Abstract
We present an inverse method to engineer uniform-intensity focal fields with arbitrary shape. Amplitude, phase, and polarization states, as adjustable parameters, are used to seek the desired focal fields in the non-iterative computational procedure. Our method can be applied to the cases with low and moderate numerical aperture (NA), in which case the feasibility and validity of our approach have been demonstrated in theory, simulation and experiment, respectively. For the case of higher NA, simulated results based on the Richards-Wolf diffraction integral are shown. We also made some discussions on the experiments with the higher NA. Our method should have wide applications in optical micro machining, optical trapping and so on.
© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Focal field shaping becomes increasingly important due to the abundant applications in various realms [1–12], in details, fabrication of surface microstructures [2–4], optical trapping of small particles [5–7], surface plasmon polariton [8], super-resolution imaging [9], femtosecond laser filamentation [10], second-harmonic imaging [11], optical storage [12], and so on. Obviously, the shaping of focal field, including amplitude, phase, and polarization states, is of great sense.
To achieve a desired focal field, we need to design the amplitude, phase, and polarization states of the input optical field. An idea is to design some diffractive optical elements [13–17], or design directly the complicated vector optical fields (VOFs) [1,18–29]. Optimization methods based on cylindrical vector beams is of particular importance [15–20, 29]. Specific distributions of the focal fields, such as the H-like shape [21], optical bubbles [16], and the optical needle [25], have been obtained. Optimized Gerchberg-Saxton algorithm also provides an approximate solution to the desired focal field [14], but it is time-consuming. Several non-iterative methods have also been proposed, but they are limited to the case of low numerical aperture (NA) [28,29]. Therefore, engineering the focal field of lens with a moderate and even high NA fast and precisely is a significant issue.
In this article, we presented an inverse method to engineer the focal field, using the Richards-Wolf diffraction integral [30] and the Fourier transform, which will be introduced in Section 2. In Section 3, we will offer an example to engineer uniform-intensity focal field with arbitrary shape and size. Simulated focal fields and the corresponding input fields will be shown here. In Section 4, we will show the experimental results and the relevant discussions. The major difference of this work from Ref. [14] is reflected in that we used all the three dimensions of amplitude, phase and polarization states to achieve a non-iterative inverse calculation method.
2. Basic principle
As is well known, the focusing process of an optical field by an aplanatic lens has two steps: the refraction of lens and the propagation in free space [31]. After passing through the lens, an input field Ei is transformed into the transmitted field Et at the exit pupil and then propagates to the focal plane, forming the focal field Ef. There exist two coordinate systems, (xi, yi, zi) and (ri, ϕi, zi) correspond to the input plane, and (xf, yf, zf) and (rf, ϕf, zf) correspond to the focal plane. The schematic of the focusing process is shown in Fig. 1.
In this section, we will show the details of our inverse calculation method in both cylindrical and Cartesian coordinate systems. We will also make some discussions on the difference and the relation between them.
2.1. Inverse calculation method in the cylindrical coordinate system
An input field in the cylindrical coordinate system can be written as , in particular, will always be zero due to the requirement of transversality condition. Upon the refraction of the lens, the radial component is deflected by θ, but the azimuthal compoment is unaffected, so the transmitted field Et has a nonzero component , we can obtain , , and [30]. This process can be expressed by a polarization transformation matrix as
with where is the apodization function of lens. The polarization transformation matrix Mp is supplemented by some elements to guarantee the energy conservation, which means the determinant of Mp must be one. After that, the focal field Ef is obtained by integrating the transmitted field Et [31] as follows where f is the focal length of the lens, λ0 is the wave length, k0 = 2π/λ0 is the wave vector. Considering the projection function ri / f = sin θ, Eq. (3) can be further simplified as When we study on the electric field Ef in the focal plane, zf = 0 and f = 1 will always be chosen. So we can obtainObviously, Eq. (5) is the form of two-dimensional (2D) Fourier transform in the polar coordinate system. Now we can integrate the apodization function of the lens and the polarization transformation matrix to implement the direct calculation from the input field to the focal field [31].
