Patent Application
20160064184
1. Field of the Disclosure
The present application relates to scanning electron microscopes, electron and X-ray detectors suitable for use in scanning electron microscopes, and systems and methods for reviewing and inspecting samples. The electron microscopes, detectors, systems and methods are particularly suitable for use in review and inspection systems including those used to review and/or inspect photomasks, reticles, and semiconductor wafers.
2. Related Art
The integrated circuit industry requires inspection tools with increasingly higher sensitivity to detect ever smaller defects and particles whose sizes may be a few tens of nanometers (nm), or less. These inspection tools must operate at high speed in order to inspect a large fraction, or even 100%, of the area of photomask, reticle or wafer, in a short period of time, e.g. one hour or less for inspection during production, or, at most, few hours for R&D or troubleshooting. In order to inspect so quickly, inspection tools use pixel or spot sizes larger than the dimensions of the defect or particle of interest, and detect just a small change in signal caused by a defect or particle. High speed inspection is most commonly performed in production using inspection tools operating with UV light. Inspection in R&D may be performed with UV light or with electrons.
Once a defect or particle has been found by high speed inspection, it is often necessary to make a higher resolution image and/or to perform material analysis to determine the origin or type of the particle or defect. This process is commonly called review. Review is usually performed with a scanning electron microscope (SEM). Review SEMs used in the semiconductor manufacturing process are typically required to review many thousands of potential defects or particles per day and so may have, at most, a few seconds per target for review. Review SEMs for the semiconductor and related industries are manufactured by KLA-Tencor Corporation (for example the eDR-7110), by Applied Materials, Inc. (for example the SEMVision G6) and other companies.
A review SEM most commonly detects secondary electrons emitted from the sample in order to form an image. An exemplary secondary electron detector for a review SEM is described in U.S. Pat. No. 7,141,791 to Masnaghetti et al. entitled “Apparatus and method for e-beam dark-field imaging”. This exemplary secondary electron detector includes electron optics for collecting secondary electrons and directing them to a scintillator. The electrons are accelerated towards the scintillator so that each electron striking the scintillator causes multiple photons to be emitted. Some of those photons are captured by a light pipe and directed to one or more photomultiplier tubes. A disadvantage of this approach is that the detector is relatively slow. The light emission from scintillator has a decay time constant of several, or many, tens of ns. Furthermore scintillators have multiple time constants. The initial response may have a time constant of a few tens of ns, but light emission will continue at a low level with a much longer time constant. Photomultiplier tubes also have responses with multiple time constants. The photocathode that emits photoelectrons has one, or more, time constants. The electrons take a significant time to travel from one dynode to another and finally to the anode, which creates an additional time constant. The electron travel time can be reduced by reducing the number of dynodes, but this reduces the gain of the photomultiplier tube and so is not a desirable trade-off as it reduces the sensitivity of the SEM in order to improve the speed.
SEMs designed for review may include electron microprobe (X-ray) analysis for material identification. In order for an SEM to have image resolution of a few nm or better to provide high-quality images of nm-sized defects and particles, the sample being inspected is placed close to the final objective lens so that it is immersed in the magnetic field of that lens, thus minimizing imaging aberrations. Placing the sample close to the objective lens prevents large detectors being placed close to the sample. In particular, X-ray detectors used for microprobe or similar analysis can only collect X-rays in a small solid angle, making such systems very slow. Many tens of seconds or minutes of data acquisition time may be required per target to capture enough X-rays to determine the material composition of the target.
The final objective lens of a review SEM also limits where secondary and backscattered electron detectors may be placed and the collection voltages that can be applied to those detectors. A small potential difference (such as less than about 2 kV) between an electron detector and the sample reduces the efficiency and sensitivity of an electron detector to collect and detect low-energy electrons from the sample.
Therefore, a need arises for a high-speed high-resolution review SEM overcoming some, or all, of the above disadvantages. In particular a need arises for a high-speed high-resolution automated SEM with capability to quickly identify at least some commonly used materials. It is further desirable that capability to quickly identify materials and/or provide improved image contrast be included in a high-speed inspection SEM.
The present invention is directed to a SEM that utilizes one or more solid-state electron detectors to achieve high-speed detection of either back-scattered or secondary electrons emitted from a sample by way of converting incident electrons into measurable charges entirely within a single integral semiconductor structure. Specifically, each solid-state electron detector includes a sensor that utilizes a p-type electron-sensitive layer to generate multiple electrons in response to each incident (detected) electron, utilizes an n-type buried channel layer to transfer at least some of the generated electrons to an n+ floating diffusion, and utilizes an amplifier controlled by a charge (voltage) collected on the floating diffusion to generate an output signal, wherein the p-type electron-sensitive layer, the n-type buried channel layer, the n+ floating diffusion and the amplifier comprise respective doped regions of the single integral semiconductor (e.g., epitaxial silicon) structure. Converting incident electrons into measurable charges entirely within an integral semiconductor structure in this manner is substantially faster than the conventional photon-based scintillator approach, whereby the present invention provides an SEM that is capable of substantially higher processing speeds (e.g., 100 MHz or higher) than is possible using conventional scintillator-based SEMs. The solid-state electron detectors are also smaller in size and generate smaller detector-to-sample potential differences, which facilitates producing SEMs including a back-scattered electron detector disposed close to the electron source (e.g., between the final (immersion) objective lens and the sample, or above the final objective lens) that operates at the same high operating speed as a secondary electron detector, whereby both secondary electron signals and back-scattered electron signals may be used in combination to provide higher resolution information about the sample's surface topography than can be obtained from either signal on its own. Moreover, by fabricating each electron sensor on a single semiconductor structure using known semiconductor processing techniques, the solid-state electron detectors can be lower in overall cost and require lower operating voltages than conventional scintillator-based sensors, thereby facilitating the production of SEMs that, in comparison to conventional scintillator-based SEM systems, are less expensive to produce, and exhibit substantially higher efficiency and are thus less expensive to operate.
