JPH0750127B2 - Electron beam test probe equipment - Google Patents

Description

Description: FIELD OF THE INVENTION The present invention relates generally to the field of scanning electron microscopes, and more particularly to scanning electron microscopes used as test probes for visualizing and testing integrated circuits. Is.

Advances in integrated circuit design and manufacturing have made it possible to form circuits with hundreds of conductors and transistors, with individual conductors and nodes being 1-2 microns. These circuits are too small and complex to be tested and analyzed by techniques that use mechanical probes. Mechanical probes tend to capacitively load the circuit under test, thus altering the behavior one wishes to measure. Moreover, mechanical probes can actually physically damage small conductors and the nodes they contact. Finally, the number of nodes that must be inspected to debug VSLI integrated circuits is rapidly increasing, making it unsuitable for manually performing one node at a time. As a result, electron beam based tests have been developed. These test probes provide a means of measuring the potential on a small conductor and a means of imaging the conductor and surrounding circuitry without any physical damage.

FIG. 1 illustrates the basic principle by which an electron beam test probe operates. When an electron beam of sufficient energy strikes a conductor, it emits secondary electrons with an energy distribution that is a function of the potential on the conductor. A typical energy distribution for a conductor at ground and -5V is shown in FIG. In addition to secondary electrons, some of the electrons from the electron beam are also scattered back by the conductor, which produces a small peak at or near the electron beam energy. These electrons are called backscattered electrons. When the conductor under impact is at negative 5V, each of the secondary electrons emerging from the conductor has an additional energy of 5V resulting from the repulsion of the electrons due to the negative potential of the conductor. Therefore, the energy distribution shown in FIG. 2 for a conductor at ground potential is shifted by 5V when the conductor is at negative 5V.
On the contrary, by observing the energy distribution of the secondary electrons emitted from the conductor under impact, it is possible in principle to infer the potential of the conductor.

Conventional electron beam test probes have been based on scanning electron microscopes. These are U.S. Pat. No. 3,628,012 to Plows et al.
Represented by the achievements of the issue. The device disclosed by Plows is shown schematically in FIG. The electron beam from the scanning electron microscope is directed onto the conductor under observation. Secondary electrons emanating from the conductor at a sufficient velocity in the direction parallel to the electron beam to overcome the disturbing electric field generated by the filter grid are measured by the detector.

This Plows device has some problems. First
In addition, the measured value for the potential of a conductor under impact depends on the electric field in the vicinity of the impact point. This is largely the result of the means chosen to measure the energy of the electrons, ie the result of choosing to measure the component of the secondary electron velocity parallel to the direction of the incident electron beam. . Referring to FIG. 4, each secondary electron can be characterized by a velocity v and an emission angle T relative to the direction of the electron beam. The number of secondary electrons emitted at an angle T is proportional to the cosine of T or the square of the cosine. The secondary electron velocity vector v can be decomposed into a component vp parallel to the electron beam direction and a component vr perpendicular to the electron beam direction. Since the disturbing electric field generated by the filter grid only affects the vp component of the velocity, the number of electrons that can successfully traverse the grid is the number of secondary electrons with vp greater than a certain value. It is a measured value, not a measured value of the number of electrons having a total velocity larger than a predetermined value. As a result, the secondary electrons emitted at a high velocity and T of 90 ° or in the vicinity thereof are not counted, and the secondary electrons emitted at a lower velocity of 0 ° or in the vicinity thereof are counted.

Another problem with conventional electron beam test probes is that the exit angle T may be changed by the electric field near the exit point of the secondary electrons, as shown in FIG. Here, the secondary electrons emitted at an angle of around 90 ° pass on the second conductor at a negative potential. This negative potential creates a local electric field that deflects the secondary electrons in question, thus reducing this effective exit angle. As a result, secondary electrons that are not counted by the detector are also counted. As a result, "crosstalk" occurs between the conductor being probed and the adjacent conductor. The signal on the adjacent conductor appears as the signal on the conductor being probed. Because the potential generated by the signal on the adjacent conductor increases or decreases the number of electrons counted by the detector when the signal becomes negative or positive, respectively. This is a serious problem in conventional devices.

In addition to crosstalk, the sensitivity of the measured potential to the local electric field is a drift in the measured potential due to the slow change in the surface potential of the circuit under investigation. The area between the conductors on the integrated circuit accumulates charge as a result of electron beam bombardment or bombardment when the electron beam is moved from one point to another. Such charge accumulation also occurs as a result of scanning the entire circuit area to form an image of the circuit before deciding which point on the circuit to probe. Since these areas of the circuit are relatively sensitive, this charge produces an electric field that decays slowly over time. Since conventional systems are sensitive to electric fields near the point under investigation, these systems exhibit long-term drift in the measured potential as a result of the slow discharge of the area between the conductors.

The conventional solution to this problem has been to place the extraction grid at a large positive potential just above the surface of the circuit under test. This grid attracts secondary electrons away from the surface of the circuit, minimizing the effect of surface electric fields. However, this grid cannot overcome all surface potential effects. In particular, it does not eliminate crosstalk between adjacent conductors.
Moreover, this grid often results in a highly positive charge on the surface of the circuit. This is because the electrons escaping the surface of the specimen are wiped out by this grid. Thus, when the electron beam is at such energy that more than one secondary electron is ejected by the sample for each absorbed electron, the surface charge reaches a value at which it neutralizes the extraction grid field. Until then, the specimen will gradually become positively charged. This surface charge can also adversely affect operation during testing.

