JP3138693U - Plasma reactor with nozzle and variable process gas distribution - Google Patents

Abstract

A plasma reactor for processing workpieces such as masks and wafers provides a plurality of gas injection nozzles adapted to be modularly coupled to a gas distribution ring.
A gas injection nozzle includes a hollow cylindrical body having a first outer diameter portion and a second outer diameter portion defining a tip, and a first outer diameter portion formed longitudinally through the body. A first passage extending at least partially through the second outer diameter portion and aligned coaxially with the first passage and longitudinally from the end of the first passage to the end of the tip And a first outer diameter portion is larger than the second outer diameter portion.
[Selection] Figure 36A

Description

Invention background

  Photolithographic mask manufacturing for ultra large scale integrated (ULSI) circuits requires even higher etch uniformity than semiconductor wafer processing. A single mask pattern typically occupies an area of 4 inches square on a quartz mask. The image of the mask pattern is focused to the area of a single die (1 inch square) on the wafer and steps are formed across the wafer to form a single image on each die. Prior to etching the mask pattern into the quartz mask, the mask pattern is written with a scanning electron beam. This makes the cost of a single mask very expensive in a time consuming process. The mask etching process is not uniform across the mask surface. Furthermore, the e-beam written photoresist pattern itself is not uniform, and a 45 nm feature size on the wafer will show as much as 2-3 nm variation in critical dimensions (eg, line width) across the entire mask. (This variation is, for example, a 3σ difference in all measured line widths.) Such non-uniformities in the photoresist critical dimension are different for different mask sources or customers. In the mask etching process, this variation cannot be increased beyond 1 nm, so the variation of the etching mask pattern cannot exceed 3-4 nm. These stringent requirements arise from the use of diffraction effects on the quartz mask pattern to obtain a clear image on the wafer. With current technology, it is difficult to meet such requirements. This is even more difficult for future technologies that will include a 22 nm wafer feature size. This problem is exacerbated by the phenomenon of etching bias, and the line width (critical dimension) in the etching pattern on the quartz mask decreases due to the consumption of the photoresist pattern during mask etching. These problems are inherent in the mask etch process. This is because the etch selectivity of common mask materials (quartz, chrome, molybdenum silicide) to photoresist is typically less than 1 because the mask photoresist pattern is etched during the mask etching process.

  Some mask patterns require periodic openings to be etched into the quartz mask at a precisely defined depth. This is important to achieve very fine phase alignment of the interfering light beam during exposure of the wafer through the mask. For example, in one type of phase shift mask, each line is defined by a chrome line, a thin crystal line is exposed on each side of the chrome line, and only one side of the crystal line is etched to a precise depth, A phase shift of 180 degrees is applied to light passing through an unetched crystal line. In order to precisely control the etching depth in the quartz, the etching process must be periodically interrupted and closely monitored to measure the etching depth in the quartz. For each such inspection, the mask is removed from the mask etch reactor chamber, the photoresist is removed, the etch depth is measured, and the remaining etch process time is estimated and the target based on the elapsed etch process time. It is necessary to reach depth, deposit a new photoresist, e-beam write the mask pattern on the resist, reintroduce the mask into the mask etch chamber and restart the etching process. Estimating the remaining etching time to reach the desired depth is not reliable because the etching rate is assumed to remain stable and uniform. Problems with such complicated procedures include low productivity, high cost, and increased possibility of contamination or defects in the photoresist pattern. However, because of the requirement for precisely controlled etch depth, there appears to be no way to avoid such problems.

  Narrow tolerance in critical dimension variations requires a very uniform distribution of etch rate across the mask surface. There are two critical dimensions in masks that require precise etching depth in the quartz material. One is the line width and the other is the etching depth. Uniformity for both types of critical dimensions requires a uniform etch rate distribution across the mask. The nonuniformity in the etching rate distribution can be reduced to some extent by using a power supply device that can change the radial distribution of plasma ion density, such as an induction power supply device consisting of inner and outer coil antennas covering the wafer. Can do. However, such an approach can only deal with symmetric non-uniformities with an etch rate distribution with a high center or a low center. In fact, the non-uniformity of the etch rate distribution can be asymmetric, for example, one corner of the mask has a high etch rate. A more fundamental limitation is that the mask etch process tends to have a very low distribution of etch rate centers. Adjustable features such as inductive power applicators with inner and outer coils cannot convert the etch rate distribution out of the low center area.

  Another problem with non-uniform etch rate distribution is that the etch rate distribution tends to be very different in different reactors of the same design and is always in the same reactor when replacing major parts or consumable components, such as replacing gas distribution nozzles. It can be very different. The etch rate distribution appears to be very sensitive to small variations in the features of the replaced part, and unpredictable changes occur when consumables are replaced. Another related problem is the ease of exchanging the main parts such as the gas distribution nozzle.

Outline of device

  A plasma reactor is described for processing workpieces such as masks and wafers, which includes a vacuum chamber having a gas distribution ring with one or more gas extraction nozzles.

  In one embodiment, a gas injection nozzle is described that is adapted to be modularly coupled to a gas distribution ring in a plasma reactor. The gas injection nozzle is formed with a hollow cylindrical body having a first outer diameter portion and a second outer diameter portion defining a tip, and formed longitudinally through the body, through the first outer diameter portion, A first passage extending at least partially to a second outer diameter portion and aligned coaxially with the first passage and extending longitudinally from an end of the first passage to an end of the tip And the first outer diameter portion is larger than the second outer diameter portion.

  In another embodiment, a gas injection nozzle is described that is adapted to be modularly coupled to a gas distribution ring in a plasma reactor. The gas injection nozzle is disposed through the first diameter portion with a hollow cylindrical body having a first outer diameter portion and a second outer diameter portion smaller than the first outer diameter diameter portion, and has a second diameter. A first passage transitioning into the second passage at a 90 degree interface to a second passage disposed through the portion and radiation coupling the first outer diameter portion to the second outer diameter portion The first outer diameter portion is approximately 50% larger than the second outer diameter portion.

  In another embodiment, a gas injection nozzle is described that is adapted to be modularly coupled to a gas distribution ring in a plasma reactor. The gas injection nozzle is a hollow cylindrical body having a first outer diameter portion and a second outer diameter portion defining a tip, wherein the first outer diameter portion is larger than the second outer diameter portion. A first passage formed longitudinally through the body and extending at least partially through the first outer diameter portion to the second outer diameter portion, and coaxial with the first passage And a second passage extending at a 90 degree interface and extending longitudinally from the end of the first passage to the end of the tip, and a second passage communicating with the first passage And a first passage having a diameter four times greater than a diameter of the second passage.