Exemplary inspection and review SEMs utilizing one or more solid-state electron detectors of the type mentioned above are described. The SEM includes an electron source, an electron optical system (electron optics), at least one solid-state electron detector, and a computer. The electron source generates a primary electron beam that is directed towards a sample. The electron optics includes lenses and deflectors configured to de-magnify, focus and scan the primary electron beam across an area of the sample to be inspected. When the primary electron beam strikes the sample, the sample absorbs many of the electrons from the primary electron beam, but scatters some of the electrons (back-scattered electrons). The absorbed energy causes secondary electrons to be emitted from the sample, along with some X-rays and Auger electrons. A back-scattered electron (first) electron detector is positioned close to the sample, whereby analog output signals having voltage levels that are proportional to the numbers and energies of detected back-scattered electrons are converted to corresponding digital values and transmitted to the computer as corresponding (first) image data signals. The secondary electrons are detected by an optional (second) solid-state electron detector whose sensor generates analog output signals that are proportional to the numbers and energies of detected secondary electrons, which are converted to corresponding digital values and transmitted to the computer as corresponding (second) image data signals. The computer receives the first and second image data signals from the first and second solid state detectors, and then processes the received image data signals to construct an image of the area of the sample over which the primary electron beam was scanned.
In a preferred embodiment, both the secondary electron detector and the back-scattered electron detector comprise solid-state detectors. In a preferred embodiment, the back-scattered electron detector has a pure boron coating on the surface that detects electrons (i.e., the front-facing surface that faces the sample or other electron source). In another embodiment, both the back-scattered electron detector and the secondary electron detector include the pure boron coatings.
An exemplary method of inspecting or reviewing a sample is described. The method includes generating a master clock signal, generating a deflection scan synchronized with the master clock, the deflection scan causing a primary electron beam to scan an area of a sample, and generating a first pixel clock synchronized to the master clock to collect and digitize back-scattered electron signals. The method further includes determining approximately the energy of each back-scattered electron from the charge generated in each pixel. The numbers and energies of the collected back-scattered electrons may be used to determine the presence or absence of a defect or defect type, to classify a defect type, or determine a material type or material class at a location in the scanned area of the sample.
In preferred embodiments of the method, the first pixel clock, or a second pixel clock synchronized to the master clock, is used to collect and digitize secondary electrons. The secondary electrons may be used to form an image of the scanned area of the sample. The image may be formed from the combined back-scattered electron and secondary electron signals. The secondary electron signals may be used in combination with the back-scattered electron signals to determine the presence or absence of a defect or defect type, to classify a defect type, or determine a material type or material class at a location in the scanned area of the sample. The combined signals may also provide more information about the surface topography of the sample than can be obtained from either signal on its own.
According to another embodiment of the invention, an electron detector comprises an array of pixels and multiple analog-to-digital converters, where each pixel functions in the manner described above to generate an analog output signal, and each analog-to-digital converter is connected to convert the analog output signal from only one associated pixel in order to facilitate both high-speed and high-resolution detection/readout operations. The pixel array includes multiple pixels arranged in rows and columns (e.g., 16×16, 32×32, 64×64 or more), thereby facilitating the detection of incident electrons over a large area. Similar to the generalized electron sensor mentioned above, each pixel includes a p-type electron-sensitive region, an n-type buried channel layer, a floating diffusion and an amplifier circuit configured to generate an analog output signal whose level corresponds approximately to the energy of an incident electron or the numbers of electrons (i.e., a back-scattered or secondary electron) entering that pixel. By utilizing multiple analog-to-digital converters, each configured to process the output signal from one pixel, the present invention provides a multi-pixel (i.e., 4×4 or greater) electron detector capable of substantially higher operating speeds (e.g., 100 MHz or higher sampling rate for each pixel) than is possible using conventional detector arrangements. Moreover, because the output signals from multiple pixels disposed in a matrix (array) are simultaneously converted for processing, the multi-pixel electron detector of the present invention facilitates both measuring the energy of more than one incident electron received during a given detection/readout operation, and also facilitates determining the path of incident electrons by way of the location of the detecting pixel in the matrix.
