Autofocusing method using fluorescence detection for precise two-photon nanofabrication

We propose a method capable of focusing a laser beam on a substrate automatically via fluorescence detection from the resin of a twophoton nanofabrication system. When two-photon absorption (TPA) occurs by focusing the laser beam in the resin, fluorescence is emitted from the focusing region in the visible range. The total pixel number above the threshold value of the fluorescence images obtained by a CCD camera is plotted on a graph in accordance with the focus position. By searching for the position when the total pixel number undergoes an abrupt change in the pre-TPA region, the correct configuration of the focused laser beam can be found. Through focusing tests conducted at four vertices of a 500 μm x 500 μm square placed arbitrarily inside SCR500 resin, the errors of the autofocusing method were found to range from −100 nm to + 200 nm. Moreover, this method does not leave any polymerized marks. To verify the usefulness of the autofocusing method, the fabrication of a pyramid structure consisting of 20 layers was attempted on a coverglass. It was completely fabricated without losing a layer. ©2011 Optical Society of America OCIS codes: (190.4180) Multiphoton processes; (220.4241) Nanostructure fabrication. References and links 1. E. S. Wu, J. H. Strickler, W. R. Harrell, and W. W. Webb, “Two-photon lithography for microelectronic application,” Proc. SPIE 1674, 776–782 (1992). 2. S. Maruo, O. Nakamura, and S. Kawata, “Three-dimensional microfabrication with two-photon-absorbed photopolymerization,” Opt. Lett. 22(2), 132–134 (1997). 3. S. Kawata, H.-B. Sun, T. Tanaka, and K. Takada, “Finer features for functional microdevices,” Nature 412(6848), 697–698 (2001). 4. H.-B. Sun, K. Takada, M.-S. Kim, K.-S. Lee, and S. Kawata, “Scaling laws of voxels in two-photon photopolymerization nanofabrication,” Appl. Phys. Lett. 83(6), 1104–1106 (2003). 5. J. Serbin, A. Egbert, A. Ostendorf, B. N. Chichkov, R. Houbertz, G. Domann, J. Schulz, C. Cronauer, L. Fröhlich, and M. Popall, “Femtosecond laser-induced two-photon polymerization of inorganic-organic hybrid materials for applications in photonics,” Opt. Lett. 28(5), 301–303 (2003). 6. D.-Y. Yang, S. H. Park, T. W. Lim, H. J. Kong, S. W. Yi, H. K. Yang, and K.-S. Lee, “Ultraprecise microreproduction of a three-dimensional artistic sculpture by multipath scanning method in two-photon photopolymerization,” Appl. Phys. Lett. 90(1), 013113 (2007). 7. H. J. Kong, S. W. Yi, D.-Y. Yang, and K.-S. Lee, “Ultrafast laser-induced two-photon photopolymerization of SU-8 high-aspect-ratio structures and nanowire,” J. Korean Phys. Soc. 54(1), 215–219 (2009). 8. M. Malinauskas, A. Žukauskas, G. Bičkauskaitė, R. Gadonas, and S. Juodkazis, “Mechanisms of threedimensional structuring of photo-polymers by tightly focussed femtosecond laser pulses,” Opt. Express 18(10), 10209–10221 (2010). 9. J. Hesse, M. Sonnleitner, A. Sonnleitner, G. Freudenthaler, J. Jacak, O. Höglinger, H. Schindler, and G. J. Schütz, “Single-molecule reader for high-throughput bioanalysis,” Anal. Chem. 76(19), 5960–5964 (2004). 10. K.-S. Lee, R. H. Kim, D.-Y. Yang, and S. H. Park, “Advances in 3D nano/microfabrication using two-photon initiated polymerization,” Prog. Polym. Sci. 33(6), 631–681 (2008). #154418 $15.00 USD Received 8 Sep 2011; revised 21 Oct 2011; accepted 21 Oct 2011; published 26 Oct 2011 (C) 2011 OSA 7 November 2011 / Vol. 19, No. 23 / OPTICS EXPRESS 22659 11. J.-J. Park, P. Prabhakaran, K. K. Jang, Y. Lee, J. Lee, K. Lee, J. Hur, J.-M. Kim, N. Cho, Y. Son, D.-Y. Yang, and K.-S. Lee, “Photopatternable quantum dots forming quasi-ordered arrays,” Nano Lett. 10(7), 2310–2317 (2010). 12. N. Tétreault, G. von Freymann, M. Deubel, M. Hermatschweiler, F. Pérez-Willard, S. John, M. Wegener, and G. A. Ozin, “New route to three-dimensional photonic bandgap materials: silicon double inversion of polymer templates,” Adv. Mater. (Deerfield Beach Fla.) 18(4), 457–460 (2006). 13. I. Staude, G. von Freymann, S. Essig, K. Busch, and M. Wegener, “Waveguides in three-dimensional photonicbandgap materials by direct laser writing and silicon double inversion,” Opt. Lett. 36(1), 67–69 (2011). 14. L. Li, R. R. Gattass, E. Gershgoren, H. Hwang, and J. T. Fourkas, “Achieving λ/20 resolution by one-color initiation and deactivation of polymerization,” Science 324(5929), 910–913 (2009).