where ℱp is the polar Fourier transform [32,33]. With Eq. (5), we can easily obtain the relation from the focal field to the transmitted field using the inverse Fourier transform as follows and the direct calculation from the focal field to the input field as with where is the inverse polar Fourier transform [32, 33]. With Eq. (8), we can find . As mentioned above, a real input field must have no longitudinal component such that the transversality condition of optical field is guaranteed, so will always be zero. Finally, we obtain a significant relation between and :Equations (8) and (10) are the basic formulae of the inverse calculation method in the cylindrical coordinate system. In this case, we should use the polar Fourier transform.
2.2. Inverse calculation method in the Cartesian coordinate system
An input field in the Cartesian coordinate system can be written as , in particular, will always be zero. The first step of the focusing process is the refraction of lens, which has been shown in Eq. (2). By applying Er = Ex cos ϕ + Ey sin ϕ and Eϕ = Ey cos ϕ − Ex sin ϕ [30], Eq. (2) can be rewritten as
withThe focal field Ef is obtained by integrating the transmitted field Et, the projection function ri / f = sin θ is used repeatedly for simplification, we can obtain
Choose the same condition as above : zf = 0 and f = 1, we can giveObviously, Eq. (14) is the form of 2D Fourier transform in the Cartesian coordinate system. Now we can derive the direct calculation from the input field to the focal field [31].
where ℱc is the Fourier transform in the Cartesian coordinate system. With Eq. (14), we can also obtain the relation from the focal field to the transmitted field using the inverse Fourier transform as follows and the direct calculation from the focal field to the input field as with where is the inverse Fourier transform in the Cartesian coordinate system. With Eq. (18), we can find . Since and ri = sin θ, we can obtain the relations among , and as followsEquations. (17) and (19) are the basic formulae of the inverse calculation method in the Cartesian coordinate system.
2.3. Procedure of the inverse calculation method
The inverse calculation methods in the two different coordinate systems are essentially the same one, because they describe the same physical process. However, they are different in form, because they correspond to different components of optical field with different orthogonal unit vectors. Equations (8) and (10) are the basic formulae of the inverse calculation method with (êr + êϕ + êz), corresponding to radial, azimuthal and longitudinal components of optical field. Equations (17) and (19) are the basic formulae of the inverse calculation method with (êx, êy, êz), corresponding to horizontal, vertical and longitudinal components of optical field.
Another difference on those two methods is the Fourier transform process. The operator in Eqs. (8) and (10) is the polar Fourier transform, which means that the sampling frequency points are equispaced in the polar coordinates [32]. The operator in Eqs. (17) and (19) is the common Fourier transform in the Cartesian coordinate system, which means that the sampling frequency points are equispaced in the x- and y-dimensions. Those two kinds of the Fourier transform are identical in the ideal condition, but they have some difference in actual discrete sampling process [32].
As is well known, under the tightly focusing case, the longitudinal field component is related to the radial component only, while has nothing to the azimuthal component. Only when the radial and longitudinal components of the desired focal field satisfy Eq. (10), the input field may be truely exist, and the focal field obtained is a true electromagnetic field; or only when the horizontal, vertical and longitudinal components of the desired focal field satisfy Eq. (19), the input field may be truely exist.
The two kinds of the inverse calculation methods are both important and useful, beucase any one has its own advantages. For the first one, the form is the most concise and explicit, and we can obtain a direct computing relation between and , thus is an independent variable, but programing a fast and accurate algorithm in the polar Fourier transform is an unsolved problem [32,33]. For the second one, the form is more complicated, but it is widespreadly used, because we can use the fast Fourier transform (FFT) method.
The inverse calculation method for the focal field can be divided into the following procedure:
- Generate the input field experimentally, which is actually vector optical field (VOF).
- Produce the focal field through the objective illuminated by the input field.
