Chapter 2 Fundamentals of VCSEL and Fabry-Perot Resonator
2.6 Work Review
Optically pumped VCSEL[39]
In our previous work, the characteristics of a GaN-based VCSEL with 25 pairs AlN/GaN DBR (with reflectivity about 95%) and 8 pairs Ta2O5/SiO2 DBR (with reflectivity about 97.5%) was successfully fabricated and investigated. The schematic diagram and SEM image of the overall VCSEL structure are shown in the Fig 2.10 (a) and (b). A narrow PL emission with full width at half maximum of 1.4nm corresponds to the cavity resonant mode at 448nm was observed, as shown in Figure 2.11. The cavity quality factor, estimated from the emission linewidth of 1.4nm, was about 320. Figure 2.12 shows the laser action was achieved under the optical pumping at room temperature with a threshold pumping energy density of about 53mJ/cm2. The GaN VCSEL emits 448nm blue wavelength with a linewidth of 0.17nm. The estimation of the carrier density and gain at the threshold were about 3×1020 cm-3 and 1.45×104 cm-1, respectively. The NFP and FFP showed that the beam width and divergence angle were about 3.0µm and 7.6o, respectively. The NFP and FFP also indicated that the shape of laser emission was close to a circle. The laser beam showed a degree of polarization of about 84% suggesting strong polarization property of the laser emission. The characteristic temperature of fabricated VCSEL was about 243K suggesting good temperature tolerance.
10 pairs of In0.2Ga0.8N/GaN MQW
25 pairs of AlN/GaN DBR
2μm GaN
25 pairs of AlN/GaN DBR
2μm GaN Sapphire
n-GaN p-GaN
10 pairs of In0.2Ga0.8N/GaN MQW
25 pairs of AlN/GaN DBR
2μm GaN
Figure 2.10 (a) The schematic diagram of the overall VCSEL structure (b) The SEM image of the overall VCSEL.
425 430 435 440 445 450 455 460 465
448nm
1.4nm
Wavelength (nm)
Figure 2.11 PL emission of the overall VCSEL structure.
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0
5000 10000 15000 20000 25000
Emission intensity (a.u.)
Excitation energy (μJ/pulse)
430 440 450 460 470 480
1.71Eth 1.43Eth 1.14Eth
0.86Eth 1.00Eth
Emission intensity (a.u.)
Wavelength (nm)
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0
5000 10000 15000 20000 25000
Emission intensity (a.u.)
Excitation energy (μJ/pulse)
430 440 450 460 470 480
1.71Eth 1.43Eth 1.14Eth
0.86Eth 1.00Eth
Emission intensity (a.u.)
Wavelength (nm)
Figure 2.12 The excitation energy - emission intensity curve (L-I)
GaN based Micro-cavity light emitting diode (MCLED)
With the achievement of optically pumped GaN-based VCSEL, the realization of electrically-injected GaN-based VCSEL has become promising. So far, the electrically injected GaN based VCSEL has not been realized. However, the micro-cavity light emitting diodes (MCLEDs) [15-22], with a quasi VCSEL structure, could be served as an initial step toward electrically injected GaN VCSEL. They mainly utilized an epitaxially growth nitride DBR as the bottom mirror and a dielectric DBR as the top mirror. This kind of device has several advantages comparable to VCSEL, such as circular beam shape, light emission in vertical direction, fully monolithic test and two dimensional arrays. Some GaN-based MCLEDs with an in-situ epitaxially grown nitride-based DBRs and dielectric DBRs as the bottom and upper mirror of the cavity were reported. Recently, Diagne et al. [15] employed 60 pairs of GaN/Al0.25Ga0.75N DBR as bottom mirror (with reflectivity of 99%) and SiO2/HfO2 as top mirror (with reflectivity of 99.5%). The schematic diagram and EL spectrum of this MCLED are both shown in Figure 2.13. The emission peak wavelength of the MCLED was located at 413nm with a narrow line width of 0.6nm. It means the Q factor is about 690. This result was the best value compared with the Q factor of the MCLED published in the recent literature.
Figure 2.13 The schematic diagram and EL spectrum of the GaN
Chapter3
Fundamentals of Ion Implantation
3.1 Introduction to Ion Implantation[40-43]
3.1.1 Introduction
For much of the past decade, GaN has been a subject of extensive research due to very important technological applications of this material. As is well documented in the literature, current applications of GaN include light-emitting diodes (LEDs), laser diodes, UV detectors, and microwave power and ultra-high power switches. In the fabrication of such GaN-based devices, ion implantation represents a very attractive tool for several technological steps, such as electrical and optical selective-area doping, dry etching, electrical isolation, quantum well intermixing, and ion-cut. It is well-known that a successful application of ion implantation depends on understanding the production and annealing of radiation damage. Thus, detailed studies of ion implantation damage in GaN are not only important for investigating
fundamental defect processes in solids under ion bombardment but are also essential for the fast developing GaN industry.
