewidth as a function of the µ-heater current.
0 50 100 150 200 250 300
SOA section current (mA) 70
60
50
40
30
20
10
0
Output power (mW)
0 20 40 60 80 100 120 140 160 μ-heater current (mA) 10
8 6 4 2 0
Linewidth (MHz)
Fig. 5. Tuning spectra of the SOA-DML: (a) spectra of the initial, minimum, and maximum beat frequencies, and (b) spectra over the entire tuning range (we defocused the lensed fiber to avoid unwanted feedback from the tip of the fiber).
–20 FWM
–40 –60 0 –20 –40 –60 0 –20 –40 –60
Output power (dBm) Output power (dB scale)
1,292 1,296 1,300 1,304 1,308 1,293 1,296 1,299 1,3021,3051,3081,311 DFB1 DFB2 Initial
FWM
FWM FWM
DFB1 DFB2
FWM Minimum
FWM DFB1
DFB2 Maximum
μ-heater2
μ-heater1 DFB1 DFB2 FWM
(a) (b) 0
Wavelength (nm) Wavelength (nm)
mode state are approximately 4 MHz, and are well maintained when changing the μ-heater currents. The linewidths of the DFB2 mode (right-side mode) as a function of the μ-heater currents are shown in the inset of Fig. 4. The operating current of the SOA is 250 mA.
Figure 5 shows the tuning spectra of the SOA-DML. By using integrated μ-heaters, we are able to tune the optical beat frequency from 380 GHz to 1,120 GHz. As shown in Fig. 5,
the output power of the SOA-DML is maintained owing to the integrated SOA, and strong FWM signals can be seen throughout the tuning range. Figure 5(a) shows the spectra at the initial, minimum, and maximum optical beat frequencies.
The side mode suppression ratio at the initial beat frequency was 48 dB and remained above 35 dB throughout the entire tuning range.
The tuning range is limited owing to a lasing of the –1 mode in each DFB section. This phenomenon most likely results from the reduced optical gain and the change of the modal power distribution inside each DFB LD section owing to the dissipated power of the integrated μ-heaters. Thermally tuned DFB LDs typically exhibit lower optical gain because the gain peak shifts an order of magnitude faster than that in the lasing mode [23]. The output power of the DFB LD then decreases slightly and becomes sensitive to the external feedback of the phase variation. Although the output powers of the lasing modes in the SOA-DML are maintained owing to the integrated SOA, the gain peak shift in response to wavelength tuning cannot be avoided. This problem might be solved by introducing an active layer with a wide optical gain bandwidth, such as multi-quantum wells with varying thickness. This limited tuning range, however, does not reduce the importance or potential applications of SOA-DMLs. A tuning range of larger than 700 GHz is sufficient for thickness measurements and sensing applications. Furthermore, the initial optical beat frequency of the SOA DML can be easily adjusted to satisfy the demands of specific applications.
IV. Design and Fabrication of Pin-PD
Figure 6(a) shows a three-dimensional schematic diagram of the PD designed to operate at a wavelength of 1.3 μm. The PD chip is mainly comprised of a single-mode ridge waveguide, a pin-mesa with an extended optical matching layer, and a broadband THz antenna with the bias pads on the benzocyclobutene (BCB) layer. The epitaxial layers and structures were designed using the Harold program and BPM method.
For efficient fiber-to-chip coupling, a shallow ridge-type InP/InGaAsP (core) passive optical waveguide is integrated at the bottom layer of the active PD section, as shown in Figs. 6(a) and 6(b) [25]. The width and height of the upper clad in the ridge waveguide are 3 μm and 0.5 μm, respectively, and the thickness of the core is 0.25 μm. Figures 6(b) and 6(c) show the fundamental and asymmetric mode profiles of the shallow ridge waveguide owing to the thin upper-clad thickness. The calculated coupling efficiency from the lensed fiber (mode field diameter of 3.2 μm) to the waveguide is about 70% because of the asymmetric mode profile in the vertical
Horizontal direction (µm)
Horizontal cut of mode profile at Y = 2.12
–5 –4 –3 –2 –1 0 1 2 3 4 5 1.0
0.0 0.6 0.8 0.4 0.2
1.0
0.0 0.6 0.8 0.4 0.2
Fig. 6. (a) Schematic diagram of the PD, and BPM simulation data of (b) single mode and (c) mode profiles with a shallow ridge waveguide, and (d) simulated optical propagation for the PD region, and (e) Epi. layers and fabricated PD chip.