3. Engineering the uniform-intensity focal fields with arbitrary shape
With Eqs. (8) and (10) [Eqs. (17) and (19)], we can obtain the desired focal fields with structured amplitude, phase and polarization states. In many applications, however, the total intensity is more important, so we give several examples of uniform-intensity focal fields. Our method is based on the Richard-Wolf vector diffraction integral, which is an extension of scalar diffraction theory, and it contains both the low and high NA cases. Considering the universality of the theory, we choose parameter NA = 0.9 in the following discussion in this section.
The longitudinal component is the most uncertain factor especially in the case of high NA, so we make a pre-procedure based on Eq. (10) before performing the inverse calculation. Here we choose the simplest way, the corresponding pre-procedure can be divided into the following steps:
- Select (zf = 0) = A, where A = A(x, y) is the function of desired uniform shape.
- Calculate the corresponding (zf = 0) by Eq. (10).
- Set (zf = 0) that acts as a supplement, making the total intensity uniform.
In this case, the polar Fourier transform should be used in Step 2, but we find that we can also obtain a kind of uniform-intensity focal field when using the FFT.
The sampling points in the FFT process is set to be 1080 × 1080, corresponding to the 1920 × 1080 resolution of the spatial light modulator (SLM) used in our experiments. Those sampling points are padded with zeros to 215 × 215 (see Section 3.4 and Ref. [31]), for the accuracy of FFT algorithm.
3.1. Simulations under the ideal condition
The ideal condition means that we can achieve an ideal modulation of the amplitude, phase and polarization states of light, that is to say, we can generate arbitrary perfect VOF. As the simulated example shown in Fig. 2, the focal field is a uniform-intensity rhombus pattern. For this case, we have A(x, y) = rect(x + y)rect(x − y), where rect(x/a) is the so-called rectangular function as . Radial, azimuthal and longitudinal components are designed to lead to a uniform total intensity with NA = 0.9. The longitudinal component dominately contributes in the center of rhombus pattern, which is exactly located at the polarization singularity of the transverse components. There is a spiral distribution of polarization states in the focal field, on account of the equal proportions of and . The distribution of polarization states can be changed by adjusting the relative ratio of and in the pre-procedure, what is more, can be more complicated so long as Eq. (10) [Eq. (19)] is satisfied. The global phase is uniform and we can add the desired phase distribution during the inverse calculation process easily.
Figure 3 shows the simulated input field corresponding to the focal field in Fig. 2. We can find that the total intensity exhibits a four-fold rotational symmetry in Fig. 3(a), the distribution of polarization states has a two-fold rotational symmetry in Fig. 3(b), and the distribution of phase has also a two-fold rotational symmetry with a change of π in Fig. 3(c), respectively. Those specific distributions lead to a phenomenon of destructive interference, which corresponds to the dark point of the transverse components, as shown in Figs. 2(b) and 2(c).
3.2. Simulations under the existing experimental condition
Since the space domain and the frequency domain in the Fourier transformation are relative, we can treat the input field as the equivalent frequency domain. As shown in Fig. 3(a), the input field has quite a lot of low-frequency information, because the focal field is uniform. Under the existing experimental condition, we are hard to achieve such an extreme amplitude modulation, and we will obtain the edge-enhanced images [34–36]. However, we can multiply a random phase hologram to avoid this edge-enhancement effect [36], which means the function A(x, y) is replaced by A(x, y)Rand(x, y) in Step 1 of the pre-procedure, where Rand(x, y) is a random function. Simulated example is shown in Fig. 4 and the corresponding input field is shown in Fig. 5.
Compared Fig. 5 with Fig. 3, we can find that the low-frequency information is restrained, implying that the operation is effective. We also find some changes of phase and polarization states, and partial loss of symmetry. Due to the introdution of the random phase, the resultant focal fields will have some difference for each simulation.
