Tài liệu hạn chế xem trước, để xem đầy đủ mời bạn chọn Tải xuống
1
/ 37 trang
THÔNG TIN TÀI LIỆU
Thông tin cơ bản
Định dạng
Số trang
37
Dung lượng
498,47 KB
Nội dung
Chapter Introduction Chapter Introduction III-nitride materials including gallium nitride (GaN), indium nitride (InN), aluminum nitride (AlN), their ternary (InGaN, AlGaN and AlInN) and quaternary alloys (AlGaInN), are wide-bandgap III-V compound semiconductors During the last two decades or so, III-nitride semiconductor material system has attracted considerable attention due to their unique and excellent optical, electrical and material properties as compared with the conventional III-V compound semiconductors, such as III-arsenides and III-phosphides The highly promising applications of III-nitrides include optical storage, laser printing, high brightness & general illumination, and wireless base stations Section 1.1 will highlight the currently most important applications of GaN-based III-nitrides; and Section 1.2 will review the research development of GaN and related III-nitride materials Typically, the growth techniques and the major properties of InGaN/GaN quantum wells and quantum dots will be discussed in Section 1.3 In this work, the GaN-based quantum structures will be applied in the fabrication of saturable absorbers and semiconductor saturable absorber mirrors (SESAMs) Chapter Introduction operating in the blue region Thus, Section 1.4 will outline the essential theories of the saturable absorber and the SESAM, and provide an overview of the development of passive mode-locking by SESAMs 1.1 Current applications of GaN-based III-nitrides In this session, the currently most important applications of GaN-based III-nitrides, including the short-wavelength optoelectronic devices [Akasaki1991; Nakamura1995] and the high-power high-frequency high-temperature electronic devices [Khan1993; Pearton2000; Mishra2002; Zhang2002], will be discussed Figure 1.1 Bandgap versus lattice constant of wurtzite III-nitrides at room temperature Chapter Introduction The great potential of GaN-based III-nitrides in fabricating short-wavelength optoelectronic devices is mainly owing to their wide range of bandgaps, which spans from the infrared spectrum and extends well into the ultra-violet (UV) As shown in Figure 1.1, the wurtzite polytypes of GaN, AlN and InN form a continuous alloy system whose bandgaps range from ~ 0.7 eV for InN, to 3.4 eV for GaN and to 6.2 eV for AlN [Nakamura1995] Furthermore, all the wurtzite III-nitrides have direct bandgaps, which are essential for producing light radiation efficiently The above features enable the III-nitrides to be good candidates for light emitting diodes (LEDs), laser diodes (LDs), and detectors operating in green, blue and UV wavelength regions [Nakamura2000; Akisaki2007], which are essential for developing full-color displays and coherent short-wavelength sources required by high density optical storage technologies [Pearton2000] In recent years, the introduction of GaN-based bright blue LEDs have paved the way for full color displays and raised the probability of mixing primary colors - red, green and blue - to obtain white light sources for illumination Most excitingly, when used in place of incandescent light bulbs, these GaN-based blue LEDs can provide higher brightness and longer lifetime while consuming only ~ 10 - 20% of the power for the same luminous flux [Mahammad1995] Therefore, GaN-based III-nitride LEDs have immense potentials as the candidates of the next-generation illuminating source Additionally, the strong absorption coefficient (4×104 cm-1) of GaN at wavelengths shorter than 365 nm makes it also a good candidate for visible-blind UV photodiodes [Yagi2000] Extension of the photodiode spectral range from UV to visible Chapter Introduction can be achieved by the employment of an InGaN active layer with a high In composition (up to 67%) Furthermore, the integration of the efficient UV/blue semiconductor light sources and the visible-blind photodiodes on a single chip has been realized [Pauchard2000] In the last few years, various types of GaN-based sensors have been demonstrated, such as p-n junction diodes, Schottky diodes, and metal-semiconductor-metal photodiodes [Razeghi1996; Li2003; Chang2007; Yam2007] These structures can be used for several sensor applications, such as lifetime sensors, drug testing sensors and portable fluorescence sensors Besides the potential for fabricating short-wavelength optoelectronic devices, GaN-based III-nitrides are very suitable for