Search

A nondeforming chucking technique for EUV lithography has been developed.Temperature distribution and shear and tension forces during freezing were calculated by FEM.Shear strength of 0.2N per pin and tensile strength of 0.34N per pin, and solidification deformation below ?0.15µm were measured.The new freezing pin chuck has the pitch of 2mm and are cooled with coolant of 5?C.The new chuck can clamp the low expansion glass mask without deformation at temperatures below 50?C. Extreme ultraviolet (EUV) lithography has edged closer to practical use as the next-generation exposure technique for replicating 10-nm-order patterns. Since this replication takes place in a vacuum, the mask and wafer are usually clamped by electrostatic force. However, an electrostatic chuck has the following limitations: the force is smaller than the vacuum force, dust adheres to the tops of the chuck pins, and deformation occurs during chucking. The authors have developed a new nondeforming chucking technique that does not use electrostatic force or vacuum force and does not damage the contact surface. This paper describes the principle and features of this nondeforming chucking technique and evaluates the cooling temperatures required to solidify the freezing liquid and fix the quartz substrate using the finite element method. The various forces between the quartz specimen and the pin and the deformation of the quartz wafer due to chucking are also discussed. By lapping the pin tops after coating with oil repellent, the average shear strength of a 10-mm-square quartz specimen on a 0.5-mm-diameter, 2-mm-pitch pin chuck was increased to 5.1?0.9N/cm2, which is more than 1000 times for the acceleration force. The deformations after being fixed with the above pin chuck were less than ?0.15µm for a 100-mm-diameter, 1.2-mm-thick quartz wafer, with a maximum convex warpage of 8µm. It was concluded that the 6-mm-thick, 152-mm-square quartz mask with a thermal expansion coefficient of 5i?10-9/K can be clamped without deformation at temperatures below 50?C using the new freezing pin chuck with the pitch of 2mm.

A site flatness of less than 26nm will be required when fabricating next-generation devices. A pin chuck is used to flatten warped wafers. This chuck has lift holes for loading and unloading wafers and a ring seal around its periphery. The lift holes and seal deteriorate the local flatness. This paper describes the influence of local dynamic deformation on wafer flatness, which is caused by flattening of the warped wafer, and the deterioration in the wafer flatness due to lift holes and due to a polishing step. The repeatability of a stitching method was examined to accurately measure the flatness over a wide region; its value was less than +/-3nm. This method was used to measure the local dynamic deformation. The fluctuation in the raised deformation above the lift hole when the wafer was stationary was +/-4nm. On the other hand, when the wafer moved it was plus/minus several tens of nanometers and the local dynamic deformation greatly affects the flatness. In addition, it was demonstrated that the polishing step and the deformation difference between the seal and the pin greatly deteriorate the flatness; a 10-mm-diameter ring seal and a 2-mm-pitch pin array cause a wafer deformation of approximately 0.23@mm.

A site flatness of less than 32nm will be required when fabricating next-generation devices. To flatten warped wafers, a pin chuck is used. This chuck has lift holes for loading and unloading wafers and a ring seal around its periphery. It is known that the lift holes and seal deteriorate local flatness. This paper describes the theoretical local deformation of a wafer due to the ring seal and the steps between the pins and ring seals generated by polishing the pin chuck, and the influences of these deformations on wafer flatness. Reducing the pin pitch from 2mm to 1mm or doubling the diameter of the pin from 0.2mm to 0.4mm decreases the buckling amount due to the ring seal to approximately 25% or 40%, respectively. Increasing the diameter of the ring seal by only 2mm decreases the buckling from 36nm to 27nm. However, deterioration of the wafer flatness caused by several tens of nanometer steps is very large and a step of 30nm causes a buckling deformation of approximately 100nm. However, it was determined that local flatness below 10nm can be achieved by using a static-pressure seal-type pin chuck with a pitch of 1mm.

CMP is the standard planarization method for silicon wafers. For environment reasons, semiconductor manufacturers have to reduce the consumption of materials, such as slurry. The oscillation-speed-control-type polishing machine uses a small tool and supplies slurry from a pad center, so the slurry spreads to the polishing area easily. This report shows the polishing characteristics under low slurry flow rate conditions. In non-oscillation polishing, it is difficult to reduce the slurry flow rate to less than 200 ml/min since a shortage of slurry results in a damaged area on a wafer. In uniform oscillation speed polishing, no damaged area was found to occur even with a slurry flow rate of only 5 ml/min. In addition the experimental results corresponded with the simulated ones. It was demonstrated that in oscillation speed control polishing, the minimum oscillation speed and slurry flow rates were different for different pad materials. In experiments, a flatness of better than 0.2 μm over the entire surface was achieved using both an IC1000 pad with a slurry flow rate of 20 ml/min and a SUBA800 pad with a flow rate of 40 ml/min.

