No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
High-speed VCSELs for short reach communication
Abstract
We present the design of a high-speed 850 nm multimode vertical cavity surface-emitting laser (VCSEL) and demonstrate record performance in terms of small signal modulation bandwidth (23 GHz) and error-free operation at high bit rates (40 Gb s−1). The large bandwidth was enabled by an active region design for large differential gain and small gain compression, a low reflectivity top mirror for photon lifetime reduction and multiple oxide layers for a reduction of the capacitance. Error-free operation at 40 Gb s−1 was achieved in a back-to-back configuration with less than 0 dBm of received optical power.
... In this paper, the effects of oxidation time, oxidation temperature and oxidation anisotropy on the oxidation rate are explored and demonstrated. The ratio of oxidation rate on [0][1][2][3][4][5][6][7][8][9][10][11] to [011] crystal orientation is defined as oxidation anisotropy coefficient, which decreases with the increase of oxidation temperature and oxidation time. In order to analyze the effect of the oxidation anisotropy on the VCSEL performance, an oxide-aperture of the VCSELs with two difference shapes is designed and then fabricated. ...
... Vertical cavity surface emitting lasers (VCSELs) have a large number of applications in optical communication, optical sensors, optical storage, and optical interconnection due to their low threshold current, low power consumption, good beam quality, small far field divergence angle, high speed modulation, and high density two-dimensional integration [1][2][3] . VCSELs were the core devices of 3D sensing technology of facial recognition in the iPhone X, which means that VCSELs have been extensively applied in the field of consumer electronics. ...
... Five sizes of three mesa shapes, including the side length of diamond, the side length of square on [011] (square transverse), the side length of square on [0][1][2][3][4][5][6][7][8][9][10][11] (square longitudinal), the diameter of circular on [011] (circular transverse) and the diameter of circular on [0][1][2][3][4][5][6][7][8][9][10][11] The oxidation rate of square transverse and circular transverse are generally higher than that of diamond, while the oxidation rate of square longitudinal and circular longitudinal are lower than that of the diamond. This phenomenon is further obvious with the increase of the oxidation time, which is mainly due to the bond energy difference between [0][1][2][3][4][5][6][7][8][9][10][11] and [011] being amplified. ...
Vertical cavity surface emitting lasers (VCSELs) are widely used in optical communications and optical interconnects due to their advantages of low threshold, low power consumption and so on. Wet nitrogen oxidation technology, which utilizes H2O molecules to oxidize the Al0.98Ga0.02As, is used for electrical and optical mode confinement. In this paper, the effects of oxidation time, oxidation temperature and oxidation anisotropy on the oxidation rate are explored and demonstrated. The ratio of oxidation rate on [0-11] to [011] crystal orientation is defined as oxidation anisotropy coefficient, which decreases with the increase of oxidation temperature and oxidation time. In order to analyze the effect of the oxidation anisotropy on the VCSEL performance, an oxide-aperture of the VCSELs with two difference shapes is designed and then fabricated. The static performance of these fabricated VCSELs has been measured, whose threshold current ratio ∼ 0.714 is a good agreement with that of the theoretical calculation value ∼ 0.785. Our research on wet nitrogen oxidation and its anisotropy serves as an important reference in the batch fabrication of large-area VCSELs.
... Several methods for reducing the modulation extrinsic limiting factors have been reported. These modulation extrinsic limiting factors include parasitic capacitances and resistances reduction [12,17,[18][19][20][21][22][23], single mode promotion [21,[23][24][25] and junction self-heating reduction [4,8,16,23,26,27]. Device reported by Lott et al [28], Chang et al [21], Johnson and Kuchta [29] and Larsson et al [22] required bias current densities of 19.8, 46.0, 21.2 and 20.0 kA/cm 2 , respectively, to exhibit modulation bandwidths of 16.0 , 22.0, 19.0 for the device reported by Haglund et al. [11] and 322% for the device reported by Tan et al. [12]; thus reducing devices reliability. ...
... These modulation extrinsic limiting factors include parasitic capacitances and resistances reduction [12,17,[18][19][20][21][22][23], single mode promotion [21,[23][24][25] and junction self-heating reduction [4,8,16,23,26,27]. Device reported by Lott et al [28], Chang et al [21], Johnson and Kuchta [29] and Larsson et al [22] required bias current densities of 19.8, 46.0, 21.2 and 20.0 kA/cm 2 , respectively, to exhibit modulation bandwidths of 16.0 , 22.0, 19.0 for the device reported by Haglund et al. [11] and 322% for the device reported by Tan et al. [12]; thus reducing devices reliability. and 23.0 GHz, respectively. ...
