1. Field of the Invention
This invention relates generally to the field of charged particle optics, and more particularly to electron detector optics for high throughput large substrate electron-beam testing systems.
2. Description of the Related Art
The use of electron beams to inspect and electrically test flat panel display substrates is an established technique. The different testing strategies may be characterized by the method of obtaining the test signal from each pixel in the display: mechanical probe testing; electron-beam probe testing; and voltage imaging.
Mechanical probe testing of a flat panel display substrate is illustrated in
Electron-beam probe testing is illustrated in
Voltage imaging is illustrated in
All three of the flat panel display substrate testing systems and methods described above suffer from throughput limitations which will only get worse as the size of display substrates continues to increase. There is a need for flat panel display substrate testing systems and methods that have higher throughput and that are more readily scalable to larger substrates.
For flat panel display testing, it is also desired to test 100% of the pixels on the substrate surface since typically a display with more than a few defective pixels is unusable. In some cases, if defective pixels are detected early enough in the manufacturing process, these pixels can be repaired. In other cases, if a substrate is found to have numerous defective pixels, it is more economical to scrap that substrate prior to further processing. Substrate testing also provides process feedback: if successive substrates show increasing numbers of defective pixels, a deviation from proper process parameters (etch, deposition, lithography, etc.) may have occurred, which must be corrected quickly to restore normal production yields. There is a need for inspection systems that are able to test 100% of the pixels on a FPD substrate with a high throughput.
Electron beam systems employed for testing or inspection purposes typically generate one or more primary electron beams (or “probes”) which are focused onto the surface of a substrate by probe-forming optics. When the primary electron beam strikes the substrate surface, it generates both secondary electrons and backscattered electrons (SE/BSEs) as is familiar to those skilled in the art. The SE/BSEs are emitted over a wide angular range, spreading out over a distance which may exceed several mm in extent at the detector collection surface. The signal detection process generally involves the collection of secondary electrons (SEs) and/or backscattered electrons (BSEs). Typically, there will be at most one SE detector and/or one BSE detector per primary electron beam. When the test system comprises multiple electron beam columns, with multiple SE and BSE detectors, efficient collection of SEs and BSEs may be compromised by cross-talk between neighboring columns. Cross-talk occurs when the SEs and/or BSEs from one column are collected by the detectors associated with a neighboring column. There is a need for multiple column electron beam test systems for FPDs that are designed to minimize cross-talk between neighboring columns' detectors.
There is a need for multiple electron beam test systems which meet the three requirements of: high throughput; with 100% pixel testing; while avoiding intercolumn crosstalk between testing signals.
Disclosed herein is a charged particle optical system comprising: N charged particle beam columns in a linear array, aligned along a main scan axis; and a linear array of at least 2N electron detectors aligned parallel to and offset from the main scan axis. Each column generates a single charged particle beam which is focused onto the surface of a substrate for imaging and/or testing purposes. The charged particle beam columns are equally spaced and each beam is deflectable through a large angle along the main scan axis. There are at least two charged particle detectors per column. The charged particle detectors can be SE and/or BSE detectors. There may be more than one linear array of detectors, where the multiple linear arrays may include SE and/or BSE detectors. The multiple linear arrays of detectors may be offset on opposite sides, or on the same side, of the linear array of columns. The signals collected by the linear arrays of SE and/or BSE detectors may be combined in signal combiner circuits (one per detector array), each signal combiner circuit generating a number of output signals equal to the number of charged particle beam columns. The signal combiner circuits are configured to receive input data from the controller for the linear array of columns, providing information on the positions of each charged particle beam on the substrate surface. Using this positional information, the signal combiner determines which detectors within each linear detector array to use in forming each output signal.
In preferred embodiments, the scan fields of the charged particle beams from each column abut or overlap with the scan fields of the neighboring columns in the linear array. In other words, the charged particle beams are deflectable in the plane of a substrate a distance greater than or equal to half the spacing of the columns.
The detector optics is configured to keep the SE/BSE signals substantially separate—uncontaminated by the SE/BSE signals from neighboring columns. This is achieved by tailoring the size, spacing and number per column of detectors in each linear array. Further, the position of the detectors relative to the main axis of each column can be adjusted. The spread of the SE/BSE distribution from each charged particle beam at the linear detector array is key to optimizing the detector optics design. In preferred embodiments, there are at least two detectors per column in a linear array and the center-to-center spacing of the detectors is less than the width of the SE/BSE distribution.
