CHAPTER 1. Introduction
1.3 Advantages of VCSELs
There are many reasons why VCSELs are becoming increasing popular as light source for applications such as datacomm and optical interconnects. The monolithically integrated structure requires one single epitaxial run, making it easier to fabricate. Since the mirrors are formed during the epitaxial growth, each individual VCSEL can be tested already on the wafer, before it is cleaved into separate chips, thereby drastically reduce the production cost.
The use of DBRs eliminates the risk of catastrophic optical damage (COD) in the mirrors which can occur in edge-emitters where the active material close to the facets are depleted by surface recombination and thereby light absorbing. It also reduces the risk of mechanical mirror damage. The extremely short resonator leads to a large longitudinal mode spacing that is large compared with the gain bandwidth and leads to inherent single longitudinal mode operation. The small active volume and high mirror reflectivity contribute to the very low threshold observed in VCSELs, as low as a few microamperes[4], resulting in low power
consumption and reduced heating of the device. This feature, combined with the absence of CODs, explains the remarkable reliability of VCSELs. Lifetimes of more than 10000 hours have been reported by several groups. [5-6]
The surface emission and the small size make it possible to fabricate very dense two-dimensional arrays of VCSELs, suitable for multi-channels parallel transmission modules.
[7] VCSELs do not need to be cleaved; it is therefore possible to integrate them monolithically with other optoelectronic components such as photodetectors, modulator or hetero-bipolar transistors (HBT). [8] Because of the circular symmetry of the VCSEL structure, the light is emitted with a circular beam and very low divergence. This results in high coupling into optical fibers, up to 90% [9] and allows for relaxed tolerance in alignment, further reducing the cost of installation. For comparison, the output light emitted from an edge-emitting laser is elliptical with a transverse and lateral divergence of about 40 and 10 degrees, respectively, making it cumbersome to couple the light into an optical fiber without significant optical loss or advanced optics.
In addition, VCSELs have inherent single-wavelength structure that is well suited for wavelength engineering, making it possible to process multi-wavelength array or tuneable VCSELs. Although the manufacturing challenges are numerous, both types of devices have been demonstrated. By carefully designing the optical cavity, with the implemenatation of a small thickness variation in the bottom DBR, a record 150-wavelength VCSEL array has been reported. The thickness gradient creating a cavity thickness variation, which in turn led to laser wavelength variation, the overall wavelength span across the array being 43 nm. [10]
1.4 Drawbacks of VCSELs
However, VCSELS also have some drawbacks compared to edge-emitters. The manufacturing tolerances on VCSEL growth are much tighter than for edge-emitting lasers, the layer thickness having to be controlled within 1%. The major disadvantage with VCSELs is the strong tendency to operate on multiple transverse modes, due to the large transverse dimensions of the optical cavity. These results in emission spectra with multiple emission wavelengths, which limits the maximum achievable distance due to chromatic dispersion effects. Most commercial VCSEL of today operate multimode and are mainly used in short distance multimode fiber based optical data links[11], optical interconnects[12], optical storage[13] and laser printing[14]. A lot of efforts are made to produce high power single mode VCSELs. This include oxide confined VCSELs with current aperture small enough to support only the fundamental mode, index-guided structures such as regrown or surface relief
VCSELs, and spatial mode filtering in an external cavity or extended cavity. Although the first and last of these techniques have produced high single mode power they are difficult to implement with high uniformity and yield. A more reliable technique is to combine a large area oxidation with an etched shallow surface relief for mode selection. This implies only a small modification to the fabrication procedure but produces reasonably high single mode power, with high uniformity and yield. [15]
1.5 Applications of VCSELs
1.5.1 Data communication
Today, datacom moduels based on near-infrared VCSELs represent 95% of the VCSEL market, 80% of which are commercialized by a few companies: Agilent, Honeywell, Infineon, Furukawa and Zarlink. The remaining 20% are shared among numerous start-ups offering innovative designs. The market has recently exploded: it is evaluated to be worth USD 500 million at present, and is constantly growing due to the rapid deployment of Gigabit Ethernet and fiber channels. Most of today’s commercial datacom components are based on oxide 850 nm VCSELS. They are often package as single component or in parallel fiber modules of linear arrays, offering 4 to 12 channels at 2.6 GB/s per channel, which aggregate bandwidth up to 30 GB/s. The research and
development efforts are focusing on the next generation of high speed VCSEL, and a number of groups have reported transmission at 10 GB/s or more for distances up to 300 m of MMF. Figure 1-4 shows a 12 channel 2.5 Gb/s VCSEL array for short distance transmission applications.