The population Pearson correlation coefficient (CC) between the result and the designed one is used to evaluate the shape and uniformity of focal field, simultaneously. This definition is widely used to characterize the distinction of two functions or two images. The formula for the CC definition is
where cov is the covariance between X and Y, and σ is the standard deviation, respectively. The closer the CC value is to 1, the better the result will be. Usually, the CC value of 0.5∼1.0 is regarded as the strong correlation. In fact, CC between the focal field in Fig. 4 and the designed one is 0.9814, indicating the feasibility of our pre-procedure.3.3. Control on the shape of focal fields
The shape of the focal field can be controlled by changing the shape-function A in Step 1 of the pre-procedure. Focal fields with the other shapes are shown in Fig. 6, the corresponding input fields are shown in Fig. 7. The triangle-shaped focal field in Figs. 6(a)–6(d), with , has CC = 0.9775. The red-corss-shaped focal field in Figs. 6(e)–6(h), with A(x, y) = rect(x)rect(10y) + rect(10x)rect(y) − rect(x)rect(y), has CC = 0.9448. Due to the same pre-procedure, the polarization states of focal fields shown in Fig. 6 is similar to those shown in Fig. 4, and the global phase is also uniform.
3.4. Control on the size of focal fields
As is well known, zero-padding is widely used in the FFT by adjusting the sampling positions aiming to amplifying the frequency domain, and it can also be used in the focal field calculations for changing the size of focal field image [31]. To investigate the influence of zero-padding before, we give another simulated example in Fig. 8. Total intensity of tightly focused fields with NA = 0.9 is shown, all of them have a rhombus shape but different dimensions. Images in Figs. 8(a)–8(d), with different sizes of zero-padding operations, have different dimensions of 60λ × 60λ, 30λ × 30λ, 15λ × 15λ, and 10λ × 10λ, correspondingly, with CC = 0.9865, 0.9762, 0.9561, and 0.9382, respectively.
Figure 9 shows the simulated input fields corresponding to the focal fields in Fig. 8. We can find that, as the size of desired focal field decreases, both amplitude and phase variation will be lesser. More significantly, the extreme value of the amplitude becomes normal, leading to an easier realization in actual experiments. What is more, the CC value has a decreasing tendency as the size of desired focal field decreases, meaning that the uniformity of total intensity becomes inferior. The dominant reason is that in our inverse calculation process, the zero-padding implies the information loss in frequency domain, so we hope it to be as little as possible in experiments. The nonuniform and blurry reason is due to the diffraction limited system in our simulation. When the size becomes comparable to the diffraction limit, the difficulty of achieving a target distribution in focal field can be understood as a manifestation of the uncertainty principle [14].
3.5. Discussions on the parameter NA
According to the relation ri / f = sin θ and , after normalizing ri with , we can get sin θ = ri NA. So we have , and Eqs. (10) and (19) can be expressed by NA as
We can adjust the parameter NA in the inverse calculation process by using Eq. (21) or Eq. (22). Figure 10 shows the simulated example with different NA. Images in Figs. 10(a)–10(d), with different NA and the same size of zero-padding operations to Fig. 4, have different dimensions of 405λ × 405λ, 135λ × 135λ, 81λ × 81λ, and 58λ × 58λ, correspondingly, with CC = 0.9847, 0.9849, 0.9854, and 0.9856, respectively.The longitudinal component can be easliy weakened by phase modulation (reduce the relative NA) and polarization setting (reduce the radial component), but difficulty enhanced due to the limit of NA. Since the specific pre-procedure, there will exist a transverse polarization singularity in the focal field, and the longitudinal component may be not enough to supplement, resulting in a center dark as shown in Fig. 10(a). However, in the case of NA < 0.3, the longitudinal component can be neglected, so we can change or discard the pre-procedure to obtain another ordinary polarization distribution and a better result.
4. Experimental analysis
In this section, we will perform experiments for demonstrating the capability of our approach for engineering the focal field with the uniform intensity for a given shape. Firstly, we will show the experimental arrangement for generating the calculated input fields. Then we will show the measured results of the focal fields, which verify the capability of our method to achieve such a focal field with a moderate NA. Finally, we will analyze the reason for the imperfect experimental results, we will also give some advice and guesses on the experiments with the high-NA.