In flat panel display (FPD) testing, there are three important tool design requirements:
In preferred embodiments of the present invention the above three design requirements for a multiple charged particle beam test system are all met, namely: high throughput, 100% testing of pixels, and substantially complete separation between SE/BSE signals from neighboring columns.
The purpose of the present invention is to meet the following three important tool design requirements for FPD testing:
The first requirement can be met using a multiple electron beam column assembly such as that shown in U.S. patent application Ser. No. 11/225,376, filed Sep. 12, 2005, incorporated by reference herein. U.S. patent application Ser. No. 11/225,376 discloses a multiple electron beam column assembly generating a linear array of electron beams where the major scan axis is aligned parallel to the linear array of beams. The use of multiple electron beams simultaneously testing multiple pixels increases the system throughput roughly proportionately to the number of beams.
In order to achieve the second requirement, 100% testing, it is desirable to be able to position at least one of the electron beams generated by the linear array of columns on every location on the substrate.
The simultaneous accomplishment of both the second and third requirements is difficult. To achieve 100% pixel testing as described above, it is necessary to be able to collect SE/BSEs generated by an electron beam even when the beam position on the substrate surface is deflected off the optical axis by approximately half the intercolumn spacing (to the “intercolumn midpoint”). In general, this will require that the SE/BSE detector physically extends to near the intercolumn midpoint, as shown in
Each of the columns in array 1210 functions independently of the other columns. A typical electron beam column for this application is described in U.S. patent application Ser. No. 11/225,376. For example, column 1240 generates electron beam 1242 which is focused by column 1240 onto the surface of substrate 1212 at position 1243. Beam 1242 has been deflected away from the optical axis 1241 of column 1240 by a deflector in column 1240 (not shown). When beam 1242 strikes the surface of substrate 1212 at position 1243, it causes the emission of SE/BSEs 1244. Most of SE/BSEs 1244 are collected by detector 1245, however a sizeable fraction at the right are not detected.
Similarly, column 1246 generates electron beam 1248 which is focused by column 1246 onto the surface of substrate 1212 at position 1249. Beam 1248 has been deflected almost the maximum distance away from the optical axis 1247 of column 1246 by a deflector in column 1246 (not shown). When beam 1248 strikes the surface of substrate 1212 at position 1249, it causes the emission of SE/BSEs 1250. Because detector 1251 is narrower than the full scan width (the distance between scan limits 1261 and 1262), almost none of SE/BSEs 1250 are collected by detector 1251.
Like columns 1240 and 1246, column 1252 generates electron beam 1254 which is focused by column 1252 onto the surface of substrate 1212 at position 1255. Beam 1254 has been deflected away from the optical axis 1253 of column 1252 by a deflector in column 1252 (not shown). When beam 1254 strikes the surface of substrate 1212 at position 1255, it causes the emission of SE/BSEs 1256. Slightly more than half of SE/BSEs 1256 are collected by detector 1257.
In
The present invention is a detector optics system for an electron probe system having the ability to image and/or electrically test a number of locations simultaneously across the full width of a large substrate, while avoiding crosstalk between neighboring beams. A linear array of N electron beams causes secondary electrons (SEs) and backscattered electrons (BSEs) to be emitted from the substrate, which are then collected by a linear array of ≧2N detectors. A signal combiner circuit: (1) dynamically determines which detectors are collecting SEs and/or BSEs associated with each electron beam as the beam scans across the substrate surface; and then (2) combines the signals from the detectors to form N simultaneous output signals.
Similarly, column 146 generates electron beam 148 which is focused by column 146 onto the surface of substrate 112 at position 149. Beam 148 has been deflected almost the maximum distance away from the optical axis 147 of column 146 by a deflector in column 146 (not shown). When beam 148 strikes the surface of substrate 112 at position 149, it causes the emission of SE/BSEs 150. The SE/BSEs 150 are collected by detectors 151 and 157. Note that this is undesirable—detector 157 should collect only SE/BSEs 156 generated by beam 154 in order to avoid crosstalk between columns 146 and 152.
Like columns 140 and 146, column 152 generates electron beam 154 which is focused by column 152 onto the surface of substrate 112 at position 155. Beam 154 has been deflected away from the optical axis 153 of column 152 by a deflector in column 152 (not shown). When beam 154 strikes the surface of substrate 112 at position 155, it causes the emission of SE/BSEs 156. The SE/BSEs 156 are collected by detector 157. Note that this is the desired situation since detector 157 corresponds to column 152.