high-frequency electronic devices, especially for high-power and high-temperature applications The material properties associated with high-temperature, high-power, and high-frequency applications of GaN and several conventional semiconductors are summarized in Table 1.1 [Pearton2000] As shown in the Table, compared with the conventional III-V semiconductors, GaN with wide bandgaps can operate at much higher temperatures before going intrinsic or suffering from thermally generated leakage current GaN also has a higher breakdown field of around × 106 V cm-1 (Vbr ∝ Eg3/2), i.e., the maximum internal electric field strength before the onset of junction breakdown This allows GaN to operate as high-power amplifiers, switches, or diodes In addition, the good electron transport characteristics of GaN, including extremely high peak velocity (3 × 107 cm s-1) and saturation velocity (1.5 × 107 cm s-1), allow it to operate at higher frequencies than its conventional cousins [Khan1995] Chapter Introduction Table 1.1 Comparison of material properties of GaN, 4H-SiC, GaAs and Si (Pearton2000) GaN 4H-SiC GaAs Si Bandgap Eg at 300K (eV) 3.40 3.26 1.42 1.12 Dielectric constant ε Breakdown field EB (MV/cm) High-field Peak velocity νs (107 cm/s) Electron mobility μ (cm2 V-1 s-1) Thermal conductivity χ (W K-1 cm-1) 9.0 4.0 9.7 3.0 12.8 0.4 11.8 0.25 3.0 2.0 2.0 1.0 1350 800 6000 1300 1.3 4.9 0.5 1.5 Melting point (°C) 2791 Sublimes T>1827 1238 1412 JFOM* = EBνs / 2π 48 24 3.2 BFOM** = χ μ EB3 3686 3473 6.3 CFOM*** = χ ε μ νs EB2 489 458 * JFOM: Johnson’s figure of merit for power-frequency performance of discrete devices ** BFOM: Baliga’s figure of merit for power loss at high frequency *** CFOM: Combined figure of merit for high power/high frequency/high temperature applications (All figures of merit are normalized to Si) Arising from the superior optical, electrical and material properties, GaN-based III-nitrides have tremendous application potential in a variety of areas, including power conditioning, wireless broadband, automotive electronics, power transmission, telecomm base stations, satellite electronics, white solid-state lighting, color signs & lighting, automotive lighting, high-density optical storage, pressure sensing, heat sensing, and flame sensing [Mills2002] The rapid development of III-nitrides in the last two decades, which will be presented in detail in Section 1.2, can be considered as a breakthrough in the field of wide bandgap compound semiconductor Chapter Introduction materials and devices [Pearton1999; Jain 2000; Akasaki2007] 1.2 Research development of GaN and related III-nitride materials The earliest attempts to synthesize GaN materials were initiated more than 70 years ago In 1932, GaN was synthesized in the powder form [Johnson1932]; and in 1938 small needles of GaN were obtained by Juza and Hahn [Juza1938] However, at this early stage, the growth of high quality epitaxial nitrides was impossible and the difficulty in controlling their conductivity had also prevented the development of nitride-based devices In 1969, the deposition of large-area single crystal GaN was successfully demonstrated on a sapphire substrate by hydride vapor phase epitaxy (HVPE) [Maruska1969] Two years later, GaN was grown epitaxially via metal organic chemical vapor deposition (MOCVD) and in 1974 by molecular beam epitaxy (MBE) Since then, GaN-based III-nitrides have experienced rapid development over the past few decades, as shown in Fig 1.2 Chapter Introduction Figure 1.2 Number of publications (INSPEC) and activities related to GaN-based III-nitrides over the years The achievement of large area GaN fabrication soon led to the first demonstration of a III-nitride LED [Pankove1972] This LED was a metal/insulating-GaN:Zn/n-GaN (M-i-n or MIS) type diode Zn was doped into GaN to produce i-n junctions, as well as acting as the luminescence centers The M-i-n LED could emit blue, green, yellow or red light depending on the Zn concentration in the light-emitting region A year later, the Mg-doped M-i-n type diode emitting violet light was also demonstrated [Maruska1973] Chapter Introduction However, until the late 1970s, it was still quite difficult to grow high quality epitaxial GaN film with a flat surface and free of cracks Moreover, the conductivity of the GaN thin film could not be controlled All early GaN samples were n-type conducting even without intentional doping It was therefore impossible to produce p-type conduction or to control the conductivity of the n-type nitrides In early 1980s, the above situation