Detailed description of the invention

Cathode with Improved RF Uniformity The inventors have found that one cause of non-uniform etch rate distribution in the mask etch process is the RF electricity on the support pedestal or cathode that holds the mask in the plasma reactor in which the mask etch process is performed. It was found that there was a heterogeneity. RF bias power is applied to the pedestal to control the plasma ion energy at the mask surface. On the other hand, the RF power source is applied to the overhead coil antenna to generate, for example, plasma ions. The RF bias power controls the electric field at the mask surface that affects the ion energy. Since the ion energy on the mask surface affects the etching rate, the distribution of the etching rate on the entire mask surface becomes non-uniform due to the RF electric non-uniformity on the pedestal. It has been found that there are several causes for RF non-uniformity in the pedestal. One is a titanium screw for fixing the aluminum pedestal (cathode) and the aluminum equipment plate together. The screw creates a node of the electric field pattern over the entire pedestal surface (and thus the entire mask surface). This is because the electrical characteristics are different from those of the aluminum cathode. The other is a non-uniform distribution of conductivity between the cathode and the equipment plate. Electrical conduction between the equipment plate and the cathode is mainly limited to the periphery of the plate and the cathode. Cathode curvature during plasma treatment induced by vacuum pressure may be at least partly responsible for this. This ambient conduction can be non-uniform due to a number of factors such as uneven fastening of titanium screws and / or surface finish variations around either the plate or the pedestal. These problems were solved by introducing several features that improve the RF electrical uniformity across the pedestal. First, the RF field non-uniformity or discontinuity caused by the presence of a titanium screw on the aluminum cathode is addressed by providing a continuous titanium gas injection ring. This ring extends around the upper surface of the cathode, including all titanium screw heads. Variations in conductivity due to surface differences or uneven fastening of the titanium screws provide a highly conductive nickel plate on the opposing peripheral surfaces of the equipment plate and cathode, and between them around the equipment plate and cathode Is addressed by introducing an RF gasket that is compressed between the two.

  Referring to FIG. 1, a plasma reactor for etching a pattern with a mask includes a vacuum chamber 10 surrounded by a sidewall 12 and an overlying ceiling 14 and is evacuated by a vacuum pump 15 that controls the chamber pressure. A mask support base 16 inside the chamber 10 supports the mask 18. As described later in this specification, the mask generally comprises a quartz substrate, and may further include an additional mask thin film layer on the upper surface of the quartz substrate, such as chromium silicide or molybdenum. In addition, a pattern defining layer is present, which may be a photoresist or hard mask formed from a chrome layer. In other types of masks, the quartz substrate has no overlying layer other than the photoresist pattern.

  Plasma power is applied through each RF impedance matching circuit 28, 30 by placing the inner and outer coil antennas 20, 22 driven by each RF power generator 24, 26 on top. Although the sidewall 12 may be aluminum or other metal coupled to ground, the ceiling 14 is typically an insulating material that allows inductive coupling of RF power from the coil antennas 20, 22 to the chamber 10. Process gas is introduced from the gas panel 36 through the gas manifold 34 and from the evenly spaced injection nozzles 32 at the top of the sidewall 12. The gas panel 36 may comprise a different gas supply 38 coupled to an output valve or mass flow controller 42 coupled to the manifold 34 through each valve or mass flow controller 40.

  The mask support pedestal 16 comprises a metal (eg, aluminum) cathode 44 supported by a metal (eg, aluminum) equipment plate 46. The cathode 44 has an internal coolant or heated fluid flow path (not shown) that is supplied and discharged by a supply and drainage port (not shown) on the equipment plate 46. RF bias power is applied to the equipment plate by the RF bias generator 48 through the RF impedance matching circuit 50. RF bias power is conducted to the upper surface of the cathode 44 across the interface between the equipment plate 46 and the cathode 44. The cathode 44 has a central plateau 44a on which a square quartz mask or substrate 18 is supported. The size of the plateau is usually matched to the size of the mask 18, but the plateau 44a is somewhat small so that a small portion or edge 18a around the mask extends a short distance beyond the plateau 44a. This will be described later.

  The pedestal gas injection ring 52 surrounding the plateau 44a is divided into a cover gas injection ring 52a that forms approximately 2/5 of the gas injection ring 52 and a trap gas injection ring 52b that forms the remaining 3/5 of the gas injection ring 52. (With the wedge or pie cross section shown in FIG. 2B or FIG. 7). The trap gas injection ring 52b has a shelf 54 on which the edge 18a of the mask 18 rests. Three lift pins 56 (only one of which is visible in FIG. 1) lift the trapped gas injection ring 52b and lift the mask 18 at the edge 18a whenever it is desired to remove the mask 18 from the support pedestal 16. The pedestal gas injection ring 52 consists of material layers 53, 55 of different electrical characteristics that are selected to match the RF impedance provided by the combination of the quartz mask 18 and the aluminum plateau 44a at the frequency of the bias generator 48. (Both the cover ring 52a and the capture ring 52b are composed of different layers 53 and 55.) Furthermore, the upper surface of the capture gas injection ring 52 is flush with the upper surface of the mask 18 and extends beyond the end of the mask 18. The wide uniform surface promotes a uniform electric field and sheath voltage across the mask 18 surface during plasma processing. Generally, these conditions are satisfied when the lower gas injection ring layer 55 is made of quartz and the upper gas injection ring layer 53 is made of ceramic such as alumina. The process controller 60 controls the gas panel 36, the RF generators 24, 26, 48 and the wafer handling device 61. The wafer handling apparatus may include a lift servo 62 coupled to lift pins 56, a robot blade arm 63, and a slit valve 64 on the side wall 12 of the chamber 10.

  A series of uniformly spaced titanium screws 70 secure the cathode 44 and equipment plate 46 together around their perimeter. Due to the electrical difference between the aluminum cathode / equipment plates 44, 46 and the titanium screw 70, the screw 70 provides separate non-uniformities in the RF field on the top surface of the cathode 44. Variations in the opposing surfaces of the cathode 44 and the equipment plate 46 create non-uniformities along the periphery of the conductivity between the cathode 44 and the equipment plate 46. Corresponding non-uniformities are imparted to the RF electric field. Because the cathode 44 tends to lift at its center during plasma processing (due to chamber vacuum), the main electrical contact between the cathode 44 and the equipment plate 46 is along its periphery. In order to reduce the sensitivity of electrical conduction between the cathode 44 and the equipment plate 46 which leads to variations in tightness in various titanium screws 70 and (b) variations in surface properties, an annular thin film 72 of a highly conductive material such as nickel is used. , Deposited around the lower surface 44b of the cathode 44. Meanwhile, a matching ring-shaped thin film 74 of nickel (for example) is deposited around the upper surface 46 a of the equipment plate 46. The nickel films 72, 72 are aligned with each other, and the two annular nickel thin films 72, 74 constitute opposing contact surfaces of the pedestal 44 and the equipment plate 46 and are very electrically conductive therebetween. To be evenly distributed. A further improvement in uniform electrical conductivity is achieved by providing an annular groove 76 along the periphery of the lower surface of the cathode 44 and placing a conductive RF gasket 80 in the groove 76. Optionally, a similar annular groove 78 aligned with the groove 76 may be provided on the top surface of the equipment plate 46. Because the cathode 44 and equipment plate 46 are pressed together and the screw 70 is tightened, the RF gasket 80 may be of any suitable conventional type, such as a compressed thin metal helix. In order to reduce or eliminate point non-uniformities in the electric field distribution that tend to occur at the head of the titanium screw 70, a continuous titanium gas injection ring 82 is placed in the annular groove 84 around the top surface of the cathode 44.