3.1.2 History overview
The beginnings of ion implantation are now more than twenty years in the past. In 1957, Shockley obtained the first patent for the technique of ion implantation. In this patent, it was pointed out for the first time that annealing after implantation is necessary for
re-crystallization of the crystal lattice. This patent covers practically all aspects of
implantation. From then on, more and more articles dealing with implantation were published.
The final breakthrough began in the middle of the 1960s. For the production of MOS transistors, the use of implantation has become generally accepted; hardly any integrated MOS circuit is produced nowadays without the use of one or more implantation steps. In the field of bipolar transistors. The use of implantation is steadily increasing, while it is already being employed as standard technology for several special devices.
Implantation offers a number of technological advantages which are important in the fabrication of optoelectronic devices:
1. Speed, homogeneity, and reproducibility of the doping process.
2. Avoidance of high processing temperatures during implantation.
3. Simple masking methods, for example, with the use of thick layers of oxide, nitride, metal or photoresist.
4. Possibility of doping through thin passivating layers (e.g., SiO2, Si3N4)
5. Low penetration depth of the ions; it is possible to dope shallow layers with very high doping gradients.
6. Multiple implantation by changing the accelerator voltage during implantation makes possible a relatively free choice of the doping profile, whereby one is not limited to the Gaussian shape.
7. Because if the minimal lateral scattering, it is possible to fabricate devices with very small dimensions and to keep the parasitic capacitance low.
3.1 Principle of Ion Implantation Theory of ion stopping
Damage is caused by the ion stopping process in semiconductors. The two important effects in determining the stopping of implanted ion are in the following.
Inelastic collisions of the ions with bound electrons in the crystal. The energy loss in this case is by excitation or ionization of the target atoms. This is termed electronic stopping, and does not create atomic displacements in the material.
Elastic nuclear collisions with nuclei or whole atoms of the crystal, in which a part of the kinetic energy of the incoming ion is transferred to the nuclei that absorb the impact, termed nuclear stopping, and it leads to the creation of deep-level, compensating defects.
The nuclear stopping process of the ion can be considered as caused by collisions between two hard spheres (the ion and the target nuclei) in which the ion loses energy by transferring it to the displaced nuclei. Theoretically this is treated by a Columbic force at a distance
scattering process. One of the important parameters, therefore, is the atomic scattering potential V(r), which is not all that accurately known. The electronic stopping process can be visualized as similar to the stopping of a projectile in a viscous medium with the ion slowed by a series of “drag” interactions.
The relative importance of mechanisms (I) and (II) above depends on the energy and mass of the implanted ions, and the mass and atomic density of the crystal. To calculate stopping of ions, it is useful to introduce the concept of a cross section S for both electronic and nuclear stopping,
dE is the energy loss per unit distance for either electronic or nuclear stopping, and
N is the atomic density of the crystal. The contribution from nuclear energy loss tends to be small at high energies because fast ions have only a short time to interact with a target nucleus,
i.e. they are moving past the target nuclei too fast to efficiently transfer energy to them. At intermediate energies, the nuclear energy loss component increases, but falls again at the lowest energies where electron screening effects lower the ehhective atomic number of the target nuclei.
If both stopping power are independent, then the total energy loss per unit path length of the ion is , where R is the average range or total path length of an ion of energy E in an amorphous crystal, and is proportional to the velocity of the implanted ion, and is proportional to the atomic density of the target and to the total energy transferred in all individual collisions. Knowing and one can obtain the total path length or range, R, of the implanted ion in the target before coming to rest. It is usual to use the projected range , which is defined as the projection of R normal to the surface. For a Gaussian distribution corresponds to the point of maximum concentration of the distribution.
)
Distribution profile of introduced ions
The range – energy relation given by Eq. (2) was reformulated by Lindhard, Scharff, and Schiott (LSS) for ion implantation into amorphous materials in terms of the reduced parameters, ε, ρ, as: Whereρandεare dimensionless variables related to the range, R, and incident energy E0, by:
2
Where M1andM2 are the masses of the incident ions and target atoms, respectively; N is the
number of atoms per unit volume; a is the screening length, equal to is the Bohr radius; and and are the atomic numbers of the ion and target.
a0
Z1 Z2
The value of ρ was converted to R, and a value for was obtained from approximate expression:
And then the implanted ion concentration, n, as a function of depth, x, can be described as:
2 ] distribution (projected straggle range of the distribution in the direction of incidence of the beam). The value ofΔRpis calculated in terms of and the mass of ions and target atoms, by the approximate expression:
The LSS assumption is a well simulation of implantation in amorphous materials, but it is not good at single crystal materials due to the channeling effect in the single crystal
materials. In single crystal lattices there are some crystal directions (known as channels) along which the ions will not encounter any target nuclei, and will be channeled along such open channels of the lattice. The channeling effect in single crystal materials can be showed out by the Secondary Ions Mass Spectroscope (SIMS) measurement or the Rutherford
Backscattering Spectroscope (RBS) measurement.