The simulation of lapping to correct the surface profile of a workpiece has been made very little study. In contrast, the simulation of polishing has been developed using the gap theory and it has been demonstrated that there is good agreement between simulated and experimental results. Our study investigates the application of polishing simulation to lapping. We have reported that a lapped profile can be calculated if a suitable value of relative elastic coefficient is used. This paper describes workpiece and tool profiles lapped with a small tool, and corresponding simulation results calculated using the gap theory, for oscillation lapping to planarize a quartz glass wafer. Under non-oscillation conditions, the experimental results for workpiece and tool profiles correspond roughly with simulated results. When the tool overhangs the workpiece, to a significant extent, the experimental results do not correspond with the simulated ones. Therefore, the slope of the pressure distribution for the simulation was adjusted to increase rapidly when the amount of overhang is more than 30 % of tool diameter. As a result, the experimental and simulation results corresponded very closely. In uniform-oscillation-speed lapping, it was demonstrated that the achieved workpiece and tool profiles coincide approximately with the results of the simulation.

The size of the photo mask and mother glass used in liquid crystal display production has increased yearly. Large rectangular glass plates are difficult to planarize using rotary-type polishing machines. We have developed a rotary-type polishing machine with tool path control that is optimized by polishing simulation for a rectangular wafer. The present paper describes the planarization of a rectangular plate by simulation. The influence of tool size and the aspect ratio of the rectangular plate on the flatness are clarified. For a square plate, the flatness obtained under optimized oscillation speed is less than a quarter of that obtained under uniform oscillation speed. For rectangular plates with aspect ratios of 1:1.25 and 1:1.5, planarization using a tool having a diameter equal to half the diagonal length of the plate is shown to be difficult because the stock removal distributions in diagonal and short side of the workpiece become the different shape.

The feature size for next-generation devices is decreasing and edge exclusion widths are becoming stricter, decreasing from 2 to 1.5mm. The development of a new porous pin chuck is thus necessary for flattening large wafers, including the periphery, which has small and extremely low pins. A reverse production process from the conventional method was adopted to form pins after careful polishing of the chuck surface. It is now possible to both measure and achieve a high level of flatness for pin chucks, while reducing vacuum leakage. Average pin heights were 33@mm in the suction region and 5@mm in the static-pressure seal. The uniformity of pin heights in the seal was improved compared with the conventional static-pressure-seal pin chuck. In addition, the contact ratio between the back surface of a wafer and the pin tops was reduced to 25% that of conventional static-pressure-seal pin chucks by using pins with smaller diameters and porous material.

In next-generation lithography, the problems of nanotopography and peripheral flatness deterioration caused by the flattening of a wafer with a current vacuum pin chuck must be solved. This paper describes the factors contributing to nanotopography and flatness deterioration, and determines the flattening ability around the periphery using a current pin chuck. It is shown that the nanotopography induced by clamping is very small after polishing of the chuck surface, but may increase above 10nm due to pin wear. Since the ring seal used to prevent the leakage is 20-40nm higher than the pins, the peripheral flatness of the wafer deteriorates. In addition, it is found that flattening a heavily warped wafer up to the wafer edge is difficult. As such, the current ring-seal-type chuck should be replaced with a static-pressure-seal-type chuck.

As ULSI feature dimensions continue to shrink, wafer flatness of less than 0.1μm has become essential, and any flaws in surface topography cause serious problems in device patterning. CMP has become a standard technology to planarize surface topography. We have developed an oscillation-speed-control-type sequential grinding and polishing machine and have simulated the polishing process. A previous paper showed that experimental profiles polished with normal and reverse rotations agreed well with the simulated result calculated with a modified equation of relative velocity. However, the simulation broke down when the tool overhung the wafer by a significant amount. This paper presents measured pressure distributions for various tool overhangs, and a method to compensate for the difference between theoretical and experimental distributions, and polishing amounts produced using the compensation. The pressure distribution from air pressure in an air-bag is significantly different than that from a weight, and its characteristics depend on the machine. The profile of the pressure distribution for large overhangs becomes concave rather than linear. The simulated results with concave pressure distributions agreed well with experimental results.

To resolve ultra-fine patterns of less than 0.1 µm by ArF eximer laser lithography, a chucking accuracy of less than ±0.03 µm must be achieved over the entre wafer except 2 mm from a processed wafer edge. A new static-pressure-seal-type vacuum pin chuck has been developed, whose seal extends beyond the wafer edge. This paper describes the chuck structure, the pin characteristics and the difference of the flattening ability between this chuck and an ordinary ring-seal-type chuck and presents the flattening ability calculated for a 12-in. wafer (1 in. = 2.54 cm). Images of pin tops show that pins made of pore-free ceramics have fewer corner cracks and smaller diameters than those made of alumina ceramics. These features are useful for decreasing the influence of dust and scratching. The experimental result with an 8-in. wafer showed that the flattening ability of the new chuck for the process-induced bow of 200 µm is three-times better than that of an ordinary chuck, which agrees roughly with the theoretical result. The calculated results for a standard 12-in. wafer clarified that the static-pressure-seal-type chuck can flatten the process-induced concave bow of 600 µm within ±0.03 µm under the edge exclusion of 2 mm.