In one embodiment, the pixels are fabricated on a separate semiconductor structure from the analog-to-digital converters, and output signals are transmitted from each pixel to its associated analog-to-digital converters by way of a corresponding solder ball. In a preferred embodiment, the pixels are fabricated on a layer of lightly p-doped epitaxial (epi) silicon as part of a sensor circuit, and the analog-to-digital converters are fabricated on a second semiconductor (e.g., silicon) substrate along with other digital circuitry as part of a signal processing circuit (e.g., as part of an ASIC—application-specific integrated circuit). Preferably the thickness of the epi silicon used to form the pixels is between about 40 μm and 100 μm in order to keep the time taken for electrons to drift from the electron-sensitive region to the n-type buried channel layer limited to less than about 10 ns, while providing good mechanical strength. Depending on the mechanical support provided by the substrate that the silicon is attached to, silicon thinner than 40 μm, such as between about 10 μm and 40 μm may be acceptable. In one embodiment, a pure boron coating is disposed on the electron-sensitive surface of the epi silicon. In one embodiment, in addition to the array of analog-to-digital converters the signal processing circuit includes processing circuitry configured, for example, to calculate the approximate energy of an incident electron based on the digitized output signal (image data) received from an associated pixel of the sensor circuit. In another embodiment, the signal processing circuit also includes high speed data transmission circuitry for transmitting an image data signal to an external processing system (e.g., a computer). After the sensor circuit and the signal processing circuit are fabricated, they are connected in a stacked arrangement with solder balls connected between each pixel and an associated analog-to-digital converter. Specifically, the output signal transmitted from each pixel is connected to a first pad disposed on a surface of the sensor circuit, and from the first pad by way of an associated solder ball/bump to a second pad disposed on the signal processing circuit, from which the output signal is transmitted to the input terminal of an associated analog-to-digital converter. This connection can also provide mechanical support for the sensor circuit. The various control and power signals utilized by the pixels may be transmitted by shared metal interconnects (signal lines) from a control circuit, which in turn may be connected to the signal processing circuit or another substrate by solder bumps or wire bonds.
In a preferred embodiment, the floating diffusion of each said pixel is located in the pixel's central region, and the lateral dimensions of each pixel are limited in order to facilitate sufficient time for the electrons to transfer from the buried channel layer to the floating diffusion at the high speed data collection. By locating the floating diffusion in the center of each pixel, the path any electron is required to travel through the buried channel layer to the floating diffusion is equal to one-half of the maximum lateral (e.g., diagonal) dimension of a pixel. Limiting the nominal lateral dimensions of each pixel to about 250 μm or less facilitates operating speeds of up to 100 MHz.
In one embodiment, a separate (third) substrate is electrically and/or mechanically connected to one or more of the sensor circuit and the signal processing circuit. The separate substrate may comprise silicon or a ceramic material, and may include an integrated circuit comprising circuits for processing the analog or digital signals from each pixel, and for providing high speed data transfer of the image data to an external computer or other system. Processing functions performed by the integrated circuit may include thresholding, summing, binning and/or counting data from the individual pixels. The integrated circuit preferably includes a high-speed (such as about 10 Giga-bits per second) serial transmitter for transmitting digitized data to a computer. The integrated circuit may include a serial receiver for receiving commands from the computer. The serial receiver may operate at a lower speed than the serial transmitter.
The present invention is further directed to a novel electron sensor pixel in which a resistive gate and one or more optional additional gates are utilized to drive electrons toward the centrally located floating diffusion. The resistive gate is implemented by an amorphous or polycrystalline silicon structure disposed over most of the pixel surface. A potential difference applied between an outer periphery and an inner periphery of the resistive gate creates an electric field that drives electrons in the buried channel towards a floating diffusion that is, preferably, located near the center of the pixel to speed up charge transfer. Various additional gates are fabricated on the front surface of the pixel between the resistive gate and the floating diffusion to direct and control the transfer of the charge from the buried channel to the floating diffusion and to allow reset of the floating diffusion. The pixel's amplifier is fabricated in a p-well region on the front surface of the pixel to buffer the signal collected in the floating diffusion.
a, 3b and 3c illustrate key aspects of an exemplary solid-state electron detector comprising multiple pixels with one output per pixel in accordance with an embodiment of the present invention.
a and 4b illustrate in exploded and assembled front/top perspective views a single pixel electron sensor according to an exemplary specific embodiment of the present invention.
a and 5b are simplified cross-sectional views showing the pixel of
The present invention relates to an improvement in sensors for semiconductor inspection and review systems. The following description is presented to enable one of ordinary skill in the art to make and use the invention as provided in the context of a particular application and its requirements. As used herein, directional terms such as “top”, “bottom”, “over”, “under”, “upper”, “upward”, “lower”, “down” and “downward” are intended to provide relative positions for purposes of description, and are not intended to designate an absolute frame of reference. In addition, the phrase “integral semiconductor structure” is used herein to describe a contiguous semiconductor material (e.g., silicon) substrate entirely formed during a single fabrication process (e.g., Czochralski crystal growth, sputter deposition, plasma vapor deposition, or chemical vapor deposition), as distinguished from two separate semiconductor structures (e.g., two “chips” from the same silicon wafer) that have been connected by way of adhesive, solder, or other interconnect. Various modifications to the preferred embodiment will be apparent to those with skill in the art, and the general principles defined herein may be applied to other embodiments. Therefore, the present invention is not intended to be limited to the particular embodiments shown and described, but is to be accorded the widest scope consistent with the principles and novel features herein disclosed.
In one embodiment electron gun 140 comprises a cathode 101 such as a thermal field-emitting or Schottky cathode, a single-crystal tungsten cathode or a LaB6 cathode, and extraction and focusing electrodes 102. Electron gun 140 may further comprise a magnetic lens (not shown). Electron gun 140 generates a primary electron beam 150 with a desired beam energy and beam current.
Upper column 141 of the electron optics includes one or more condenser lenses 107 that de-magnify the primary beam to create a small spot on the sample 131. Generally spot sizes of about one or a few nm are preferred for generating high-resolution images for review of samples. Inspection of a sample may use larger spot sizes in order to scan the sample 131 more quickly. A single condenser lens 107 may suffice when the spot size is of order 100 nm or larger, but two or more condenser lenses are typically needed for spot sizes of tens of nm or smaller. Condenser lens 107 may comprise a magnetic lens, an electrostatic lens or both. Upper column 141 may also include one or more deflectors 105 that scan the primary electron beam across an area of sample 131. Deflectors 105 may be placed on either side of condenser lens 107 as shown, or within the condenser lens 107 (not shown), or after the condenser lens 107. Deflectors 105 may comprise electrostatic deflectors or a combination of magnetic and electrostatic deflectors. In one embodiment, there may be no deflectors in the upper column 141. Instead all the deflectors may be contained in lower column 142.