Fig. 1-4 Commercial 2.5 Gb/s VCSEL array for data communication application.
For long wavelength part, Metro and Access Networks are dominated by 1300nm
and 1550 nm FP and DFB lsers up to now. A long wavelength VCSEL (LW-VCSEL) would be an ideal low-cost alternative to the DFB laser, particularly for the standard IEEE 802.3ae applications, which extend the existing Gigabit Ethernet into traditional SONET markets at OC-192 data rates.[16]. However, the performance specifications for such LW-VCSELs are challenging. For low-cost transceivers, they must operate over the 0 to 70 C temperature range for indoor applications and over the -40 to 85 C range for outdoor applications, without external temperature stabilization. The laser power launched into the single mode fiber must usually be more than 0.7 mW in order to support transmission distance of 10 km at 10 GB/s. Despite intense research effort, the technology so far has not yet met these requirements
1.5.2 Optical interconnect
The optical interconnect is considered by many to be inevitable in the computer technology. The performance of massively parallel computers is usually limited by the communication bottleneck between processors. Optic provides an effective mean to line these processors because of its high capacity, low crosstalk and attenuation, and the possibility to obtain three-dimensional architectures. Other potential applications include routers, switches and storage. The VCSEL is a strong candidate as the preferred optical light source for the emerging optical interconnect mass market, meeting the requirement of low cost, high density integration and low power dissipation. A 256-channel bi-directional optical interconnect using VCSELs and photodiodes on CMOS was demonstrated.[17]
1.5.3 Sensor
Reflective optical sensors are used to sense the presence or absence of a distant object.
Examples of reflective sensors used in a variety of industrial and consumer products include barcode scanners and proximity sensors. The packaging of optical reflective sensors can be quite compact, and in the case of some LED sources, can even be packaged in a single TO can.
However, a significant disadvantage to these devices is the quantity of optical crosstalk that may degrade the signal-to-noise ratio (S/N) in the detector. Crosstalk results from the fact that LEDs emit from all surfaces and the emission subtends nearly 90°. Suppliers go to great lengths to isolate the LED and the detector by using a mechanical structure to separate the optoelectronic components. In addition, the LED optical output is not easily collimated or focused to a spot to increase the amount of reflected light from a distant object. By using the technical features of the VCSEL, integrating a phototransistor in the package, and designing
the optical element into the TO can lid, an effective reflective sensor can be developed. The advantages of the sensor include the ability to package the entire assembly in a single compact TO can, along with the focusing optics and a phototransistor. Depending on the application, a single-mode or multimode VCSEL can be used. In some cases, when coherence of the optical beam is desired, the single-mode VCSEL might be the best choice, but in other cases when total output power is more important, a multimode VCSEL might be more beneficial. For example, a multimode VCSEL can be mounted on the centerline of the lens and package, and the phototransistor mounted to the side of the VCSEL. In this configuration, the optimal signal is obtained by tilting the package with respect to the centerline of the TO.
