Diode pumped solid-state lasers

The first solid-state laser was flash-lamped-pumped chromium-doped sapphire crystal now referred to as the ruby laser. Following the demonstration of the ruby laser by Maiman [35-38], Nd:CaWO4 laser and Nd:YAG laser were developed [39].

Flashlamp-pumped solid state lasers were the preferred source for nonlinear optical experiments and applications for the next two decades. Early 1962, laser diode pumped solid-state lasers were proposed [40], and demonstrated [41]. The first laser diode pumped solid-state laser was operated at cryogenic temperatures, but the promise of the approach was recognized. It was not until 1978, with the demonstration of an 1 W cw laser diode bar [42], the laser diode pumped solid state laser began their rapid evolution that continues today.

Diode laser pumped solid state lasers are efficient, compact, all solid state sources of coherent optical radiation. The primary commercial example of a conventional solid state laser is Nd:YAG. Nd:YAG lasers at 1.064 µm and can be frequency doubled to 532 nm, tripled to 355 nm and quadrupled to 266 nm. CW Nd:YAG lasers are available in power levels up to kilowatts watts and pulsed Nd:YAG lasers are available with pulse energies up to a few joules per pulse.

Nd:YAG lasers can also be operated Q-switched or mode-locked. Q-switched Nd:YAG lasers are frequently used for laser machining processes that require rapid removal of relatively small quantities of material. Examples include trace and link blowing in the electronics industry, laser marking, and laser hole drilling. An important market niche is the use of Q-switched and frequency-doubled Nd:YAG lasers for laser marking of silicon wafers. A developing market niche is the use of Q-switched Nd:YAG lasers for removal of unwanted body hair [43].

Frequency-doubled cw Nd:YAG lasers compete with argon-ion lasers for the moderate power green laser market. Thus, Nd:YAG lasers compete with argon ion lasers in such applications as printing, display technology, stereolithography, and retinal photocoagulation. A developing application for Nd:YAG lasers is laser-enhanced bonding [44], the process uses a laser to drive a polymer adhesive into the material being joined. The process provides a replacement for more traditional sewing, taping, gluing, or ultrasonic welding. Specific applications for the technology include bookbinding, laminating, textiles, injection molding, and carpeting.

2.2 Pumping source

Solid-state laser development has been paced by the improvement and discovery of pumping sources. The most efficient laser pump lamp will produce maximum emission at wavelengths which excite fluorescence in the laser material, and produce minimal emission in all spectral regions outside of the useful absorption bands. The helical lamp, used to pump the first ruby laser, was replaced by the linear flash lamp and discharge arc lamp that are now used to pump virtually every neodymium-doped yttrium-aluminum-garnet and neodymium glass laser system in the world. The latest advance in solid-state laser technology promise to be improved pumping by means of diode lasers and diode laser arrays. The recent and rapid advances in the power and efficiency of diode lasers and diode laser arrays and their application to the pumping of solid-state lasers have led to a renaissance in solid-state laser development. Advanced solid-state lasers pumped by diode lasers will make such diverse applications as coherent radar for global wind measurements, semiconductor circuit repair, and all solid state color video projection possible.

Diode laser pumped solid-state laser provides spectral brightness

“amplification” that is essential for many applications that require a high degree of

temporal and spatial coherence. It offers a number of advantages over flash lamp pumping or the diode directly:

Firstly, flash-lamp pumping efficiency is limited by the broad spectral emission of the lamp and the less efficient absorption of the lamp radiation by the solid state laser medium, but the diode laser can efficiently emit optical radiation into a narrow spectral band. Excess heat and power fluctuations of the lamp also degrade the solid state laser performance. Secondly, the diode laser is essentially a cw device with low energy storage capability, whereas the solid-state laser can store energy in the long-lived metastable ion levels. The stored energy can be extracted to provide peak power levels that are orders of magnitude greater than from the diode laser itself, for example the Q-switched laser. Thirdly, the solid state laser can collect the output from several diode lasers to provide greater average power than is available from a single diode laser. Fourthly, the diode laser pumped solid-state laser can operate at a verity of wavelengths not accessible with diode lasers. Fifthly, the linewidth of the diode laser pumped solid-state laser is fundamentally orders of magnitude less than that of the diode laser source. Sixthly, the solid-state laser source also emits optical radiation in a diffraction limited spatial beam that is easily focused into a fiber or to a small spot.