Introduction
Nanofabrication by two-photon polymerization (TPP) is a well-known and popular technique for fabricating complicate 3-D structures [1][2][3][4][5][6][7].It is possible to fabricate specified 2-D patterns and 3-D structures under the diffraction limit using two-photon nanofabrication, as TPP only occurs at the laser focal spot (LFS) due to the intrinsic characteristics of the multiphoton absorption phenomenon [1][2][3][4].Recently, it is reported that electron generation by avalanche is more dominant than that by multi-photon absorption at the polymerization [8].By controlling the position of the LFS in the resin, various 3-D structures can be fabricated as the input designs, making it possible to fabricate 3-D structures with a sub-diffraction-limit spatial resolution of about 100 nm in our nanofabrication system [7].In more detail, the 3-D structures are built up from the bottom on the surface of a substrate layer by layer.Naturally, the first layer of the structure has to be directly above the surface of the substrate.In other words, the region in the LFS above a two-photon absorption (TPA) threshold must be immediately above the surface of the substrate before the fabrication process via TPA begins in the resin.If the LFS is far above the surface, the fabricated structure is swept away or turned during its development via the floating of the structure.On the other hand, when the LFS is below the surface, it is partially created on the surface.Therefore, focusing in the twophoton nanofabrication process is essential for the precise fabrication of the specified output.
If there is no resin on the surface of a glass substrate, it is easy to focus the focal spot on the surface by a manual focusing method, e.g., controlling the translation stage by hand, as the light of the focal spot reflected from the surface of the substrate can be ascertained distinctly by a CCD camera when such a camera is attached to the system.In contrast, if there is resin on the surface of the substrate, it is not easy to collect the light reflected from the surface when the resin and the glass have similar refractive indices.In practice, many resins utilized to the TPP have refractive indices similar to that of the glass.Therefore, it is difficult to locate the LFS on such a surface exactly by monitoring the reflected light without additional hardware, e.g., an autofocus laser [9].The time and accuracy for the focusing process in the resin depend on the experience and skill of the researcher.
In general, by changing the position of the substrate on the optical axis using a PZT-stage, we made several 2-D test patterns to estimate the position of the LFS in the resin before fabricating the desired structures.By observing the test patterns, the correct configuration of the LFS was found.The correct configuration refers to a condition in which the center of the focus at which the TPP occurs is right on the surface of the substrate and not in the resin.This determines when the test pattern creation begins.If the total time for fabricating the desired structure is very short, it is possible to fabricate the structure directly instead of making a test pattern.However, this manual focusing method takes a relatively long time and can lead to the creation of several unwanted structures on the surface by trial and error.