  FIG. 2A shows the mask support pedestal 16 and the lift assembly 90 below it. The lift assembly 90 includes a lift spider 92 driven by a pneumatic actuator or lift servo 94 and three lift pins 56 mounted on the lift spider 92. The lift pin 56 is guided by a lift bellows 96 that includes a rolling bearing 98 for very smooth and substantially frictionless movement (to reduce contamination caused by wear). FIG. 2B shows the cathode 44 with the trapped gas injection ring 52b and the mask 18 in the raised position. When the mask is raised, the robot blade can come into contact with the mask 18 by the void formed by the separation of the cover ring 52a and the capture ring 52b.

  The problem of very low center etch rate distribution across the entire surface of mask 18 is solved by changing the distribution of the electrical properties (eg, electrical permittivity) of cathode plateau 44a. This is accomplished in one embodiment by providing a central insert 102 and a surrounding outer insert 104 on the top surface of the plateau 44a. The two inserts form a flat surface continuous with the pedestal gas injection ring 52 and are made of electrically different materials. For example, in order to reduce the tendency for the etch rate distribution to be very low at the center, the center insert 102 is made of a conductive material (eg, aluminum) and the outer insert 104 is made of an insulating material (eg, a ceramic such as alumina). To do. By making the center insert 102 conductive in this manner, the impedance path for the RF current is much lower and boosts the ion energy and etch rate at the center of the mask 18. On the other hand, the insulating outer insert 104 exhibits a high impedance, which reduces the etch rate around the mask 18. This combination improves the etching rate distribution and makes it more uniform. With this feature, the etching rate distribution can be finely adjusted by adjusting the relative RF power levels applied to the inner and outer coil antennas 20,22. Changes in the radial distribution of plasma ion density necessary to achieve a uniform etch rate distribution are reduced to a very small amount, within the capability of RF power distribution between the inner and outer coils 20, 22, and uniform. Etch rate distribution is achieved. FIG. 3 is a plan view of the inner insert 102 and the outer insert 104. In an alternative embodiment, the inserts 102, 104 may be insulators having different dielectric constants (electrical dielectric constants). 4 and 5 show the details of this concept. The four concentric rings 102, 104, 106, 108 with gradually different electrical characteristics are used to make the etching rate distribution more uniform. 6 and 7 show a modified embodiment in which real-time adjustment of the distribution of the RF electrical characteristics of the cathode 44 is performed. The plunger 110 adjusts the axial position of the movable aluminum plate 112 in the hollow cylinder 114 inside the cathode 44. The aluminum plate 112 is in electrical contact with the remainder of the aluminum plateau 44a. An insulator (eg, ceramic) top film 116 can cover the top of the cathode 44. Pushing the aluminum plate 112 near the top of the cylinder 14 reduces the electrical impedance through the central region of the cathode 44 and increases the etch rate at the center of the mask 18. Conversely, if the aluminum plate 112 is moved downward from the cylinder 114 away from the mask 18, the etching rate at the center of the mask decreases. An actuator 118 that controls the axial movement of the plunger 110 can be managed by the process controller 60 (FIG. 1) to adjust the etch rate distribution to maximize uniformity or compensate for non-uniformity. .

Etch rate monitoring and endpoint detection from the back side of the mask The high manufacturing cost of periodically interrupting the etching process to measure the etching depth or critical dimension of the mask, through the cathode 44 and of the mask or substrate 18 Reduce or eliminate by using optical sensing through the backside. Performing such periodic measurements has required interruption of the etching process due to poor etch selectivity to the photoresist. Usually, the mask material is etched later than the photoresist. This problem is generally addressed by depositing a thick layer of photoresist on the mask, but high rate etching of the resist causes the photoresist surface to be randomly non-uniform or rough. This roughness affects the light passing through the photoresist and introduces noise in the optical measurement of critical dimensions or etch depth. Therefore, the photoresist is temporarily removed for each period measurement to ensure noise-free optical measurement. Prior to restarting the interrupted mask etch process, it is necessary to re-deposit the photoresist and redraw the reticle pattern on the photoresist.

  The mask etch plasma reactor shown in FIG. 8 eliminates these problems and allows continuous measurement of critical dimensions or measurement of etch depth during the entire etching process. Meanwhile, the mask or substrate 18 remains in place on the mask support pedestal 16 using the backside optical measurement device provided within the cathode 44. The back side measuring device utilizes the optically transparent nature of the mask substrate 18 which is generally quartz. The thin film deposited thereon (such as chromium silicide or molybdenum) may be opaque, but the formation of the patterned openings that define the reticle pattern of the mask 18 can be optically sensed. The change in light intensity reflected by or transmitted through such a layer is measured behind the mask through the cathode 44. This measurement may be used to perform etch process endpoint detection. When etching quartz material, optical depth measured on the back side of the mask through the cathode 44 may be sensed to perform etch depth measurements in real time during the etching process. One advantage is that the image or optical signal sensed from the back side of the mask is not affected by photoresist noise, or at least affected compared to trying to perform such a measurement from the top surface of the mask 18 (the photoresist side). There is very little.

  For these purposes, the reactor of FIG. 8 has a recess 120 in the upper surface of the cathode 44, in which a lens 122 is housed, with the optical axis facing the back side of the mask or substrate 18. ing. A pair of optical fibers 124 and 126 having a small diameter with respect to the lens 122 have end portions 124a and 126a in the vicinity of or in contact with the lens 122, and both are arranged next to each other along the optical axis of the lens 122. . Each of the optical fibers 124 and 126 shown in FIG. 8 may actually be a small bundle of optical fibers. The optical fiber 124 has a second end 124 b coupled to the light source 128. The light source emits light at a wavelength at which the mask 18 is transparent, typically a visible wavelength for a quartz mask. For interference depth measurement, the wavelength spectrum of the light source 128 is selected to promote local coherence of the reticle pattern of the mask 18. For periodic features (or periodic feature sizes less than 1 micron) in an etched mask structure of about 45 nm, this requirement is met when the light source 128 emits a visible light spectrum. The optical fiber 126 has a second end 126 b coupled to the optical receiver 130. In the case of simple end point detection, the optical receiver 130 may simply detect the light intensity. In the case of a critical dimension (eg, line width) measurement, the optical receiver 130 may sense an image of an etched line in the field of view of the lens 122. From this, the line width is determined. In the case of etching depth measurement, the optical receiver 130 may detect an interference pattern or interference fringe. From this, the etch depth is determined (ie inferred from the interference or diffraction pattern or calculated from the interference fringe count). In other embodiments, the optical receiver 130 may include a spectrometer for performing multi-wavelength interference measurements. From this, the etch depth is inferred or calculated. For such determination, the process controller 60 includes an optical signal processor 132 that can process the optical signal from the optical receiver. Such optical signal processing performs each process end point detection from ambient light intensity changes, measures critical dimensions from a two-dimensional image sensed by the optical receiver 130, calculates the etch depth by counting interference fringes, The etch depth is determined from the wavelength interference spectrum (in this case, the optical receiver 130 comprises a spectrometer) (depending on the particular implementation). Alternatively, using such a spectrometer, the end point of the etching process may be detected using light emitted by plasma from the backside of the wafer by light emission analysis and transmitted through the transparent mask 18. In this case, the light source 128 is not used.