Significant progress has been made in developing monolithic, linear laser-diode arrays which have become the building blocks for solid state laser pumps. Output power, slope efficiency, laser threshold wavelength control have all been dramatically improved due to a combination of new structures and advanced growth techniques. In particular, epitaxial growth based on metal organic chemical vapor deposition (MOCVD) allows close control of material composition, layer thickness and device geometry.

Early semiconductor lasers were constructed from n type and p type layers of the same semiconductor material. Such lasers are called homostructure lasers. There are many problems associated with the simple p-n junction lasers which can be attributed to the fact the same material for both the p and n regions were used. Two of the critical ones are (1): The injected minority carriers are “free” to diffuse where they will, a fact that dilutes the spatial distribution of recombination and thus the gain. (2):

There is very little guiding and confinement of the electromagnetic wave being amplified. There is a small bit of wave guiding caused by the slight decrease of refractive index on the n side and on the p side due to the small change in Eg with acceptor doping. However, these changes are very small, and the unpleasant fact is that the central part of the wave may be amplified with the tails, which extend into the noninverted regions, being attenuated. Both of these problems can be ameliorated by the use of heterostructures to form the active portion of the laser.

Heterostructure lasers require layering these different materials. This is a very complex problem, as the materials have different physical properties. Perhaps the most important of these properties is the lattice spacing. If the materials do not have the same lattice spacing, then dislocations can appear in the semiconductor laser. In addition to the structural difficulties imposed by the dislocations, dislocations can also be highly detrimental to semiconductor laser operation as they can serve as a nonradiative sink for carriers. The majority of commercial heterostructure semiconductor lasers are fabricated from semiconductor materials in columns III and V of the periodic table. Column III is born (B), aluminum (Al), gallium (Ga), indium (In), and thallium (Tl), Column V is nitrogen (N), phosphorus (P), arsenic (As), antimony (Sb), and bismuth (Bi). Common laser materials include virtually all combinations of Al, Ga, and In, with P and As. Some work has been done with B, Al, and Ga, with N; as well as Al, Ga, and In with Sb. Very little has been done with Tl

and Bi. These are junctions between two dissimilar materials such as GaAs with AlxGa1-xAs, with x being the fraction of gallium being replaced by Aluminum. It is a fortuitous fact of nature that GaAs and AlAs semiconductors have almost identical lattice constants as shown in Table 2.1 and thus can be mixed and can be grown on top of each other with little strain involved and a very small density of traps at the interface. This metallurgical fact is critical to the success of making the junction.

Table 2.1: Properties of common semiconducting materials

Band gap ε_r µ_e µ_h m_e^∗ m₀ Lattice

constant

Material (eV) -- (cm²/V-sec) -- (Å)

C(i) 5.47 5.7 1800 1200 0.2 3.5668 GaP(i) 2.26 11.1 1600 100 0.82 5.4512

AlAs(i) 2.16 10.9 180 -- -- 5.6605

GaAs(d) 1.43 13.2 8500 400 0.067 5.6533 InP(d) 1.35 12.4 4600 150 0.077 5.8686

Si(i) 1.12 11.9 1500 450 1.1 5.4309 GaSb(d) 0.72 15.7 5000 850 0.042 6.0957

InAs(d) 0.36 14.6 33,000 460 0.023 6.0584 InSb(d) 0.17 17.7 80,000 1250 0.0145 6.4794

Note: i stands for indirect bandgap, d stands for direct bandgap.

A very useful diagram for visualizing lattice match in heterostructure lasers is the energy versus lattice constant diagram. An example of such a diagram for III-V materials is given in Fig. 2.1. Notice that only the AlAs-GaAs system is lattice-matched across the entire compositional range. This is one major reason for the

widespread use of AlGaAs/GaAs heterostructures on GaAs substrate for semiconductor laser diodes [45].

Figure 2.1. The energy gap versus lattice constant diagram for compound semiconductors [45].

In a ternary alloy, the lattice constant is linearly dependent on the composition.

The linear relationship of the lattice constant generally holds for quaternary alloys too.

However, other parameters, of a mixed alloy do not, in general, obey this linear relationship. The bandgap, for example, is usually given by an empirical relationship,

( )

E_g = _go + + , where E is the bandgap of the lower bandgap binary, b is a _g fitting parameter, and c is called the bowing parameter, which may be calculated theoretically or determined experimentally. It is not only the variation of bandgap with composition, but also the energy variation of the higher-lying bandstructure with

composition, that is extremely important for the understanding of material properties.