4.1. Experimental arrangement for generating the input VOFs
We can generate the input optical field after figuring out the components we need. There are many ways to modulate amplitude and phase simultaneously, in this work we use the macro-pixel encoding method by a SLM [37]. The experimental arrangement is the typical generator of VOFs based on the 4f system [37], as shown in Fig. 11. After the complex amplitude modulation by the SLM, the polarization conversion by the λ/4 plates, and the interferometric superposition by the Ronchi grating, the input optical field we desired can be generated. We use objectives with different NA to focus and a CCD camera to detect. The CCD camera has a size of 1280 × 1024 pixels with a pixel pitch of 4.6 μm. The SLM has a size of 1920 × 1080 pixels with a pixel pitch of 8.0 μm. The light source used in experiment is a laser at a wavelength of 532 nm. The original input beam in experiment is a top-hat beam.
In experiment, when using the grating encoded with hologram directly, we try to generate the input field shown in Fig. 3, but we obtain only a hollow shape, which is called the edge-enhancement effect [36]. As mentioned in Section 3.2, therefore, the introduction of random phase hologram is not only necessary but also effective.
4.2. Experimental results
Figure 12 shows the measured total intensity of focal fields. We generate the input fields and detect their focal fields with different NA. From the first to third column, images have different dimensions of 1500 × 1500 μm2, 250 × 250 μm2, and 150 × 150 μm2, correspondingly, with NA = 0.1, 0.4, and 0.65, respectively. Images in Figs. 12(a)–12(c) show the rhombus-shaped focal fields, with CC = 0.9536, 0.9495, and 0.8863, respectively. Images in Figs. 12(d)–12(f) show the triangle-shaped focal fields, with CC = 0.9519, 0.9493, and 0.8367, respectively. Images in Figs. 12(g)–12(i) show the red-cross-shaped focal fields, with CC = 0.9197, 0.9383, and 0.8270, respectively. Measured intensity patterns of components are also in agreement with the simulated results, meaning the correctness of polarization.
4.3. Discussions and suggestions for experiments
Experimental results in Fig. 12 suggest that we can well engineer the shape of focal field with a moderate NA, which has a uniform total intensity. Focal fields with NA = 0.9 fails to be detected, because of the tiny working distance of objectives (∼0.2 mm) and the thick protective glass of CCD (∼1 mm).
We cannot get the arbitrary large focal field, unless using another experimental arrangement. In the actual experiment, the ±1-orders in the plane of spatial filter (Fig. 11) become the rhombus shape. Here we need two pinholes for filtering, and the two rhombus-shaped spots cannot overlap. Considering the period of the Ronchi grating, the interval between the ±1-orders in the plane of SF is ∼6 mm, so the interval between those two rhombuses is ∼4.2 mm. Hence the maximal size of the rhombus in the SF plane is ∼2.9 mm × 2.9 mm for the lens L1 with NA = 0.035. Thus we can achieve the maximal size of the focal field to be ∼100/NA μm. What is more, the period of the Ronchi grating and the NA of lens L1 can be changed to obtain a larger size.
In the detecting process by the CCD, the focal field is in fact composed of many speckles, due to the use of random phase hologram (see Section 3.2). When increasing the power of laser, a desired shape can be seen in the focal plane, and the total intensity is uniform. The center singularity is mingled with light specks, becoming undistinguished.