was drastically changed by two critical breakthroughs, i.e the development of the two-step growth method and the discovery of p-type conduction The idea of the two-step growth is, at a low temperature, to insert a very thin buffer layer with physical properties similar to both the GaN epilayer and the sapphire, so as to accommodate the large lattice and thermal mismatches between GaN and sapphire In 1983, the two-step method was first applied in the MBE growth of GaN [Yoshida1983] The electrical and luminescent properties of the GaN epitaxial films were improved with a low-temperature (LT) AlN buffer on the sapphire substrate Later, in 1986, the world’s first high-quality GaN grown by MOCVD was also demonstrated with a LT AlN buffer; and the optimized thickness and deposition temperature for the LT AlN buffer were established [Amano1986] It was later found that a thin GaN layer could also be used as the LT buffer to obtain high-quality GaN epilayers grown on sapphire substrates [Nakamura1991] Moreover, besides the single AlN or GaN LT buffer, namely single-layered buffer (SLB), a new structure for the LT buffer, namely double-layered buffer (DLB), was also proposed later [Turnbull1996] With this two-step growth method, the residual donor density in the GaN grown by MOCVD decreased to the order of 1015 cm-3; and the electron mobility increased by Chapter Introduction more than an order of magnitude As a result, the development of the two-step growth technique has greatly improved not only the crystalline quality but also the optical and electrical properties of GaN epilayers grown by MOCVD The second breakthrough of p-type conductivity realization in GaN was not demonstrated until the late 1980s With the two-step growth method, the control of the n-type conductivity of GaN by SiH4 doping was possible [Nakamura19921]; while to fabricate the practical LED and LD devices, the ability to control the p-type conductivity was also necessary Therefore, right after the first breakthrough, many research groups attempted to produce p-type GaN with Be, Mg, or Cd doping In 1988, Amano et al found that low-energy electron-beam irradiation (LEEBI) treatment largely enhanced the blue emission of Zn-doped GaN, but the crystal still did not show p-type conductivity [Amano1988] This finding suggested that the LEEBI treatment might be closely related with the activation of Zn-acceptors Hence, in 1989, the Mg-doping using Cp2Mg or MCP2Mg as a Mg-dopant was successfully achieved while the high crystal quality of the GaN was maintained Then after the electron beam irradiation, the Mg-doped GaN samples showed greatly enhanced blue luminescence as well as the p-type conductivity with low resistivity [Amano1989] This critical discovery immediately led to the fabrication of the first GaN-based p-n junction blue/UV LED in 1989 [Amano1989] In 1992, p-type GaN was also produced by annealing GaN:Mg at above 700oC in N2 or vacuum ambient [Nakamura1992] A hole concentration as high as 3×1018 cm-3 was achieved with a resistivity of only 0.2 Ω⋅cm Compared to the LEEBI treatment method, this annealing method is a more effective Chapter Introduction way to achieve p-type conduction and more favorable for mass-production Afterwards, p-type GaN was also obtained by UV or electro-magnetic wave irradiation at temperatures below 400oC [Kamiura1998; Tsai2000; Takeya2001] The activation of the Mg acceptors by either the beam radiation or the N2 annealing is related to the hydrogen passivation effect [Nakamura19922] It has also been found that the p-type conductivity is dominated by the acceptors with an activation energy of ~170 meV; and these acceptors are attributed to the Mg atoms substituting for Ga atoms in the GaN lattice [Gotz1996] Due to the high ionization energy of Mg, the acceptor activation ratio is typically in the range of 0.1% to 1% Therefore, a high Mg chemical concentration, in the range of ~1020 cm-3, has to be incorporated to achieve a good p-type conducting GaN epilayer Up to now, Mg is the only effective acceptor successfully developed for III-nitrides As shown in Fig 1.2, the above-mentioned two critical breakthroughs (two-step growth method and p-type conductivity control) have eventually led to the commercialization of high-brightness blue LEDs and long-life blue-violet LDs as well as the development of a large variety of III-nitride based optoelectronic devices To fabricate the devices with longer lifetimes, epitaxially lateral overgrown (ELO) GaN on sapphire was then developed to largely reduce the threading