  The process controller 60 responds to the process end point detection information (or etching depth measurement information) from the optical signal processor 132 in response to various configurations of the plasma reactor including the RF generators 24, 26, and 48 and the wafer handling device 61. Control elements. Generally, the process controller 60 stops the etching process and removes the mask 18 from the pedestal 16 when the etching process end point is reached.

  FIG. 9 is a graph showing ambient reflected light intensity sensed from the top (photoresist coating) side of the mask as a function of time during the chrome etching process (the chrome thin film on the quartz mask surface is etched according to the mask reticle pattern). ) The large fluctuation shown in the graph of FIG. 9 represents noise induced by the roughness of the upper surface of the photoresist layer. A broken line represents a step function signal hidden in noise. The step function is consistent with the end point of the chrome etching process. FIG. 10 is a graph of the same measurements taken from the backside of the wafer through the cathode 44 of the reactor of FIG. The optical receiver 130 senses the reflected light level. The photoresist induced noise is greatly reduced and the endpoint definition step function is clearly visible in the optical data. The end of the step function indicates the transition point where the reflected light intensity drops when the etching process reaches the bottom of the chromium thin film. At this point, the reflective surface area of chrome suddenly decreases.

  FIGS. 11 and 12 are graphs of light intensity over time (or equally spaced), as sensed by the optical receiver 130 in FIG. The periodic peak of light intensity corresponds to an interference fringe, the spacing of which is the depth of the etch or the thickness between different surfaces of the periodically spaced periodic features etched in the transparent quartz mask substrate 18. Determine the difference. FIG. 11 shows the intensity sensed through the photoresist from the top of the mask with severe photoresist induced noise components that impair interference fringe detection. FIG. 12 shows the intensity sensed through the back of the mask by the optical receiver 130 of FIG. There is little photoresist induced noise.

  FIG. 13 is a graph showing the light intensity as a function of wavelength when the optical receiver 130 is a spectrometer and the light source 128 generates a spectrum of wavelengths. The behavior of the intensity spectrum in the graph of FIG. 13 is unique when interference effects occur between light reflected from different depth surfaces in sub-micron features periodically spaced by the transparent mask 18. At low wavelengths, the peaks are fairly periodic and are evenly spaced. A significant optical effect is interference. At high wavelengths, the local coherence between the periodic features of the mask 18 is not strong and the effects of diffraction become increasingly greater with increasing wavelength. As a result, the intensity behavior at high wavelengths becomes more complicated rather than very uniform as shown in FIG. The peak spacing in FIG. 13 is a function of the etching depth, especially at low wavelengths, and is inferred from the spacing between peaks.

  FIG. 14 shows an embodiment of the reactor of FIG. The optical receiver 130 is an ambient light intensity detector and the optical signal processor 132 is programmed to look for a large curve (step function) in the total reflected light intensity corresponding to the endpoint detection graph of FIG. The light source 128 of the present embodiment can be any suitable light source. Alternatively, the light source 128 can be omitted and the light sensor 130 simply responds to light from the plasma transmitted through the transparent mask or substrate 18.

  FIG. 15 is an interference fringe detector in which the optical receiver 130 is sufficiently focused by the lens 122 to determine the interference fringes, and the optical signal processor 132 is programmed to count the interference fringes. The etch depth in the transparent quartz mask 18 is calculated (eg, from intensity versus time data of the type shown in FIG. 12). This calculation gives an almost instantaneous etch depth. This is compared by logic 200 with a user-defined target depth stored in memory 202. The logic 200 detects consistency between the stored value and the measured depth value using conventional numerical matching or minimal routines. Due to the matching, the logic 200 flags an etching end point in the process controller 60.

  FIG. 16 shows an embodiment of the reactor of FIG. 8 using the interferometry technique of FIG. 13 to measure or determine the etch depth in the transparent quartz mask or substrate 18. In this case, the light source 128 emits multiple wavelengths or spectra in the visible range (for periodic mask feature sizes of about a few hundred nanometers or less). The optical receiver 130 is a spectrometer. A signal conditioner and analog to digital converter 220 converts the spectral information collected by the spectrometer 130 (corresponding to the graph of FIG. 13) into digital data that can be processed by the optical signal processor 132. One mode in which end point detection can be performed is to calculate the etching depth from the interval between periodic peaks in the low wavelength range of the data shown in FIG. 13, as described above. Comparison logic 200 may compare the instantaneous measured etch depth with a user-defined target depth stored in memory 202 to determine if the etch process endpoint has been reached. In other modes, the comparison logic 200 converts the digitally represented wavelength spectrum (corresponding to the graph of FIG. 13) representing the instantaneous output of the spectrometer 130 into a known spectrum corresponding to the desired etch depth. It is strong enough to compare with. This known spectrum may be stored in the memory 202. An etch process end-point flag is sent to the process controller 60 by a match or approximate match between the measured spectrum and the stored spectrum detected by the comparison logic 200.

  FIG. 17 shows an embodiment of the reactor of FIG. 8, wherein the optical receiver 130 is an emission spectrometer that can distinguish emission lines from the optical radiation emitted by the plasma in the chamber, and emission analysis (OES). Is implemented. The processor 132 is an OES processor programmed to track (or detect disappearance) the intensity of a selected optical line corresponding to a chemical species indicative of the material of the layer being etched. Upon a predetermined transition (eg, the disappearance of the chrome wavelength line in the OES spectrum during the chrome etch process), the processor 132 sends an etch process endpoint detection flag to the process controller 60.

  FIG. 18 shows an embodiment constructed by having a pair of lenses 230, 232 in recesses 231, 233 spaced apart on the surface of the cathode 44. The lenses 230 and 232 are collected to determine an interference fringe, and the collected light is transmitted by each optical fiber 234 and 236 that faces or contacts each lens 230 and 232. The optical fibers 234, 236 are coupled to an interference detector 238 (which can be either a fringe detector or a spectrometer). Detector 238 has an output coupled to process controller 60. The lenses 230 and 232 receive light from the light source 240 through the optical fibers 242 and 244. This light is reflected back from the upper surface of the mask 18 to the lenses 230 and 232 and transmitted to the detector 238 through the optical fibers 234 and 236. Further, the embodiment of FIG. 18 has a third recess 249 on the cathode surface that houses a third lens 250 that is coupled through an optical fiber 252 to the input of an OES spectrometer 254. The OES processor 256 processes the output of the OES spectrometer 254 to perform end point detection and communicate the result to the process controller 60. The cathode 44 of the embodiment of FIG. 18 is shown in FIG. Three recesses 231, 233, 249 that house each lens 230, 232, 250 are shown. FIG. 20 shows corresponding holes 260, 261, 262 for receiving in an equipment plate 46 optic (not shown) that supports the lenses 230, 232, 250. FIG. 21 is a cross-sectional view showing coupling of an optical fiber to a lens inside the pedestal 16.

  Although the reactors of FIGS. 16, 17 and 18 have been described as using spectrometers 130 (FIGS. 16 and 17) and 254 (FIG. 18), one or more spectrometers 130 or 254 are tuned to a predetermined wavelength. The optical wavelength filter may be replaced. Each of these optical wavelength filters may be combined with a photomultiplier tube to improve the signal amplitude.

