The laser light source is generally Gaussian intensity distribution, and its energy concentration features uneven surface energy density of the sample, material damage, and other problems in the fields of micro-precision machining such as laser drilling, MicroLED repair and transfer, and laser cutting. In this paper, the micron-scale uniform light spot is prepared based on diffractive optical elements (DOEs), and the shape and size of the uniform light spot can be flexibly designed with a high degree of freedom. The difficulty in DOE design for preparing small-sized uniform light spots is to achieve steep and sharp transition areas and characteristic parameters with uniformity of less than 5% in the flat-top area. If the DOE is designed based on the analytical method, the transition area of the uniform light spot changes slowly, and the uniform area effectively used is quite smaller than the diameter. The analytical method is not suitable for preparing small-sized uniform light spots. Therefore, the numerical analysis method is used to design the DOEs. The most effective one is the iterative Fourier algorithm, including the Gerchberg-Saxton (GS) algorithm and its various improved algorithms, which are effective tools in DOE design. Usually, the parameters of the optical system are not easily changed after being determined, but the detector position is prone to deviation. In order to analyze the application of this DOE in the engineering environment, the influence of the defocus on the beam shaping is analyzed through simulation and experiments.

Before the DOE element is introduced into the optical system, the diffraction limit of the system should be calculated, and the DOE can only prepare a uniform light spot larger than the diffraction limit. Based on diffraction limit constraints, an ultraviolet (UV) short wavelength light source is used. This paper combines the basic GS algorithm and the modified GS algorithm. Firstly, the basic GS algorithm is used to calculate the primary phase distribution of DOE. The initial phase will affect the number of iterations and the convergence speed of the GS algorithm. Moreover, it is easy to fall into the local optimal solution. Based on the principle of spatial filtering, the initial phase is calculated, which can reduce the number of iterations. Secondly, this phase is used as the initial phase of the modified GS algorithm. In the frequency spectrum range, the range of the uniform light spot is defined as the signal area S, and the rest of the range is defined as the noise area. The frequency domain amplitudes both in the signal area and the noise area are limited [Eq. (11)]. The beam shaping is realized through the spatial DOE phase and frequency domain amplitude constraints. The schematic diagram of the comparison between the basic GS algorithm and the improved GS algorithm is shown in Fig. 4. According to the beam shaping system (Fig. 6), the analytical method, the basic GS algorithm, and the basic GS algorithm combined with the modified GS algorithm are compared to analyze their effects of beam shaping.

The size of the uniform light spot prepared based on the analytical method is limited by that of the incident light spot, the parameters of the optical system, etc., and it has strong constraints and can only prepare regular shapes, and simulation results are shown in Fig. 1. Because the size of the uniform spot is close to the diffraction limit, the basic GS algorithm has an unsatisfactory shaping effect and does not change the Gaussian intensity distribution. As a result, it cannot meet the parameter requirements, and the Gaussian beam cannot be shaped into a uniform spot only by the phase of the blazed grating. Therefore, in addition to the phase degree of freedom, the amplitude degree of freedom is added to modify the GS algorithm and improve the beam shaping effect. The basic GS algorithm combined with the modified GS algorithm designs the uniform light spot by DOE, and the light intensity stability and uniformity are greatly improved and tend to become the ideal uniformity (Fig. 5). According to the simulation and experimental results, it is concluded that the defocus leads to the deformation of the uniform spot, which destroys the well-defined light intensity distribution in the transition area and the flat-top area. When the detector is far away from the geometric focus along the beam transmission direction, the spot size becomes slightly larger. The flat-top area shrinks, and the energy in the central area accounts for a large proportion. When the detector is close to the DOE direction, or in other words, the detection surface is in front of the geometric focus, the energy is concentrated in the four corners of the square edge, and the collapsed structure with weak light intensity is in the flat-top area (Fig. 8 and Fig. 9). The square uniform spot with a side length of 28 μm is more sensitive to defocus error. When |Δz|<20 μm, the flat-top beam structure can still be maintained, uniformity (Uni) drops to about 7.31%, and root-mean-square error (RMSE) drops to about 10.62%. The strength uniformity is seriously deteriorated and cannot meet the requirements of use. In practical applications, the influence of detector errors should be reduced as much as possible, and focal plane calibration should be performed before the DOE is used.

In this paper, the DOE phase distribution is designed by combining the basic GS algorithm and the modified GS algorithm, and the Gaussian beam of UV light source with a diameter of 3 mm is shaped into a square uniform spot with a side length of 28 μm. The effectiveness and reliability of the method are verified by simulation and experiments. Within a certain error range, the experimental results are consistent with the theoretical simulations. In the follow-up, research work to improve the beam shaping effect of DOE in engineering applications will be carried out. This paper provides data reference for the design method of uniform spot shaping DOE and the application of DOE in an engineering environment.