dislocations [Kapolnek1997; Nam1997] Additionally, to further eliminate the effects of sapphire, pure GaN substrates, fabricated by removing the sapphire substrate after the growth of thick GaN layers on the ELO GaN template, were even fabricated for the growth of GaN-based epilayers [Nakamura1998] Furthermore, in order to further improve the 10 Chapter Introduction (b) (a) leading edge pulse modified by a fast absorber pulse modified by a slow absorber Figure 1.8 Illustration of pulse modification by saturable absorbers The pulses modified by the absorbers have been normalized to the original ones for easier comparison (a) Fast saturable absorber The pulse is temporally narrowed (b) Slow absorber Only the pulse front gets shortened [Xiang2003] In addition, it should be noted that the amplifying medium has gain saturation characteristics, which can make the leading part of the pulse experience stronger gain than the rear of the pulse Thus, the combined effects of the slow absorber and the saturable amplifying medium on the pulse shortening can be sketched qualitatively as shown in Fig.1.9 [Xiang2003] After many round-trips, the pulse is effectively narrowed and the peak intensity is greatly enhanced The shortened pulse can be symmetrical or asymmetrical, depending on the relative effectiveness of the above two effects 23 Chapter Introduction Figure 1.9 Pulse shortened by a simultaneous action of a saturable absorber and an amplifying medium [Xiang2003] The development of semiconductor technology has made it possible to use semiconductors as absorbers operating in a broad range of wavelengths and to accurately control the device parameters, such as operation wavelength, saturation energy, and recovery time Compared with the bulk semiconductors, quantum well structures are more favorable to be used as saturable absorbers, because of the convenience to control the wavelengths and their enhanced nonlinear property To achieve the effective pulse shaping, the saturable absorption of a saturable absorber should recover to its initial state in a short time (a few picoseconds to a few tens of picoseconds) However, the absorption recovery times of those epitaxially grown compound semiconductor quantum wells normally fall in the nanosecond range Therefore, many methods have been developed to purposely reduce the recovery times of the saturable absorbers The most common methods are low-temperature (LT) 24 Chapter Introduction growth [Gupta1992], ion implantation [Delpon1998], and proton bombardment [Gopinath2001] The gist of these methods is to introduce deep levels in the bandgap of the active layer so as to reduce the carrier lifetime through the enhanced non-radiative recombination 1.4.2 Key properties of a SESAM A SESAM integrates the semiconductor saturable absorber with a mirror More specifically, as shown in Fig 1.6, if a semiconductor saturable absorber and the mirror on the right side are fabricated as a single structure, this structure is called as a SESAM For a SESAM, the mirror can either be a metallic mirror or a distributed Bragg reflector (DBR) The general structure of a SESAM using DBR is plotted schematically in Fig 1.10 The barriers between the quantum wells, the cap layer, and the spacers are all transparent at the operation wavelength of the SESAM If the incident light has a longer wavelength than that of the absorption edge of the absorber, it will pass through the absorber and be bounced back to the air by the DBR without any significant absorption But the light at a proper wavelength will be absorbed The absorption is saturated for high-intensity light, because electrons in the valance band are exhausted, due to the absorption [Xiang2003] Figure 1.10 Schematic structure of a SESAM [Xiang2003] 25 Chapter Introduction To reduce the adverse non-saturable optical losses during the mode-locking by a SESAM, an ideal DBR with a broad and highly reflective stopband is necessary A high-reflectivity DBR is made by stacking pairs of alternating layers of high and low refractive indices with thicknesses of λ / 4n (where λ is the center wavelength in air and n is the refractive index of the layer material) The reflectivity R at λ is then expressed as: ns n1 N ⎛ ⎜ 1− ( ) R = ⎜ n0 n ⎜ + ns ( n1 ) N ⎜ ⎝ n0 n 2 ⎞ ⎟ ⎟ , ⎟ ⎟ ⎠ (1.3) where n1 and n2 are the refractive indices of high-n and low-n materials, ns and n0 are those of the substrate and incident medium (usually air with n0 = 1), and N is the number of pairs of the layers Eq (1.3) is valid for normal incidence In this work, the alternating SiO2 