When TPA occurs in the resin by a focused laser beam, the photon energy is used to polymerize the small focused region of the resin.However, some of the excited electrons in the resin relax to their ground state with the emission of fluorescence [10].In other words, if LFS above the TPA threshold exists in the resin, it is possible to observe fluorescence from the resin by the TPA phenomenon.As a result, the correct configuration of the LFS can be found by measuring the position on the optical axis when the emission of the fluorescence begins to be observed while translating the z-axis stage.In this paper, we demonstrate that the fluorescence of SCR500 resin is changed by the z-axis position of the substrate and propose an autofocusing method using the detection of the fluorescence for two-photon nanofabrication.SCR500 resin contains a photoinitiator and a photosensitizer, like most other TPA resins used for photopolymerization [10].The photosensitizers absorb two photons of near-IR light and emit fluorescence in the visible range.The photoinitiators are excited from a ground state to an excited state by absorbing some of the fluorescence, as shown in Fig. 1(a).Some rate of the excited photoinitiators yields a chemical reaction for photopolymerization, but some of the photoinitiators relax to the ground state upon the emission of light.Also, avalanche absorption can play a major part in free electron generation for photopolymerization [8].The fluorescence caused by the photosensitizer and the photoinitiator is used for autofocusing.In addition, the photoinitiators absorb two photons and emit photon energy as fluorescence in the visible range itself [10].The photosensitizer in the SCR500 used in the experiments was a TP-MOSF-TP (0.1 wt%) [11].Its absorption and emission spectra are shown in Fig. 1(b).If the LFS moves up inside the resin, fluorescence by TPA will be observed by the CCD camera.Figure 2 shows the LFS, which is located at the boundary between the resin and the glass substrate.Generally, the lateral size of the region above the TPA threshold is smaller than the diameter of the focal spot; however, the size of the region depends on the laser power and the exposure time.A unique region which is slightly larger than the region above the TPA threshold surrounds that.This region has a very low probability of TPA such that photon energy absorption does not occur to the extent that it can polymerize the resin.However, only weak fluorescence is observed by TPA.This region is defined as the pre-TPA region because the fluorescence occurs in this region with virtually no polymerization prior to the process reaching the above-TPA threshold region when the LFS starts to move into the resin.Moreover, fluorescence can occurs in the region above the TPA threshold, causing emitted light to be observed in both regions.Therefore, the positions at which the fluorescence is measured are closely related to the focus position of the laser beam for the fabrication of structures.By measuring the fluorescence along the position of the LFS on the optical axis (zaxis), it is possible to locate the LFS on the surface of the substrate.The experimental setup for the nanofabrication via TPP is shown in Fig. 3.The system consists of a mode-locked Ti:Sapphire laser (Mai Tai, Spectra Physics), an isolator, two halfwave plates, polarizing beam splitter, galvano shutter, pinhole, high numerical-aperture objective lens (100 × , N.A. 1.40, oil), XYZ-PZT stage, CCD camera, and computer.The laser delivers 100 fs and 780 nm laser pulses at a repetition rate of 80 MHz.The isolator prevents the reflected laser light from going back to the laser.The polarizing beam splitter and half-wave plates are used to control the laser power.The galvano shutter rotates to control the on/off state of the laser beam with a pinhole under a maximal frequency of 1.0 kHz.The position of the focused laser beam is changed on x, y, and z axes by the PZT stage to fabricate three-dimensional structures.The PZT stage has the resolution of 10 nm on x and y axes, and 100 nm on z axis.The laser beam is focused inside the resin on a glass substrate by the objective lens.The CCD camera has 1024 x 768 pixels.The fluorescence from the resin is obtained by the CCD camera above the objective lens.The substrate and the resin are inverted.If the substrate moves up along the z-axis, after which the relative position of the LFS is higher from the substrate.The computer controls the galvano shutter, XYZ-PZT stage, and CCD camera.To analyze the change in the fluorescence intensity as the LFS moves along the z-axis, we measured the fluorescence via the following steps:

Autofocusing method
(i) The initial position of the LFS is located below the resin after the preparation of the resin on the substrate.
(ii) At the same position, a CCD image is recorded to employ it as the background noise image after a shutter using a galvano mirror is activated, as shown in Fig. 3. Figure 4(a) shows the background noise image.