Vertical cut of mode profile at X = 0
–3 –2 –1 0 1 2 3 4 5 6 7
Air InGaAsP -Core
InP-Substrate InP-Clad Computed transverse mode profile
(m = 0, neff = 3.215)
4 3 2 1 0 –1
–2–5 –4 –3 –2 –1 0 1 2 3 4 5 5
Antenna BCB
140 120 100 80 60 40 20
0 –4 –2–02 4 6 1 0.5 0 Matching
Absorber
Waveguide
BCB
(a) (b)
(c) (d)
(e)
Mode amplitude
Y (µm)
X (µm)
Vertical direction (µm)
Y (µm) Monitor value (a.u.)
Z (µm)
Au electroplating
P-metal N-metal
P-clad BCB
Active Matching layer Spacer Waveguide core
Au electroplating Waveguide
Matching Absorber
direction (Y-axis). To reduce the reflection loss, an anti- reflective coating layer is deposited on the input facet of the ridge waveguide. Actually, to obtain the maximum coupling efficiency from fiber to chip, thick diluted waveguide structures can be used. However, owing to the burdens of the deep etching over a few micrometers in the fabrication and the thick epitaxial layers in the growth process, the waveguide structures should be considered with the maximum input power from the optical source and the coupling efficiency of the waveguide. To couple the optical beam evanescently from the base waveguide to the active region (absorber), the matching layer is important in achieving a high quantum efficiency in the PD. The uniform absorption along the absorber and the maximized coupling efficiency can be obtained based on the design parameters of the matching layer, such as the thickness, refractive index, and protrusion length. Actually, this design should be considered using the base waveguide structures determining the mode profile of the optical beat signal. In our structure, a 300-nm thick n+-doped InGaAsP layer, which is used as a low-
resistance n-type contact layer, is grown on the ridge waveguide. In addition, the protrusion length is 14 μm, and the starting taper width is 1 μm, with an angle of 14°. The active material of intrinsic In0.53Ga0.47As is used as the absorber for 1.3-μm light. For high-speed operations and a high absorption efficiency, we designed the thickness, width, and length of the active layer as 240 nm, and 5 μm, and 7 μm, respectively.
Figure 6(d) shows the simulated optical propagation for the PD region. Owing to the matching layer, the calculated responsivity in the active layer is about 0.25 A/W. Using these geometrical parameters, the p-clad and p-contact layers are stacked up on the active mesa layer with a 500-nm thick p+ doped InGaAsP (p-clad) and a 150-nm thick p+ doped In0.53Ga0.47As p-contact layer. In addition, to reduce the potential spike preventing the current flow, an InGaAsP space layer is inserted between the p- contact layer and absorption layer [25].
The absorbed optical beat source induces the photocarriers (holes and electrons), which are transported by the externally applied reverse bias voltage and then radiated by a log-periodic toothed planar antenna, as shown in Fig. 6(e). To reduce the antenna resistance and propagation loss of the THz signals, the antenna is plated with Au with a thickness of about 2 µm. In addition, we use BCB materials with a low dielectric constant (εr = 2.67) compared with the InP substrate under the bias/antenna lines instead of an air bridge structure, such that the capacitance and propagation loss can also be reduced.
The materials of the InGaAsP/InGaAs/InP on the InP substrate are grown using metalorganic chemical vapor deposition. To fabricate the devices, standard III-V processing techniques are used. This includes photolithography (stepper), thermal/E-beam evaporation, annealing, wet etching, dry etching, Au plating, lapping, and scribing.