and Si3N4 dielectric layers grown by plasma enhanced chemical vapor deposition (PECVD) were used for the DBR fabrication, which will be discussed in Section 5.1 The refractive index for our SiO2 and Si3N4 layers were 1.45 and 1.97, respectively To design a DBR, the following factors should be considered: 1) DBR materials should be transparent at the operation wavelength; 2) The index contrast Δn = n − n1 should be as large as possible; 3) Higher N yields higher R; 4)The reflectivity bandwidth of the bottom DBR may limit the bandwidth of a SESAM A typical nonlinear reflectivity property of a SESAM is shown in Fig 1.11 As can be observed, when the incident light fluence (the energy per unit area) increases, R 26 Chapter Introduction increases nonlinearly, and saturates at high intensities Then the reflectivity difference between the saturated and the low-intensity values ( ΔR ) is called the optical modulation depth of a SESAM, which is an important parameter for a SESAM Figure 1.11 Nonlinear reflectance from a SESAM [Xiang2003] It has been shown that the width of a mode-locked pulse ( τp ) is inversely proportional to the modulation depth: τp ∝ (ΔR) − β ( β > ), (1.4) where β is a numerical factor [Jung1997] It indicates that the larger the modulation depth, the shorter the pulses will be The modulation depth can be increased by increasing the number of quantum wells once the quantum well composition and width are fixed Another important parameter for a SESAM is the saturation fluence, which is analogous to the saturation intensity Isat in Eqs (1.1) and (1.2) Eq (1.1) is actually valid for fast saturable absorbers only, and the Isat value for a fast saturable absorber can be obtained by curve-fitting Whereas for slow saturable absorbers, the saturation 27 Chapter Introduction behavior is described in terms of the saturation fluence Fsat by the following equation: qp ( Fp ) = q 0[1 − exp( − Fp Fsat )]( ) Fsat Fp (1.5) where Fp is the energy fluence, q0 is a constant, and qp ( Fp ) is the saturable loss of the absorber q0 is almost identical to the modulation depth of the absorber The qp value can be obtained experimentally from the plot of R versus Fp as shown in Fig 1.11 using the following equation: R=1- qp - qns (1.6) where qns is the non-saturable loss, which is caused by scattering on interfaces or surfaces, and/or the additional absorption from defect states or reflectivity losses from the bottom mirror Hence, given an R versus Fp plot, the value of Fsat can be obtained by curve-fitting according to Eqs (1.5) and (1.6) [Paschotta2001; Paschotta2003] If a saturable absorber, without the mirror and/or the AR coating, is studied, its nonlinear property can then be characterized through the nonlinear transmittance behavior As indicated in Fig 1.12, the modulation depth of a saturable absorber is now the transmittance difference ( ΔT ) between the saturated and the low-intensity values For slow saturable absorbers, similarly, the saturation behavior can be described in terms of the saturation fluence Fsat by Eq (1.5) The qp value is now obtained experimentally from the plot of T versus Fp as shown in Fig 1.12 using the following equation: T=1- qp - qns (1.7) where qns is the non-saturable loss 28 Chapter Introduction Figure 1.12 Nonlinear transmittance from a saturable absorber [Xiang2003] Sections 1.4.1 and 1.4.2 have introduced the basic concept of a saturable absorber, the principle of the passive mode-locking by a saturable absorber and the key properties of a SESAM In the next section, we will briefly review the historical development of the passive mode-locking by SESAMs 1.4.3 Historical review on the passive mode-locking by SESAMs During the last two decades, passive mode-locking by SESAMs has been successfully developed in solid state lasers, semiconductor lasers and fiber lasers in a wide region of wavelengths to produce ultra-short optical pulses The initial idea of passive mode-locking by saturable absorbers was demonstrated in semiconductor diode lasers in the early 1980’s [Ippen1980; Silberberg1984; Ziel1981] The recovery times of the saturable absorbers were reduced by the damage induced either during the aging process [Ippen1980], by proton bombardment [Ziel1981], or by multiple quantum wells [Silberberg1984] Pulses as short as 650 fs with the center wavelength of 813 nm were achieved in GaAs buried 29 Chapter Introduction optical guide semiconductor lasers [Ziel1981] But at this early stage, no SESAM structure was fabricated and external mirrors were used Also, at that time, it