The imperfect experimental results in Fig. 12 are due to the reasons as follows: (i) To avoid the edge-enhancement effect [36], a random function is multiplied on the initial desired components of the focal field, which may influence the uniformity of the focal field in the actual experiment. (ii) We use a phase-only SLM to change the amplitude and phase of the light simultaneously, the coding method contains of one macro-pixel and four sub-pixels [37], and the size of macro-pixel is 16 μm × 16 μm. In this work, the polarization states of the input VOF are quite complicated, such a big size of pixel may influence the clarity of boundary in the focal field. (iii) The pixel size of the CCD camera (4.6 μm) is smaller than the macro-pixel of the SLM (16 μm). This may magnify the defects of focal field to get a bad detection. (iv) The working distance of the objectives with NA = 0.65 is 0.6 mm, which approximates to the thickness of the protective glass of the CCD, so images taken by the CCD [see Figs. 11(c), 11(f) and 11(i)] may not the precise image plane of the focal field. (v) Replacement of the Fourier integrals by discreet transformations is known to difficultly approximate these integrals [38], so the FFT process may be inaccurate. An optimized interpolation operation may improve the accuracy to get a better experimental result.
For the experiments with the higher NA, to our knowledge, the difficulties mainly come from two points: the short working distance of the objectives and the longitudinal components. To solve the problem caused by the former, an imaging system (relay optical system) may be used to magnify the target images, a complicated equivalent lens system with long working distance may also be useful. As to the latter, it is indeed difficult to measure the intensity profiles for a high-NA lens, because the objectives cannot map the longitudinal electric fields, and Novotny et al. made an attempt on that [39]. For our case, however, the longitudinal field is not so strong meanwhile it occupies a small region as shown in Fig. 4(d), which provides a convenience in the actual experiment. What is more, if the longitudinal component in experiment is strong, we can adjust the pre-setted NA of the backward calculation process bigger than the observed NA, which may be effective.
5. Conclusions
We presented a non-iterative inverse method to engineer a focal field with uniform intensity. The shape and size can be precisely controlled, by adjusting the phase and polarization states in the pre-procedure and the inverse calculation process. Our theory and simulations are verified by the experiments with the moderate NA, in fact, which requires VOFs. Experiments with the higher NA are also discussed. We have demonstrated the feasibility and validity of our method at least under the moderate NA, which should have wide applications in optical storage, optical micromachining, optical coherent tomograghy, optical trapping and so on.
The simulations in Section 3 correspond to an input top-hat beam in actual experiment. For the input cases of Gaussian, Laguerre-Gaussian or Bessel-Gaussian beams, the amplitude modulation needs to be changed, by dividing the preceding amplitude by the function of a given beam. The phase adjustment is similar.
Although our method has been demonstrated to be successful for engineering the uniform-intensity focal fields, it should also have important reference for achieving the nonuniform-intensity focal fields. Besides, focusing shaping with aberrations may be a new research direction [26].
Funding
National Key R&D Program of China (2017YFA0303800, 2017YFA0303700); National Natural Science Foundation of China (11534006, 11774183, 11674184); Natural Science Foundation of Tianjin (16JCZDJC31300); 111 Project (B07013).
Acknowledgments
We acknowledge the support by Collaborative Innovation Center of Extreme Optics.