(iii) After the noise input, the substrate moves up along the z-axis at 100-nm intervals once per second due to the resolution limit of the z-axis-PZT stage in our TPP setup, after which the relative position of the LFS is higher from the substrate into the resin, as shown in Fig. 3.
(iv) As the substrate moves up, optical images are accepted by a CCD camera at each fixed position in 100-nm intervals.When the LFS exists in the resin, fluorescence occurs and is superposed onto the background noise in these images, as shown in Fig. 4(b).
(v) After subtracting the background noise image from the optical images, the total intensity values of the remaining fluorescence images, as shown in Fig. 4(c), are input into a computer.The total intensity is the total sum of the pixel values in the image.Figure 4(d) shows that the total intensity increases as the position of the LFS is farther from the surface of the substrate along the z-axis.An abrupt change of the total intensity is observed at a specific position, and this position indicates the point at which the TPA region (the pre-TPA region or the above-TPA threshold region) of the LFS moves into the resin from the substrate, although the position on the graph is not exact compared to the actual boundary between the substrate and the resin.This change is caused by the TPA inside the resin.After the change, the total intensity is oscillated at a high value.Fluctuation of the laser power dominates this phenomenon, although inhomogeneity of the resin caused by the photosensitizer and impurities contained in the resin contribute slightly to it.The laser beam pulse intensity at the sample fluctuated in the range of 0.58 TW/cm 2 to 0.64 TW/cm 2 in processes above.The polymerization threshold of the SCR500 resin used in experiment is 0.13 TW/cm 2 .
(vi) To clarify the position of the abrupt change and eliminate the oscillations in the graph, the total intensity is integrated as the position of the LFS increases on the z-axis, as shown in Fig. 4(e).The position of this change is determined as the point where the slope of the graph changes in Fig. 4(e).
It is possible to determine the position of the LFS immediately above the surface of the substrate in the resin by the linear fitting of the graph in Fig. 4(e) on both sides of the abrupt change point.This simple manipulation, integration and fitting can be used as an autofocusing method to focus the LFS on the boundary between the substrate and the resin.However, this method leaves elongated-shape patterns on the substrate caused by TPP at the test points, as shown in Fig. 4(f).When fabricating bulky 3-D structures without an empty space inside the structures, this method is useful because the remaining pattern will be buried in the structure.However, if the designs of the structures are 2-D patterns, low 3-D structures, or center-hole 3-D structures, these remaining patterns can be a problem.In addition, the total processing time of this method is relatively long, at over 1 minute, even if it is very efficient compared to the manual focusing method.Therefore, we propose the following autofocusing method to overcome the problems mentioned above: (i') -(iii') This is identical to the processes (i)-(iii) that are described above.
(iv') As the substrate moves up, optical images are detected by the CCD camera at 100-nm intervals.When the LFS exists in the resin, fluorescence occurs and is superposed onto the background noise in these images, as described in process (iv).Simultaneously, all images have the background noise image subtracted as soon as the images are obtained at each position.These images are the fluorescence images utilized in the following calculations.
(v') To enhance the SNR, the 10 fluorescence images are superimposed onto one another at each position and go through a thresholding process.The 10 images are continuously obtained at 20 fps by a CCD camera at each fixed position at 100-nm intervals.
(vi') The threshold value is set and the number of pixels above the threshold value in each image is recorded at each position.
(vii') When the number of pixels is larger than the specific value, the process is stopped, as shown in Fig. 5(a).This specific value is set so that the stopped position is where the surface of the substrate exists inside the pre-TPA region.
(viii') The final focusing position is determined by correcting the stopped position by as much as the compensation factor.The LFS is then automatically moved to the determined position by a computer.

method (process (i') -(viii'
)) executed at four vertices of a 500 µm x 500 µm square placed arbitrarily inside the SCR500 resin and the focus positions determined by the autofocusing method and the theoretical focus position were compared on each other.There are no polymerized marks in the autofocusing test point (circle) in a SEM image.(c) A SEM image of a minimum sized voxel fabricated using SCR500 resin in our TPP system.(d) A 2-D window pattern completely fabricated by using the autofocusing method.(e) A fabricated 3-D pyramid structure which is composed of 20 layers by using the autofocusing method.A SEM image on the right shows the counting of 20 layers of the structure.All 20 layers were successfully fabricated.