In
In the example in
Scan limits 260-263 correspond to scan limits 160-163, respectively, in
In order to determine how to select which detectors to use with each column, it is first necessary to introduce terminology relating to the scan width and the SE/BSE distribution at the detector. Define the width of the beam scan to be S. The description of
For the embodiment of the present invention shown in
Number of Columns≧(width of substrate)/(width of scan field).
If the number of columns calculated with this formula is not an integer, it should be rounded up to the next highest integer. Assuming two detectors per column, the number of detectors would then be twice the number of columns.
Table IIA shows parameters for column 240, including the definitions of four regions A1-A4, which together constitute the full width, S, of the scan range for position 243 for the case where L/S is between 0 and 0.25. Assume that the columns have priorities starting with column 240 (1st priority), column 246 (2nd priority), column 252 (3rd priority), etc. where “priority” refers to which column has the most freedom to determine where it can position its respective beam. For example, position 243 can always be anywhere between scan limits 260 and 261, corresponding to regions A1-A4 in Table IIA, independent of where position 249 is. Table IIA shows that if position 243 is in region A1, only detector 200 is required to collect all of the SE/BSEs 244. In this case, position 249 can be anywhere between scan limits 261 and 262, corresponding to regions B1-B5 in Table IIIA. If, however, position 243 is in regions A2-A3, both detectors 200 and 201 are needed to collect SE/BSEs 244. Because any particular detector can only be used for collecting one group of SE/BSEs, if detector 201 is needed to collect SE/BSEs 244, then detector 201 cannot be used to collect SE/BSEs 250. This places a restriction on where position 249 can be. Position 249 cannot be in region B1 since Table IIIA shows that in this case, some of SE/BSEs 250 will fall on detector 201, causing unacceptable intercolumn crosstalk. Thus, position 249 must be in regions B2-B5. If position 243 is in region A4, then both detectors 201 and 202 are required to collect SE/BSEs 244. In this case, neither detector 201 nor detector 202 is available for the collection of SE/BSEs 250, and position 249 must be in regions B4-B5, where only detectors 203 and 204 are needed for the collection of SE/BSEs 250.
Similar considerations apply to the interaction of columns 246 and 252 in the case when L/S is between 0 and 0.25. When position 249 is in regions B1 or B2 in Table IIIA, position 255 may be anywhere between scan limits 262 and 263, corresponding to regions C1-C5 in Table IVA. If position 249 is in regions B3-B4, detector 203 is unavailable for the collection of SE/BSEs 256, and thus position 255 must be in regions C2-C5. If position 249 is in region B5, then detectors 203 and 204 are unavailable for the collection of SE/BSEs 256, and position 255 must be in regions C4-C5.
Tables IIB-IVB correspond to Tables IIA-IVA, for the case where L/S is in the range 0.25-0.50. A larger value for L/S corresponds to a proportionately wider distribution of SE/BSEs at the detectors, thus typically the SE/BSEs will tend to fall on more than one detector a larger percentage of the time. This can be seen by comparing Tables IIA-IVA with IIB-IVB—each “X” corresponds to the use of the detector shown at the top of the column for the detection of SE/BSEs falling within the region shown at the left of that row. A larger number of X's in a table corresponds to more detectors being used more of the time—this increases the chances for conflicts between neighboring columns vying for the use of any particular detector. Increased conflicts will result in a higher percentage value for the figure of merit, potentially leading to reduced system throughput.
For example, Table IIB shows that if position 243 is in region AA1, only detector 200 is required to collect all of the SE/BSEs 244. In this case, position 249 may be anywhere between scan limits 261 and 262, corresponding to regions BB1-BB5 in Table IIIB. If, however, position 243 is in region AA2, both detectors 200 and 201 are needed to collect SE/BSEs 244. Then position 249 must be within regions BB3-BB5 as shown in Table IIIB. If position 243 is in regions AA3-AA4, then both detectors 201 and 202 are required to collect SE/BSEs 244. In this case, position 249 must be within region BB5, where only detectors 203 and 204 are needed for the collection of SE/BSEs 250.
If L/S>0.5, detector 201 is always needed to collect SE/BSEs 244, so it is never possible for position 249 to be near separator 261 (which would require the use of detector 201 to collect SE/BSEs 250)—this in an unresolvable intercolumn conflict. Thus, when the scan fields of neighboring columns abut, it is always necessary for L/S to fall below 0.5. System performance is generally improved with lower L/S values. Lower L/S values can be obtained in several (nonexclusive) ways, such as reducing the gap between the substrate surface and the detector optics or by applying a higher attractive bias voltage on the detector optics relative to the substrate voltage (see U.S. patent application Ser. No. 11/093,000 filed Mar. 28, 2005).