To meet the demands of next-generation ArF excimer laser lithography—a site flatness better than 0.03 μm in a 30-mm square field—a new vacuum pin chuck was developed. A 2-mm wide static-pressure seal, which extends beyond the wafer edge, makes it possible to almost completely flatten the wafer periphery. Since the wafer back-surface contacts only the tips of the pins, the chuck can significantly reduce the influence of dust. In addition, conductive pore-free ceramics and a coating of SiC, as well as round pin corners and small roughness on the pin tops, prevent dust adhesion. It was found experimentally and theoretically that the flattening ability of the new chuck around the wafer periphery is several times higher than that of an ordinary ring-seal chuck. Such superior flattening was achieved by expanding the vacuum region by the wafer edge. Calculations showed that the new chuck can flatten an 8-inch concave wafer with a bow of 200 μm to a residual bow of less than 0.03 μm.

Mass production of ULSI devices requires a 12-inch silicon wafer to have a site flatness better than 0.1 μm/30mm. In addition, film thickness, which varies across the wafer, must be polished to within ±5% uniformity. In order to meet these requirements, the machining process has been replaced from batch to single-wafer and from one side to both sides. This paper shows that all ordinary oscillation methods in abrasive machining can be converted to a linear oscillation method where a tool oscillates on the center line of a workpiece. For linear oscillation polishing with a viscoelastic and wear polisher, equations for surface generation were obtained by using the clearance between a workpiece and a tool. A polishing simulation program was newly developed on the basis of these equations. The calculation procedures for stock removal and optimum oscillation speed, and polishing conditions for high flatness are shown. It is also clarified that it is possible to polish within ±1% uniformity with the optimum oscillation speed control.

Polishing of CVD diamond plate with metallic disk wheel in air is proposed as a practical and simple polishing method. The effects of polishing temperature and disk materials on removal rate and surface roughness are examined experimentally, and the optimum conditions and the polishing mechanism are discussed. When disk materials of mild steel, cast iron, nickel alloy and titanium alloy are used, the adhesion between the CVD diamond and the metallic disk happens. Removal rate and surface roughness are the best when stainless steel is used, because various oxides produced in the disk prevent adhesion between the CVD diamond and the disk. At polishing temperature below 400°C, the diamond is polished mechanically. As polishing temperature reaches from 500°C to 600°C, the diamond is oxidized and removal rate increases extremely.

The fabrication of ULSI devices on a 12" silicon wafer requires a site flatness of within 0.1μm. However, the flatness is affected by the vacuum pin chucks used for planarizing wafers in optical lithography or for final polishing. This paper describes the contact between the pin tops and the wafer back-surface during clamping. Complete contact between the back-surface and the pin tops is difficult to achieve with a 775-μm-thick standard 12" wafer, even when the back is the mirror surface. The contact ratio between the wafer and the pin tops was measured. It depends on the wafer thickness. These experimental results closely agreed with theoretical results. The pin deformation that degrades the flatness was also analyzed. If the pressure loaded on all pins was uniform, the pin deformation is only 8nm. However, when the pressure loaded on each pin varied in peak to valley values by a factor of 16, the effect of the pin deformation on the flatness became significant.

The fabrication of ULSI devices requires a wide focus margin. A newly developed vacuum pin chuck has been used to decrease the margin reduction caused by wafer chucking. This paper describes the influences of pin support and back-surface waviness on site flatness during clamping. Pin support reduces the effect of dust, but degrades the site flatness. It is shown that the deformation between pins with a 2-mm pin-pitch chuck for an 8″ standard wafer is negligibly small, but the back-surface waviness of the etched wafer degrades the site flatness by 0.05 μm in a 30-mm-square field. Reducing the vacuum pressure and polishing for a short time only slightly improves the site flatness. Therefore, it is difficult to use the present type of etched wafers and it is necessary to reduce wavinesses onvinesses on both sides of wafers for next-generation lithography of 0.13 μm.

Fabrication of ULSI devices requires a 12-in. silicon wafer to have a site flatness better than 0.1 μm /30 mm. Vacuum pin chucks have been developed for flattening processed wafers. This paper describes to reduce the influence of the back-surface waviness on wafer flatness resulting from vacuum clamping. It was found that the contact ratio between the back-surface and the pin tops decreases with an increase in pin pitch or waviness, which degrades wafer flatness. Reducing the vacuum pressure improves flatness only slightly, but polishing for a short time produced good results. It is also shown that flatness better than 0.05μm/30 mm is obtained by reducing the waviness to better than 0.1μm /30 mm in clamping a 775 -μm-thick silicon wafer.

The beginning and ending position on the working surface of grinding wheel when involute helical gear is ground with rack-tooth worm wheel are geometrically analyzed. The effects of operating parameters on the grinding mechanism of gear tooth surface are discussed. Moreover, it is confirmed that the tendency of the experimental results agree well with that of the analytical results. As the results, the following facts are clarified. (1) The ending position on the working surface of grinding wheel is the generating position of gear tooth profile in leading side and is behind it in trailing side. (2) The working length of grinding wheel, when infeed of wheel is large and feed of workpiece is small, is longer in leading side than in trailing side. (3) The wheel wear is apt to increase when the working length is shorter.