V. Design of Single and Dual PD
The measured dark current of the fabricated PD with a log- periodic antenna was 80 nA under a 1.4 V reverse bias condition. When the input optical power is around 30 mW at a 1.3-μm wavelength, the measured responsivity is 0.3 A/W corresponding to a photocurrent of 9 mA under a reverse bias of 1.4 V. A lensed fiber with a mode field diameter of 3.2 μm is used for the measurement. As mentioned before, to reduce the resistance, propagation loss, and capacitance of the PD, which are related to the f3dB bandwidth and output power, we fabricated a Au-plated antenna on BCB polymer with a low dielectric constant, and thus the output power is increased by about four-fold at 400 GHz, as shown in Fig. 7(a). A homodyne measurement system based on an In0.53Ga0.47As detection photomixer grown at low temperature is used when a log-spiral antenna is utilized for the receiver [9], [26]. In
Fig. 7. (a) Measured amplitudes of THz photocurrent and (b) HFSS simulation data as the frequency with and without BCB and Au plating, (c) images of the fabricated single and dual PDs, and (d) measured power ratio of dual to single PDs.
0.05 0.2 0.4 0.6 0.8
100 101 102
THz photocurrent (nA)
Frequency (THz)
Single Waveguide
Active region
Dual
Dual / Single
(c)
S11 parameter (dB)
(a)
Frequency (GHz) (d)
(b)
+ _
_ +
–8
–12
–16
–20
–24200 250 300 350 400 450 500 Without BCB and plating
With BCB and plating
Without BCB and plating With BCB and plating
220 245 270 295 320
1.8 1.6 1.4 1.2 1.0 0.8 Enhanced power ratio (Dual/single)
Frequency (GHz)
addition, resonance peaks near 0.25 THz originate from the antenna structure. The blue-shifted resonance is due to the BCB material with a low dielectric constant compared with the III-V materials. Figure 7(b) shows the simulated resonance property of the log-periodic antenna, which is related to the BCB material under the antenna, using commercial HFSS software. The thickness of the BCB layer is 2 μm. The resonance frequencies are shifted about 50 GHz, which originates from the BCB layer. This effect should be considered when we try to design the resonance antenna.
A typical Pin-photodiode (pin-PD) suffers from an output current saturation resulting in a saturation of the output power through the space charge effect. We designed the dual active PD to overcome the power limit, as shown in Fig. 7(c).
Identical absorbers are connected in parallel with the antenna, such that the generated in-phase THz signals in each of the absorbers are summed and radiated through the antenna. In addition, to couple the optical signal from the waveguide to the dual absorbers without a phase difference, Y-type waveguide bends are used. The design concept of the dual active PD is similar to that described in [27], which uses a horn antenna and matching circuit with an operating frequency of around 300 GHz. Figure 7(d) shows the measured power ratio of the dual to single PDs, which results in a 1.7-fold increase in output power at 220 GHz. However, the enhanced power is gradually decreased as the frequency increases. Because the total capacitance of the dual PDs can be increased by the mutual influence of both absorbers, the RC reaches its limit. A commercial Schottky-barrier diode (SBD) with a horn antenna,
Fig. 8. Schematic diagram of the PD packaging and photograph of the packaged PD.
(7) Metal cover
(6) Urethane (5) Si lens (4) Submount (3) Chip (2) Lens (1) Body
(8) SMA connector
(9) Fiber assembly
(0) Packaged photodiode
which limits the measurement bandwidth from 220 GHz to 325 GHz, is used for the output power measurement.
VI. Package Design of the PD
To utilize the fabricated PD chip as a CW THz emitter in various THz application fields, a mobile PD package is needed.
Figure 8 shows a schematic of a PD package designed using the commercial software SolidWorks. The package is comprised of nine parts: the body, optical lens, PD chip, submount, Si lens, polyurethane spacer, metal cover, SMA connector, and fiber assembly.
Initially, the PD chip is bonded to the AlN submount using a flip-chip bonding process. The submount material should have a high thermal conductivity. Because a thermal failure can be generated during a high current operation, the output power can be limited by the saturation current. The thermal conductivity of the AlN material is 285 W/(m·k). [26] Then, using the die-bonding technique, the bottom of the submount is attached to the body. During this process, the centers of the log- periodic antenna and the silicon lens must absolutely be same for the symmetric THz radiation patterns. In addition, the SMA connector is fixed to one side of the body and connected with the bias pads of the submount to supply the operation voltage.