As the percentage of aluminum is increased, the band gap increases and the index of refraction goes down, and this asynchronous behavior is true for quaternary alloy combinations also. Figure 2.2 illustrates the dependence of the band gap and the index of refraction on the mole fraction of Al substituted for gallium. These graphs are plots of the analytic expressions for Eg and the index given by reference [46-49].

(

)

[

(

)

]

Wavelength (nm) Band gap Eg(eV)

GaAs x AlAs

Figure 2.2. (a) Compositional dependence of direct and indirect conduction band minima in the AlxGa1-xAs mixed crystals, (b) dependence of the index of refraction on the fraction, x, of aluminum in the composition.

Figure 2.3 illustrates a laser using a double heterostructure geometry and is shown biased in the forward direction. Shown also is the variation of the refraction index and the light intensity in a plane perpendicular to the junction. Note that the electrons injected from the n-type AlxGa1-xAs material are confined to and recombine in lower band-gap p-type GaAs.

2.9

Figure 2.3. (a) The band diagram for a forward-biased heterostructure, (b) the refractive index, and (c) a sketch of the light intensity in the vicinity of the active region.

(a)

(b)

(c)

In designing a diode-array-pumped laser system, we must select the array configuration for the proposed application. Several common configurations are available. These include small linear arrays with a length of 100 or 200 µm, 1-cm long array bars, and stacked diode bars. For high power lasers, the 1-cm linear bars must be combined into modules at some level to reduce complexity of the electronic drivers, heat removal and mechanical structure. Stacking of diode arrays is done manually or in a semi-automated mode at the present time. This process allows large-area arrays with dimensions up to about 1 cm² to be fabricated. However, the process is labor intensive. The cost of the arrays also increases faster than linearly as the size of the array increases. This is because the selection of wavelength and current threshold of linear bars that constitute the arrays becomes more critical as the size of the array increases. All of the bars in the array must be matched for optimum performance. A trade off exists between increased design complexity using arrays which consist of only a few bars and reduced efficiency and increased cost with large area arrays comprised of a large number of bars. As a result of this trade off, a common module for high power lasers is an array consisting of 5 one-cm bars. Such an array has an aperture of about 10×1 mm².

Semiconductor laser technology has produced an amazing variety of new device structures in the past decade. Overviews of this technology made possible by sophisticated growth techniques such as metallorganic chemical vapor deposition and molecular beam epitaxy can be found in [50-51]. The key design features of diode lasers by describing the single quantum well separate confinement heterostructure (SQW-SCH) will be illustrated, because it is the most widely employed design for solid state lasers pumps.

The active layer shown in Fig 2.4 is sandwiched between two pairs of layers having a different concentration of Al. The main purpose of the innermost pair of

layers is to confine the carriers to the active region, whereas the purpose of the outer layers is to confine the optical beam. In a diode laser, recombination of current carriers takes place in a thin active layer which separates p- and n-doped regions.

Figure 2.4. Gain-guided, single quantum well separate confinement heterostructure stripe laser. Thickness of layers is greatly exaggerated. Optical mode cross-section is actually 5µm wide and 0.5 µm high.

Progress in high power laser diodes for solid state laser pumping has emphasized the development of quantum well structures in which laser emission is produced in very thin epitaxial layers (quantum wells) less than 0.02 µm thick.

A quantum well is a thin layer of semiconductor located between two layers with larger bandgap. Electrons in the quantum well layer lack the energy to escape, and cannot tunnel through the thicker surrounding layers. The quantum well layer in Fig 2.4 has a composition of Ga and Al indicated by x which defines the emission wavelength. A higher Al concentration increases the bandgap and shifts the output towards shorter wavelength. The quantum well structure is sandwiched between two

thick layers of composition which contains a higher concentration (y>x). The higher Al concentration increases the bandgap, thereby defining the quantum well, and the large thickness of the layer prevents tunneling of the carriers out of the quantum well.

Single quantum well devices exhibit slightly lower lasing thresholds and slightly higher differential quantum efficiencies, making them preferable for high power lasers.