References and links
1. Q. Zhan, “Cylindrical vector beams: from mathematical concepts to applications,” Adv. Opt. Photon. 1, 1–57 (2009). [CrossRef]
2. A. M. Morales and C. M. Lieber, “A laser ablation method for the synthesis of crystalline semiconductor nanowires,” Science 279, 208–211 (1998). [CrossRef] [PubMed]
3. C. Hnatovsky, V. Shvedov, W. Krolikowski, and A. Rode, “Revealing local field structure of focused ultrashort pulses,” Phys. Rev. Lett. 106, 123901 (2011). [CrossRef] [PubMed]
4. K. Lou, S. X. Qian, X. L. Wang, Y. N. Li, B. Gu, C. H. Tu, and H. T. Wang, “Two-dimensional microstructures induced by femtosecond vector light fields on silicon,” Opt. Express 20, 120–127 (2011). [CrossRef]
5. D. G. Grier, “A revolution in optical manipulation,” Nature 424, 810–816 (2003). [CrossRef] [PubMed]
6. X. L. Wang, J. Chen, Y. N. Li, J. P. Ding, C. S. Guo, and H. T. Wang, “Optical orbital angular momentum from the curl of polarization,” Phys. Rev. Lett. 105, 253602 (2010). [CrossRef]
7. Y. Roichman, B. Sun, Y. Roichman, J. Amato-Grill, and D. G. Grier, “Optical forces arising from phase gradients,” Phys. Rev. Lett. 100, 013602 (2008). [CrossRef] [PubMed]
8. P. Wróbel, J. Pniewski, T. J. Antosiewicz, and T. Szoplik, “Focusing radially polarized light by a concentrically corrugated silver film without a hole,” Phys. Rev. Lett. 102, 103902 (2009). [CrossRef]
9. X. Li, P. Venugopalan, H. Ren, M. Hong, and M. Gu, “Super-resolved pure-transverse focal fields with an enhanced energy density through focus of an azimuthally polarized first-order vortex beam,” Opt. Lett. 39, 5961–5964 (2014). [CrossRef] [PubMed]
10. S. M. Li, Y. N. Li, X. L. Wang, L. J. Kong, K. Lou, C. H. Tu, Y. J. Tian, and H. T. Wang, “Taming the collapse of optical fields,” Sci. Rep. 2, 1007 (2012). [CrossRef] [PubMed]
11. G. Bautista, M. J. Huttunen, J. Mäkitalo, J. M. Kontio, J. Simonen, and M. Kauranen, “Second-harmonic generation imaging of metal nano-objects with cylindrical vector beams,” Nano Lett. 12, 3207–3212 (2012). [CrossRef] [PubMed]
12. Y. Kawata, H. Ueki, Y. Hashimoto, and S. Kawata, “Three-dimensional optical memory with a photorefractive crystal,” Appl. Opt. 34, 4105–4110 (1995). [CrossRef] [PubMed]
13. G. Sinclair, J. Leach, P. Jordan, G. Gibson, E. Yao, Z. Laczik, M. J. Padgett, and J. Courtial, “Interactive application in holographic optical tweezers of a multi-plane Gerchberg-Saxton algorithm for three-dimensional light shaping,” Opt. Express 12, 1665–1670 (2004). [CrossRef] [PubMed]
14. T. G. Jabbour and S. M. Kuebler, “Vectorial beam shaping,” Opt. Express 16, 7203–7213 (2008). [CrossRef] [PubMed]
15. K. Prabakaran, K.B. Rajesh, and T.V.S. Pillai, “Focus shaping of tightly focused TEM11 mode cylindrically polarized Laguerre Gaussian beam by diffractive optical element,” Optik 124, 5039 (2013). [CrossRef]
16. Y. Zhao, Q. Zhan, and Y. P. Li, “Design of DOE for beam shaping with highly NA focused cylindrical vector beam,” Proc. SPIE , 5636, 56 (2005). [CrossRef]
17. B. Hao and J. Leger, “Polarization beam shaping,” Appl. Opt. 46, 8211–8217 (2007). [CrossRef] [PubMed]
18. Q. Zhan and J. R. Leger, “Focus shaping using cylindrical vector beams,” Opt. Express 10, 324–331 (2002). [CrossRef] [PubMed]
19. W. Chen and Q. Zhan, “Three-dimensional focus shaping with cylindrical vector beams,” Opt. Commun. 265, 411–417 (2006). [CrossRef]
20. L. Rao, J. Pu, Z. Chen, and P. Yei, “Focus shaping of cylindrically polarized vortex beams by a high numerical aperture lens,” Opt. Laser Technol. 41, 241–246 (2009). [CrossRef]