... Reliability tests on 850 nm VCSELs stressed under the same temperature demonstrated that 50% of the devices under test failed after around 900 and 350 hours of operation when biased at 7.5 and 9.0 mA, respectively [40]. VCSELs reported in the literature [21,22,28,29] required bias current densities of 19.8, 46.0, 21.2 and 20.0 kA/cm 2 to exhibit modulation bandwidths of 16,22,19, and 23 GHz, respectively. Since the failure rate of VCSELs is proportional to the bias current density squared [30], the VCSEL reliability is significantly reduced at such high bias current density. ...
An oxide-confined, dielectric-planarized, metal-plated, n-side up verticalcavity surface-emitting laser (VCSEL) with 850 nm emitting wavelength was fabricated and characterized. 30 μm mesa diameter VCSELs with 10 μm active area diameter and 4 μm thick Cu plated heat sink displayed a maximum frequency modulation bandwidth, f-3dB, of 16.4 GHz and a resonance frequency, fR, of 11.7 GHz at a low bias current density, Jb, of only 7.5 kA/cm², which corresponds to a high f-3dB 2/Jb ratio of 35.6 GHz2/kA/cm². The achieved 7.5 kA/cm² current density is less than the industrial benchmark current density for reliability, which is 10.0 kA/cm², by 25%. The reported devices demonstrated a modulation current efficiency factor of 9.7 GHz/mA1/2 and a D-factor of 6.5 GHz/mA1/2. Examination of the analogue modulation response (f-3dB/fR ∼1.41) demonstrated that the presented VCSELs did not suffer from high damping typically produced by self-heating and gain suppression. The measured thermal resistance was 1.01 °C/mW, which corresponds to an improved distributed Bragg reflectors thermal conductivity of 0.50 W/cmK and a reduced VCSEL thermal resistance by approximately 62% compared to previously reported oxide-confined polyimide-wrapped VCSELs with similar epitaxial structure and dimensions. Threshold current density and series resistance were as low as 1.4 kA/cm² and 60 Ω, respectively, with a maximum optical output power of 0.9 mW.
... Several methods for reducing the modulation extrinsic limiting factors have been reported. These modulation extrinsic limiting factors include parasitic capacitances and resistances reduction [12,17,[18][19][20][21][22][23], single mode promotion [21,[23][24][25] and junction self-heating reduction [4,8,16,23,26,27]. Device reported by Lott et al [28], Chang et al [21], Johnson and Kuchta [29] and Larsson et al [22] required bias current densities of 19.8, 46.0, 21.2 and 20.0 kA/cm 2 , respectively, to exhibit modulation bandwidths of 16.0 , 22.0, 19.0 for the device reported by Haglund et al. [11] and 322% for the device reported by Tan et al. [12]; thus reducing devices reliability. ...
... These modulation extrinsic limiting factors include parasitic capacitances and resistances reduction [12,17,[18][19][20][21][22][23], single mode promotion [21,[23][24][25] and junction self-heating reduction [4,8,16,23,26,27]. Device reported by Lott et al [28], Chang et al [21], Johnson and Kuchta [29] and Larsson et al [22] required bias current densities of 19.8, 46.0, 21.2 and 20.0 kA/cm 2 , respectively, to exhibit modulation bandwidths of 16.0 , 22.0, 19.0 for the device reported by Haglund et al. [11] and 322% for the device reported by Tan et al. [12]; thus reducing devices reliability. and 23.0 GHz, respectively. ...
... Reliability tests on 850 nm VCSELs stressed under the same temperature demonstrated that 50% of the devices under test failed after around 900 and 350 hours of operation when biased at 7.5 and 9.0 mA, respectively [40]. VCSELs reported in the literature [21,22,28,29] required bias current densities of 19.8, 46.0, 21.2 and 20.0 kA/cm 2 to exhibit modulation bandwidths of 16,22,19, and 23 GHz, respectively. Since the failure rate of VCSELs is proportional to the bias current density squared [30], the VCSEL reliability is significantly reduced at such high bias current density. ...
... Six layers are used for forming an oxide aperture (dark shaded region). Other details of the device design can be found in [20]. ...
... 1, is grown on undoped GaAs substrates. It employs an oxide-confined configuration optimized for high speed modulation [20]. The top and bottom DBRs are fabricated with graded interfaces and modulation doping to reduce their electrical resistance [13]. ...
... The bottom DBR is partly composed of AlAs to lower its thermal impedance [8]. The active region is made of five strained InGaAs quantum wells for improved differential gain [7] and is surrounded by a SCH designed for efficient carrier trapping and low gain-compression [9], [20]. As indicated with dark shading inFig. ...