The thin active region incorporating a quantum well active layer structure provides low threshold and high electrical-to-optical efficiency. However, such a very small emitting surface poses one problem since the power from a diode laser is limited by the peak flux at the output facet. One can increase the output from these devices by spreading the beam over an area which is larger than the active layer or gain region. The standard approach is to deposit layers next to the active layer, each of which has a slightly lower refractive index than the active layer, thus making a wave guide of the active layer. In the separate carrier and optical confinement heterostructure design shown in Fig. 2.4, the refractive index boundary is abrupt, at the layer boundary. Alternatively, the refractive indices of the surrounding layers may be graded, forming a graded-index and separate carrier and optical confinement heterstructure.

In the SQW-SCH structure, the overlap between the optical mode and the gain region is only about 4%, which results in a large effective aperture of the laser on the order of 0.3 to 0.5 µm. This substantially reduces the energy density at the output facet and enhances reliability by minimizing characteristic facet damage.

In the GaAlAs diode laser, the wavelength changes with temperature according to 0.3 nm/^oC. More powerful arrays with outputs up to 3W are on the market, however the greater aperture width of these devices makes coupling of the pump beam into the small mode volume of the solid state laser fairly inefficient.

The key concept in diode-pumped solid-state lasers is to make the output mode of the diode laser precisely match the operating mode (usually TEM00) of the laser.

The simplest possible configuration is to end pump a solid-state laser rod with a single laser diode as shown in Fig. 2.5.

Figure 2.5. In the end pumped geometry, the light is directed down the length of the rod.

Commercially available diode arrays suitable for endpumping contain a small number of stripe lasers on a single chip, as shown in Fig. 2.6.

Figure 2.6. Multistripe semiconductor laser array geometry.

A typical example is a 20 stripe gain guided, SQW-SCH laser array made by Spectra Diode Laboratories (Model SDL-2460) as shown in Fig. 2.7. The diode array contains twenty 5-µm-wide stripe lasers on 10 µm centers which produce a total output of 1W in the cw mode. The output is emitted from an area of 200 * 1.0 µm with a beam divergence of 40^o by 10^o.

Figure 2.7. The (a) outlook and (b) cross-section view of 1-W laser diode (SDL-2460).

Because laser diodes tend to lase with an output mode promising a highly elliptical beam due to the difference between the horizontal and vertical dimensions of the lasing region of the diode, there are two major problems associated with this output mode. The first problem is the large astigmatism. The second is the huge divergent angle. Typically, a large-aperture lens is placed close to the laser in order to capture as much energy as possible. Then, some type of astigmatism compensation is used. Finally, the beam is refocused into the optimal configuration for mode matching into the rod. Traditional techniques for compensating for the astigmatism include coupling the beam into a fiber or using a GRIN lens. However, both of these techniques intrinsically strip off part of the beam as the beam mode matches into the optical element. Thus, fiber or GRIN lens coupling techniques are inefficient. More efficient techniques include using aspherical lenses (expensive) or anamorphic prisms [52] (cheaper but more complex to assemble). The anamorphic prism technique is

(a) (b)

probably the most commonly used in commercial diode pumped lasers because it is highly efficient and relatively low cost as shown in Fig. 2.8.

Figure 2.8. Anamorphic prisms are often used for compensating for the astigmatism of a laser diode.

Some microlens options are also offered. One is a simple lens (actually a round optical fiber). The other is an aspheric design that provides diffraction limited performance. These lenses are cylindrical lenses that modify the divergence of the fast (high divergence) axis of the laser diode: the lenses have no effect on the slow (low divergence axis). The lenses may be focused to attain best collimation, or may be focused to equalize the divergence of the slow and fast axes as shown in Fig. 2.9.

Figure 2.9. (a) A standard offering of the round simple cylindrical microlens of 125 µm diameters, (b) the aspheric microlens which designed to provide a diffraction limited 8X magnification of laser diode fast axis (Polaroid laser diode manufacturing

& development product literature).

End-pumping of laser rods with diode laser arrays (rather than single diodes) is also done. However, some care must be taken in these designs, as it is much more difficult to mode match laser arrays into the relatively small end face of the rod.

Therefore, the additional power advantage achieved with the laser diode array may be offset by the mode matching disadvantages of the geometry. End-pumping of laser rods with fiber-coupled laser arrays is an effective solution to the design problem of mode matching laser arrays into the small end face of the rod. However, creating the

submount

laser

cylindrical lens beam

(a)

Aspheric lens

(b) submount

fiber bundle requires coupling a fiber into the end of each individual laser diode in the

fiber bundle requires coupling a fiber into the end of each individual laser diode in the