21. Y. Pan, S. M. Li, L. Mao, L. J. Kong, Y. N. Li, C. H. Tu, P. Wang, and H. T. Wang, “Vector optical fields with polarization distributions similar to electric and magnetic field lines,” Opt. Express 21, 16200–16209 (2013). [CrossRef] [PubMed]
22. F. Kenny, D. Lara, O. G. Rodriguez-Herrera, and C. Dainty, “Complete polarization and phase control for focus-shaping in high-NA microscopy,” Opt. Express 20, 14015–14029 (2012). [CrossRef] [PubMed]
23. B. Hao and J. Leger, “Numerical aperture invariant focus shaping using spirally polarized beams,” Opt. Commun. 2811924–1928 (2008). [CrossRef]
24. J. Hao, Z. Yu, Z. Chen, H. Chen, and J. Ding, “Shaping of focal field with controllable amplitude, phase, and polarization,” Chin. Opt. Lett. 12, 090501 (2014). [CrossRef]
25. H. Wang, L. Shi, B. Lukyanchuk, C. Sheppard, and C. T. Chong, “Creation of a needle of longitudinally polarized light in vacuum using binary optics,” Nat. Photon. 2, 501–505 (2008). [CrossRef]
26. B. R. Boruah and M. A. A. Neil, “Focal field computation of an arbitrarily polarized beam using Fourier transforms,” Opt. Commun. 282, 4660–4667 (2009). [CrossRef]
27. H. Chen, Z. Zheng, B. F. Zhang, J. P. Ding, and H. T. Wang, “Polarization structuring of focused field through polarization-only modulation of incident beam,” Opt. Lett. 35, 2825–2827 (2010). [CrossRef] [PubMed]
28. Z. Chen, T. Zeng, and J. Ding, “Reverse engineering approach to focus shaping,” Opt. Lett. 41, 1929–1932 (2016). [CrossRef] [PubMed]
29. G. Rui, J. Chen, X. Wang, B. Gu, Y. Cui, and Q. Zhan, “Synthesis of focused beam with controllable arbitrary homogeneous polarization using engineered vectorial optical fields,” Opt. Express 24, 23667–23676 (2016). [CrossRef] [PubMed]
30. B. Richards and E. Wolf, “Electromagnetic diffraction in optical systems II. Structure of the image field in an aplanatic system,” Proc. Roy. Soc. A 253, 358–379 (1959). [CrossRef]
31. M. Leutenegger, R. Rao, R. A. Leitgeb, and T. Lasser, “Fast focus field calculations,” Opt. Express 14, 11277–11291 (2006). [CrossRef]
32. A. Averbucha, R. R. Coifmanb, D. L. Donohoc, M. Eladd, and M. Israelid, “Fast and accurate polar Fourier transform,” Appl. Comput. Harmon. Anal. 21, 145–167 (2006). [CrossRef]
33. S. A. Abbas, Q. Sun, and H. Foroosh, “An exact and fast computation of discrete Fourier transform for polar and spherical grid,” IEEE Trans. Signal Process. 65, 2033–2048 (2017). [CrossRef]
34. J. L. Horner and P. D. Gianino, “Phase-only matched filtering,” Appl. Opt. 23, 812–816 (1984). [CrossRef] [PubMed]
35. J. L. Horner and J. R. Leger, “Pattern recognition with binary phase-only filters,” Appl. Opt. 24, 609–611 (1985). [CrossRef] [PubMed]
36. J. A. Davis, S. W. Flowers, D. M. Cottrell, and R. A. Lilly, “Smoothing of the edge-enhanced impulse response from binary phase-only filters using random binary patterns,” Appl. Opt. 28, 2987–2988 (1989). [CrossRef] [PubMed]
37. Z. Chen, T. Zeng, B. Qian, and J. Ding, “Complete shaping of optical vector beams,” Opt. Express 23, 17701–17710 (2015). [CrossRef] [PubMed]
38. W. H. Press, S. A. Teukolsky, W. T. Vetterling, and B. P. Flannery, Numerical Recipes in C: The Art of Scientific Computing (Cambridge University, 1992).
39. L. Novotny, M. R. Beversluis, K. S. Youngworth, and T. G. Brown, “Longitudinal field modes probed by single molecules,” Phys. Rev. Lett. 82, 5251–5254 (2001). [CrossRef]