Backside Endpoint Detection Mask Etching Process FIGS. 22A and 22B illustrate a process for etching a reticle pattern in the mask quartz material. In FIG. 22A, the quartz mask substrate 210 is covered with a photoresist layer 212 having a periodic structure of spaced lines 214 and openings 216 defined in the photoresist layer 212. In the reactor of FIG. 15 or 16, a CHF ₃ + CF ₄ + Ar crystal etching process gas is introduced into the chamber 10, power is applied by the RF generators 24, 26 and 48, and the crystal material is formed in the photoresist layer 212. Etching is performed in the opening 216. The crystal etch depth is continuously measured by the interference between the light 218 reflected from the etched upper surface of the quartz substrate 210 and the light 219 reflected from the unetched upper surface. The etching process is stopped as soon as the desired etching depth is reached (FIG. 22A). Next, the photoresist is removed to form a desired mask (FIG. 22B).

FIGS. 23A-23E illustrate a photoresist layer having an underlying quartz mask substrate 210, a molybdenum silicide layer 260 (containing silicon molybdenum oxynitride), a chromium layer 262, a chromium oxide antireflective coating 264, and an opening 268. A process for etching a three-layer mask structure of 266 is shown (FIG. 23A). In the step of FIG. 23B, the chrome layer 262 and the anti-reflective coating 264 are Cl ₂ + O ₂ + CF in a plasma reactor chamber with simple reflectance endpoint detection (chamber of FIG. 14) or OES endpoint detection (chamber of FIG. 17). Etching using a chrome etching process such as ₄ . The photoresist layer 266 is removed (FIG. 23C). The molybdenum silicide layer 260 is etched as shown in FIG. 23D using a process gas which is an etching solution of molybdenum silicide such as CF ₆ + Cl ₂ and using the chromium layer 262 as a hard mask. This step is performed in a plasma reactor with an endpoint detector, such as by the simple ambient reflectivity, or by OES endpoint detection, such as the chamber of FIG. In FIG. 23E, the chromium layer 262 and the chromium oxide anti-reflective coating 264 are removed using a chromium etching process such as CH ₃ + CF ₄ + Ar. This step can be performed using the reactor of FIG. 14 or 17 with a simple endpoint detector without etching depth measurement. This leaves a quartz mask substrate with a layer overlying the molybdenum silicide defining the reticle pattern.

24A-24E illustrate a process for manufacturing a binary mask made of periodic chrome lines at the exposed quartz transparent quartz mask side-face periodic intervals. The exposed quartz spacing is alternately etched to a depth where the transmitted light is phase shifted by a desired angle (eg, 180 degrees). FIG. 24A shows an initial structure consisting of a quartz mask substrate 300, a chromium layer 302, a chromium oxide antireflective coating 304 and a photoresist layer 306. In the step of FIG. 24B, the chromium and chromium oxide layers 302, 304 are etched in a process gas of Cl ₂ + O ₂ + CF ₄ in a reactor chamber such as the chamber of FIG. In the step of FIG. 24C, the photoresist layer 306 is removed and then the exposed portion of the quartz mask substrate 300 is etched in a CHF ₃ + CF ₄ + Ar quartz etching process gas, as shown in FIG. 24D. The crystal etching step of FIG. 24D is performed in a reactor chamber that can sense or monitor the etching depth of the crystal mask substrate 300, such as the chamber of FIG. 15 or 16. During the etching process, the instantaneous etching depth is continuously monitored and as soon as the target etching depth is reached with the mask 300, the etching process is stopped. The final result is shown in FIG. 24E.

Continuous Monitoring of Etch Rate Distribution over the Mask Surface FIGS. 25 and 26 show an embodiment of the wafer support pedestal 16 of FIG. On the top surface of the cathode 44 is provided a matrix of backside etch depth sensing elements (lenses and optical fibers). An instantaneous image or sample of the etch rate distribution or etch depth distribution is continuously provided over the entire surface of the mask or substrate during the etching process without interrupting the etching process or otherwise interfering with the mask substrate. The aluminum plateau 44 a has a matrix of openings 320 on its upper surface, and each opening holds a lens 322 facing the back side of the mask substrate 300. A light source 324 provides light through an output optical fiber 326 coupled to each lens 322. Lens 322 provides sufficient light collection to determine interference fringes. An interference detector 328, which can be either a sensor that facilitates fringe counting or a spectrometer, is coupled to an input optical fiber 330 that is coupled to each lens 232. A switch or multiplexer 332 continuously enters light from each input optical fiber 330 into the detector 328. There are three modes in which the apparatus of FIGS. 25 and 26 operates. In the first mode, the etching depth in the field of view of one of the lenses 322 is calculated from the spacing between the interference fringes. In the second mode, the detector 328 is a spectrometer and the etch depth in the field of view of one of the lenses 322 is calculated from the low wavelength peak spacing of the multi-wavelength interference spectrum (corresponding to FIG. 13). In the third mode, the multi-wavelength interference spectrum is detected instantaneously and compared with a library of spectra 340 with known corresponding etch depths. The etching rate distribution is calculated from the etching depth and elapsed time. This distribution records the etch non-uniformity of the process and is provided to the process controller 132. The controller 132 can reduce non-uniformities in the etch rate distribution in response by adjusting the tunable features of the reactor.

  The embodiment of FIGS. 25 and 26 is shown as having a 3 × 3 matrix of etch depth sensors or lenses 322 on the top surface of the plateau 44a, but using any number of rows and columns of such sensors. The matrix may be an nxm matrix. Here, m and n are suitable integers.

  In one embodiment, the process controller 132 may be programmed to estimate whether the etch rate distribution is high center or low center (from etch rate distribution information provided by the spectrometer or sensor 130). In response to this information, the process controller 60 can reduce non-uniformities by adjusting certain adjustable features of the reactor. For example, the process controller 60 may change the RF power distribution between the inner and outer coils 20,22. Alternatively or in addition, the process controller 60 may change the height of the movable aluminum plate 112 in the reactor of FIGS. Feedback from an array or matrix of etch depth sensing elements in plateau 44a allows process controller 60 to improve etch rate distribution uniformity through continuous trial and error of reactor tunable elements.

Real-time configurable process gas distribution FIGS. 27 and 28 illustrate the embodiment of the plasma reactor of FIG. 1 having an array of independently controllable gas injection orifices or nozzles 32. By separately controlling the different nozzles 32, the gas distribution within the chamber 10 can be altered to compensate for the non-uniform distribution of etch rates across the workpiece or mask 18. In the illustrated embodiment, the array of gas injection nozzles 32 is located on the side wall 12 near the ceiling 14. To this end, the reactor includes an upper gas injection ring 338 held between the top of the sidewall 12 and a removable lid 342 having a bottom surface that forms the ceiling 14. An outer shoulder 344 on the bottom surface of the upper gas injection ring 338 is disposed on the upper portion of the side wall 12. An internal shoulder 346 on the top surface of the gas injection ring receives the end of the lid 342. An outer shoulder 348 is provided on the bottom surface of the lid 342 disposed on the inner shoulder 346 of the gas injection ring 338. The gas injection orifice or nozzle 32 is formed in the vertical inner surface 349 of the gas injection ring 338 and is defined as the inner diameter of the gas injection ring 338. The gas flow to each injection nozzle 32 is individually controlled by a separate valve 350. There is one valve 350 for each nozzle 32. The process gas supplied from the gas panel 36 flows through the gas supply line 352. This is coupled to an input port 354 formed in the gas injection ring 338. A gas supply outlet 356 formed in the gas injection ring 338 outputs the process gas received at the input port 354. A series of severable gas flow lines 358 form a continuous connection around the periphery of the gas injection ring 338. This conveys process gas from each gas supply outlet or port 356 to a corresponding set of valves 350.