We study the impact of device parameters, such as inner-aperture diameter and cavity photon lifetime, on ther-mal rollover mechanisms in 850-nm, oxide-confined, vertical-cavity surface-emitting lasers (VCSELs) designed for high-speed operation. We perform measurements on four different VCSELs of different designs and use our empirical thermal model for calculating the power dissipated with increasing bias currents through various physical processes such as absorption within the cavity, carrier thermalization, carrier leakage, spontaneous carrier recombination, and Joule heating. When reducing the top mirror reflectivity to reduce internal optical absorption loss we find an increase of power dissipation due to carrier leakage. There is therefore a trade-off between the powers dissipated owing to optical absorption and carrier leakage in the sense that overcompensating for optical absorption enhances carrier leakage (and vice versa). We further find that carrier leakage places the ultimate limit on the thermal performance for this entire class of devices. Our analysis yields useful design optimization strategies for mitigating the impact of carrier leakage and should thereby prove useful for the performance enhancement of 850-nm, high-speed, oxide-confined VCSELs.
... In order to provide carrier injection in VCSELs, a CPW access line is most commonly used [14], [15]. To avoid EM coupling with the highly doped bottom n-contact layer close to the active region, the n-layer is etched away under the pad line which constitutes an additional process step. ...
... A highly doped substrate also acts as a poor metal layer that shields the CPW line and increases its equivalent capacitor C p compared to NID substrate. Undoped substrates are preferred in recent high-speed VCSELs for all these reasons [14]. Finally, the performance of CPW access lines is strongly dependent on the substrate conductivity. ...
Vertical-cavity surface-emitting lasers with vertically integrated electro-absorption modulators (EAM-VCSELs) potentially can reach higher modulation bandwidths than directly current-modulated VCSELs. The aforementioned device modulation capabilities are, however, currently restricted by their electrical contact parasitics. It, thus, becomes critical to optimize their access line to improve performance. In this paper, we numerically and experimentally demonstrate that a microstrip (MS) access line using a planarized benzocyclobutene (BCB) layer exhibits improved high-frequency characteristics compared to the coplanar waveguide (CPW) access line used to-date, since it reduces the losses induced by the doped substrates. We also use this opportunity to introduce an innovative technique for BCB layer planarization, which is not only compatible with the EAM-VCSEL double-mesa structure, but is also independent of the device pitch. The resulting BCB layer dielectric permittivity and losses are measured up to 100 GHz, and a side-by-side comparison of the electrical response of three-electrodes MS and CPW access lines is subsequently carried out. Finally, using the measured pad and access line RF responses and the EAM modulation characteristics, the devices with MS access are shown to be no longer limited by their electrode parasitics, but by the modulator internal impedance.
... Vertical-cavity surface-emitting lasers (VCSELs) have been developed for over forty years and have attracted considerable attention due to many advantageous features, such as the light beam emitting from the top surface, circular and low-divergence beam, low threshold operation, compact size, and high output power brought by array of lasers. To date, VCSELs are considered as established light sources to be widely used in numerous applications, e.g., laser printing [1], laser display [2], optical communication [3,4], optical clock [5], gas sensor [6,7], three-dimensional (3D) sensing, and light detection and ranging (LiDAR) application [8][9][10], which operate in red to mid-infrared (MIR) wavelength bands. To have even shorterwavelength devices, GaN-based VCSELs have quickly evolved in recent years [11][12][13][14]. ...
In this paper, a pulsed electrically pumped GaN-based vertical-cavity surface-emitting laser (VCSEL) with one dielectric distributed Bragg reflector and one n-GaN monolithic high-index contrast grating (MHCG) mirror was demonstrated at room temperature. The reflectance of the n-GaN MHCG and cavity mode behaviors of the VCSEL with MHCG for varying n-GaN thickness, MHCG pattern diameter, and current aperture size were numerically investigated. Measured characteristics of the fabricated device showed that the lasing action started at an injection current of 10.2 mA, corresponding to a current density of about 15.1 kA/ cm 2 . Above threshold, the measured slope efficiency was 6.2 × 10 − 3 W / A , and the output power was 0.13 mW at 30 mA. Moreover, the measured lasing peak occurring at 403.4 nm and the longitudinal mode spacing of 5.6 nm were in good agreement with simulations. The incorporation of an n-GaN MHCG mirror not only greatly simplified the fabrication but also substantially improved the lasing characteristics in comparison to the previous work applying TiO 2 HCG mirrors.
... And the larger D-factor will be along with smaller threshold current. Furthermore, the VCSELs with smaller oxide aperture diameters (5 μm) and shorter cavity length (λ/2) are especially well-suited for data transmission at low energy per bit [30][31][32]. We expect the VCSEL can achieve error-free operation rate up to 50 Gb/s. ...
We have studied the characteristics of frequency response at 850-nm GaAs high-speed vertical-cavity surface-emitting lasers (VCSELs) with different kinds of oxide aperture sizes and cavity length using the PICS3D simulation program. Using 5-μm oxide aperture sizes, the frequency response behavior can be improved from 18.4 GHz and 15.5 GHz to 21.2 GHz and 19 GHz in a maximum of 3 dB at 25 °C and 85 °C, respectively. Numerical simulation results also suggest that the frequency response performances improved from 21.2 GHz and 19 GHz to 30.5 GHz and 24.5 GHz in a maximum of 3 dB at 25 °C and 85 °C due to the reduction of cavity length from 3λ/2 to λ/2. Consequently, the high-speed VCSEL devices were fabricated on a modified structure and exhibited 50-Gb/s data rate at 85 °C.