Lower column 142 includes a final (immersion) lens 110 for focusing the primary electron beam to a small spot on the sample 131. Final lens 110 may comprise a magnetic lens (as shown) or a combination of a magnetic lens and an electrostatic lens (not shown). In order to achieve a small spot size at the sample 131, final lens 110 is placed close to the sample 131, so that the sample is immersed in the magnetic field of the lens. This can reduce aberrations in the electron spot on the sample 131. Lower column 142 also includes deflectors 109 that work in combination with deflectors 105 (if present) to scan the primary electron beam across an area of the sample 131.
Sample 131 is placed on a stage 130 in order to facilitate movement of different regions of sample 131 underneath the electron column. Stage 130 may comprise an X-Y stage or an R-θ stage, and in one embodiment is configured to support and position numerous sample types typically reviewed by the integrated circuit industry (e.g., an unpatterned semiconductor wafer, a patterned semiconductor wafer, a reticle or a photomask). In preferred embodiments, stage 130 can adjust the height of sample 131 during inspection to maintain focus. In other embodiments, final lens 110 can be adjusted to maintain focus. In some embodiments, a focus or height sensor (not shown) may be mounted on or proximate to final lens 110 in order to provide a signal to adjust the height of sample 131 or to adjust the focus of the final lens 110. In one embodiment, the focus sensor or height sensor may be an optical sensor.
Secondary electrons and back-scattered electrons are emitted from an area of the sample 131 when the primary electron beam 150 is scanned by the electron optics across the area. Secondary electrons may be collected and accelerated by electrodes 120 and directed to secondary electron detector 121. Electron optics for collecting, accelerating and/or focusing secondary electrons are described in U.S. Pat. No. 7,141,791 to Masnaghetti et al., entitled “Apparatus and method for e-beam dark-field imaging”. This patent is incorporated by reference herein. As described in the '791 patent, the electron optics for the secondary electron detector may include de-scanning optics for, at least, partially canceling the effects of the deflectors 109 on the trajectories of the secondary electrons. In some embodiments of this invention the de-scanning electron optics are not needed and may be omitted, as the de-scanning may be approximately achieved by an ASIC included within the secondary electron detector as described herein. Secondary electron detector 121 is preferably a solid-state electron detector, such as one of the solid-state electron detectors described herein, and is configured to generate an image data signal ID2 in accordance with detected secondary electrons, wherein image data signal ID2 is transferred to computer 160 and utilized to generate an image of the associated scanned sample area, whereby visual inspection of a defect D is facilitated. Other electron optics and detector configurations and methods for detecting and analyzing secondary electrons that may be used in combination with the systems and methods described herein are described in U.S. Pat. No. 7,838,833, to Lent et al., entitled “Apparatus and method for e-beam dark imaging with perspective control”, and U.S. Pat. No. 7,714,287 to James et al., entitled “Apparatus and method for obtaining topographical dark-field images in a scanning electron microscope”. Both of these patents are incorporated by reference herein.
Back-scattered electrons may be detected by a back-scattered electron detector, such as those shown at 122a and 122b, which is implemented by one of the solid-state electron detectors described herein, and is configured to generate an image data signal ID1 in accordance with detected back-scattered electrons, where data signal ID1 is transferred to computer 160 and is also utilized to generate the image of the associated scanned sample area. Preferably the back-scattered electron detector is placed as close as possible to the sample 131, such as at location 122a (i.e., between final lens 110 and sample 131). However the gap between the sample 131 and the final lens 110 may be small, such as about 2 mm or less, and clearance may be needed, for example, for a focus or height sensor, and so it may not be practical to place the back-scattered electron detector at location 122a. Alternatively the back-scattered electron detector may be placed at location such as 122b, on the other side of the final lens 110 pole pieces from the sample 131. Note that the back-scattered electron detector must not block primary electron beam 150. The back-scattered electron detector may have a hole in the middle or may comprise multiple detectors (such as two, three or four separate detectors) disposed around the path of the primary electron beam 150 so as not to block that path, while being efficient for capturing back-scattered electrons.
The landing energy of the primary electron beam 150 on the sample 131 depends on the potential difference between the cathode 101 and the sample 131. In one embodiment, the stage 130 and sample 131 may be held close to ground potential, and the landing energy adjusted by changing the potential of the cathode 101. In another embodiment, the landing energy on the sample 131 may be adjusted by changing the potential of the stage 130 and sample 131 relative to ground. In either embodiment, the final lens 110 and the back-scattered electron detector 122a and/or 122b must all be at potentials close to one another and close to that of the sample 131 and stage 130 (such as less than about 1000V relative to the sample 131 and stage 130) in order to avoid arcing to the sample 131. Because of this small potential difference, back-scattered electrons from the sample 131 will be accelerated only a small amount, or not all, from the sample to the back-scattered electron detector 122a and/or 122b. Since the landing energy on the sample 131 may be quite low (such as between about 500 eV and 2 keV) for some semiconductor samples in order to avoid damage to those samples, the energy of the back-scattered electrons as they land on the back-scattered electron detector 122a and/or 122b will be quite low. Thus it is important for the sensitivity of the SEM that the back-scattered electron detectors 122a and 122b generate many electron-hole pairs from a single low-energy back-scattered electron (such as an electron having an energy of about 2 keV or less). Conventional silicon detectors unavoidably have a thin oxide, such as a native oxide, coating on the surface of the silicon, which blocks most electrons with energies below about 2 keV from reaching the silicon, or alternatively have a thin metal (such as Al) coating on the surface which scatters and absorbs a significant fraction of incident low-energy electrons. In a preferred embodiment, solid-state electron detectors described herein have a pin-hole free pure boron coating on their surface. A pin-hole free pure boron coating prevents oxidation of the silicon and allows efficient detection of low energy electrons (including electrons with energies less than 1 keV). Methods for fabricating silicon detectors with pin-hole free pure boron coatings and the design of such detectors are described in U.S. Published Patent Application 2013/0264481 entitled “Back-illuminated Sensor With Boron Layer”, and filed by Chern et al. on Mar. 10, 2013. This patent application is incorporated herein by reference.
