• Ei tuloksia

2 THEORETICAL BACKGROUND

2.2 Semiconductor lasers

The semiconductor lasers can be named as the most ubiquitous group of lasers. The commonly used laser pointers, lasers utilized in CD players and optical fiber communications belong to this group. They can be classified as a part of a solid-state laser cluster due to the nature of the active media. In most cases semiconductor lasers are electrically pumped, however in certain cases optical pumping is applied instead.

VCSEL arrays used in this research are an example of electrically pumped semiconductor lasers. Their properties are reviewed in the following subsections.

2.2.1 VCSELs structure and design

Operating the theoretical background defined in the previous section we would now be able to characterize VCSELs. Laser diodes (LDs) are semiconductor lasers meaning the gain is provided by the flow of electrical current through p-n junctions. This flow causes the electron – holes recombination and consequent creation of photons. All LDs can be divided into two groups (Figure 4) based on the direction of the emission relative to the wafer surface. In the most common type, EELs, the emission takes place parallel to the wafer surface. Reflection takes place at the coated end facets of the semiconductor wafer which serve as mirrors. In VCSELs, however, emission takes place perpendicular to the wafer surface.

Figure 4: Types of lateral confinement in laser diodes [7]

Active region in VCSELs is thus clearly limited by the thickness of the semiconductor wafer, which is a thin slice of semiconductor [10] formed by highly pure and nearly defect free crystalline material. The active region is often realized as a quantum well structure (quantum well - 1D confinement, quantum wire – 2D, quantum dot – 3D). It is formed as a thin semiconductor medium embedded between two semiconductor layers of a material with higher band gap. For instance, InGaAs embedded in GaAs or GaAs embedded in AlGaAs. The thickness of a quantum well would normally lie in the range of 5-20 nm [11]. Such structures can be grown with epitaxy growth techniques, such as:

molecular beam epitaxy (MBE), characterized by sharp layers interfaces of high quality, or metal-organic chemical vapor deposition (MOCVD), which produces excellent quality of layers purity and crystallinity [12].

Another consequence of vertical confinement would be a rather short pathlength, which limits the round-trip laser gain. The resonator with low losses would thus be required.

Normally, distributed Bragg reflectors (DBR) are used as resonators in VCSELs. DBR is a mirror structure consisting of a sequence of layers of alternating two optical materials with high and low refractive indices. The design of the DBR is specific for each wavelength. Most frequently quarter-wave mirrors are used, meaning each optical layer thickness corresponds to ¼ of the laser emission wavelength [13]. The reflectivity of the mirror reaches up to 99%. The fact of DBRs being embedded into the wafer structure provides VCSELs with a set of superior traits in comparison to EELs.

2.2.2 VCSELs characteristics

The structure specifications discussed in the previous subsection lead to certain advantages of vertical emission structure. Firstly, monolithic wafer design i.e., the absence of the necessity for additional facet coating eliminates the need for waver cleavage before testing. This speeds up the processing, testing and overall development

process. The difficulty of handling chips with cleaved outer sides is also eliminated. The overall production costs are thus decreased significantly.

Another advantage lies in a superior laser emitting area. Its surface fully depends on the mask design and is not limited to the wafer thickness as in EELs. Combined with the generally lower power of a single VCSEL the optical intensity defined with

𝐼 = 𝑃 𝜋𝑤2

2

, (2)

where 𝐼 is optical intensity, 𝑃 is optical power, 𝑤 is Gaussian beam radius, is lower as well.

This eliminates the challenge of catastrophic optical damage (COD), which is a limitation factor for EELs caused by an overload in power density. The melting recrystallization of the semiconductor material at the laser facets is thus not a threat to VCSELs.

Lower optical intensities also enable operation at higher temperatures. Thus, the temperature sensitivity of the whole laser structure decreases. This eases the testing and operating condition requirements as well as increases the scope of application possibilities. Thin active area contributes to a narrow emission spectrum. This leads to the impossibility of several modes formation. Single mode operation is a valuable property required in various applications. As an example, they can be utilized for optical pumping in the cases where energy supplied by the laser should correspond very precisely to the energy gap of the pumped material.

The aperture design plays a key role in defining beam’s profile properties. Since the aperture is established completely by the mask design – it can be set to be circular in contrast to rectangular aperture in EELs (Figure 4). This ensures low beam divergence and high beam quality as seen from Figure 5. Moreover, the beam with such properties can be easier coupled to an optical fiber which increases its applicability.

Figure 5: VCSEL beam profile compared to LED and EEL [14]

Finally, one of the most appealing VCSELs properties contributing to their market popularity is the possibility to arrange them into 2D arrays. In this work we are going to assess the need for measurement setup for VCSEL arrays and thus we would take a look at their properties in more detail in the following subsection.

2.2.3 VCSEL arrays

Generally, despite VCSELs advantages, the power produced by a single emitter is quite limited and normally lies in the range of a couple of mW. The possibility of 2D arrays production, however, enables to increase the total output power to the order of 10 W [15].

Array formation techniques have already been applied to EELs (Figure 6), the power output of diode stacks composed of diode bars can reach the order of 100 W. However, their production is much more laborious and time consuming as it requires a sequence of consecutive processing steps.

a) b)

Figure 6: Schematic structure of (a) diode bar and (b) diode stack [16]

VCSELs 2D arrays production is much faster and more cost-effective. The produced power output is directly proportional to the number of single emitters in the array. In Figure 7 we can observe VCSEL arrays designed by II-VI Inc.

Figure 7: High Power VCSEL arrays (II-VI Inc.) [17]

While such qualities as high beam quality and low beam divergence apply to VCSEL single emitters, they are inversely proportional to the number of emitters in the array.

Meaning that by increasing the number of emitters with the aim of achieving higher power output we lose the beam quality. One of the possible solutions is to optimize the efficiency of the single emitters first and only then proceed to array formation. That way maximizing the power and simultaneously minimizing the number of single emitters used.

Another solution is to minimize the spacing between the single emitters since the area of the produced laser beam is higher than simply the sum of all single emitters areas due to the spacing being present. Spacing between single emitters defines pitch of the array.

While as small as possible spacing is desired due to higher beam quality, it might have a poor effect on the power output and efficiency conversion as the density of heat generation would be increased [18]. The optimal solution should thus be found to obtain the best possible efficiency with the lowest beam divergence simultaneously.

The efficiency of VCSEL arrays is comparable to that of EELs and reported to be about 50 %. The highest achieved efficiency for EELs is so far higher, cases of over 70 % have been reported [19], while the highest achieved value for VCSEL array is 63 % [20].

The possible beam profile divergence should be considered when performing laser testing. Beam collimation should be thus performed to avoid losses in power measurements. The absence of COD allows testing with high peak powers in pulsed operations.