... The achieved values for both energy-to-data rate ratio (EDR) and HBR are the lowest among all reported high-speed VCSELs at 40Gbit/sec operation. 5,6 The device remains error-free at 33Gbit/sec under operating temperatures of 85 ı C; and a bias current of 2.8mA can be achieved: see Figure 2(b). Figure 3 shows the BER, measured at 25Gbit/sec operation, versus the optical power and injected bias current of our device, respectively. ...
This paper presents a method to evaluate the impact of temperature characteristics on vertical cavity surface emitting laser (VCSEL) module. As one of the core modules in the optical communication system, the performance of VCSEL strongly influences the communication quality of the high-speed optical communication system. However, it is difficult to directly analyze the temperature change of VCSEL. In order to solve this problem, batches of laser sources have been integrated into the optical communication module, the physical properties of the laser beams then can be easily measured at different temperatures (low temperature −5 °C, room temperature 25 °C and high temperature 70 °C). By analyzing the wavelength, ext. ratio and the margin of eye diagram of these laser beams, we calculate the percentage value which referrers to an engineering experience standard value as the evaluator, to describe the quality of the optical communication system. The performance of communication quality is evaluated under different parameters, including amplitude, emphasis, mode and bias etc. Several tests have been preceded which all obtained the satisfactory results.
The simulation can play an important role in all phases of the development of communication systems, from the earliest step of design, to the final steps of realization, testing and implementation of the system. Our thesis has been achieved through the application of different modeling techniques possible with VHDL-AMS for the creation and improvement of the practices and reusable models of the optoelectronic components, such as the VERTICAL CAVITY SURFACE EMITTING LASER (VCSEL), optical fibers and photodiode PIN. We have developed a library of the optoelectronic models well documented and configurable for they are be use in simulation of three blocks of optical transmission line. In the simulation of each block, we introduced the main disturbances of the optical chain, namely the temperature, the intrinsic noise of the VCSEL, the effect of attenuation in the optical fiber and the influence of chromatic and modal dispersion. We do not pretend to replace the specialized software in the field of optics and photonics, but our model is designed for designers and developers of optical transmission systems.
Keyword : VHDL-AMS; VCSEL; Graded inex optical fiber; Step index optical fiber; single mode optical fiber, Top-Down, Photodiode PIN, Test-Bench.
The main problems of providing a high-speed operation semiconductor lasers with a vertical microcavity (so-called “vertical-cavity surface-emitting lasers”) under amplitude modulation and ways to solve them have been considered. The influence of the internal properties of the radiating active region and the electrical parasitic elements of the equivalent circuit of lasers are discussed. An overview of approaches that lead to an increase of the cutoff parasitic frequency, an increase of the differential gain of the active region, the possibility of the management of mode emission composition and the lifetime of photons in the optical microcavities, and reduction of the influence of thermal effects have been presented. The achieved level of modulation bandwidth of ∼30 GHz is close to the maximum achievable for the classical scheme of the direct-current modulation, which makes it necessary to use a multilevel modulation format to further increase the information capacity of optical channels constructed on the basis of vertical-cavity surface-emitting lasers.
850-nm vertical-cavity laser diodes with copper-plated heat sinks were fabricated and tested. Vertical-cavity surface-emitting lasers with 10μm aperture diameter and 4μm of electroplated copper demonstrated a maximum −3dB modulation bandwidth of more than 16GHz and a resonance frequency of more than 11GHz at bias current of only 5.4mA, which correspond to a bias current density of only 7.5kA/cm 2 . Thus, reported devices exhibited a high −3dB modulation bandwidth squared over bias current density ratio of more than 35GHz 2 /kA/cm 2 . The displayed 7.5 kA/cm 2 bias current density is less than the 10 kA/cm 2 current density standard for reliability by 25%. Devices also demonstrate a reduced thermal resistance of 1.01 °C/mW, a threshold current of 1.1mA, a series resistance of only 60Ω, and a maximum output power density of about 1.2mW/cm 2 .