  In the preferred embodiment, each valve 350 is pneumatically controlled and has an input flow port 350a, an output flow port 350b, a controlled gas outlet port 350c, and a pneumatic control input port 350d. The output port 350c provides a controlled process gas flow to the corresponding single nozzle 32. Process gas flows freely from the input flow port 350a to the output flow port 350b. The compressed air pressure at the control input port 350d determines whether the process gas passing through the flow ports 350a, 350b has turned to the gas outlet port 350c. Since such pneumatic control valves are well known, their internal structure need not be disclosed here. The gas flow lines 358-1 and 358-2 are connected from the gas supply outlets 356-1 and 356-2 to the input flow ports 350a of the valves 350-1 and 350-2. The remaining gas flow lines 358 are connected from the output flow port 350a of one valve 350 to the input flow port 350b of the continuous valve 350, respectively. Thus, the gas flow through the series of valves 350 on the left side of the diagram of FIG. 28 is counterclockwise, while the gas flow through the series of valves 350 on the right side of the diagram of FIG. 28 is clockwise. .

  Gas flow from each output port 356 to a series of valves 350 connected thereto is not blocked by a series of intervening valves 350. Each valve 350 can be “on” to turn on and off the other valves 350 to provide gas flow to the corresponding gas injection orifice 32, which ends the gas flow to that injection orifice. can do. The valve structure processor 360 controls all of the valves 350 and can turn any combination of the valves 350 on and off via the valve control link 362. As described above, in the preferred embodiment, the valve 350 is a pneumatic valve and the control link 362 is a pneumatic (air) tube, eliminating the presence of electrical conductors near the coil antennas 20,22. In the embodiment of FIG. 28, the compressor 364 provides air under pressure to an array of solenoidal (ie, electrically controlled) valves 365, and the pressurized air to the pneumatic control input 350a of each pneumatic valve 350. Control application. The valve structure processor 360 controls the solenoid valve 365 through an electrical link remote from the coil antennas 20, 22.

  FIG. 29 shows a variation of the embodiment of FIG. 28 in which the valves 350 are each electrically controlled rather than pneumatically controlled. In FIG. 29, each control link 362 is an electrical line that extends directly from the controller 360 to a corresponding one valve 350, and the array of air compressors 364 and compressed air solenoid valves 365 is eliminated.

  Referring again to FIGS. 27 and 28, each orifice 32 is formed from a radial cylindrical passage 366 through a gas injection ring 338. The hollow cylindrical sleeve 368 is housed in the passage 366 and the tip 368a of the sleeve 368 forms a gas injection orifice. Each sleeve 368 may be formed of a ceramic material and removable. Details of the gas injection orifice or nozzle 32 are described in FIGS. 36A and 36B. A controlled gas outlet port 350 c of each valve 350 is connected from the short gas supply line 370 to the outer end of the corresponding radial passage 366. The entire gas distribution assembly is modular and immediately disassembled by connection (separation) by an outer gas supply line 358 and a short gas supply line 370, respectively. The sleeve 368 can be removed separately from the hole 366. In this way, the gas distribution component and assembly support of the gas injection ring 338 can be easily facilitated separately without the need to remove or replace more expensive components of the reactor, such as the gas injection ring 338, for example. Exchanged.

  30A-30D are graphs of the etch depth distribution across the mask 18 obtained over a period of time for the etch process performed in the reactor of FIGS. 27 and 28 for different valve configurations. The etching distribution in FIG. 30A is obtained when all the valves 350 are opened, and is usually an etching distribution having a low center. The entire mask surface has a high non-uniformity or variation of 0.51%. The distribution of FIG. 30B is obtained by closing the pair of proximity valves 350a and 350b and opening the remaining valves 350, and is a substantially uniform distribution with a non-uniformity or variation of only 0.38%. FIG. 30C was obtained by returning the valve configuration with all valves 350 open. The distribution in FIG. 30C has a lower center. The distribution of FIG. 30D was obtained by closing different pairs of proximity valves 350c, 350d. The resulting distribution was more uniform, not very low in the center, and the variation was only 0.40%.

  FIG. 31 shows an alternative embodiment in which the gas injection nozzles 32 are arranged in a gas injection ring 338 in a zigzag or “W” pattern. Each nozzle is controlled independently as in the previous embodiment. The injection pattern may be moved relative to the ceiling by operating only the upper row 32a or only the lower row 32b of the nozzles. The distance between the nozzles may be changed by operating only the selected nozzle 32 (for example, every third nozzle or every fourth nozzle). FIG. 32 is a cross-sectional view of a portion of the gas injection ring 338 showing how the nozzles 32 may be arranged to spray in different directions. A large change in gas distribution is obtained, for example, by a valve structure controller 360 that turns on only nozzles 32 oriented in a particular direction. For example, all of the nozzles 32c angled toward the right in FIG. 32 may be turned on at the same time except for the other nozzles. For example, turning all of the nozzles 32d angled to the left on, while turning off all others, including all the nozzles 32c angled to the right, can result in significant changes or corrections.

Control of Adjustable Reactor Elements with Feedback from an Array of Backside Etch Depth Measurement Sensors Referring to FIGS. 33 and 34, the output of the two-dimensional array of backside etch depth sensors of FIGS. Feedback control of the adjustable elements of the reactor takes place. The adjustable element or elements may include the array of separately controlled gas injection nozzles 32 of FIGS. Alternatively, or in addition, an adjustable element controlled by such a feedback loop is the RF power distribution between the inner and outer coils 20, 22, or the height of the movable aluminum plate 112 in the reactor of FIGS. May be included.

  With feedback from an array or matrix of etch depth sensing elements 130 in FIGS. 25 and 26, the process controller 60 can improve etch rate distribution uniformity through continuous trial and error adjustment of the reactor tunable elements. . In FIG. 33, the feedback loop begins with an array 400 of backside etch depth sensors 130 of FIGS. The process controller 60 is programmed to use an instantaneous etch depth measurement image of the entire mask 18 to estimate the location and magnitude of the non-uniformity in the etching rate of the mask 18 and reduce or eliminate such non-uniformities. Estimate the most likely change in a particular tunable element in the reactor. This information is converted by the processor 60 into a command (or commands) to be sent to one, some or all of the adjustable elements of the reactor. Thus, FIG. 33 shows the output signal path from the processor controller 60 to any one or all adjustable elements present in the reactor. These are the inner and outer antenna RF power generators 24, 26 (for inner and outer RF power distribution), the actuator 110 for the movable aluminum plate 112, the nozzle array controller 360 of an array of adjustable nozzles 32.