The optical system is made by including a melt-formed glass lens in the TO lid. The lid can be designed to accept other lenses, and the height can be varied, which allows for the design of a wide variety of optical sensors. In addition to the reduced power consumption and single-package interface, the recurring theme in the application is the ability of the VCSEL sensor to provide higher S/N in environments where the LED sensor is not able to adequately perform. Other application areas include the sensing of diffuse reflective surfaces such as paper in a printing system, or low-reflectivity surfaces such as glasses or plastics. The small focal spot also has significant advantages in optical encoding applications such as barcode
reading or positioning equipment. Figure 1-5 illustrate the VCSEL sensor module used in laser mouse application. The sensitivity and resolution of the laser mouse is 20 times higher than the conventional LED mouse. In 2004, the VCSEL based biosensor also demonstrated by C. J. Chang-Hasnain et al. [18-19]
VCSEL
Fig. 1-5 A laser mouse module using VCSEL.
References
[1] Iga and Y. Suematsu, “GaInAs/InP surface emitting injection lasers”, Jpn. J.
[2] and T. Nishimura, “Room temperature pulsed oscillation
[3] ing
[4] , M. H. MacDougal and P. D. Dapkus, “Ultralow threshold current
[5] Penner, R. H. Johnson, R. A. Hawthorne, J. R.
[6]
[7] . Chandramani,
[8] ithic
[9] , Y. H. Hang, B. Weir, J. Gamelin and A. Y.
[10] anical
[11] edler, G. Reiner, P. Schnitzer and K. J. Ebeling, “Top surface-emitting
[12] ensen, P. Milojkovie, J. Ekman, P. Chandramani, R. Rozier, F.
H. Soda, K.
Appl. Phys. v.18, 2329, (1979) K. Iga, Sishikawa, S. Ohkouchi
of GaAlAs/GaAs surface-emitting injection laser”, App. Phys. Lett. v45, 348, (1984) F. Koyama, S. Kinoshita and K. Iga, “Room-temperature continuous wave las characteristics of GaAs vertical cavity surface-emitting lasers”, App. Phys. Lett. V44, 221, (1989)
G. M. Yang
vertical-cavity surface-emitting lasers obtained with slelective oxidation”, Electron. Lett., Vol. 13. pp. 886-888, 1995
J. K. Guenter, J. A. Tantum, A. Clark. R. S.
Biard, Y. Liu, “Commericalisation of Honeywell’s VCSEL technology:further developments”, Proceedings of SPIE’s Optoelectronics 2001, Vol. 4286, pp. 1-14, 2001.
J. S. Span, Y. S. Lin, C. F. Li, C. H. Chang, J. C. Wu, B. L. Lee, Y. H. Chuang, S. L. Tu, C. C. Wu, “Commericailised VCSEL components fabricated at Truelight Corporation”, Proceedings of SPIE’s Optoelectronics 2001, Vol. 4286, pp. 15-21, 2001.
A. V. Krishnamoorthy, K. W. goossen, L. M. F. Chirovsky, R. G. Rozier, P
S. P. Hui, J. Lopata, J. A. Walker, L. A. D’Asaro, “16x16 VCSEL array flip-chip bonded to CMOS VLSI circuit”, IEEE Photon. Technol. Lett., Vol. 12, pp. 1073-1075, 2000 U. Eriksson, P. Evaldsson and K. Streubel. “A novel technology for monol integration of VCSELs and heterojunction bipolar transistors at 1.55um”, CLEO Pacific Rim ’97, Chiba, Japan, paper PD2.8, 1997
K. Tai, G. Hasnain, J. D. wyn, R. J. Fisher
Cho, “90% coupling of top surface-emitting GaAs/AlGaAs quantum well laser output into 8 um diameter core silica fiber”, Electron. Lett., Vol. 26, pp. 1628-1629, 1990
M. Y. Li, W. Yuen, G. S. Li and C. J. Chuang-Hasnain, “Top-emitting micromech VCSEL with a 31.6nm tuning range”, IEEE Photon. Technol. Lett., Vol. 10, pp. 18-20, 1998
U. Fi
vertical-cavity laser diode for 10GB/s data transmission”, IEEE Photon. Technol. Lett., Vol. 8, pp. 746-748, 1996
M. W. Haney, M. P. Christ
Kiamilev, Y. Liu and M. Hibbs-Brenner, “Multichip free-space global optical
interconnection demonstration with integrated arrays of vertical-cavity surface-emitting lasers and photodetectors”, Appl. Opt., Vol. 38, pp. 6190-6200, 1999
K. Goto, “Proposal for ultrahigh density optical disk system usin
[13] g a vertical cavity
[14] roc. SPIE,
[15] M. Grabherr, R. Michalzik, R. Jager, K. J. Ebeling and A.