Intensity modulation with direct detection (IM-DD) dominates the commercial short-reach optical communications. However, when upgrading the data-rate distance product to 1000 Gb/s·km per wavelength and beyond, IM-DD faces severe performance barrier. Although a linear mapping exists between the intensity of transmitter and receiver in IM-DD back-to-back system, it becomes nonlinear under fiber channel impairments; for instance, the dominant chromatic dispersion. Coherent detection overcomes this fundamental obstacle by recovering a linear replica of the optical field instead of intensity. The local oscillator provides a reference carrier which mixes with the received signal, from which both intensity and phase of the signal can be recovered. Borrowing the idea from coherent detection, in DD system, a carrier can be sent along with the signal so that the receiver utilizes the reference carrier originated from transmitter; namely, the receiver conducts self-coherent (SCOH) detection. In this paper, we review a variety of optical SCOH subsystems, and reveal how they realize linear channels, while maintaining the simple and low-cost DD. SCOH can readily reach a data-rate distance product beyond 10000 Gb/s·km, making it suitable for the future high-speed short- and medium-reach applications, such as the data center interconnect, passive access network, and metropolitan area network.
Due to the chirp effect, the modulation information is carried by both intensity and phase of the optical signal generated by a directly modulated laser (DML). Compared to the signal recovery based on intensity alone, the receiver sensitivity is significantly improved when phase is also involved for signal recovery. In this paper, we provide thorough analysis of two schemes to recover the directed modulated signal based on the combined information of intensity and phase. The intensity and phase combined recovery for polarization-division-multiplexed (PDM) directly modulated signal is demonstrated for the first time, achieving a data rate of 40 Gb/s over a transmission distance of 160 km. Compared with the conventional intensity-only recovery, the receiver sensitivity is improved by 5.5 dB and 10 dB respectively by using two intensity and phase combined recovery schemes.
Vertical Cavity Surface Emitting Laser (VCSEL)-based data links are attractive due to their low-power dissipation and low-cost manufacturability. This chapter reviews the foundations for this technology, as well as the device and module design challenges of extending the data rate beyond the current level. We begin with a review of data communications from the business perspective, and continue with a brief discussion of the current and future standards. This is followed by a survey of recent advances in VCSELs, including data links operating at 28 Gb/s. We review the recent efforts on ultra-fast data links and discuss the advantages of the different approaches. We also examine key design aspects of optical transceiver modules and we focus our discussion on novel applications in high-performance computing using both multi-mode and single-mode fiber optics. We highlight the importance of the device/component-level and system-level modeling and show some modeling examples with comparison to measured data. We conclude with a comparison of the VCSEL-based data links with other competing technologies, including silicon photonics and short-cavity edge emitting lasers.
Inverted-polarity (n-up), high-speed, oxide-confined, polyimide-planarized, copper-plated 850-nm vertical-cavity surface-emitting lasers with various aperture sizes were fabricated and characterized. The reported devices demonstrated intrinsic, parasitic, and thermal maximum bandwidth limitations of 39.3, 24.6, and 22.9 GHz, respectively. VCSELs with 7 μm active area diameter and 4 μm of plated copper exhibited a maximum -3 dB frequency modulation bandwidth (f-3 dB max) of 18.8 GHz and a resonance frequency of 14.8 GHz at a bias current density (Jbias) of only 8.8 kA/cm2, limited by thermal effects, with a high modulation current efficiency factor of 17.0 GHz/mA1/2 for quantum well VCSELs. The presented VCSELs also demonstrated a record high f-3 dB max2/Jbias ratio of 40.2 GHz2/kA/cm2 which represents an 11% increase compared with the highest previously reported ratio. Rate-equation-based thermal VCSEL model and three-pole transfer function approximation were applied to extract several VCSELs' parameters, which enabled the estimation of the VCSELs internal temperature and major bandwidth limitation. Particle swarm optimization was utilized in parameter extraction and optimization.
Directly modulated laser (DML) with direct detection (DD) dominates the commercial optical short reach applications. When upgrading the DML-DD to higher data rate, the frequency chirp is commonly regarded as a performance barrier. On the contrary, in this letter, we take advantage of the chirp to generate phase modulation along with the intensity, namely, realize complex modulation at transmitter using a single DML. Coherent detection (CD) is applied to recover this 2 dimensional (2-D) signal. A modified Viterbi algorithm (VA) conducts maximum likelihood decoding utilizing the signal intensity and differential phase in receiver digital signal processing (DSP). By making decision using 2-D information, VA provides more than 9-dB optical signal to noise ratio (OSNR) sensitivity advantage over the conventional DML-CD with intensity only decision, in a 10-GBaud pulse-amplitude modulation (PAM)-4 system. We demonstrate the first PAM-8 experiment over 160-km standard single mode fiber using the phase diversity DML.
This paper investigates methods of suppressing higher order transverse modes in a GaInNAs/GaAs QW GaAs-based VCSEL, with the aid of a three-dimensional self-consistent model. The inverted surface relief design, which creates the anti-phase condition, proved efficient at suppressing higher order transverse modes, in spite of unfavorable thermal effects.