  The feedback loop is operated continuously throughout the entire mask etch process to reduce the non-uniformity read by the processor 60 from the “image” of the etch rate distribution across the mask 18, thereby reducing the etch rate across the mask 18. Improve distribution uniformity. The feedback can be managed by the processor 60 software to perform trial and error correction. Alternatively, the software of the processor 60 can incorporate commercially available neural training and feedback learning techniques that allow the processor 60 to read the non-uniformities in the etch rate distribution with a more intelligent response. . Such software technology does not form part of the present invention.

  In one embodiment, feedback commands for an adjustable element (or elements) are generated to reduce variations in the array of etch depth sensors. In other embodiments, feedback may be selected to address specific non-uniformities. For example, the etch rate distribution sensed by the array of sensors 130 is very high in the quadrants or corners of the mask 18, where the valve structure processor issues a command to limit the gas flow in that quadrant. Reduce to the amount of (trial) done. This measure may increase this adjustment in the gas flow distribution if the subsequent image obtained from the array of backside sensors 130 yields limited results. This adjustment and correction cycle may continue until there is no further improvement in etch rate distribution uniformity.

  Other non-uniformities may be addressed on a similar basis after correcting the first one. For example, assume that the etching rates at different positions are very high. In this case, for a number of samples of the etch rate distribution “image” from the array of backside sensors 130, the gas flow to that location is reduced as long as this non-uniformity results in some reduction.

  In case of symmetric etch rate non-uniformity (eg high at center or low distribution at center), the height of the aluminum plate 112 or the RF power distribution between the inner and outer coils 20, 22 can be adjusted symmetrically Various elements may be used by processor 60 to reduce non-uniformities using a feedback control loop. For example, a low etch rate distribution at the center does not add much non-uniformity by the processor 60 and either raises the aluminum plate 112 or increases the distribution of RF power to the inner coil 20 (relative to the outer coil 22). This (or both) increases the etch rate at the center of the mask 18. In the feedback loop, this change is initially small and as the etch distribution image from the array of backside sensors 130 improves uniformity, the position of the aluminum plate and / or the distribution of power to the inner coil 20 further increases. This cycle may continue until no further improvement is observed. All of the techniques described above may be incorporated into software executed by the process controller 60.

  FIG. 35 shows one possible example of a feedback cycle implemented by the process controller 60 in the embodiment of FIGS. First, the processor 60 obtains at least a two-dimensional image of the etch rate across the mask surface from the array of backside sensors 130 (block 380 in FIG. 35). From this image, the processor 60 estimates a non-uniform pattern of the etch rate distribution (block 382) and selects an adjustment to one of the reactor's tunable elements from a list of options that reduce the non-uniformity (block). 384). After making this adjustment (block 386), the processor 60 obtains the latest etch rate distribution image (block 388) and compares it with the previous image taken before the adjustment. If there is an improvement (reduction in non-uniformity), the processor 60 repeats the same cycle, so that the same smooth adjustment is probably further increased. If there is no improvement (NO in block 390), remove the selected adjustment from the list of options (block 392) and return to the step in block 384 to select a different adjustment.

FIG. 36A is a cross-sectional side view of one embodiment of a gas injection orifice or nozzle 32 inserted into the radial cylindrical passage 366 of the gas injection ring 338 of FIGS. As described above, the nozzle 32 is configured to be easily removed from the radial cylindrical passage 366 for easy replacement. The nozzle 32 includes a hollow cylindrical sleeve or hollow body 368, which may be made of a material compatible with a process such as metal or ceramic. In one embodiment, the hollow body 368 is made of an aluminum or ceramic material, such as an aluminum oxide (Al ₂ O ₃ ) material. The hollow body 368 has a first outer diameter 3620 and a smaller second outer diameter 3615 forming a tip 368a. In one embodiment, the first outer diameter 3620 is about 50% larger than the second outer diameter 3615.

  One challenge presented by other nozzles relates to the removal or installation of the nozzle being replaced. This is because the previous nozzle includes a surface, thread or other coupling mechanism and tends to stick or fasten the nozzle to the passage 366. In addition, because these conventional nozzles must be rotated, turned, pulled or otherwise separated from the passageway, the gas injection ring 338 (FIG. 27) must be removed from the chamber. This creates a long downtime and adversely affects throughput. The nozzle 32 allows for easy replacement by reducing the surface and / or minimizing the connecting connections that tend to stick and fasten. Thus, replacing the nozzle 32 does not require extensive chamber disassembly as previously and minimizes tool downtime.

  The hollow body 368 also includes a longitudinal passage formed therethrough, which includes a first passage 3605 and a second passage 3610. In one embodiment, the hollow body 368 includes a first passage 3605 that begins at a first outer diameter 3620 portion, the first passage 3605 extending at least partially through a second outer diameter 3615 portion of the tip 368a. ing. The first passage 3605 has a first diameter, and the second passage 3610 has a second diameter that is smaller than the first diameter. The first passage abruptly transitions to the second diameter at an interface of about 90 degrees. In one embodiment, the second passage 3610 has a diameter that is about one-fourth smaller than the diameter of the first passage 3605. The second outer diameter 3615 also includes a side passage 3618 that opens to the first passage 3605. The side passageway 3618 may have a diameter that is substantially equal to the diameter of the second passageway 3610 and generally disposed normal to the second outer diameter 3615 or the first passageway 3605.

  36B is an exploded side view of a portion of the body 368 shown in FIG. 36A. The interface between the first outer diameter 3620 portion and the second outer diameter 3615 portion is coupled by a radiation surface 3625. This surface extends orthogonally from the second outer diameter 3615 portion to the first outer diameter 3620 portion. In one embodiment, the radiating surface 3625 has a relief 3630 that may be a chamber, bevel, or radius that transitions to the first outer diameter 3620. Surface 3625 includes an annular groove 3635 formed therein. This may be configured as an o-gas injection ring groove that receives an o-gas injection ring (not shown). In one embodiment, the portion of the first outer diameter 3620 that includes a radially outward surface 3625 from the annular groove 3635 may be removed, as indicated by the dashed line 3622. In this configuration, the surface 3625 may be stepped and the o-gas injection ring is aligned with the stepped portion rather than the annular groove. In certain embodiments, the surface 3625 includes a radius 3640 that transitions to a second outer diameter 3615. The annular groove 3635 (or the stepped portion after the portion indicated by the broken line 3622 is optionally removed) may include a surface without scratches or marks by tools. In one embodiment, the surface may be finished to have an average surface roughness (Ra) of about 16 μm. The surface of the hollow body 368 except for the annular groove 3635 may also be finished to Ra of about 32 μm. In one embodiment, the tip 368a includes a relief 3612 that may be a chamber, bevel, or deformation that relaxes the end of the face of the tip 368a.

  Although the above description relates to an embodiment of the present invention, other and further embodiments of the present invention can be created without departing from the basic scope thereof, the scope of which is a utility model registration request. It is determined based on the range.

  In order that the exemplary embodiments of the present invention may be understood in detail, the invention briefly summarized above will be more specifically described with reference to the embodiments shown in the accompanying drawings. Certain well-known processes are not described herein in order not to obscure the present invention.