To transfer the optical power from the DML to the PD chip, the single-mode fiber is aligned with an optical lens by monitoring the photocurrent of the PD. It is then joined with the body using laser beam welding for long-term stability of the package.
Finally, the Si hyper-hemispherical lens is mounted on the PD chip using a polyurethane spacer and a metal cover in a compact body, as shown in Fig. 8. Owing to the high dielectric constant of the InP substrate, the generated CW THz signals are mainly emitted toward the backside of the chip, and the signal divergence is thus properly suppressed by the Si lens.
VII. Compact CW THz Spectroscopy System
Figure 9 shows a schematic diagram of a simple THz
Fig. 9. Schematic diagram of a simple THz frequency-domain spectroscopy system.
SBD Sample Controller board
Polarization controller
Lock-in amp.
Ref Bias
THz
Function generator PD
PD module
SOA-DML module
frequency-domain spectroscopy system. The SOA-DML module is used by a typical butterfly package. Optical beat signals provided by the SOA-DML are focused on the input waveguide of the PD chip using a polarization controller. Most of the generated THz fields are radiated onto the InP side because of the high refractive index, n = 3.5, at 1 THz. The PD is mounted on the silicon lens used for the collimation of the THz radiation. A radiated THz signal is detected using a quasi- optical broadband SBD detector after a free-space transfer for a distance of 1-in below the silicon lens side. The function generator supplies a –1.4 V signal modulated by 1.03 kHz with a 50% duty-cycle square wave for the PD and the modulation frequency for a lock-in amplifier. This simple CW THz spectroscopy system based on the heterodyne configuration eliminates the use of passive components, such as a 3-dB optical coupler, polarization-maintaining fiber, or polarization controllers, and thus the system size, power consumption, and system configurations are significantly enhanced.
To demonstrate a simple THz frequency-domain spectroscopy system, we prepared an α-lactose pellet from a 30% α-lactose and 70% polyethylene mixture with a 1-mm thickness. The characteristic absorption peak of α-lactose monohydrate below 1 THz, associated with the molecular vibration of α-lactose in the crystal, can be identified at 0.53 THz (17.1 cm–1) [26]. The α-lactose pellet was placed between the emitter (PD) and detector (SBD), as shown in Fig. 9. Figure 10(a) shows the frequency-domain spectra measured without (free space, line) and with the sample (dots) using a lock-in technique. We tuned the optical beat frequency with a 1-GHz resolution from 380 GHz to 900 GHz using the integrated μ-heaters. The measured data are from only one scan, and the values of the time constant, acquisition time, and reference frequency in the lock-in amplifier are 300 ms, 0.7 s, and 1.03 kHz, respectively.
The photocurrent of the PD is 5.7 mA at a bias voltage of –1.4 V, and the input optical power is only about 20 mW, which is within the output power capability of the SOA-DML.
The measured broad dip near 650 GHz was caused by the Tx/Rx antenna pair with log-periodic antenna structures. A commercially available Schottky barrier diode is used for the
Fig. 10. (a) Measured output voltages in free space (line) and with an α-lactose pellet (dots) from the SBD detector and (b) comparison of α-lactose resonances measured using CW THz and pulse spectroscopy systems.
–4.0 –4.2 –4.4 –4.6 –4.8 –5.0 –5.2 –5.4
–5.6400 500 600 700 800 900 Frequency (GHz)
Measured voltages (V, rms)
1.20
1.15
1.10
1.05
1.00
Ratio of measured voltages
5
4
3
2
1
0 –1Absorption coefficient (cm) Free space
α-lactose
CW Pulse
400 450 500 550 600 650 Frequency (GHz)
(a) (b)
receiver. Figure 10(b) shows the ratio of the voltages (blue line) and the absorption coefficient (red line) of the α-lactose pellet, as measured by the THz-TDS system [26]. The same absorption peaks measured near 530 GHz indicate the effectiveness of our simple frequency-domain spectroscopy system.
VIII. Conclusions
In conclusion, we developed an SOA-DML module demonstrating a high output power of over 15 dBm, a wide optical beat frequency tuning range of 380 GHz to 1,120 GHz, and an SMSR above 35 dB throughout the entire tuning range.