The autofocusing method does not leave any polymerized marks because the process is closed before the above-TPA threshold region of the LFS moves into the resin.To apply the method to the TPA nanofabrication system, two variables have to be determined.The first is the intensity threshold value I th , and the second is the number of pixels N pix to close the autofocusing process in the method outlined above.
In Eq. ( 1), P x and P y are the total pixel numbers on the x-axis and y-axis of the CCD, respectively.I xy is the intensity value corresponding to its (x, y) position on the CCD array.The function N[f] is 1 if the equation f is true; otherwise it is 0. As the substrate moves up along the z-axis, the autofocusing process is closed at the position where Eq. ( 1) is true.If the initial z position which satisfies Eq. ( 1) is Z i (N pix ), this is solved by Eq. ( 2).
In Eq. ( 2), A z is the final focusing position determined by the autofocusing method to start the fabrication of the structure via TPP.As mentioned above, the Z i (N pix ) position at which the autofocusing process is closed is always lower than A z , and no polymerized mark remains.This occurs because Z i (N pix ) is in the pre-TPA region, whereas A z is the center of the LFS, as shown in Fig. 5(a).Therefore, the difference in the position between Z i (N pix ) and A z has to be corrected, and the compensation factor D c in Eq. ( 2) is necessary for this.The value of D c is not only determined by N pix , but it is also determined by the laser power; as the laser power is higher, the pre-TPA region and the region above the TPA threshold become broader, especially on the optical axis.
The value of N pix has to be determined by considering the slope of the fluorescence graph, which depends on the resin used and the elements in the experimental setup, for example, the detector type.Through pre-experimental process (i)-(vi), we set N pix equal to 10,000, which was intentionally chosen to minimize the error at which the slope of the total intensity undergoes the steepest increase.Next, we found that D c is about 0.8 µm in our system when the laser beam pulse intensity at the sample is 0.61 ± 0.03 TW/cm 2 .Consequently, the actual focusing process is automatically closed earlier than the theoretical focus position by as much as 0.8 µm.

Experimental results
To measure the accuracy of the autofocusing method (process (i')-(viii')), the method was executed inside the SCR500 resin.Subsequently, the theoretical focus position was determined via test patterns.The test patterns were voxel arrays which were fabricated by increasing the exposure time of the laser and the LFS position on the z-axis.In a test pattern, the exposure time increases exponentially from 4 ms to 256 ms at both sides of the test point of the autofocusing method and the LFS position increases from (A z -0.6 µm) to (A z + 0.6 µm) at 100-nm intervals, as shown in Fig. 5(b).The distance between the voxel arrays is 2 µm.A change in the laser exposure time creates various sizes of voxels, and several voxels fall due to the advance in the LFS position.Therefore, the vertical size of a voxel can be measured using the fallen voxels.The theoretical focus position is determined where the length of the fabricated voxel is half of the entire vertical size.The focus positions determined by the autofocusing method and the theoretical focus position were compared at four vertices of a 500 µm x 500 µm square placed arbitrarily inside the resin, as shown in Fig. 5(b).500 µm is distant enough to change the theoretical focus position.The table in Fig. 5(b) shows that the errors are from −100 nm to + 200 nm compared to the theoretical focus position.This autofocusing method did not leave any polymerized mark, as shown in the SEM image of Fig. 5(b).Moreover, the total processing time was less than 30 seconds. Figure 5(c) shows a minimum sized voxel fabricated using the SCR500 resin in our TPP system of Fig. 3.This voxel has the lateral size of about 100 nm and the vertical size of about 400 nm.Because the errors of the autofocusing method fall within the vertical size of the minimum voxel in our TPP system, the autofocusing method adopted to it is suitable for fabricating a specified structure.This method can be effectively applied to the fundamental set-ups that feature a photosensitizer-containing resin, optimized laser conditions, and no additional processes as our experimental setup, because the voxels fabricated in this environment come out in an ellipsoidal shape whose size is generally assumed to be over 400 nm [4,12].Moreover, this method demonstrates the feasibility to apply it to other cases whose resolution is higher than 400 nm if the fluorescence is emitted during the polymerization process [13,14].
