Research on normal temperature imprint lithography
Abstract: Aiming at the requirements of the development of modern integrated circuit manufacturing industry in pursuit of low cost and smaller feature size, an imprint lithography process at normal temperature is proposed. It adopts mold conversion process to realize large mold stamping with high generation efficiency; The normal temperature imprinting is realized by UV curing to eliminate the material thermal deformation and positioning error caused by temperature change; Low viscosity and high photosensitivity resist is used as the graphic conversion medium; The double loop servo control system is used to realize the control requirements of high-precision positioning and stamping dynamic process. The research results of normal temperature imprint lithography process show that its advantages of low cost and great potential of micro replica are a strong successor to the next generation of integrated circuit lithography process
key words: integrated circuit, ultraviolet, embossing, lithography
preface
at present, the mainstream process of international IC manufacturing is optical lithography technology, but due to the limit of optical lithography, international semiconductor manufacturing equipment manufacturers and academia have invested resources in the research of NGL (next generation lithography) principle and the industrialization of existing NGL reserve technology. The existing NGL mainly includes: 157nm duv+psm, electron beam projection etching technology, X-ray projection etching, 13nm EUV etching, ion beam projection etching and maskless etching manufacturing. These technologies face such problems as the development of high-performance beam sources, the manufacture of ultra numerical aperture and high transmittance optical lenses, the development of high transmittance mask materials, the requirements for extremely high environmental conditions, and the high cost of solving these technical problems and supporting equipment [1~ 3]. In order to avoid these technical difficulties and reduce the high cost, it will produce a new technical route to learn from the embossing technology that has been successfully applied in the manufacture of optical discs and holographic anti-counterfeiting marks. Its main attraction lies in its low cost and high potential for feature size replication up to 6nm. At present, IC lithography technology saves 20% - 60% of power than traditional oil pressure system; Equipped with the optimized hydraulic oil circuit system of two board machine and the current development level of integrated circuits in China, this paper discusses an innovative IC manufacturing process route: normal temperature imprint lithography, which is called IL (imprint lithography). Although IL technology has been proposed in IC lithography internationally, this method mostly refers to hel (hot embossing lithography). Because hel needs to be heated to reduce the viscosity of the resist, but the heat source will cause the thermal deformation of the mechanism and the loss of lithography accuracy and nesting accuracy. The use of IL will completely solve the problems faced by hel [4][5]. Of course, the research of imprint lithography process, the selection of resist materials, the selection of curing mode, the design of imprint mechanism and the formulation of high-precision positioning control strategy are also the technical problems that Il has to face and solve
1 formulation of IL process
il process is generally divided into mold manufacturing and embossing production. As shown in Figure 1, mold manufacturing consists of three parts: the preparation of template substrate, the manufacture of small template and the production of large template. Embossing production includes substrate cleaning, resist coating, embossing, plasma etching, impurity infiltration, deposition, etc. The whole process can not only replace the existing IC lithography process, but also realize a good interface with the existing IC manufacturing process, thus avoiding the change of the existing IC manufacturing process and the injection of a large amount of funds
Figure 1 principle of cold embossing etching process
in order to successfully apply the embossing process to the IC lithography process, the manufacturing of molds is the first problem to be solved, and it is also the bottleneck of IC manufacturing. The reason is that the better way to generate high-quality nano graphics on small molds is to write directly by electron beam and scribe directly on the template with scanning probe microscope. The disadvantage of these two methods is low efficiency, so the motherboard cannot be too large (generally 10 × 10~25 × 25mm or less). Using such a small mold for stamping, the number of unit chips contained in the template is too small, which is obviously not suitable for batch generation. Therefore, the transformation from small mold to large mold is the way to solve this contradiction, and it is also one of the innovations of this process. Figure 2 shows the process of making large templates from small templates. Generally, a small template can only contain the graphic information of a chip graphic or a chipset φ 300mm silicon chip can be filled with thousands of chip units or chipsets. We choose φ A 50mm large template can be distributed with 20-25 chip units or chipsets, which can improve the efficiency of subsequent imprinting lithography by 20-25 times. Figure 2a is a cross-sectional view of the reproduction process of the large template. The graphics of the small template are transferred to the large template through the distributed imprinting technology. The specific process of conversion is that the graphics are first transferred to the resist film formed on the large template substrate by rotating coating, and then the large template is finally generated through the process of plasma etching and cleaning the residual resist. Figure 2B shows the two degree of freedom operation process of distributed imprinting of small template on large template
Figure 2 small template reproduction large template
in the whole IL process, the most critical process is template generation and imprint lithography. In fact, the generation process of large templates also includes imprinting technology. Compared with hel, IL is a normal temperature embossing process, and the whole process is realized at normal temperature. The curing of resist is fundamentally different from hel by reducing the temperature. The curing method of IL is UV curing, so the template substrate adopts quartz glass with high light transmittance. This is also an innovation of IL process. At the same time, the thermal expansion of silicon wafer and mechanism caused by temperature change is eliminated, which leads to the loss of positioning accuracy and the loss of the benchmark of repeated positioning and multi-layer imprinting. Of course, IL also brings corresponding technical problems, such as the selection of UV curing materials and the analysis and research of cold embossing process