The integrated SSC ensures a high coupling efficiency with an SMF and a narrow spectral linewidth owing to the reduced facet reflection. We believe that a single-chip THz emitter or compact CW THz emitter module with a wide tuning range of the optical beat frequency can be fabricated using the SOA- DML approach described herein. A hybrid package composed of an SOA-DML and a high-speed PD without any other devices may be the first step toward the development of a single-chip CW THz emitter based on photonics technology.
Additionally, we successfully demonstrated the simple implementation of CW THz frequency-domain spectroscopy.
These results indicate that the proposed system constitutes a promising development toward the realization of a compact and cost-effective CW THz system.
References
[1] B. Ferguson and X.-C. Zhang, “Materials for Terahertz Science and Technology,” Nature Mater., vol. 1, Sept. 2002, pp. 26–33.
[2] P.U. Jepsen, D.G. Cooke, and M. Koch, “Terahertz Spectroscopy and Imaging – Modern Techniques and Applications,” Laser
Photon. Rev., vol. 5, no. 1, Oct. 2011, pp. 124–166.
[3] J. Xu et al., “Limit of Spectral Resolution in Terahertz Time- Domain Spectroscopy,” China Physics Lett., vol. 20, no. 8, 2003, pp. 1266–1268.
[4] B. Sartorius et al., “All-fiber Terahertz Time-Domain Spectrometer Operating at 1.5 µm Telecom Wavelengths,” Opt.
Express, vol. 16, no. 13, June 2008, pp. 9565–9570.
[5] K. Moon et al., “Generation and Detection of Terahertz Waves Using Low-Temperature-Grown GaAs with an Annealing Process,” ETRI. J., vol. 36, no. 1, Feb. 2014, pp. 159–162.
[6] M. Asada, S. Suzuki, and N. Kishimoto, “Resonant Tunneling Diodes for sub-Terahertz and Terahertz Oscillators,” Jpn. J. Appl.
Physics, vol. 47, no. 6, June 2008, pp. 4375–4384.
[7] M. Bessou et al., “Three-Dimensional Terahertz Computed Tomography of Human Bones,” Appl. Opt., vol. 51, no. 28, Oct.
2012, pp. 6738–6744.
[8] E.R. Brown, “THz Generation by Photomixing in Ultrafast Photoconductors,” Int. J. High Speed Electron. Syst., vol. 13, no. 2, June 2003, pp. 497–545.
[9] I.-M. Lee et al., “Frequency Modulation Based Continuous-Wave Terahertz Homodyne System,” Opt. Express, vol. 23, no. 2, Jan.
2015, pp. 846–858.
[10] E.R. Brown, F.W. Smith, and K.A. Mclntosh, “Coherent Millimeter-Wave Generation by Heterodyne Conversion in Low- Temperature-Grown GaAs Photoconductors,” J. Appl. Physics, vol. 73, no. 3, 1993, pp. 1480–1484.
[11] T. Nagatsuma, H. Ito, and T. Ishibashi, “High-Power RF Photodiodes and their Applications,” Laser Photon. Rev., vol. 3, Feb. 2009, pp. 123–137.
[12] B. Sartorius et al., “Continuous Wave Terahertz Systems Exploiting 1.5 µm Telecom Technologies,” Opt. Express, vol. 17, no. 17, Aug. 2009, pp. 15001–15007.
[13] M. Theurer et al., “Photonic-Integrated Circuit for Continuous- Wave THz Generation,” Opt. Lett., vol. 38, no. 19, Oct. 2013, pp.
3724–3726.
[14] P.J. Moore, Z.J. Chaboyer, and G. Das, “Tunable Dual- Wavelength Fiber Laser,” Opt. Fiber Technol., vol. 15, no. 4, Aug.
2009, pp. 377–379.
[15] M.Y. Jeon et al., “Widely Tunable Dual-Wavelength Er3+-Doped Fiber Laser for Tunable Continuous-Wave Terahertz Radiation,”
Opt. Express, vol. 18, no. 12, May 2010, pp. 12291–12297.