The bubble located in the lower-left portion of
The various circuits and systems of SEM 100 are described above in a simplified form for brevity, and it is understood that these circuit and systems include additional features and perform additional functions. For example, although back-scattered electron detectors 122a/122b and secondary electron detector 121 of SEM 100 are described above as including simplified sensor 123 to introduce certain key features of the present invention with brevity, it is understood that back-scattered electron detectors 122a/122b and secondary electron detector 121 are preferably implemented using the multi-pixel electron detectors described below. Moreover, in addition to generating images of scanned sample areas, computer 160 may be configured to perform additional functions, such as determining the presence of a defect and/or the type of the defect based on incident electron energy values indicated by the image data signals using the methods described below.
For each area on the sample to be inspected or reviewed, the exemplary method 200 starts at step 201. A master clock signal is generated at step 202 that is used to control the timing of the scanning of the primary electron beam and the acquisition of image data.
A beam-deflection scanning pattern is generated at step 204. This beam-deflection scanning pattern generates voltages and/or currents that go to beam deflectors such as those shown at 105 and 109 in
A first pixel clock signal is generated at step 206. The first pixel clock signal is synchronized with the master clock signal. The first pixel clock signal may be at the same frequency as the master clock signal, a frequency that is a multiple of the master clock signal, a frequency that is a sub-multiple of the master clock signal (i.e. the master clock signal frequency divided by an integer), or a frequency that is a rational multiple of the master clock signal frequency.
In step 208, on each period of the first pixel clock signal, signals collected in the back-scattered electron detector are read out and digitized.
In step 210, a second pixel clock signal is generated that is synchronized to the master clock signal. The second pixel clock signal may be at the same frequency as the master clock signal, a frequency that is a multiple of the master clock signal, a frequency that is a sub-multiple of the master clock signal (i.e. the master clock signal frequency divided by an integer), or a frequency that is a rational multiple of the master clock signal frequency. The second pixel clock signal may be at the same frequency as the first pixel clock signal. In one embodiment the first pixel clock signal is used for both the first and second pixel clock signals and no separate second pixel clock signals is generated.
In step 212, on each period of the second pixel clock signal (or the first pixel clock signal if no second pixel clock signal is used), signals collected in the secondary electron detector are read out and digitized.
In step 214, the digitized back-scattered and secondary electron signals are used to determine the presence of one or more defects in the scanned area. A defect may comprise the presence of material (such as a particle) that is not supposed to be there, the absence of material that is supposed to be there (such as may happen with an over-etched condition), or a malformed pattern.
In an optional step 216, for each defect found in step 214, the defect type or the material type of the defect may be determined. For example, high atomic number elements generally scatter a greater fraction of incident electrons than low atomic number elements. The back-scattered electron signal may be used to infer the presence or absence of a high-atomic number element (such as a metal). In step 216, when an area is being reviewed that has previously been inspected, the prior inspection data (optical and/or e-beam) may be used in combination with the digitized back-scattered and secondary electron signals in order to better determine the defect or material type. In one embodiment, steps 214 and 216 may be combined into a single step that simultaneously determines the presence and type of a defect.
Method 200 may be repeated from the beginning for each area on the sample to be reviewed or inspected.