Oxide-relief and Zn-diffusion structures are realized in 850 nm VCSELs. Both record-low energy-to-data-rate ratio (96 fJ/bit at 25 Gbit/sec; 135 fJ/bit at 37 Gbit/sec) and energy-to-data distance ratio 175 fJ/bit.km at 25 Gbit/sec are demonstrated.
This paper presents an overview of our recent work on high speed, oxide confined, 850 nm vertical cavity surface emitting lasers (VCSELs). With proper active region and cavity designs, and techniques for reducing capacitance and thermal impedance, we have reached a modulation bandwidth of 23 GHz and demonstrated 40 Gbps transmission. Using an integrated mode filter for reducing the spectral width we have extended the reach on multimode fiber at 25 Gbps from 100 to 500 m. Improved link capacity was also demonstrated using a more spectrally efficient multi-level modulation format (4-PAM). Finally, a MEMS-technology for wafer scale integration of tunable high speed VCSELs was developed, enabling a tuning range of 24 nm, a 6 GHz modulation bandwidth, and 5 Gbps transmission.
We demonstrate novel structures of a vertical-cavity surface-emitting laser (VCSEL) for high-speed (~40 Gbit/s) operation with ultralow power consumption performance. Downscaling the size of oxide aperture of VCSELs is one of the most effective ways to reduce the power consumption during high-speed operation. However, such miniaturized oxide apertures (~2 μm diameter) in VCSELs will result in a large differential resistance, optical single-mode output, and a small maximum output power (<; 1 mW). These characteristics seriously limit the maximum electrical-to-optical (E-O) bandwidth and device reliability. By the use of the oxide-relief and Zn-diffusion techniques in our demonstrated 850-nm VCSELs, we can not only release the burden imposed on downscaling the current-confined aperture for high speed with low-power consumption performance, but can also manipulate the number of optical modes inside the cavity to maximize the E-O bandwidth and product of bit-rate transmission distance in an OM4 fiber. State-of-the-art dynamic performances at both room temperature and 85 °C operations can be achieved by the use of our device. These include extremely high D-factors (~13.5 GHz/mA1/2), as well as record-low energy-to-data ratios (EDR: 140 fJ/bit) at 34 Gbit/s operation, and error-free transmission over a 0.8-km OM4 multimode fiber with a record-low energy-to-data distance ratio (EDDR: 175.5 fJ/bit.km) at 25 Gbit/s.
In the current work, we show a detailed analysis of the transverse beam
profile and polarization characteristics of devices with one and three
oxide apertures. A Gaussian transverse beam profile is achieved with an
oxide aperture diameter of less than 6 μm. The laser light is
linearly polarized with a high degree of polarization (> 97 %) in the
complete current range. The stable polarization direction can be
attributed to ordering effects ocurring during epitaxial growth of the
GaInP material system and a reduction in crystal symmetry. Different
oxide aperture diameters can be implemented in one device due to high
oxidation selectivity of the AlxGa1-xAs layer depending on aluminum
content. These deep oxidation layers lead to a reduction of the
parasitic capacitance, while beam profile and polarization
characteristics are not affected.
We present an empirical thermal model for VCSELs based on extraction of
temperature dependence of macroscopic VCSEL parameters from CW
measurements. We apply our model to two, oxide-confined, 850-nm VCSELs,
fabricated with a 9-μm inner-aperture diameter and optimized for
high-speed operation. We demonstrate that for both these devices, the
power dissipation due to linear heat sources dominates the total
self-heating. We further show that reducing photon lifetime down to 2 ps
drastically reduces absorption heating and improves device static
performance by delaying the onset of thermal rollover. The new thermal
model can identify the mechanisms limiting the thermal performance and
help in formulating the design strategies to ameliorate them.
This paper presents a review of recent work on high speed tunable and
fixed wavelength vertical cavity surface emitting lasers (VCSELs) at
Chalmers University of Technology. All VCSELs are GaAs-based, employ an
oxide aperture for current and/or optical confinement, and emit around
850 nm. With proper active region and cavity designs, and techniques for
reducing capacitance and thermal impedance, our fixed wavelength VCSELs
have reached a modulation bandwidth of 23 GHz, which has enabled
error-free 40 Gbps back-to-back transmission and 35 Gbps transmission
over 100 m of multimode fiber. A MEMS-technology for wafer scale
integration of tunable high speed VCSELs has also been developed. A
tuning range of 24 nm and a modulation bandwidth of 6 GHz have been
achieved, enabling error-free back-to-back transmission at 5 Gbps.
Oxide-free lithographic vertical-cavity surface-emitting lasers (VCSELs) are characterized for their thermal properties. VCSELs with minimum threshold temperatures < 20 °C demonstrate lasing at heat sink temperatures ≥ 190 °C, with high rollover cavity temperatures.