[16] A.J. Seeds et al., “Coherent Terahertz Photonics,” Opt. Express, vol. 21, no. 19, Sept. 2013, pp. 22988–23000.
[17] N. Kim et al., “Widely Tunable 1.55-μm Detuned Dual-Mode Laser Diode for Compact Continuous-Wave THz Emitter,” ETRI J., vol. 33, no. 5, Oct. 2011, pp. 810–813.
[18] N. Kim et al., “Monolithically Integrated Optical Beat Sources toward a Single-Chip Broadband Terahertz Emitter,” Laser Physics Lett., vol. 10, no. 8, June 2013, p. 085805.
[19] H.-C. Ryu et al., “Simple and Cost-Effective Thickness
Measurement Terahertz System Based on a Compact 1.55 μm λ/4 Phase-Shifted Dual-Mode Laser,” Opt. Express, vol. 20, no. 23, Nov. 2012, pp. 25990–25999.
[20] J.-I. Hashimoto, Y. Nakano, and K. Tada, “Influence of Facet Reflection on the Performance of a DFB Laser Integrated with an Optical Amplifier/Modulator,” IEEE J. Quantum Electron., vol.
28, no. 3, Mar. 1992, pp. 594–603.
[21] N. Yoshimoto et al., “Highly Efficient Coupling Semiconductor Spot-Size Converter with an InP/InAlAs Multiple-Quantum-Well Core,” Appl. Opt., vol. 34, no. 6, Feb. 1995, pp. 1007–1014.
[22] H.-S. Kim et al., “1.55 μm Spot-Size Converter Integrated-Laser Diode Fabricated by Selective-Area Metalorganic Vapor-Phase Epitaxy,” Microw. Opt. Technol. Lett., vol. 25, no. 5, June 2000, pp. 300–302.
[23] J. Renaudier et al., “Phase Correlation between Longitudinal Modes in Semiconductor Self-Pulsating DBR Lasers,” IEEE Photon. Technol. Lett., vol. 17, no. 4, Apr. 2005, pp. 741–743.
[24] T. Okoshi, K. Kikuchi, and A. Nakayama, “Novel Method for High Resolution Measurement of Laser Output Spectrum,”
Electron. Lett., vol. 16, no. 16, July 1980, pp. 630–631.
[25] E.S. Lee et al., “Design and Characterization of Evanescently Coupled Photodiodes for 1.3 µm Wavelength,” Int. Conf. Infrared, Millimeter THz. Waves, Tucson, AZ, USA, Sept. 14–19, 2014, pp.
1–2.
[26] S.-P. Han et al., “Compact Fiber-Pigtailed InGaAs Photoconductive Antenna Module for Terahertz-Wave Generation and Detection,” Opt. Express, vol. 20, no. 16, July 2012, pp. 18432–18439.
[27] H.-J. Song et al., “Uni-Travelling-Carrier Photodiode Module Generating 300 GHz Power Greater Than 1 mW,” IEEE Microw.
Wireless Compon. Lett., vol. 22, no. 7, July 2012, pp. 363–365.
Eui Su Lee received his PhD in electronics and electrical engineering from the Korea Maritime University, Busan, Rep. of Korea in 2013. He joined the THz Photonics Creative Research Center of ETRI, Daejeon, Rep. of Korea in 2013. His interests include THz time-domain spectroscopy, the application of THz parallel- plate metal waveguides, and the fabrication of photodiodes for THz generation and Schottky-barrier diodes for THz detection.
Namje Kim received his PhD in physics from Chungnam National University, Daejeon, Rep.
of Korea in 2007. He has worked in the field of optoelectronic devices, particularly quantum dot optical amplifiers. After graduation, he joined the THz Photonics Creative Research Center of ETRI, Daejeon, Rep. of Korea. He is currently working on the development of a compact THz emitter.
Sang-Pil Han received his BS, MS, and his PhD degrees in optoelectronics engineering from the University of Seoul, Rep. of Korea in 1992, 1994, and 1998, respectively. He joined the Optical Interconnection Module Team at ETRI, Daejeon, Rep. of Korea, in 2000. He has been engaged in the development of compact photomixer modules for terahertz-wave generation and detection, fiber- coupled terahertz spectroscopy and imaging systems, and optical and terahertz devices for data communications.