a illustrates an exemplary simplified multi-pixel electron detector 300 for use in a review SEM or other SEM system, such as SEM 100 shown in
Referring to the lower portion of
According to an aspect of the invention, each pixel of sensor circuit 310 includes electron-sensitive, buried channel, floating diffusion and amplifier circuit structures similar to those described above with reference to
Sensor circuit 310 is depicted in
b is a simplified diagram showing an exemplary pixel 315-41 of
According to a preferred embodiment of the present invention depicted in
Referring to the upper portion of
In one embodiment, in addition to the array of analog-to-digital converters 325-xx the signal processing circuit 320 includes processing circuitry 328-1 configured, for example, to calculate the approximate energy of an incident electron based on the digitized output signal (image data) received from an associated pixel of the sensor circuit. In another embodiment, the signal processing circuit 320 also includes high speed data transmission circuitry 328-2 for transmitting an image data signal ID to an external processing system (e.g., a computer)
Referring again to the lower portion of
As explained herein, each pixel has multiple signals or electrical connections, such as gates, control signals, power supply and ground. The interconnect density would be too high for practical and cost-effective assembly to individually connect each of these signals to each pixel. Preferably most, or all, of these signals are connected together between neighboring pixels and brought to a convenient location, such as near the edge of the sensor where an external electrical connection can be made. For example, as indicated in
In contrast to the shared signal lines of sensor circuit 310, as indicated at the upper portion of
c illustrates exemplary electron detector 300A comprising electron sensor 310A, an ASIC (signal processing circuit) 320A and a substrate 301. Substrate 301 provides mechanical support for the electron detector 300A and allows external electrical connections (not shown) to the electron detector 300A. Substrate 301 may comprise silicon or a ceramic material. Electron sensor 310A and ASIC 320A are fabricated on separate silicon substrates (dies or chips) that are then stacked on top of one another, as shown. Alternatively, electron sensor 310A and ASIC 320A may be placed on opposite sides of substrate 301, or side-by-side on the substrate 301 (not shown). Electron sensor 310A is preferably a multi-pixel electron sensor similar to that illustrated in
Electron sensor 310A is electrically connected to ASIC 320A by solder balls or bumps 306. In a preferred embodiment, the output signal generated by each pixel 315 of electron sensor 310A is transmitted by way of an associated solder ball/bump 306 to an associated analog-to-digital converter 325 of ASIC 320A. For example, output signal OS11 generated by pixel 315-11 is transmitted by way of an associated conductor to a first pad 309 disposed on the lower surface of sensor 310A, and from first pad 309 by way of associated solder ball/bump 306-11 to a second pad disposed on ASIC 320A, from which output signal OS11 is transmitted to the input terminal of associated analog-to-digital converter 325-11. One or more solder balls/bumps 306 may also be used to transmit signals from ASIC 320A (e.g., from circuit 328) to control circuit 318 of sensor 310A. These balls or bumps also provide mechanical support for electron sensor 310A and provide thermal conductivity to electron sensor 310A. Solder balls or bumps may instead be used to directly mount electron sensor 310A to substrate 301 (not shown). Metal pads may also be provided on electron sensor 310A to enable wire bonds to provide electrical connections to electron sensor 310A, for example to surface 312-ES of electron sensor 310A.
ASIC 320A may be mounted directly to substrate 301 as shown, or may be mounted and electrically connected to substrate 301 by solder balls or bumps (not shown). If ASIC 320A includes through-silicon vias, then solder balls or bumps may be used on both sides of ASIC 320A. Metal pads 307 and 327 and/or wire bonds 339 may be used to make electrical connections between ASIC 320A and substrate 301. Similar wire bond connections may be made between sensor 310A and substrate 301, or all connections between substrate 301 and sensor 310A may be made through ASIC 320A. ASIC 320A may comprise a single ASIC or two or more ASICs. For example, in one embodiment, ASIC 320A may comprise two ASICs, one ASIC containing primarily analog functions and the other ASIC containing primarily digital functions. Additional integrated circuits, such as a fiber-optic transmitter or a fiber-optic receiver (not shown) may also be mounted on substrate 301.
ASIC 320A preferably includes analog-to-digital converters 325 configured to digitize output signals from pixels 315 of electron sensor 310A. In one embodiment, ASIC 320A includes one analog-to-digital converter 325 for each pixel 315 so that all pixels 315 can be digitized in parallel at high speed, such as at a speed of 100 MHz or higher. With a high digitization rate, such as 100 MHz or higher, each pixel 315 may detect, at most, a few electrons per clock period, so each analog-to-digital converter 325 may only need 8, 6 or fewer bits. It is easier to design a converter with a smaller number of bits to operate at high speed. An analog-to-digital converter with a small number of bits may occupy a small area of silicon making it practical to have a large number, such as 1024 or more on one ASIC.
ASIC 320A preferably implements part of the method shown in
When the electron detector 300A is used as a secondary electron detector, ASIC 320A may implement a de-scanning of secondary electrons similar in result to that implemented by electron optics in the '791 patent cited above. ASIC 320A may sum signals from a group of pixels corresponding to secondary electrons emitted from the sample into one range of angles and output that sum as one signal. As the beam deflection changes, ASIC 320A may sum a different group of pixels under the changed deflection that corresponds to approximately the same range of angles. Since the same master clock is used to generate or synchronize the beam deflection and to generate or synchronize the first and second pixel clocks, the ASIC 320A has the necessary timing information to adjust which groups of pixels are summed together in synchrony with the beam-deflection scan.
When the electron current is low and the pixel clock rate is high enough such that the average number of electrons per pixel is much less than one, then the charge collected in a single pixel in a single period of the pixel clock period can be used to determine whether an electron was detected in that pixel in that clock period, and, if detected, to determine an approximate energy of that electron. The boron coating on the electron sensor surface is necessary to enable this capability. Without a boron coating, few, or no, electrons are created per incident electron when the incident electron energy is less than about 1 keV. With an approximately 5-nm thick boron coating, about 100 electrons are created per incident 1 keV electron. Such a signal can be detected above the noise level if the floating diffusion capacitance is small enough to generate more than about 10 μV per electron. In one embodiment, the floating diffusion capacitance is small enough that the floating diffusion generates more than about 20 μV per electron. For such low level signals, coupling each pixel by as short a path as possible to the corresponding analog to digital converter is important for keeping the noise level low and the stray capacitances low. Attaching the electron sensor directly to the ASIC allows for a very short path from each pixel to the corresponding analog-to-digital converter.