The detector selection process shown in Tables IIA-IVB would be implemented in either the analog signal combiner 1035 in
Scan limits 360-363 correspond to scan limits 160-163, respectively, in
Table VI shows parameters for column 340, including the definitions of five regions D1-D5, which together constitute the full width, S, of the scan range for position 343 for the case where L/S is between 0 and 0.25. Assume that the columns have priorities starting with column 340 (1st priority), column 346 (2nd priority), column 352 (3rd priority), etc. where “priority” has the same definition as for
Similar considerations apply to the interaction of columns 346 and 352. In systems with more than three columns, the column priorities would descend in order across the linear array of columns and more tables could be generated similar to Table VII. The same beam scanning restrictions would apply to all of these columns as outlined above for columns 340 and 346. The detector selection process shown in Tables VI-VII would be implemented in either the analog signal combiner 1035 in
As in
Scan limits 460-463 correspond to scan limits 160-163, respectively, in
Table X shows parameters for column 440, including the definitions of six regions G1-G6, which together constitute the full width, S, of the scan range for position 443 for the case where L/S is between 0 and ⅙. Assume that the columns have priorities starting with column 440 (1st priority), column 446 (2nd priority), column 452 (3rd priority), etc. where “priority” has the same definition as for
Similar to the case for
Scan limits 560-563 correspond to scan limits 160-163, respectively, in
Table XIV shows parameters for column 540, including the definitions of seven regions J1-J7, which together constitute the full width, S, of the scan range for position 543 for the case where L/S is between 0 and ⅙. Assume that the columns have priorities starting with column 540 (1st priority), column 546 (2nd priority), column 552 (3rd priority), etc. where “priority” has the same definition as for
Similar to the case for
Scan limits 660-663 correspond to scan limits 160-163, respectively, in
Table XVIII shows parameters for column 640, including the definitions of eight regions M1-M8, which together constitute the full width, S, of the scan range for beam 642 for the case where L/S is between 0 and 0.125. Assume that the columns have priorities starting with column 640 (1st priority), column 646 (2nd priority), column 652 (3rd priority), etc. where “priority” has the same definition as for
Similar to the case for
Table XXII for column 246 is similar to Table XXI for column 240, except now we must take into account the interaction between the higher priority column 240 and the next lower priority column 246—there are times when position 249 cannot be placed where it normally would be due to conflicts with column 240. The second column from the left in Table XXII is similar to the right column in Table XXI—since L<<S, position 249 is usually in regions B2 and B4. The third column from the left shows the “% of time not allowed”—this is the fraction of time that position 243 falls within a region which prevents position 249 from being placed in the region (left column) corresponding to each row. In quantifying the level of intercolumn conflicts, the relevant factor is the product of these two percentages, as shown in the rightmost column. For example, position 249 would normally be in region B1 3.2% of the time, but 52.3% of the time, position 243 is in conflict—if we take a worst case assumption that beam 248 remains blanked until position 243 moves out of regions A2-A4, then the product gives a contribution of 1.70%. Similarly, position 249 would normally be in region B2 43.6% of the time, but 3.2% of the time, position 243 would be in conflict, giving a product of 1.40%. Summing the percentages for all regions B1-B5 gives 3.30% which serves as a figure of merit for the first embodiment in
Using the formulas in Tables IIA-IVB and it is possible to derive the analytical formula for curve 703:
Figure of Merit=(L/S)+(L/S)2 for 0≦L/S≦0.50
Similarly, Tables VI-VIIIB show that for curve 704 the formula is:
Figure of Merit=( 1/16)+0.5 (L/S)+(L/S)2 for 0≦L/S≦0.25
The analytical formula for curve 803 is:
Figure of Merit=(⅔) (L/S)+(L/S)2 for 0≦L/S≦⅓
The analytical formula for curve 804 has two parts:
Figure of Merit=( 1/36)+(⅓) (L/S)+(L/S)2 for 0≦L/S≦⅙
Figure of Merit=(− 1/12)+(L/S)+(L/S)2 for ⅙≦L/S ≦0.5
Comparison of curves 803 and 804 with curves 703 and 704 in
The analytical formula for curve 904 has two parts:
Figure of Merit=0.5 (L/S)+(L/S)2 for 0≦L/S≦0.25
Figure of Merit=(−⅛)+(L/S)+(L/S)2 for 0.25≦L/S≦0.5
Table XXV is a comparison of the figures of merit for the cases where the detector length, D, meets the following two requirements: D=L, and D=2 L. As expected, the figure of merit is improved (has lower values) for increasing numbers of detectors per column. Also, the larger the detector length, D, is relative to the SE/BSE distribution, the lower the figure of merit.