Donghun Lee received his BS degree from Dong-A University, Busan, Rep. of Korea, in 1999, and his MS degree in physics from Seoul National University, Rep. of Korea in 2001. He joined the Basic Research Laboratory of ETRI, Daejeon, Rep. of Korea, where he is currently a senior researcher in the Photonic/Wireless Convergence Components Research Department. His current research interests include metalorganic chemical vapor phase epitaxy of III-V semiconductors and the development of semiconductor lasers.
Won-Hui Lee received his MS and PhD degrees in electronics and information communication engineering from Konkuk University, Seoul, Rep. of Korea, in 2000 and 2003, respectively. From 2002 to 2008, he was with the LG Electronics Digital Appliance Laboratory, Seoul. Rep. of Korea. From 2008 to 2009, his postdoctoral research was conducted at Pohang University of Science and Technology, Rep. of Korea. Since 2009, he has been with ETRI, Daejeon, Rep. of Korea. His current research interests include microwave, millimeter-wave, and terahertz system designs.
Kiwon Moon received his BS, MS, and his PhD degrees in electrical engineering from Pohang University of Science and Technology, Rep. of Korea, in 1998, 2000, and 2011, respectively. He worked for Samsung Electro- Mechanics, Suwon, Rep. of Korea, from 2000 to 2005, specializing in the design and epitaxial growth of visible laser diodes and blue light emitting diodes. He is currently working for ETRI, Daejeon, Rep. of Korea, where his main research interests are in spectroscopic terahertz near-field microscopes, real-time terahertz imaging using the asynchronous optical sampling technique, and terahertz devices including photo-conductive antennas and photomixers using plasmonic nano-structures.
Il-Min Lee received his BS and MS degrees in electrical engineering and computer science from Seoul National University, Rep. of Korea.
From 2000 to 2003, he worked as an engineer in the Optoelectronics Division, Telecommunication Network Business, of Samsung Electronics, Suwon, Rep. of Korea. In 2009, he received his PhD degree in electrical engineering and computer science from Seoul National University. Since then, he has been a postdoctoral researcher at the National Creative Research Center for Active Plasmonics Applications Systems in Seoul National University. From 2011 to 2013, he was a BK21 contract professor of Seoul National University. Since July 2013, he has been working in the Terahertz Basic Research Sector of ETRI, Daejeon, Rep. of Korea, as a senior researcher. His current interests include terahertz photonics related with terahertz communications, imaging, and spectroscopy, particularly using terahertz nano-photonics.
Jun-Hwan Shin received his BS degree in electrical engineering from Kyungpook National University, Daegu, Rep. of Korea, in 2009, and his PhD degrees in the Department of Advanced Device Technology from the University of Science and Technology, Daejeon, Rep. of Korea in 2016. His research interests include the fabrication and characterization of broadband THz emitters and detectors.
Kyung Hyun Park is the Director of the Terahertz Photonics Creative Research Center, ETRI, Daejeon, Rep. of Korea. In 1996, he received his PhD degrees in physics from Yonsei University, Seoul, Korea. From 1996 to 1997, he worked on large-scaled photonic integrated circuits at the Microelectronic Science Lab of Columbia University, NY, USA, as a postdoctoral research fellow. From 1990 to 1999, he was with the Korea Institute of Science and Technology, Seoul, Rep. of Korea, as a member of the research staff, where he was engaged in research on functional photonic devices. Since 1999, he has been at ETRI, continuing his research on photonic devices, and has been in charge of the WDM Photonic Devices Team and the Next-Generation Photonic Device Team. He has demonstrated various photonic devices for optical communications. His current focus is on the realization of THz technology using photonic devices. Dr. Park has published more 20 peer-reviewed scientific papers in the area of THz photonics, and was a recipient of the Minister Award at Nano Korea 2013, which was sponsored by the Ministry of Science, ICT, and Future Planning, for his innovative research contributions. He is also a committee member of the SPIE Photonics West and THz-Bio International Workshop.