To verify the usefulness of the autofocusing method, a 2-D window pattern first was fabricated using SCR500 resin on a coverglass by this method.As a result, the pattern was successfully fabricated, as shown in Fig. 5(d).Moreover, we fabricated a 3-D structure, as shown in Fig. 5(e).The fabricated structure was designed as a pyramid shape with 20 layers, and the gap between each layer was 500 nm.If the error of the focus is greater than 250 nm under the first layer, the pyramidal structure will be swept away during its development process via the floating of the structure.In addition, by counting the total layers of the fabricated structure after the development process, it is possible to estimate the error of the focus over the first layer.The SEM images of Fig. 5(e) shows the fabricated pyramid structure, which consists of all 20 layers with none left out.Hence, the autofocusing method using the detection of the fluorescence can be sufficiently adopted to determine the position to focus a laser beam on the surface of a substrate.

Conclusions
To fabricate the structure on the substrate via TPP precisely, the process of focusing a laser beam on the surface of the substrate is shown to be significant in this study.Thus far, the focusing process has been manually executed by trial and error inside the resin.This makes it inaccurate and inefficient in terms of time.In this paper, we proposed an autofocusing method that can be applied to nanofabrication by TPP to overcome the aforementioned disadvantages.This method utilizes the detection of the fluorescence from the resin along the optical axis.In detail, the focus-searching process of the method is efficiently completed without leaving any polymerized marks on the substrate by means of fluorescence detection in the pre-TPA region, intensity thresholding, and pixel counting in that order.We tested the autofocusing method at four vertices of a 500 µm x 500 µm square inside the resin.The errors ranged from −100 nm to + 200 nm compared to the theoretical focus position.Because the errors were smaller than the minimum vertical size of a voxel, the method can be applied to a nanofabrication system using TPA.Finally, a pyramid structure consisting of 20 layers was completely fabricated by utilizing the proposed autofocusing method.
The proposed autofocusing method can utilize the laser power above the polymerization threshold.Therefore, it has a high SNR due to the strong fluorescence emitted from the resin.Especially, this method is effective when the sensitivity of a CCD camera is low.Moreover, this method doesn't need to control the laser power after the autofocusing process.Because it uses the laser power above the polymerization threshold, the specified structures can immediately be fabricated after the autofocusing process.This method is particularly useful for fabricating structures when the focus position changes considerably over a wide area.

Fig. 1 .
Fig. 1.(a) Schematic energy level diagram of TPP process, and (b) the absorption and emission spectra of a TP-MOSF-TP in SCR500 resin.

Fig. 2 .
Fig.2.Scheme of the region above the TPA threshold, the region of pre-TPA, the region of fluorescence, and isophotes in a LFS.The LFS is between the resin and the glass substrate.

Fig. 4 .
Fig. 4. (a) A background noise image, (b) a fluorescence image with the noise, and (c) a fluorescence image without the noise obtained by the CCD camera.The graph of (d) z position versus total intensity, and (e) z position versus integrated intensity.The z position is the relative position of the LFS from the surface of a substrate on the optical axis.Total intensity is the total sum of the pixel value of the fluorescence image without the noise, as like (c).(f) The integrated intensity is the integration of the total intensity as the position of the LFS increases on the z-axis.Scanning Electron microscope (SEM) image of nine remaining polymerized marks after process (i)-(vi).The process is tested nine times at 10-µm spaces on a coverglass.

Fig. 5 .
Fig. 5. (a) The graph of z position versus total number of pixel and schematic diagram of the LFS in the autofocusing method.The z position is the relative position of the LFS on the optical axis by moving up the glass substrate.In the experimental setup, the substrate and the resin are inverted.The total number of pixel is the total sum of the pixels above the threshold value Ith in the fluorescence image.(b) The results of test experiments of the autofocusing