When individual electrons can be detected, ASIC 320A may use the signal level to determine an approximate energy of that electron. ASIC 320A may further threshold, count or bin incident electrons according to their energies in order to detect or classify one or more types of defect or material on the sample.
a and 4b are exploded and assembled perspective views, respectively, showing a simplified pixel 400 of an electron sensor (e.g., sensor 310 described above with reference to
Referring to
Buried channel layer 455 and electron-sensitive layer 457A are disposed in an epitaxial silicon layer 457 such that an upper extent of buried channel layer 455 coincides with (forms) a top (first) surface 457-S1 of epitaxial silicon layer 457, and electron-sensitive layer 457A comprises a portion of epitaxial silicon layer 457 disposed between buried channel layer 455 and a bottom (electron-sensitive) surface 457-S2 of epitaxial silicon layer 457. Epitaxial silicon layer 457 has a thickness preferably between about 10 μm and 100 μm, and is lightly p-doped such that the resistivity is, in one embodiment, between about 10 and 2000 Ωcm. A thicker epi layer provides more mechanical strength, but may generate more dark current. A lower doping level (higher resistivity) may be required for a layer thicker than about 20 μm or 30 μm in order to maintain a fully depleted state in the bulk of the silicon. Too low a doping level is not preferred as that will lead to a higher dark current.
Buried channel layer 455 is created under top surface 457-S1 of epitaxial silicon layer 457 by n-type doping diffused using known techniques. The doping concentration of buried channel layer 455 must be orders of magnitude greater than the doping concentration in epitaxial silicon layer 457, so that epitaxial silicon layer 457 is fully depleted during operation. In one preferred embodiment, the concentration of n-type dopants in the buried channel layer 455 is between about 1016 and 5×1016 cm−3.
Floating diffusion FD comprises a relatively small n+ doped region disposed in buried channel layer 455 that is configured to collect electrons generated in pixel 400 in response to incident back-scattered or secondary electrons. In one preferred embodiment, floating diffusion FD has a nominal lateral size between about 1 and 5 μm, and the concentration of n-type dopants in floating diffusion FD is between about 1019 and 1021 cm−3. A connection for transmitting stored charges to amplifier 410 is made to floating diffusion FD using known techniques.
Pure boron layer 460 is preferably deposited on the back or bottom surface 457-S2 of epitaxial silicon layer 457. Boron layer 450 is preferably between about 2 nm and 10 nm thick, such as a thickness of about 5 nm. As explained in U.S. patent application Ser. No. 13/792,166 (cited above), during the boron deposition process, some boron diffuses a few nm into the epitaxial silicon layer 457 to form a thin, very highly doped p+ layer adjacent to the pure boron layer 460. This p+ layer is important for optimal operation of the sensor. This p+ layer creates an electric field that drives electrons towards the buried channel 455, reduces dark current from the back surface of the epitaxial silicon layer 457, and increases the conductivity of the silicon surface allowing the sensor to function at high incident electron currents as well as low currents. In one embodiment, during deposition of the pure boron layer 460, additional boron is allowed to diffuse into the silicon. This can be done by one of several methods. In one exemplary method, a thicker layer of boron is deposited than the final desired thickness (for example an 6 nm to 8 nm layer may be deposited when a 5 nm final thickness is required) and then the boron is allowed to diffuse into silicon epitaxial layer 457 by keeping the sensor at the deposition temperature or a higher temperature (such as between about 800° C. and 950° C.) for a few minutes. In another exemplary embodiment, a few nm thick layer of boron may be deposited on the silicon, then the boron may be driven in at the deposition temperature or a higher temperature, and then the final desired thickness of boron (such as 5 nm) may be deposited.
According to an aspect of the present embodiment, amplifier 410 is formed in and over an elongated p-well region 459 that extends vertically from top surface 457-S1 into electron-sensitive layer 457A, and extends outward from a point adjacent to the pixel's central region C (i.e., toward peripheral outer edge 455-OPE of n-type buried channel layer 455). Note that p-well region 459 is shown separated from silicon epitaxial layer 457 in
As indicated in
According to another aspect, pixel 400 further includes a resistive gate 451 comprising one or more polycrystalline or amorphous silicon gate structures 470 disposed on dielectric layer(s) 454 and configured to cover most of upper surface 457-S1. As indicated in
According to another aspect, pixel 400 further includes one or more optional additional gate structures disposed between resistive gate 451 and floating diffusion FD to further drive electrons onto floating diffusion FD, or to control when the electrons are collected/accumulated on floating diffusion FD. For example, pixel 400 includes a C-shaped highly-doped polycrystalline gate structure 453 disposed on dielectric layers 454 and inside inner peripheral edge 451-IPE of resistive gate 451. Constant or switched voltages may be applied to gate structure 453 to control and ensure efficient charge transfer from the portions of buried channel layer 455 under resistive gate 451 to floating diffusion FD. In one embodiment described below with reference to
b shows simplified pixel 400 in a partially assembled state. As indicated, most of pixel 400 (i.e., most of top surface 457-S1) is covered by amorphous silicon or poly-silicon gate structure 470, which forms resistive gate 451. The exposed region (indicated by dashed-line box) above p-well region 459 is indicated for illustrative purposes as being empty, but in fact includes various connection structures and gates associated with the transistors forming amplifier 410 and a reset transistor RT. An exemplary layout of these structures and gates is provided below with reference to
a and 5b are simplified cross-sectional views showing pixel 400 during an exemplary detection/readout cycle (operation), where
Referring to
In the example depicted in
Note that the voltage values cited in the example above are merely examples. Different values may be used, and optimal values depend on many factors including the desired speed of operation of the sensor, the geometries of the one or more gates, the doping profiles and the thicknesses of the dielectric layer(s) 454. Note also that it is typically convenient to define the backside (i.e., electron sensitive side) of a sensor as 0V (note that this voltage may be far from ground potential if the electron detector is floated at some potential other than ground), and conductor 471 will preferably be connected to a similar potential.