Existing problems in the development of vertical-cavity surface-emitting lasers with photonic crystals (PC-VCSELs) for transverse mode control are reviewed. The theoretical background of VCSELs with PCs incorporated within their top distributed Bragg reflector (DBR) (type I PC-VCSELs) and VCSELs with PCs incorporated within their cavity (type II PC-VCSELs) is introduced. Their fabrication using focused ion-beam (FIB) etching is discussed. It is shown that type I single-mode PC-VCSELs suffer from low optical power and poor small-signal modulation performance due to the reduction of the reflectivity of the top DBR introduced by the PC. The problem is expected to be reduced in type II PC-VCSELs that are being investigated now. One approach to the fabrication of type II PC-VCSELs is described.
Microcavity lasers (μ CLs ) , reduced-size ( <sup>></sup><sub>~</sub>3 μ m aperture) vertical cavity surface-emitting lasers (VCSELs) defined by the buried-oxide process for current and field confinement (thus wide mode spacing), are demonstrated with low threshold current, sharp turn-on L-I characteristics, and wide bandwidth operation. Due to the enhanced spontaneous recombination rate at reduced mode and improved photon density, μ CLs exhibit lower charge-field resonance peaks at a modulation bandwidth f<sub>-3 dB </sub>=18.7 GHz , thus permitting open-“eye” operation at 20 and 40 Gb/s data rates ( I <sup>></sup><sub>~</sub>3 mA ) . The energy efficiency for 20 Gb/s data transmission is measured to be 4.84 Gb/s/mW, which is eight times better than 7 μ m aperture VCSELs.
The vertical-cavity surface-emitting laser (VCSEL) has become a light source of great importance for industrial and consumer applications. This includes communication and sensing in particular, where dynamics and optical mode behavior are key performance characteristics. This tutorial treats relevant VCSEL basics, performance requirements and recent progress toward higher speed, higher single-mode power, and polarization control.
We use an empirical model together with experimental measurements for studying mechanisms contributing to thermal rollover in vertical-cavity surface-emitting lasers (VCSELs). The model is based on extraction of the temperature dependence of threshold current, internal quantum efficiency, internal optical loss, series resistance and thermal impedance from measurements of output power, voltage and lasing wavelength as a function of bias current over an ambient temperature range of 15-100 °C. We apply the model to an oxide-confined, 850-nm VCSEL, fabricated with a 9-μm inner-aperture diameter and optimized for high-speed operation, and show for this specific device that power dissipation due to linear power dissipation (sum total of optical absorption, carrier thermalization, carrier leakage and spontaneous carrier recombination) exceeds power dissipation across the series resistance (quadratic power dissipation) at any ambient temperature and bias current. We further show that the dominant contributors to self-heating for this particular VCSEL are quadratic power dissipation, internal optical loss, and carrier leakage. A rapid reduction of the internal quantum efficiency at high bias currents (resulting in high temperatures) is identified as being the major cause of thermal rollover. Our method is applicable to any VCSEL and is useful for identifying the mechanisms limiting the thermal performance of the device and to formulate design strategies to ameliorate them.
Higher speed short-wavelength (850 nm) VCSELs are required for future high-capacity, short-reach data communication links. The modulation bandwidth of such devices is intrinsically limited by the differential gain of the quantum wells (QWs) used in the active region. We present gain calculations using an 8-band k·p Hamiltonian which show that the incorporation of 10% In in an InGaAs/AlGaAs QW structure can approximately double the differential gain compared to a GaAs/AlGaAs QW structure, with little additional improvement achieved by further increasing the In composition in the QW. This improvement is confirmed by extracting the differential gain value from measurements of the modulation response of VCSELs with optimized InGaAs/AlGaAs QW and conventional GaAs/AlGaAs QW active regions. Excellent agreement is obtained between the theoretically and experimentally determined values of the differential gain, confirming the benefits of strained InGaAs QW structures for high-speed 850-nm VCSEL applications.
The influeuce of mirror reflectivity on laser performance of InGaAs-GaAs vertical-cavity surface-emitting lasers fabricated by selective oxidation is investigated by the stepwise change of the number of pairs in top mirror stack after device fabrication. Devices with 18-pair stacks in the top mirror, which is the optimized number of pairs in this structure, show an output power over 1.9 mW and a slope efficiency of 55% while maintaining a low threshold current of 212 /spl mu/A. The analysis of the threshold current and differential efficiency related to mirror reflectivity shows an internal quantum efficiency of 95%, an internal round-trip loss of 0.072, and a transparency current density of 71 A/cm/sup 2/.< >
We explored in detail the dynamics of multitransverse-mode
vertical-cavity surface-emitting lasers (VCSELs), which are important
for optical interconnection applications. We found that the small-signal
modulation response exhibits a rich structure including multiresonance
peaks as well as response notches. The detailed theoretical and
experimental study enabled us to attribute the various features to the
coupling level between the participating modes. The generality of the
predictions was tested and approved for several types of
multitransverse-mode VCSELs and even for a coherent VCSEL array
Oxide-confined 850 nm vertical-cavity surface-emitting lasers operating at 40 Gbit/s at current densities ~10 kA/cm<sup>2</sup> are realised. The deconvoluted rise time of the device is below 10 ps and remains hardly temperature sensitive up to 100degC.