[16] ps/802/3/ae/index.html
surface emitting laser array”, Jpn. J. Appl. Phys., Vol. 37, pp. 2274-2278, 1998 R. L. Thornton, “Vertical-cavity lasers and their application to laser printing”, P Vol. 3003, pp. 112-119, 1997
H. Martinsson, J. A. Vukusic,
Larsson, “Transverse mode selection of large-area oxide-confined vertical-cavity surface-emitting laser using a shallow surface relief”, IEEE Photon. Technol. Lett., Vol.
11, pp. 1536-1538, 1999 http://grouper.iee.org/grou
. Razavi, M. Chateauneuf, A. G. KirK,
[18] B. Cunningham, SRU
[19] al cavity surface emitting laser
[17] D.V. Plant, M. B. Venditti, E. Laprise, J. Faucher, K
J. S. Ahearn, “256-channel bi-directional optical interconnect using VCSELs and photodiodes on CMOS”, J. Lightwave Techno. 19(8) 1093, 2001
F. Mateus, M. C. Huang, Univ. of California/Berkeley; P. Li,
Biosystems; C. J. Chang-Hasnain, “High sensitivity label-free biosensor using VCSEL” , Proceedings of SPIE’s BIOS 2004 vol. 5328-22, 2004.
D. Kumar, H. Shao, Kevin Lear, "Dependence of vertic
diodes with integrated micro-fluidic channels on fluid refractive index" Optical Information systems II in series Proceedings of SPIE, vol. 5557, Aug 2004
Chapter 2 Rate Equations and laser dynamics
The rate equations provide the most fundamental description of the laser. It describes the time-evolution of carrier- and photon densities in a laser cavity as a function of the pump rate, material gain and parameters associated with the material properties and laser construction. In its simplest form, it consists of a pair of coupled nonlinear differential equations Eq.(2-1), one for the carrier density, and one for the photon density. Therefore it is well suited for modeling simulation. By this technique, many of properties of VCSELs can be investigated in this paper.
The phase does not enter into Eq.(2-1), since the optical power does not depend on the phase, it depends on the optical magnitude only. However, some effects are highly related to the laser phase, such as mode locking, injection locking and polarization switching, etc, in this situation, the phase has to be taken into account. Before proceeding further, it is important to clear up the fundamental mechanisms and assumption in the rate equations, such as stimulated and spontaneous emission, stimulated absorption, non-radiative recombination, and so on. When referring to
“carriers”, the ambipolar assumption is applied, that means there is no difference between electrons and holes.
2.1 Carrier Rate Equation
The rate of change of the electrons (or holes) density comes from electron-hole generation and recombination. And these changes must be related to the number of photons produced (for a direct bandgap semiconductor). We assume that the electrons and holes remained confined to the active region having volumeV . It is intuitive to obtain the basic form of carrier rate equation:
a
Rnonr where Ginj is pumping energy, Rstim is stimulated recombination which is the difference between stimulated emission and stimulated absorption, Rspon is spontaneous emission, and Rnonr is non-radiative recombination. In Figure 2.1, it shows all the individual terms contribute to the carrier rate equation. Now, we will carefully examine each term as the section proceeds.
Carrier Rate Equation
Fig 2.1 The individual terms considered by general carrier rate equation.
2.1.1 Pumping Energy
The pump consists of either bias current or optical flux. The pump term describes the density of electron-hole pairs produced in the active volume in each second, which is given by
Va
a
whereGinjis pumping current density per second (carriers/cm3/ sec), ηi is injection efficiency or internal quantum efficiency (ideal is 1), is injection current per second, is the elementary charge which changes the units from coulombs to the
“number of electrons”. Pump increases the number of electrons and holes in the conduction band (CB) and valence band (VB), respectively, is shown in Figure 2.2.