In an alternative embodiment, instead of switching the reset transistor and the voltages on the various gates of each pixel, the reset transistor and the various gates are held at fixed potentials so that electrons generated in the epitaxial silicon (electron sensitive) region 457A can flow continuously to the floating diffusion FD. In this mode, the voltage on the reset gate RG (
In this embodiment, the gate or gates between the inner peripheral edge of resistive gate 470 and the floating diffusion FD much each be held at successively higher voltages, all higher than that of conductor 472, so that electrons in the buried channel 455 will be driven towards floating diffusion FD. For example, if conductor 472 is at a potential of 5V, then summing gate 453 could be held at a voltage of 6V. If there were another gate (not shown) between the inner peripheral edge of resistive gate 471 and the summing gate 453, then that other gate could, for example, be held at 6V and the summing gate 453 at 7V. The reset drain RD must be held at a voltage significantly more positive than the inner-most gate (such as summing gate 453) so as to keep the floating diffusion FD at a sufficiently high potential relative to all the gates to attract the electrons in the buried channel. For example the reset drain RD might be held at 15V.
As will be readily understood, the channel of the reset transistor RT and the capacitance of the floating diffusion FD form an RC time constant that determines how quickly a voltage on the floating diffusion FD decays back to the reset drain RD voltage after electrons arrive at floating diffusion FD. For example, if the analog-to-digital converters are sampling each pixel at 100 MHz (i.e. once every 10 ns), then an RC time constant of approximately 20 ns or 30 ns might be appropriate. In this example, if the capacitance of the floating diffusion is about 10 fF, then the reset gate RG voltage should be set so that the resistance of the channel of the reset transistor RT is about 2.5 MΩ so as to give a time constant of about 25 ns.
This embodiment is made possible by the sensor disclosed herein because each pixel is connected to its own analog to digital converter. In a conventional two-dimensional CCD or CMOS image sensor, charges need to be stored and read out serially as the number of analog to digital converters is fewer than the number of pixels. Furthermore conventional CMOS image sensors use transistors and gates with surface channels rather than buried channels. Surface channels, in contrast to buried channels, generate noise and cannot transfer small charges without loss.
Referring to the upper portion of
In one embodiment, reset transistor RT is controlled to discharge floating diffusion FD to reset voltage RD using a reset gate voltage RG that is sufficiently positive to turn on reset transistor RT. RD should be more positive than the voltages applied to the various pixel gates (e.g., resistive gate 451 and summing gate 453, described above with reference to
As mentioned above, the amorphous or polycrystalline gate structure utilized to generate the resistive gate (and any additional gates such as summing gate 453, discussed above) in each pixel essentially entirely covers the pixel area except for the central region (i.e., to allow for access to the floating diffusion) and the area in which a p-well is formed. In the example described above with reference to
To accommodate the extended M3 transistor shape, the pixels of sensor 700 are configured to share a portion of their space with an adjacent pixel. Specifically, to provide space for both its own elongated p-well region 759-1 and the portion of p-well region 759-0 extending downward from the pixel above (not shown), resistive gate structure 770-1 of pixel 740-1 is formed in a generally “H” shaped pattern. Similarly, resistive gate structure 770-2 of pixel 740-2 is formed in the same “H” shaped pattern to accommodate the lower portion of p-well 759-1 and the upper portion of p-well region 759-1.
As also discussed above, pixels in each row of sensor 700 share common signal lines that extend along the entire row to a peripherally positioned control circuit (not shown). In the case shown in
In an embodiment, the electron detectors described herein may also detect X-rays. If an X-ray emitted by the sample has enough energy, such as an energy of about 1 keV or higher, it may generate enough electrons when absorbed in the electron sensor to be detected.
The systems and methods described herein may be used with any of the systems and methods described in published U.S. Published Patent Application 2014/0151552, entitled “Tilt-Imaging Scanning Electron Microscope” and filed by Jiang et al. on Mar. 18, 2013, U.S. Published Patent Application 2013/0341504, entitled “Auger Elemental Identification Algorithm”, and filed by Neill et al. on Jun. 7, 2013, U.S. Published Patent Application 2011/0168886, entitled “Charged-particle energy analyzer” and filed by Shadman et al. on Mar. 17, 2011, and U.S. Published Patent Application 2010/0208979, entitled “Use of design information and defect image information in defect classification” and filed by Abbott et al. on Feb. 16, 2009. All these applications are incorporated by reference herein.
The various embodiments of the structures and methods of this invention that are described above are illustrative only of the principles of this invention and are not intended to limit the scope of the invention to the particular embodiments described. For example, the size, shape and layout of the structures within a pixel may differ significantly from those shown herein. The amplifier in a single pixel may comprise one, two or three stages for example. More, or fewer, gates may be used to control the transfer of charge to the floating diffusion. The ASIC within the electron detector may further comprise an FPGA or a digital signal processor to implement algorithms for processing or analyzing the signals from the detector. The ASIC may also include serial transmitter circuits and/or serial receiver circuits to send data to an image processing computer and/or receive commands.
Therefore, the scanning electron microscopes, sensors and methods described herein are not intended to be limited to the particular embodiments shown and described, but are to be accorded the widest scope consistent with the principles and novel features herein disclosed.
The present application claims priority to U.S. Provisional Patent Application 62/043,410, entitled “Scanning Electron Microscope And Methods Of Inspecting”, filed on Aug. 29, 2014, and incorporated by reference herein.
| Number | Date | Country | |
|---|---|---|---|
| 62043410 | Aug 2014 | US |