2 IL process analysis
il imprinting process is basically the same as the process of small template reproducing large template. The difference lies in the different imprinting area and the different materials used for the substrate. The material used in the template reproduction process is quartz glass, while the IL process is silicon. In the process of IL, the resist is the intermediate medium of pattern transfer. The parameters of the resist, such as viscosity coefficient, shrinkage, elasticity coefficient and UV curing time, directly determine the residual thickness, stamping speed and final stamping quality of the resist film after stamping. The photoresist selected by IL is polymer photosensitive curing resin. In order to achieve the required coating uniformity, replica resolution and accuracy, and curing and film removal, the photoresist should have high photosensitivity, low viscosity coefficient, as well as the smallest possible elastic deformation and curing shrinkage. Based on the above conditions, the dynamic analysis of the stamping process can be further studied by means of hydrostatics
for the convenience of research, the stamping process can be simplified to the simplest model shown in Figure 3. Where, s is the embossing area, W is the width of the groove of the template pattern, D is the depth of the groove, and H (T) is the thickness of the resist. Assuming that the template and the substrate are completely parallel, and the two contact surfaces are completely flat without wave fluctuations, ignoring the surface tension and capillarity during imprinting, it is considered that the resist is an incompressible Newtonian fluid in the whole imprinting process. According to Navier Stokes equation, the stress analysis, film thickness and stamping time of the stamping process are shown in equations (1), (2) and (3)
where: F is the force applied during stamping, TF is the stamping time, η Is the viscosity of the resist, P is the effective pressure during embossing, H0 is the thickness of the resist before embossing, HR is the thickness of the residual resist after embossing, HS is the groove depth of the embossing template, and V is the embossing speed. From equations (1), (2) and (3), it can be concluded that the smaller HR, the larger f, and the viscosity coefficient of the resist η The smaller the better; At the same time, stamping time and viscosity coefficient η So is the relationship between η The smaller the better, and the smaller the effective pressure P, the better. In practice, the template figure is shown in Figure 2. Compared with figure 3, the number of convex and concave of the figure is far more than one, and the feature size is aperiodic. In the process of imprinting, the main change of imprinting characteristics focuses on the change of effective pressure. This is because the characteristic dimensions of the template grooves are different, and the filling speed and time of the resist are not constant. This directly affects the change of effective pressure in the stamping process, making the effective pressure a variable. When the effective pressure changes, the whole mechanical process in the stamping process also changes, making the relationship between the pressure, speed, time and the thinning speed of the resist film change. Therefore, the complex imprinting process control is a technical difficulty of the whole IL, and it is also a technical innovation of IL
Figure 3 Analysis of imprinting process
with the analysis of mechanics, speed and time of dynamic imprinting process, in order to ensure the ultra-high precision micro replica performance of IL, the technical research of nano positioning and repeated positioning and positioning detection links, such as the double loop high-precision positioning feedback link composed of laser interferometer and grating alignment, is another technical difficulty and innovation of IL
3 IL high precision positioning analysis
il ultra-high precision positioning system is shown in Figure 4, and figure 4a is the block diagram of nano positioning dual servo control system. The system is driven by macro and micro levels to realize IL ultra-high precision positioning. The stroke of macro drive is 100mm, and the step accuracy is 1 μ m. Full stroke straightness is less than 1 μ M linear motor. The real-time detection is realized by laser interferometer and constitutes the feedback link of the system. The micro drive adopts a stroke of 60 μ m. Piezoelectric ceramic actuator with step accuracy of 1nm. The micro drive adopts the exact model matching (EMM) control strategy, so that the linearity error is less than 3%. At the same time, the full stroke detection of micro drive adopts the positioning and alignment system with the spatial amplification of molar fringe, as shown in Figure 4B. In the distributed imprinting process, alignment marks are pre engraved on both the master and the silicon wafer in the form of grating stripes. In the alignment process, the grating fringe is irradiated by the laser to produce the molar fringe, which realizes the spatial amplification, converts the change of micro displacement into the change of light energy, and then realizes the acquisition and conversion of the alignment signal through the PD receiver. Finally, the alignment error is less than 30nm.. In practice, the master uses two groups of 52.5 μ M and 50.0 μ This is due to: 1) m grating array with different directions, and the silicon chip adopts 10.0 μ M and 10.5 μ M with different directions
Figure 4 IL ultra-high precision positioning system
il macro micro two-stage stroke L is 10 μ The control result of M is shown in Figure 5. Figure 5A shows the switching process of macro and micro levels. It can be seen that the macro drive using PID control strategy reaches 9 in L μ M, the existing control mode is transformed into the change process of EMM control strategy by the dynamic switching algorithm of control strategy. Figure 5B shows the dynamic process of micro drive control after stabilization. It is obvious that the positioning error after stabilization is within 8nm, which is also the positioning accuracy index of the whole system
(a) switching process of macro micro two-stage drive (b) local response in micro drive process
Figure 5 macro micro two-stage control
another basis of the above research results is that the design of the overall mechanism is accurate to 0.1mm. In practice, in order to eliminate the errors caused by the thermal expansion of materials, friction between moving parts of the mechanism and other factors, indium steel with small thermal deformation coefficient is used on the whole mechanism from the frame to the embossing driving parts, so that the thermal deformation stress field of the whole mechanism is consistent and offset each other. At the same time, the temperature change of the super clean room is also controlled at ± 0.5 ℃. In addition, the micro positioning platform of the system adopts flexible hinge mechanism to realize no gap
LINK
Copyright © 2011 JIN SHI