Damage of ion implantation
When the energetic ions strike the GaN target, they lose their energy in a series of nuclear and electronic collisions, and rapidly come to rest some hundreds or some thousands of atom layers below the surface. Only the nuclear collision result in displaced atoms. An individual nuclear collision can different types of displacement events, depending on the magnitude of the energy transferred. If the energy of nuclear collision (△En) transferred to the Ga or N atom is less than the energy required to displace it from its lattice site, Ed, no
displacement event results, if 2Ed>△En>Ed, a single displacement and simple isolated point defects are created. If △En>2Ed, multiple secondary displacements and defect cluster are produced. According to the reports of ion implantation, the single displacement energy of III-V chemical compound is about 20eV. There are about three types of defect made by point defect and cluster defect after ion implantation:
(i) Isolated point defects or point defect clusters in crystalline GaN layer.
(ii) Local amorphous layer in an otherwise crystalline GaN layer.
(iii) Continuous amorphous layers in all epitaxial GaN layer.
All the three types of defect need to be annealed out by thermal annealing or another thermal-produced process (RTA) in GaN. The following section is describing the common types of thermal annealing process: rapidly thermal annealing (RTA).
RTA post-annealing
In the semiconductor industry, ion implantation used for electrical and optical doping is always followed by an annealing step. Such annealing is necessary:
(i) To remove implantation-produced lattice disorder.
(ii) To electrically/optically activate implanted species by stimulating their migration into energetically favorable lattice sites.
Post-implantation annealing is a very important technological step since device performance is highly dependent on the efficiency of such annealing.
Isolation
Ion implantation can be employed for two applications in compound semiconductors, namely, the creation of doped regions by implantation and activation of dopant species or the reverse process of creation of high resistance regions by formation of deep traps or
compensating centers. The latter process, named implant isolation, is used for inter-device isolation or to produce current guiding. There is a strong need for an understanding of the implant isolation process in GaN because of the emerging applications for high temperature, high power electronics based on this materials system. Prototype devices such as
hetero-structure field-effect transistors, hetero-junction bipolar transistors, junction field effect transistors, and metal– oxide–semiconductor field-effect transistors have all been
demonstrated, with impressive high temperature (>.300 °C) performance. There are two types of defect-formation mechanisms that are found for implant isolation in semiconductors:
(I) Damage-related isolation: The creation of midgap, damage related levels, which trap the free carriers in the material. This type of compensation is stable only to the temperature at
which these damage-related levels are annealed out. For damage compensation, the resistance typically goes through a maximum with increasing post-implantation annealing temperature as damage is annealed out and hopping conduction is reduced. At higher temperatures, the defect density is further reduced below that required to compensate the material, and the resistivity decreases.
(II) Chemically-induced isolation: The creation of chemically induced deep levels by implantation of a species that has an electronic level in the middle of the band gap. This type of compensation usually requires the implanted species to be substitution and hence annealing is required to promote the ion onto a substitutional site. In the absence of out diffusion or precipitation of this species, the compensation is thermally stable. For chemical compensation, the post-implantation resistance again increases with annealing temperature with a reduction in hopping conduction but it then stabilizes at higher temperatures as a thermally stable compensating deep level is formed. Typically, there is a minimum dose (dependent on the doping level of the sample) required for the chemically active isolation species to achieve thermally stable compensation. Thermally stable implant isolation has been reported for n- and p-type AlGaAs where an Al-O complex is thought to form and N in GaAs (C) where C–N complexes are thought to form.
In this study, we use the Mg ion implantation to isolate the GaN for achieving current confinement. The main mechanism to form high resistivity layer is damage induced isolation.