We report on the design, fabrication, and evaluation of large-aperture, oxide-confined 850 nm vertical cavity surface emitting lasers (VCSELs) with high modulation bandwidth at low current densities. We also compare the use of InGaAs and GaAs quantum wells (QWs) in the active region. Both VCSELs reach an output power of 9 mW at room temperature, with a thermal resistance of 1.9deg C/mW. The use of InGaAs QWs improves the high-speed performance and enables a small-signal modulation bandwidth of 20 GHz at 25degC and 15 GHz at 85degC. At a constant bias current density of only 11 kA/cm<sup>2</sup>, we generate open eyes under large-signal modulation at bit rates up to 25 Gbit/s at 85degC and 30 Gbit/s at 55degC.
The dependence of the performance of high-speed 850 nm vertical cavity surface-emitting lasers (VCSELs) on photon lifetime is investigated. The photon lifetime is controlled by shallow surface etching. It is demonstrated that a reduction of photon lifetime by ~50% leads to a significant improvement of efficiency and speed. A record high 3 dB bandwidth of 23 GHz is achieved.
We have developed InGaAs-based VCSELs operating around 1.1µm for
high-speed optical interconnections. By applying GaAsP barrier layers,
temperature characteristics were considerably improved compared to GaAs
barrier layers. As a result, 25Gbps 100°C error-free operation was
achieved. These devices also exhibited high reliability. No degradation
was observed over 3, 000 hours under operation temperature of 150°C
and current density of 19kA/cm2. We also developed VCSELs
with tunnel junctions for higher speed operation. High modulation
bandwidth of 24GHz and a relaxation oscillation frequency of 27GHz were
achieved. 40Gbps error-free operation was also demonstrated.
High-efficiency, high-speed, tapered-oxide-apertured 980 nm vertical-cavity surface-emitting lasers (VCSELs) with 35 Gbit/s error-free operation have been demonstrated. This is the highest data rate reported for directly modulated VCSELs to date. The devices are also highly efficient, showing a record-high data-rate/power-dissipation ratio of 3.5 Gps/mW.
Strained-layer In<sub>0.3</sub>Ga<sub>0.7</sub>As/GaAs multiquantum well lasers fabricated by chemically assisted ion beam etching have achieved a record 3 dB modulation bandwidth of 28 GHz. The low frequency rolloff and nonlinear gain coefficient have been improved substantially by employing a separate confinement heterostructure design that has a fast carrier capture time.
Relative intensity noise (RIN) of selective oxidized multimode VCSELs has been investigated. The VCSELs show an excellent noise performance with a RIN⩽-140 dB/Hz at 500 MHz and only one resonance peak in the RIN spectrum, as long as the total light power is detected. The situation deteriorates if the transmission link contains mode-selective losses. Introducing external filtering of the emitted light by an aperture or a polarizer, one may observe an increase of RIN by several orders of magnitude and multiple-resonance peaks in the RIN spectrum. However, even in spite of a very restrictive external filtering a RIN of -120 dB/Hz can still be achieved, which enables a 1.25 Gb/s transmission. The multiple-resonance peaks in the filtered RIN spectra do not describe the different relaxation resonance frequencies of transverse modes. We believe there is only one relaxation resonance frequency, which is proportional to the square root of the photon number in the active layer, irrespective of the number of existing modes. The other peaks in the RIN spectrum ran be considered as “mode partition frequencies,” which result from the carrier interchanges between the modes. Using a simplified numerical model, which takes two modes into account, analytical expressions describing the RIN of each mode, as well as of the total power have been derived. It has been found that the larger the overlap between the modes, the smaller the “mode partition frequency” and the larger is the maximum mode partition noise (MPN)
We present a quasi-3-D dynamic model of vertical-cavity
surface-emitting lasers (VCSELs). The interdependent processes of
carrier transport, heat generation and dissipation, and optical fields
are solved self-consistently for each point in time and space. An
effective index model is adopted for the evaluation of the optical
fields in the complex layer structure. The inclusion of a temperature-
and carrier-density-dependent refractive index, and its time dependence,
allows us to study the evolution of the transverse optical field
distributions under dynamic conditions. The model is applied to a
typical index-guided structure with a 7-μm oxide aperture. A direct
comparison is made using "cold" cavity modes, which is a normal
technique when modeling the dynamics of VCSELs. Significant
discrepancies are demonstrated both at smalland large-signal modulation,
which indicates the need of a more sophisticated model for accurately
predicting and understanding the geometry-dependent modal
evolution