The ADC outputs 1049-1052 can be either serial or parallel digital signals, suitable for further processing by a data acquisition computer (not shown). The control signal input 1036 comes from the system control computer (not shown) and conveys information to the ASC 1035 about the positions of the beams on the substrate surface—this tells the ASC 1035 which detectors are collecting the various SE/BSE signals caused by each beam as described in
Optimization Principles for the Design of the Detector Arrays
Optimization of the detector design array involves the balancing of a number of factors, including the number of detectors (relative to the number of columns), the width of the separators, the maximum expected value for L/S, and the choice of signal combination circuit (either analog or digital). Table XXVII summarizes how variations in each of these parameters affect the overall system performance, cost and reliability.
While the present invention has been described with reference to a single linear array of detectors, this description is solely for the purpose of illustration and is not to be construed as limiting the scope of the present invention claimed below. For example, the embodiments of the present invention contained hereinabove show a single linear array of detectors configured to collect one of the following three populations of electrons:
The detector optics configuration illustrated in
The detector optics configuration illustrated in
The detector optics configuration illustrated in
The present invention may also be embodied with a linear array of ion beam columns generating a linear array of ion beams, instead of electron beams. As is well known in the art, ion beams also cause the emission of secondary electrons which can be collected by a linear detector array as described in
The detectors shown in
Although the five embodiments described herein show the separators with minimal thickness, alternative embodiments may utilize finite-width separators in order to minimize electrical (capacitive) coupling between neighboring detectors. Widening the separators will inevitably cause some loss in signal due to electrons which strike the separators rather than either of the two neighboring detector collection surfaces, this will have some deleterious effect on signal uniformity across the width, S, of the beam scan.
A more sophisticated beam scanning control could be employed which would image and/or electrically test pixels on the substrate in a different sequence in order to reduce the amount of time that beams are blanked to avoid intercolumn conflicts. This control would employ feedback which would determine the sequence of pixels on the substrate to be tested based on the availability of detectors for SE/BSE collection. For example, in Tables IIA and IIIA, whenever position 243 is in region A4, position 249 would be kept in regions B4-B5 to avoid detector utilization conflicts between columns 240 and 246—this will typically require scanning of pixels on the substrate in a nonsequential order.
This application is a continuation-in-part of U.S. application Ser. No. 11/225,376 filed Sep. 12, 2005, which claims the benefit of U.S. Provisional Application Ser. No. 60/608,609 filed Sep. 10, 2004, and is a continuation-in-part of U.S. application Ser. No. 11/093,000 filed Mar. 28, 2005 now U.S. Pat. No. 7,227,142. The foregoing applications are incorporated by reference herein.
Number | Name | Date | Kind |
---|---|---|---|
4342949 | Harte et al. | Aug 1982 | A |
4742234 | Feldman et al. | May 1988 | A |
4902898 | Jones et al. | Feb 1990 | A |
5276330 | Gesley | Jan 1994 | A |
5430292 | Honjo et al. | Jul 1995 | A |
5466940 | Litman et al. | Nov 1995 | A |
5608218 | Sato et al. | Mar 1997 | A |
5644132 | Litman et al. | Jul 1997 | A |
5982190 | Toro-Lira | Nov 1999 | A |
6075245 | Toro-Lira | Jun 2000 | A |
6627886 | Shachal et al. | Sep 2003 | B1 |
6734428 | Parker et al. | May 2004 | B2 |
6777675 | Parker et al. | Aug 2004 | B2 |
6797953 | Gerlach et al. | Sep 2004 | B2 |
6878936 | Kienzle et al. | Apr 2005 | B2 |
6943351 | Parker | Sep 2005 | B2 |
20050001165 | Parker et al. | Jan 2005 | A1 |
20050285541 | LeChevalier | Dec 2005 | A1 |
Number | Date | Country | |
---|---|---|---|
20060169899 A1 | Aug 2006 | US |
Number | Date | Country | |
---|---|---|---|
60608609 | Sep 2004 | US |
Number | Date | Country | |
---|---|---|---|
Parent | 11225376 | Sep 2005 | US |
Child | 11355256 | US | |
Parent | 11093000 | Mar 2005 | US |
Child | 11225376 | US |