was believed that pure continuous-wave (CW) mode-locking by saturable absorbers could not be achieved in solid state lasers This was owing to the long upper state lifetimes (microsecond to millisecond range) of most solid state lasers, which would result in high tendency for Q-switching instabilities [Keller19961] With the subsequent advent of bandgap engineering and the modern semiconductor growth technique, accurate parameter control during the saturable absorber fabrication was realized In 1992, Keller U et al reported the first self-starting and stable passive mode-locking of the diode-pumped solid state laser with an intracavity SESAM; and a cavity design named antiresonant Fabry–Perot saturable absorber (A-FPSA) was demonstrated [Keller19921] This was also the first demonstration of SESAM-based passive mode-locking Since then, diode-pumped solid-state lasers using SESAMs have become dominant in the area of high-performance passive mode-locking; and many new SESAM designs [Keller1999] have been developed These new designs provide not only saturable absorption but also negative dispersion compensation For example, it is known that overall dispersion of an optical resonator is a major factor in determining the duration of the pulses emitted by the laser; and a pair of prisms are usually arranged to produce net negative dispersion to balance the positive dispersion of the laser medium But for the dispersion compensation saturable absorber mirror (D-SAM), the structure is more compact It replaces the traditional prism pairs for dispersion by special cavity resonator designs, 30 Chapter Introduction combining both saturable absorption and dispersion compensation within one integrated device [Kopf1996] Today, a large variety of all-solid-state ultrafast lasers with pulse durations ranging from picoseconds (ps) to less than 100 fs have been successfully fabricated Table 1-2 [Keller2006] summarizes the high-repetition-rate results achieved for various solid-state lasers by passive mode-locking with SESAMs In these lasers, the center lasing wavelengths vary from ~ 850 nm to ~ 1.55 μm Apparently, these wavelengths are all beyond the visible region However, considering the applications in high density data storage, ultra-short laser pulses in much shorter wavelengths are preferred Table 1.2 Passively mode-locked diode-pumped solid-state lasers: high-repetition rate results (i.e >1 GHz) [Keller2006] Gain ML λ0 τp Pav (mW) frep (GHz) Reference Cr:LiSAF SESAM 857 nm 146 fs Kemp2001 Cr:YAG KLM 1.54 μm 115 fs 150 2.64 Tomaru2001 1.52 μm 68 fs 138 2.33 Tomaru2003 1.52 μm 200 fs 82 0.9, 1.8, 2.7 Collings1997 1.52 μm 75 fs 280 Mellish1998 SESAM Nd:YLF SESAM 1.34 μm 21 ps 127 1.4 Zeller2006 Nd:YVO4 SESAM 1.064 μm 8.3 ps 198 13 Krainer1999 6.8 ps 81 29 Krainer19991 4.8 ps 80 39, 49, 59 Krainer2000 2.7 ps 288 40 Lecomte2005 2.7 ps 65 77 Krainer20001 2.7 ps 45 157 Krainer2002 13.7 ps 2.1 10 Lecomte2004 ps 45 Spühler2005 7.3 ps 40 10 Spühler2005 18.9 ps 3.46 0.37–3.4 Kong2004 12 ps 500 9.66 Krainer2004 4.4 ps ~ 60 2.5–2.7 Krainer2004 1.534 μm 3.8 ps 12 10 Krainer20021 full C-band 1.9 ps 25 25 Spühler2003 1.34 μm Nd:GdVO4 Er:Yb:glass SESAM SESAM 1.064 μm 31 Chapter Introduction 1.534 μm 4.3 ps 18 40 Zeller2003 1.534 μm 1.7–1.9 ps > 20 8.8–13.3 Erny2004 1.533 μm ps 7.5 50 Zeller2004 1.536 μm 3.2ps 10 77 Zeller2007 ML: mode-locking techniques λ0: center lasing wavelength τp: measured pulse duration Pav: average output power frep: pulse repetition rate For semiconductor lasers, while the initial idea of mode-locking by saturable absorbers was demonstrated in such lasers, the first demonstration of SESAM mode-locked semiconductor lasers was only reported in 2000 after the introduction of the vertical cavity surface emitting lasers (VCSELs) [Hoogland2000] This was mainly because the early edge-emitting semiconductor lasers had limited average and peak powers due to the limited mode area; and the newly developed VCSELs have large gain cross-section, which suppresses the Q-switching instabilities Therefore, VCSELs are ideally suitable for high-repetition-rate mode-locking in combination with high average output powers [Keller2006] Taking advantage of the SESAM designs already developed in solid state lasers, semiconductor lasers mode-locked by SESAMs have experienced rapid development During the past few years, the pulse width and output power of the SESAM mode-locked VCSELs have improved to 486-fs pulses at 10 GHz with 30 mW at the center-wavelength of 1034 nm [Hoogland2005] and 4.7-ps pulses at GHz with 2.1 W at