) essential condition to reach population inversion.
The injection efficiency ηi represents the fraction of injected current that flow into the active region. In practical, due to the surface leakage or other carrier loss mechanisms, the ηi is always less than 1, it is defined by
2.1.2 Stimulated absorption and stimulated emission
The reason we discuss stimulated absorption and emission in the same section is due to their interaction will be used to present the concept of gain medium, it is the major part of laser device ( recall laser = Light Amplified by the Stimulated Emission of Radiation ). Both of stimulated absorption and stimulated emission results from stimulated recombination, therefore, need photons to conduct its process (see Figure 2.3). In contrast, the spontaneous emission does not need photons to conduct its process, will be discussed in next section(2.1.3). According to their various physical mechanisms, please refer to [1][2][3] for further explanation, again, the main purpose of this paper is to demonstrate how to build up a device model based on well-known electro-optics theory. In the simplest way to merge stimulated absorption and recombination mechanisms into to rate equations is to understand the stimulated absorption process would decrease photons number and electron-hole pairs increased.
On the opposite, the stimulated recombination process would multiple increase photons (photons amplified) and electron-hole pairs decreased, that’s why the pumping current should maintain in a certain level to keep the emission process continuously in the active region.
hv>Eg
CB
VB
CB
VB hv
CB
VB
CB
VB Stimulated absorption Stimulated emission
before after befor after
hv
Fig. 2.3 Stimulated absorption generates more electron-hole pairs. Stimulated emission produced coherent light (same energy, phase, propagation direction, highly monochromatic).
The term Rstim is stimulated recombination which is the difference between stimulated emission and stimulated absorption, usually, the stimulated recombination produces more photons than them absorbed (this is one of the essential conditions for lasing, otherwise, the gain will be less than one, no amplified). The Figure 2.4 shows single photon incident on the left side of the gain medium. This photon enters the material and interacts with the carriers. Some of the processes emit photons (stimulated emission) while some of them absorb photons (absorption or sometimes called stimulated absorption) and some do nothing. It implies the gain is the ratio of output photons over the input photons. The gain describes here only the stimulated emission and absorption processes and does not include photon losses through the side of the laser or through the mirrors.
hv
CB
VB
hv
hv hv hv hv hv hv hv hv
hv
CB
VB
hv
hv hv no photon emission
hv
CB
VB
hv
hv hv hv hv
gain=4/1=4
(lasing condition)
negative gain
(no lasing)
gain=1/1=1
(transparency condition)
Fig. 2.4 The interaction of stimulated absorption and stimulated emission contribute the optical gain of material. The “gain” represents in this figure is just the ratio of photons that is not the same as “material gain” we will discuss later. It also presents the gain is the function of photons traveling length.
The lower one of Figure 2.4 brings up one terminology, called “material transparency”. The semiconductor material becomes “transparent” (“material transparency”) when the rate of stimulated absorption just equals the rate of stimulated emission. One incident photon produces exactly one photon in the output.
The transparency density (number per unit volume) represents the number of excited carriers per volume required to achieve transparency. The material gain required to achieve lasing will be much larger than zero since the gain must offset other losses besides stimulated absorption (typically G=150 cm
Ntr
-1 ). The relation between material gain and carrier density is depicted in Figure 2.5. This yields the simplest expression of material gain:
(2-5) )
( )
(N g N Ntr
G = 0 −
where is linear gain coefficient or differential gain at , is transparency density. In this case, we assume the photon density S is not high. Therefore, the gain curves can be approximated by a straight line at (refer to Figure 2.5).
g0 Ntr Ntr
Ntr
When photon density is high enough, the factor (1+εS)is taken into account. This factor accounts for nonlinear gain saturation, where ε is gain suppression coefficient.
This yields another material gain expression:
This yields another material gain expression: