In-Process Measurement of Thickness of Cured Resin in Evanescent-Wave-Based Nano-stereolithography Using Critical Angle Reflection
Stereolithography is one of the most powerful ways to fabricate complex three-dimensional polymeric-based structures layer-by-layer using optical power. Evanescent-wave-based nano-stereolithography using ultra-thin field distribution of evanescent wave to solidify photosensitive resin can provide a sub-micrometer vertical resolution of each layer. To meet strong demands for in-process thickness measurement of cured resin in evanescent-wave-based nano-stereolithography, a measurement method that utilizes variations of resin’s refractive index after polymerization and a high sensitivity of total internal reflection at the critical angle has been proposed. By launching the a measurement light from a substrate to resin at the critical angle and detecting reflections from the resin, slight change in refractive index and thickness of cured resin that have great influences on reflectivity can be in-process measured. This method has been firstly examined by simulation using the rigorous coupled wave analysis method. Here, we show that an increase in cured resin’s thickness induces a decrease in the reflectivity. In experiments, the largest reflectivity contrast between cured and uncured resins has been proved at the critical angle. In addition, the relationship between reflectivity drop and thickness has been calibrated and a linear relationship within a certain thickness range has been experimentally confirmed. Furthermore, the difference of curing process between continuous and discontinuous exposure has been investigated by using the proposed measurement method. At last, image subtraction and a median filter have been applied in imaging processing to remove influences of uneven illumination and background noises.
KeywordsPhotosensitive resin Nano-stereolithography Evanescent light Critical angle In-process measurement RCWA method
Significance of our investigation is based on an inevitable fact that some fabrications errors often appear in EWNSL. These unavoidable fabrication errors can be simply classified into projection errors that resulted from the scattering and diffraction, curing errors generated by environment disturbances in vulnerable photo initiation reactions, polymerization errors caused by chain reactions of monomer molecules and shrinkage errors due to post-curing of cured resin [19, 20, 21, 22, 23, 24]. All of above fabrication errors will finally lead to uneven thickness of each solidified layer, which as a serious problem directly influences the fabrication quality and limits the applications of EWNSL. Therefore, the in-process measurement on the thickness of the cured resin that detects the fabrication errors in time and assists us to further investigate origin of them in EWNSL is of great significance and value. In addition, the in-process measurement makes it possible to develop an intelligent fabrication system that uses measurement results as feedbacks to dynamically adjust irradiation, control the curing process and correct the fabrication errors.
Difficulties of the in-process measurement in EWNSL are mainly caused by following problems: Thickness of each layer is only several hundreds of nanometers; cured resin is submerged in uncured (liquid) resin in the fabrication process; there is gradient-refractive-index boundary between the cured and uncured resins due to uncompleted polymerization. Due to above problems, methods for thickness measurement based on interferometry cannot be used in EWNSL since the gradient boundary between the cured and uncured resins greatly weakens reflections. In addition, because curing of resin is a quick process, some thickness measurement methods based on a scanning probe that require a certain sweep time cannot be used. Other measurement methods like FTIR (Fourier transform infrared spectroscopy) [25, 26], differential scanning calorimetry (DSC) and Raman spectroscopy [27, 28] have been frequently used to monitor the curing process of resin; however, these methods are hard to obtain spatial information of the cured resin and therefore cannot meet the demands of thickness measurement. Differential interference contrast (DIC) microscopy has a theoretical potential to measure the thickness of cured resin in EWNSL, but it is hard to be applied as the in-process measurement in realistic productions. This is because measurement systems need to be supplemented in fabrication systems, and there is limited space for measurement systems. DIC needs to occupy both spaces over and under resin for detecting light transmitting through resin, which leads to a spatial conflict with the fabrication system.
In our previous investigation, surface plasmon resonance (SPR) [29, 30] has been applied for the in-process measurement in EWNSL [15, 16]. Based on a theory that the thickness of cured resin determines a resonance condition and significantly affects the reflectivity, the in-process measurement of the thickness has been achieved by detecting and analyzing the reflection. To generate SPR, resin has to be placed on a substrate with a metal filming layer, and a measurement light needs to be incident from the bottom of resins in a particular incident angle. On the downside, measurement accuracy is limited by a low lateral resolution in SPR measurement which is determined by propagation lengths of polaritons rather than optical diffraction limit.
In this research, we proposed a measurement method utilizing the high sensitivity of total internal reflection at the critical angle. Once resin is cured, the refractive index and the thickness change of resin will disturb the total internal reflection condition and result in the reflectivity drop. The in-process measurement can be simply achieved by detecting the reflection. Just as SPR measurement, only space below the substrate is needed and the measurement system shares a prism or an oil-immerging objective with the fabrication system. The difference is that in this method metal coating is not needed any more and the incident angle of measurement light is in the critical angle rather than the SPR angle.
The following contents of this paper consist three sections: theory, simulation and experiment. In the theory section, a study on the formation of resin layers will be briefly explained; after that, principles of our proposed method will be demonstrated. Its feasibility will be further discussed in the simulation section. In the experiment section, relationship between the reflectivity and the thickness of cured layer will be confirmed.
2 Theoretical Background
In EWNSL, field strength of the evanescent wave drops off exponentially away from the substrate; the closer the resin to the substrate, the higher exposure energy and therefore the faster curing speed . As a result, the resin near the substrate will be cured firstly and the thicknesses of cured layer increase as exposure time expands. It is notable that, in the curing process, interfaces between the cured and the uncured resins are not clear. These gradient boundaries are made of half-cured resin in a state of uncompleted polymerization [22, 32]. As we mentioned in the Introduction section, the gradient boundary with low reflection brings some difficulties for the measurement. More seriously, due to the existence of the gradient boundary, the thickness of cured layers is finally determined after a washing and drying treatment. Therefore, it is impossible to measure absolute thickness in the fabrication since the absolute thickness does not exist before removing the uncured and half-cured resins. What we measure in the fabrication process are effective thicknesses of cured resin. They represent the thickness of cured layers that will probably appear after the cleaning process.
2.1 Curing Process
2.2 In-Process Measurement
3 Simulation on Optical Response in Critical Angle Illumination
Figure 4b plots computed results recorded by the monitor when the thickness of cured resin increases from 0.3 to 0.6 μm. It is shown that the reflectivity drops below cured resin enlarged obviously with the thickness increase. This confirmed the feasibility of thickness measurement using the proposed method. In addition, we found that the reflection distribution below the cured resin was not as flat as the top surface of cured resin; and at region that was not directly below the cured resin (in the area of x > 25 μm, right of cured resin) also appeared reflectivity drop. These unexpected oscillations were enhanced with the thickness increase as well. In our point of view, they were caused by multiple reflection in the cured resin. The measurement light that is reflected into substrate after interference will reflect again from the up surface of cured resin to substrate. Therefore, except leaking from the edge on right sides of cured resin, all of the light was finally reflected in substrate. The oscillations of reflection curve were caused by the supersession of multi-reflection.
To confirm this point, time-averaged intensity distribution of E field was plotted when a cured resin was in a thickness of 0.4 μm, as shown in Fig. 4c. It can be seen that the field distribution near the up surface of cured resin increases from left to right showing the accumulation of multi-reflection. In the consideration of experiment, a lot of factors including gradient boundary, uneven surface and non-uniform distribution of refractive index of cured resin will destroy the susceptible total internal reflection on up surface. As a result, the multi-reflection should be greatly weakened and there will be no obvious oscillations in physical experiments. In a word, the impact of thickness on reflectivity has been proved; the relation between thickness and reflectivity needs to be measured by actual experiments.
4.1 Experiment Setup
Regarding the performance of the measurement system, few points are worth discussing. First of all, it is hard to clarify imaging resolution of the current equipment since resin was measured in a large viewing angle. In our future work, sub-microresolution can be achieved by introducing oil immersion objective in a high numerical aperture. However, considering the measurement light has a larger wavelength than the fabrication light, detail loss of the measurement on cured resin is unavoidable in case of the fabrication and the measurement section shares a same objective lens. Furthermore, measurement range of thickness is not specifically defined; the intensity of reflection will change in a periodic of 450 nm and the multi-reflection will be largely enhanced when thickness keeps on increasing. Considering the cured resin produced by the evanescent-wave-based stereolithography is normally in a thickness within 500 nm, the measurement range of thickness is enough for our investigations. In addition, measurement rate of current system, mainly determined by the CMOS camera, is 90 frames per second (FPS). This rate is sufficient for our experiment as laser used in the experiment was weak (about 0.5 mW) and generated relatively low curing speed. The measurement rate can be further improved by using a high-speed camera to match fast fabrication speed in industrial production.
4.2 Measurement of Bulk Sample
The cross section of gray value near the boundary region stretches from cured to uncured resin in the various incident angles is plotted in Fig. 6e. The gray value of image that directly indicates the intensity of reflection was used to evaluate the reflectivity. In order to remove the noise of images, every curve in these figure are averaged from ten lines stretching a same distance in cured and uncured resins. It is obvious that the gray values of cured resin and uncured resin vary with incident angle. The reflectivity contrast became maximum when the incident light was near the critical angle. At the boundary, there were small peaks that show the increments of reflectivity, which was caused by the scattering at the boundary. The reflectivity drop of cured resin was calculated by dividing averaged gray value drop by the gray value in total internal reflection condition (around 96 gray value units). The reflectivity drop obtained by experiment was plotted with the theoretical value gained by calculating the reflectivity difference between cured and uncured resins, as shown in Fig. 6f. It shows that the experiment results well agree with the theoretical value, which confirms the refractive index increase of resin and also proves the feasibility of our experiment equipment.
4.3 Measurement of the Relation Between Thickness of Cured Resin and Reflectivity
In this experiment, in order to confirm the influence on reflectivity caused by thickness variation and experimentally find the relationship between thickness and reflectivity, cured resin in a thickness of sub-micrometer was cured by evanescent light and measured by the proposed method.
In order to find the relation between the reflectivity drop in in-process measurement and thickness of cured resin, the thickness of sample was measured by AFM (atomic force microscope) after drying process. Figure 7g plots the thickness and the reflectivity drop distribution in same cross section of cured resin. It is obvious that the reflectivity drop shows the same tendency with the thickness of cured resin. The slope of the reflectivity distribution was not as gentle as the curve of thickness. This is because in the in-process measurement, interference and dust caused uneven distribution of measurement light which directly influenced the measurement results. The uneven distribution can be removed by making subtraction between the gray value distributions obtained before and after curing. The imaging process and its results will be illustrated in the next experiment. In addition, the reflectivity drop varied linearly with the thickness of cured resin according to the measurement results. For this reason, the reflectivity-position curve was coinciding with the thickness position curve after adjusting the range of the two vertical axes in Fig. 7g. To confirm this point, the reflectivity drop as a function of the thickness is plotted in Fig. 7h. A linear relation between them can be clearly seen. In the experiments, the relation between the reflectivity drop and thickness was caused by many factors including profile of the gradient boundary, flatness of the cured layer and the refractive index of cured resin. The linear relation that was even out of our expectation might not be a standard situation. To explain the linear relation between the reflectivity drop and the thickness, the mathematical relation will be investigated in the future by simulating the curing process and modeling the gradient boundary. Compared with the simulation results, the obvious reflectivity drop on one side of cured resin was not shown in the experiment. This is because the shape of cured resin in the experiment is different from ideal rectangular shape in the simulation, and therefore, multi-interference between the top and bottom boundaries of cured resin was largely weakened.
4.4 In-process Measurement and Imaging Processing
Furthermore, the image processing techniques were applied to remove the uneven distribution of measurement light. It was achieved by the subtraction between the reflection distribution before and after exposure. The imaging filter was adopted after subtraction to remove the background noise. The processed images are shown in Fig. 8c, d, representing the continuous and discontinuous exposure condition corresponding to Fig. 8a, b. In processed image, the variation of shape and reflectivity drop of cured resin in the curing process are more obvious than in the original image; the difference between continuous and discontinuous exposure as we discussed above can also be clearly observed. Obviously, imaging process benefits the in-process measurement at lot; however, the usage of filter results in some loss of image information and impacts the measurement. More detailed research of the types and parameters of image filter will be done in our future work.
In conclusion, we proposed the in-process measurement in evanescent-wave-based nano-stereolithography to monitor curing process and measure effective thickness of cured resin. The measurement method using the refractive index increase of resin and the high sensitivity of total internal reflection when incident angle is at the critical angle was proposed to develop a suitable in-process measurement system in nano-stereolithography. The feasibility of proposed measurement method was firstly proved by simulation using the RCWA method. The influence of thickness of cured resin on reflectivity was confirmed. The experiment system including both evanescent wave curing and measurement sections was built to experimentally examine our method. There experiments were done using the developed system. In the first experiment, by measuring a bulk sample, the variation of resin’s refractive index and the reliability of experiment was confirmed. In the second experiment, relationship between reflectivity drop and effective thickness was measured. The linear relationship was experimentally found. In the third experiment, sample in the conditions of continuous and discontinuous exposure was in-process measurement and compared. The image processing that extracts the variation of resin and removes the background noise was demonstrated by this experiment. The difference between continuous and discontinuous exposure was clearly observed. It also effectively identified our method as a powerful way of in-process measurement in nano-stereolithography.
- 17.Kong D, Michihata M, Takamasu K, Takahashi S (2017) In-process measurement on the thickness of photosensitive resin in evanescent wave-based nano-stereolithography. In: The 13th international symposium on measurement technology and intelligent instruments, No. 17Google Scholar
- 18.Kong D, Michihata M, Takamasu K, Takahashi S (2017) Monitor resin’s curing degree for in-process measurement in micro-stereolithography. In: The 7th international conference of Asian society for precision engineering and nanotechnology, MET-0-08Google Scholar