the center-wavelength of 957 nm [Aschwanden2005] The shortest center lasing wavelength achieved in SESAM mode-locked semiconductor lasers is 950 nm, which was achieved by an In0.15Ga0.85As single quantum well saturable absorber [Häring2002] But this shortest wavelength of 950 nm is still far away from the visible region 32 Chapter Introduction Moreover, since 2004, the novel SESAMs based on quantum dot saturable absorbers (QD-SESAMs) have been developed In QD-SESAMs, the saturable absorbers are integrated into the VCSELs’ gain structures Passive mode-locking with the same mode area in the gain and the absorber has been demonstrated for the full wafer-scale [Lorenser2004] For fiber lasers, the issues of high repetition-rate, high peak energy and large material dispersion were initially the critical challenges However, with the recent development of SESAMs, such problems have been circumvented [Okhotnikov2004] Fiber lasers equipped with SESAMs can easily generate ultra-short pulses with high peak energy; and the material dispersion is compensated with the proper SESAM designs Up to the present, fiber lasers passively mode-locked by SESAMs can cover a wide wavelength range of 0.9 μm < λ < 4.8 μm, using environmentally stable tunable light sources [Barnett1995; Collings19971; Guina2001; Xiang2002; Herda2006; Erny2007] Another advantage is that, a fiber laser system can be packaged in a compact form For example, a compact fiber laser package equipped with a SESAM has been fabricated by Fianium Ltd (UK), with a 120 × 100 mm2 footprint and 110 g in weight, producing 500 fs mode-locked pulses at an average power of 15mW In contrast to the great development of the SESAM mode-locking in the above solid state lasers, semiconductor lasers and fiber lasers, the development of SESAM mode-locking in dye lasers have been very limited Although there are a few reports regarding the dye lasers with saturable absorbers [Dietel1980; Kuhlke1983; Michailov1989; Hébert1992], to the best of our knowledge, there is so far only one 33 Chapter Introduction demonstration of the SESAM mode-locked dye laser [Noh2000] This is probably because this type of laser can easily generate short pulses and the use of liquid dye is inconvenient and would result in a short lifetime due to degradation Nevertheless, it should be noted that, a 56-fs pulse at a center-wavelength of 580 nm has been generated from a synchronous pumped Rh6G dye laser using a GaAs-based SESAM [Noh2000] This is a successful demonstration of direct generation of ultra-short visible pulses by SESAM mode-locking While the ultra-short pulses have been generated in a wide range of wavelengths by SESAM mode-locking, the direct generation of ultra-short pulses in the visible wavelength has only been reported in dye lasers Moreover, it should be noted that the direct generation of ultra-short blue/UV pulses by passive mode-locking has not been demonstrated yet; and the blue/UV ultra-short pulses have so far been obtained mainly by frequency conversion methods from infrared solid state lasers, such as Ti: sapphire lasers [Wilhelm1997] and Cr:LiSAF lasers [Agate2004] One of the major problems is the difficulty in monolithically fabricating broadband high-reflective GaN-based DBRs, due to the lack of suitable semiconductor DBR materials lattice-matched to GaN On the other hand, for data storage applications, in order to achieve high speed operation as well as high density storage, high-repetition-rate ultra-short optical pulses in the blue/UV wavelength region have become indispensable Therefore, the fabrication of the SESAMs applicable for the blue/UV wavelength region is essential If successfully demonstrated, such SESAMs can be used to passively mode-lock GaN-based semiconductor lasers to generate ultra-short blue/UV 34 Chapter Introduction pulses 1.5 Motivations and the synopsis of the thesis Considering the applications in data storage mentioned in Section 1.4.3, there is an intense need for direct generation of ultra-short blue/UV optical pulses by passive mode-locking It has also been highlighted that mode-locking by SESAMs is one of the best methods to achieve the ultra-short pulses The main aim of this work was to design and fabricate saturable absorbers and SESAMs operating in the blue region GaN-based materials were chosen to fabricate such saturable absorbers, because, as mentioned in Section 1.1, such materials are good candidates for fabricating optoelectronic devices operating in short-wavelength region In view of the large absorption of GaN-based materials