Chapter4
GaN-based Micro-cavity Light Emitting Diodes (MCLEDs) using ITO as transparent contact layer
4.1 Recent Status
4.1.1 Introduction to Conventional GaN-based MCLED [44]
With the achievement of optically pumped GaN-based VCSEL, the realization of electrically-injected GaN-based VCSEL has become promising. So far, we have successfully fabricated the GaN-based MCLED with hybrid structure, composed of high reflectivity, crack-free, wide stop-band width in-situ grown AlN/GaN bottom DBRs (with reflectivity of 95%) and ex-situ deposited SiO2/Ta2O5 top DBRs (with reflectivity of 97.5%). Such device could be used as an initial step toward the GaN-based VCSEL and has several advantages, such as circular beam shape, light emission in vertical direction, fully monolithic test and two dimensional arrays. Figure 4.1 shows the structure and characteristics of the conventional device. The turn on voltage and resistance were about 3.5V and 530Ω, respectively. The device showed the emission wavelength of 458.5nm and FWHM of 6.7nm at 20mA injected current. The previous work was accepted and published by Japanese Journal Applied Physics letters, Vol. 45, No. 4B, 2006, pp. 3446–3448.
4.1.2 Issue of Q factor
In our previous work, the Q factor of our conventional MCLED could be calculated to be about 68, which is much low than that of our optically pumped VCSEL structure (Q = 320).
The main difference between optically pumped VCSEL structure and conventional MCLED structure is the insertion of Ni/Au transparent contact layer. According to laser theory, we can find the Q-factor is inversely proportional to the total loss δ inside the cavity. It suggests that our relatively low Q factor could be attributed to the insertion of Ni/Au transparent contact layer. In order to improve the Q factor of micro-cavity, replacing the Ni/Au transparent
contact layer by another one with high transmittance and low absorption is required. Over few past years, high electrical conductivity and transparency to visible light have made indium tin oxide (ITO) a useful material for transparent contacts to many optoelectronic devices [45]. Rectifying contacts to silicon-, GaAs-, and InP-based solar cells and photo-detectors have already been demonstrated. In the next section, we discuss the effect on Q factor while using
ITO as transparent contact layer for GaN based MCLED, and demonstrate the theoretical calculation to further understand the effect.
Ti/Al/Ni/Au n-contact 8 pairs of Ta2O5/SiO2DBR Ni/Au
Ni/Au p-contact 5-30μm
15-40 μm
n-GaN
25 pairs of AlN/GaN DBR stack Collimated output beam
Sapphire substrate
(a)
0 5 10
0 2 4 6 8 10 12 14 16
Voltage (V)
Current (mA)
(c) (b)
400 450 500
458.5nm
FWHM = 6.7nm
Figure 4.1 The conventional MCLED using Ni/Au as transparent contact layer.
(a) The structure of conventional MCLED. (b) The EL spectrum of the conventional MCLED.
(c) The I-V curve of the conventional MCLED
4.2 GaN based MCLED using ITO as transparent contact layer 4.2.1 Introduction to indium tin oxide (ITO)
Indium tin oxide (ITO, or tin-doped indium oxide) is a mixture of indium (III) oxide (In2O3) and tin (IV) oxide (SnO2), typically 90% In2O3, 10% SnO2 by weight. It is transparent and colorless in thin layers. In bulk form, it is yellowish to gray. Indium tin oxide's main feature is the combination of electrical conductivity and optical transparency. However, a compromise has to be reached during film deposition, as high concentration of charge carriers will increase the material's conductivity, but decrease its transparency. Thin films of indium tin oxide are most commonly deposited on surfaces by electron beam evaporation, physical vapor deposition, or a range of sputtering techniques.
ITO is mainly used to make transparent conductive coatings for liquid crystal displays, flat panel displays, plasma displays, touch panels, electronic ink applications, light-emitting diodes, and solar cells. ITO is also used for various optical coatings, most notably
infrared-reflecting coatings for architectural, automotive, and sodium vapor lamp glasses.
Other uses include gas sensors, antireflection coatings, and Bragg reflectors for VCSEL lasers.
4.2.2 Optimization of the thickness of ITO transparent contact layer
In order to let the PL emission peak locate at the cavity mode, the thickness of ITO film must be optimized for providing a suitable cavity length. First, an N&K measurement for this ITO film on glass substrate yielded a refractive index of 1.928. Figure 4.2 shows the results of ITO (240nm) within the cavity using TFCal simulation software. The incident angle of
illumination and wavelength of the reference light used in the simulation were 0o and 460nm, respectively. It shows the 240nm thick ITO film would be the optimization to make the cavity mode locate at around 460nm, which corresponds with the PL emission peak of this sample.
Figure 4.2 The results of ITO (240nm) within the cavity using TFCal simulation software
4.2.2 Comparison of the absorption coefficient of ITO and Ni/Au film
Figure 4.3 shows the transmittance (T) and reflectance (R) of Ni/Au (5/5nm) and ITO (300nm) films deposited on glass substrate. These two films were both thermally treated to
Figure 4.3 shows the transmittance (T) and reflectance (R) of Ni/Au (5/5nm) and ITO (300nm) films deposited on glass substrate. These two films were both thermally treated to