1 is a schematic cross-sectional view of one embodiment of a plasma reactor for performing a mask etching process. FIG. 2 is an isometric cutaway view of the lower portion of the reactor of FIG. 1. FIG. 2 is an isometric partial cutaway view of the mask support pedestal of the reactor of FIG. 1 in the raised position. It is a top view of the cathode of the reactor of FIG. ~ FIG. 6 is a plan view and a side view of a modified embodiment of the cathode of the reactor of FIG. 1. ~ FIG. 6 is a plan view and a side view of another modified embodiment of the cathode of the reactor of FIG. 1. It is a schematic side view of the plasma reactor which has a back side end point detection apparatus. ~ It is a graph of the optical end point detection signal obtained from the front side and the back side of the mask. ~ It is a graph of the interference fringe optical signal obtained from the front side and the back side of the mask. FIG. 9 is a graph of a multi-wavelength interference spectrum signal obtained in one embodiment of the reactor of FIG. FIG. 11 is a schematic side view of one embodiment of the reactor of FIG. 8 with backside endpoint detection based on the overall reflected light intensity corresponding to FIG. FIG. 13 is a schematic side view of one embodiment of the reactor of FIG. 8 with backside endpoint detection based on interference fringe counting corresponding to FIG. FIG. 9 is a schematic side view of one embodiment of the reactor of FIG. 8 with backside endpoint detection based on multi-wavelength interferometry. FIG. 9 is a schematic side view of one embodiment of the reactor of FIG. 8 with backside endpoint detection based on luminescence analysis (OES). FIG. 9 is a schematic side view of the reactor of FIG. 8 illustrating an example of backside endpoint detection based on both OES and interference. ~ FIG. 19 is an isometric view of the cathode and equipment plate of the embodiment of FIG. FIG. 20 is a cross-sectional view of the cathode of FIG. 19. ~ It is a figure which shows a series of processes in the quartz mask etching process using back side end point detection. ~ FIG. 6 shows a series of steps in a chromium-molysilicide-quartz mask etching process using backside endpoint detection. ~ FIG. 6 shows a series of steps in a chrome-quartz mask etching process using backside endpoint detection. ~ FIG. 6 is a schematic side view and an end view of an embodiment in which a real-time etching rate distribution is continuously measured from the back side of the mask. ~ FIG. 3 is an isometric and plan view of an embodiment having an array of separately controllable gas injection nozzles. FIG. 29 is a plan view of an implementation of the embodiment of FIGS. 27 and 28 utilizing a pneumatic valve. ~ FIG. 29 is a graph of etch depth distribution across a mask obtained with a different array of valves of FIGS. 27 and 28 operated. FIG. 29 is an isometric view of the alternative embodiment of FIGS. 27 and 28; FIG. 29 shows another modified embodiment of the reactor of FIGS. 27 and 28. ~ 2 is a block diagram and perspective view of a plasma reactor capable of implementing real-time feedback control of a reactor tunable element based on an instantaneous two-dimensional image of an etch rate distribution. FIG. FIG. 35 is a block diagram of a feedback control process that can be implemented in the reactor of FIGS. 33 and 34. It is a sectional side view of one embodiment of a gas injection nozzle. FIG. 36B is an exploded side view of a part of the main body of the nozzle of FIG. 36A.

  To facilitate understanding, identical reference numerals have been used, where possible, to designate identical elements that are common to the figures. The components and features of one embodiment are not specifically listed, but are believed to be advantageously incorporated into other embodiments. However, it should be noted that the accompanying drawings illustrate only typical embodiments of the present invention and are not considered to limit the scope thereof, and the present invention includes other similarly effective embodiments. It is.

Claims (20)

In a plasma reactor, a gas injection nozzle adapted to be modularly coupled to a gas distribution ring, comprising:
A hollow cylindrical body having a first outer diameter portion and a second outer diameter portion defining a tip; and formed longitudinally through the body and at least partially through the first outer diameter portion. A first passage extending to the second outer diameter portion and a coaxial alignment with the first passage, and extending in a longitudinal direction from an end of the first passage to an end of the tip A gas injection nozzle, wherein the first outer diameter portion is larger than the second outer diameter portion.   The nozzle according to claim 1, wherein a side passage disposed at the second outer diameter communicates with the first passage.   The nozzle of claim 1, wherein the first passage has a diameter that is four times larger than the diameter of the second passage.   The nozzle according to claim 1, wherein the hollow cylindrical body includes a radiating surface, and the radiating surface extends orthogonally from the second outer diameter portion to the first outer diameter portion.   The nozzle according to claim 4, wherein the radiation surface includes an annular groove.   The nozzle according to claim 1, wherein the hollow cylindrical body is made of a ceramic material. In a plasma reactor, a gas injection nozzle adapted to be modularly coupled to a gas distribution ring, comprising:
A hollow cylindrical body having a first outer diameter portion and a second outer diameter portion smaller than the first outer diameter diameter portion, and disposed through the first diameter portion, wherein the second diameter portion is A first passage that transitions to a second passage at a 90 degree interface to a second passage disposed therethrough and radiation that couples the first outer diameter portion to the second outer diameter portion A gas injection nozzle, wherein the first outer diameter portion is about 50% larger than the second outer diameter portion.   The nozzle according to claim 7, wherein the radiation surface is disposed orthogonal to the second outer diameter portion.   The nozzle according to claim 7, wherein the radiation surface includes an annular groove.   The nozzle according to claim 9, wherein the annular groove has an average surface roughness of 16 μm.   The nozzle according to claim 9, wherein the hollow cylindrical body is made of a ceramic material.   The nozzle of claim 9, wherein the first passage has a diameter that is approximately four times larger than the diameter of the second passage.   The nozzle according to claim 9, wherein a side passage disposed in the second outer diameter portion communicates with the first passage. In a plasma reactor, a gas injection nozzle adapted to be modularly coupled to a gas distribution ring, comprising:
A hollow cylindrical body having a first outer diameter portion and a second outer diameter portion defining a tip, wherein the first outer diameter portion is larger than the second outer diameter portion; A first passage formed longitudinally through the body and extending at least partially through the first outer diameter portion to the second outer diameter portion; and coaxial with the first passage A second passage that is aligned and transitioned at an interface of 90 degrees and extends longitudinally from the end of the first passage to the end of the tip; and communicates with the first passage And a side passage disposed in the second outer diameter portion, wherein the first passage has a diameter four times larger than the diameter of the second passage.   The nozzle according to claim 14, wherein the first outer diameter portion and the second outer diameter portion include surfaces having an average surface roughness of 32 μm.   The nozzle according to claim 14, wherein the hollow cylindrical body includes a radiating surface extending orthogonally from the second outer diameter portion to the first outer diameter portion.   The nozzle of claim 16, wherein the radiating surface includes an annular groove.   The nozzle according to claim 17, wherein the annular groove has an average surface roughness of 16 μm.   The nozzle according to claim 14, wherein the hollow cylindrical body is made of a ceramic material.   The nozzle according to claim 14, wherein the hollow cylindrical body is made of an aluminum oxide material.

Images