Micro-optics is one of the important components of future micro-electro-mechanical systems (MOEMS) (the other two components are microelectronics and micromechanics), sometimes called optical MEMS. Micro-optical elements have the advantages of small size, light weight, flexible design, array realization and easy large-scale replication, and have been successfully applied in various fields of modern optics, such as correcting aberration of optical system, improving imaging quality of optical system and reducing system weight. It is widely used in the field of laser optics to change the wavefront of laser beam and realize beam transformation, such as beam collimation, shaping, optical exchange and optical interconnection. Micro-optical elements can be simply divided into two types according to the propagation mode of light: diffractive optical elements (DOEs) and refractive optical elements (ROEs). Binary optical element (BOEs) is a commonly used diffractive micro-optical element and an important micro-optical element, which approximates a continuous optical surface with a multi-step surface shape. The corresponding design methods of micro-optical elements include diffraction method and geometrical optics method of refraction propagation, such as Fresnel zone method, G-S algorithm, genetic algorithm, ray tracing and so on. At present, mature commercial software such as CODE V, ZEMAX, OSLO, etc. all have the function of optimizing the design of micro-optical components and systems.

2. Manufacturing method of micro-optical element

There are two manufacturing methods of micro-optical elements: mechanical processing and optical processing. The main processing methods are [1]: fiber lens drawing, ultra-precision grinding, molding, diamond turning and so on. The optical processing method is photolithography. The advantage of machining method is simple process, but the disadvantage is that it is difficult to realize array devices and large-scale cheap replication, and it is difficult to make non-rotationally symmetric micro-optical elements, such as cylindrical lenses and micro-optics with arbitrary irregular surfaces. The advantage of optical processing method is that it can realize any irregular lens (especially binary micro-optical elements) and can be copied on a large scale, but the disadvantage is that the process is complex and the environmental requirements are high. Optical lithography can realize binary diffractive micro-optical elements and continuous surface micro-optical elements, mainly including binary optics method, mask moving method, gray mask method, hot melt method and gradient refractive index method. Figure 1 is the processing principle of 8-step binary diffraction micro-optical element in lithography. Using three masks with different frequencies, a micro-optical element with 95% diffraction efficiency is realized through three processes: glue throwing, exposure, development and etching. Fig. 2 is a continuous plane micro-optical array element made by mask moving method. Firstly, the mask is designed according to the required surface shape, and then different exposure amounts of each part are realized by moving the mask during the exposure process. Finally, the surface shape of the photoresist is transferred to the optical surface material through development and reactive ion etching. Gray mask method is to encode the mask according to the surface type required by micro-optical elements, and form the corresponding light intensity transmittance distribution function. Through one exposure and development, the corresponding photoresist surface shape can be obtained, and finally the surface shape on the optical material can be obtained through etching, as shown in Figure 3. The hot-melt method forms the surface shape by shrinking the surface tension of the photoresist after exposure, as shown in Figure 4. Among these methods, the application field of hot melt method is limited, because the surface shape is not easy to control and it is difficult to manufacture irregular surface shape. Although binary diffraction method can realize all kinds of complex surface shapes and is widely used, it is impossible to make large-value micro-optical elements due to the limitation of lithography line width resolution. The mask moving method can make components with larger numerical aperture, but it is difficult to make components without central symmetry or rotational symmetry. Gray-scale mask method is flexible in design, and can be used to fabricate micro-optical elements with arbitrary surface shape. However, due to the large amount of data in the mask fabrication process, it is difficult to accurately control the surface shape. Generally speaking, binary diffraction method is suitable for micro-optical elements with small numerical aperture, and continuous surface method is suitable for fabricating micro-optical elements with large numerical aperture.

3. Characteristics of1semiconductor laser and fiber coupling method

Laser diode and its array are considered as the most promising lasers because of their small size, light weight, high luminous efficiency and easy modulation and integration. High-power semiconductor laser requires that the laser is not a single light-emitting area structure, but that these single light-emitting areas are arranged in a strip chip or stacked array according to certain rules. Fig. 5 shows a typical high-power strip array semiconductor laser.

Schematic diagram of luminous cross section of device. The special structure of semiconductor laser makes its divergence angle larger and astigmatism exists, which brings a lot of inconvenience to the use and restricts the application of semiconductor laser. Except for a few applications, such as the side face of DPSL, most applications, such as the end face of all-solid-state laser (DPSL) pumped by semiconductor laser, fiber laser and side-pumped laser with high requirements, need to shape LDA beam to form fiber-coupled laser output with small core diameter, small numerical aperture and high brightness. The early method was to form a bundle of optical fibers by one-to-one correspondence between one optical fiber and each light emitting area of LDA. This method needs to use a large bundle of optical fibers at high power, but the brightness is not high, so it is difficult to further shape the beam to improve the brightness, so this method has tended to be eliminated. Considering the miniaturization and array characteristics of micro-optical elements and high-power semiconductor laser arrays, it is considered that collimating, shaping and coupling semiconductor laser beams with micro-optical elements is the most promising.