in the UV region, this study is only focused on the saturable absorbers and SESAMs operating in the blue region The study consists of the following four parts: 1) Saturable absorbers with GaN-based quantum structures (Chapter 3) GaN-based quantum structures, comprising of quantum wells and quantum dots, were used as the active region of the saturable absorbers The nonlinear transmittance properties of InGaN/GaN quantum wells were studied The quantum wells grown under optimal conditions behave as slow saturable absorbers because of their relatively good crystal quality In order to achieve efficient pulse shaping, InGaN/GaN quantum well saturable absorbers with reduced GaN buffer thickness were also fabricated Dislocations were purposely introduced into their active regions so as to reduce the absorption recovery time Dislocation density, nonlinear transmission 35 Chapter Introduction property and recovery times of these saturable absorbers were studied Also, because of the reduced buffer thicknesses, these InGaN/GaN quantum well samples showed high density of large V-pits, which could be related with their ultra-short absorption recovery time So the effects of V-pits on the morphologies and emission properties of InGaN/GaN multiple quantum wells were studied Owing to the unique “zero-dimension” property of quantum dots, they were introduced into the active region of the saturable absorbers A GaN-based multi-layer quantum dot saturable absorber was fabricated by a simple self-assembly method The nonlinear property and recovery dynamics of the quantum dot saturable absorber were studied and compared with those of the InGaN/GaN quantum well saturable absorbers 2) Design and fabrication of GaN-based SESAM (Chapter 4) Because of the small refractive index difference between the GaN-based materials, GaN-based monolithic DBRs exhibit narrow stopbands and unsatisfactory maximum reflectivity Therefore, instead of the monolithic DBR, a non-monolithic SiO2/Si3N4 dielectric DBR was deposited onto the saturable absorbers to form the SESAM The DBR and anti-reflective (AR) coatings were studied and calibrated separately before being incorporated with the saturable absorber The characteristics of the non-monolithic SESAMs were studied 3) Optimization of GaN-based SESAM (Chapter 5) To deal with the drawbacks of the as-grown GaN-based SESAM, a modified SESAM structure was first designed by simulation Experiments were then conducted according to the simulation The SESAM after structure modifications was 36 Chapter Introduction characterized to examine whether the modifications affected the characteristics of the SESAM The results of the present study should be of great importance as there is so far no report regarding saturable absorbers and SESAMs operating in the blue wavelength region The characterization results of the saturable absorbers should contribute to a better understanding of the structural property, optical property and nonlinear property of GaN-based quantum wells and quantum dots Also, the results of SESAM fabrication and structure modifications may provide useful information for GaN-based optoelectronic device processing Moreover, this study regarding saturable absorbers and SESAMs in the blue region may provide a basis for the future research on passively mode-locking GaN-based blue lasers by such SESAMs In this work, the quantum well and quantum dot saturable absorbers were grown by metal organic chemical vapor deposition (MOCVD), and the SiO2 and Si3N4 dielectric layers were deposited by plasma enhanced chemical vapor deposition (PECVD) Many characterization techniques, such as photoluminescence (PL) spectroscopy, X-ray diffraction (XRD), transmission electron microscopy (TEM), were used to study the optical property, structural property and the crystal quality of the samples The next chapter will introduce these growth and characterization techniques in detail 37 ... properties of the InGaN quantum structures, including InGaN/ GaN quantum wells and InGaN quantum dots 1. 3 InGaN/ GaN quantum wells and quantum dots InGaN is an important alloy for fabricating III-nitride... Electron mobility μ (cm2 V -1 s -1) Thermal conductivity χ (W K -1 cm -1) 9.0 4.0 9.7 3.0 12 .8 0.4 11 .8 0.25 3.0 2.0 2.0 1. 0 13 50 800 6000 13 00 1. 3 4.9 0.5 1. 5 Melting point (°C) 27 91 Sublimes T >18 27 12 38... following equation: T =1- qp - qns (1. 7) where qns is the non -saturable loss 28 Chapter Introduction Figure 1. 12 Nonlinear transmittance from a saturable absorber [Xiang2003] Sections 1. 4 .1 and 1. 4.2