Beam shaping method of microlens array. Firstly, the LDA beam is collimated into a collimated beam by using a microlens array, then the beam is further shaped, and finally the shaped beam is focused and coupled to an optical fiber, as shown in Figure 6.

3.2 principle analysis of optical fiber coupling LDA module

The main parameters of laser beam output by optical fiber coupling are not only power, but also core diameter and numerical aperture. For fiber-coupled beams with a certain core diameter and numerical aperture, the whole coupling process satisfies the principle that the product of optical parameters is invariant [3]. The product of optical parameters is defined as the product of spot diameter and divergence angle in this direction. For a circularly symmetric beam with diameter d, the far-field divergence angle is θ, and the optical parameter product of the beam is

BPP LDA and BPP fiber are the products of optical parameters of coupled beam and fiber respectively. For the high-power semiconductor laser array shown in fig. 5, the optical parameter products in the fast and slow axis directions are 0.70mm*mrad and 1745mm*mrad, respectively, but if the divergence angle is defined as 1/e2, the divergence angle of the laser is larger. In fact, there is a gap between the light-emitting regions of the array semiconductor laser, and the duty ratio is 0.3 instead of 1. Therefore, one-to-one collimation with microlens array can improve the duty cycle and reduce the optical parameter product in the slow axis direction, so that the optical parameter product becomes19× 0.15×10×17.

3.3, collimation and shaping of collimated beam

For an optical fiber with a core diameter of 800μm and a numerical aperture of 0.22, the optical parameter product is 352mm*mrad, and the optical parameter product in the fast axis direction is enough to meet the coupling requirements, and the optical parameter product in the slow axis direction is too large for the traditional optical system to change, so it is necessary to shape the beam. Beam shaping is to reduce the spot size in one direction and increase the spot size in the other direction by rearranging the beams on the fast and slow axes, thus realizing the balance of optical parameter products in the two directions. Assuming that the product of optical parameters in the slow axis direction is BPPslow and the product of optical parameters in the fast axis direction is BPP fast, the number of beam shaping times n can be calculated by equation (3).

In fact, because the increase of folding times will inevitably lead to the loss between beam splitting gaps, it is only necessary to satisfy that the product of optical parameters in the fast and slow axis directions is less than that in the coupled fiber. At present, there are three kinds of beam shaping: reflective, refractive and catadioptric. The refracted and catadioptric beams still have a certain divergence angle in the slow axis direction after collimation, which will inevitably produce large reflection loss on several refraction surfaces and deviate from the optical path, thus reducing the coupling efficiency of the whole system. Reflection is an ideal method, so choosing reflection is beneficial to improve the coupling efficiency of the system.

For the stripe array LDA with an aspect ratio of 0.3, the products of optical parameters in the fast axis and slow axis directions are 0.70 mm*mrad and 497 mm*mrad, respectively. If it is necessary to couple into an optical fiber with an optical parameter product of 352 mm*mrad and a wavelength of 800um and 0.22NA, the slow-axis beam only needs to be shaped and folded twice.

3.4, calculation and simulation

Using ZEMAX EE non-sequential ray tracing optical design software, the luminous model, beam collimation, shaping and focusing of the light source are simulated, and the light field distribution and efficiency of each step are obtained. Fig. 8 shows the light intensity distribution at several important optical surface positions, where A is the light intensity distribution at the light emitting surface of the stripe array laser. As can be seen from the figure, there are 19 light-emitting regions, and the output power of each light-emitting region is 2W, so the total output light power is 38W, where b is the light intensity distribution after collimation in the fast and slow axes, the power is 37.9, c is the shaped light intensity distribution, and the laser power is 365438+. D is the light intensity distribution of the end face of the focused fiber. From the simulation results, the light spot is less than 150um×720um, the output power is 26W, and the coupling efficiency is 68.5%. If the total power is 40W, the output power of optical fiber coupling is 27.4W W..

3.5, experimental results and analysis

40W stripe array semiconductor laser is coupled with microlens array through optical fiber. The laser consists of 19 light-emitting regions, and the length of each light-emitting region is 150? M, the distance between luminous areas is 500? M, so the length of the light-emitting area of the strip array is 10mm, and the divergence angles in the fast and slow axis directions are 8 and 36°(FWHM) respectively. After being collimated by the fast-slow axis microlens array, the divergence angles of the collimated beam in the fast-slow axis direction are 2.3mrad and 42.5mrad, respectively, and the spot is about10 mm× 0.6 mm. After being folded twice, it becomes a spot of1.2 mm× 5 mm. Fig. 9 shows the P-I curve of strip array semiconductor laser and the P-I curve of optical fiber coupling output, and Figure 10 shows the actual optical fiber coupling semiconductor laser. The power loss is mainly in the following aspects: the measured efficiency is low, first, the collimation envelope energy is fast, and the slow axis direction is only 90%; Secondly, the reflected energy loss of each lens accounts for about 5-8%; During the shaping process, the wavefront is divided and rearranged, and the edge loss is about 5 ~ 8%. Finally, about 10% energy is lost due to the reflection and leakage of the end face of the coupled fiber.