Stray electric fields in other parts of the device also affect the measured potential, either by changing the trajectory of secondary electrons or by changing the trajectory of the electron beam. For example, the extraction and filter grid must be positioned and supported by the insulator. These insulators collect charge in a non-reproducible manner as a result of being hit by electrons that have not been completely collimated by the electron beam, by backscattered electrons, and by secondary electrons. The prior art solves this charge storage problem by using a complex ground shield around the insulator in question, which shields stray electrons from impinging on the insulator and the electron beam and the shield. Secondary electrons are passed through these shields to shield them from the electric field generated by the charge stored on the insulator in question.

Similarly, stray electric fields can affect the electron beam deflection scheme. The electron beam is often swept across the circuit by magnetic deflection. The electron beam must pass through the sweep magnet in the tube, which must be sufficiently conductive to prevent charge accumulation, but prevents eddy currents from interfering with the sweep magnet's response time. Must be sufficiently resistant to do so. One prior art solution to this problem is to use an insulating tube with a thin carbon layer on the inner surface.

The significant space required between the specimen and the last magnetic lens in the electron beam optics to house the filter and extraction grid assembly in conventional systems is that these conventional systems There are two major drawbacks. This space is determined by two points of view. First, the electron beam and secondary electrons must pass through these grids, so the space between the wires used to form the grid must be large compared to the diameter of the wires. Otherwise, a significant portion of the electron beam and secondary electrons will be intercepted by the grid. Second, for such an open grid to be an ideal potential barrier, the distance between the extraction grid and the filter grid must be large compared to the spacing between the wires forming the grid. As a result of these two considerations, a significant distance D must exist between the last focusing magnet and the specimen in the electron beam column, as shown in FIG. This large space limits the size of the electron beam spot on the conductor being probed.

Referring to FIG. 6, the last magnet in the electron beam column focuses the electron beam onto a spot on the specimen at the point under investigation. The minimum size of this spot is determined by the chromatic aberration of this magnet lens. The chromatic aberration of a magnet lens is roughly proportional to its focal length. The focal length of a conventional magnet lens used for this purpose is at least as long as the distance from the bottom of the lens to the specimen. Therefore, the space required for housing the extraction grid and the filter grid forces the conventional method to use a magnet lens with a long focal length, which causes a large chromatic aberration. . This chromatic aberration limited the spatial resolution of the conventional method.

The fact that the conventional method is basically a modified scanning electron microscope gives rise to a third problem. Conventional scanning electron microscopes are optimized to have the highest spatial resolution possible. A resolution of 5/1000 microns is normal. This requires a high energy electron beam, typically 10-20 KeV. The problem generated is, firstly, that this high energy beam can interfere with the operation of the circuit under test. Second, it creates a high background of electrons from the backscattered electrons striking the walls of the containment vessel. These electrons limit the signal to noise ratio of the probe. Finally, it prevents the manufacture of small devices.

More specifically, a 20 KeV electron beam can penetrate a significant distance into the surface of the integrated circuit under test. This penetration causes charging or discharging of the spaced gate elements within the circuit under test. Such charging and discharging can cause permanent circuit damage.

Second, a significant portion of the electron beam is backscattered by the specimen under test. These backscattered electrons strike the walls of the vacuum vessel and produce secondary electrons with energies within the range detected by the electron detector. The number of secondary electrons generated by each such backscattered electron increases as the energy of the backscattered electrons increases. At 20 KeV, these secondary electrons are a significant source of noise in test probe equipment.

Finally, the physical dimensions of the device are constrained by the energy of the electron beam, the higher the electron beam energy, the greater the physical dimensions of the device. In many scanning electron microscopes, an electron beam is made to scan the specimen under test by magnetic deflection coils. The size and power dissipation of the deflection coil and the magnetic lens used to focus the electron beam are directly related to the energy of the electron beam. At 20 KeV, these coils consume considerable power, so they have to be cooled, which complicates the physical configuration. To incorporate this cooling, magnetic deflection coils are usually located outside the vacuum chamber. This requires that the vacuum chamber have a tube through each magnetic deflection coil. This geometry complicates the vacuum vessel and increases its cost. Moreover, as the electron beam energy increases, so does the distance from the magnetic lens to the specimen. This constrains the minimum height of the electron beam column.

Conventional test probe systems are also not designed to be used by untrained persons in maintenance and use of electron microscopes. Usually it is a complex instrument and requires a great deal of tuning work by the operator before the desired image is obtained.

Scanning electron microscopes are optimized to form images of small areas on a specimen. The test probe scheme must be optimized to measure the rapid change in voltage at a particular point anywhere in the integrated circuit.

This raises two additional problems when such a device is adapted for use as a test probe.
First, the physical dimensions of a typical VLSI integrated circuit are significantly larger than the "field of view" of a typical scanning electron microscope. As a result, only a small part of the circuit is probed at any one time. Second, the time required to measure the potential at a particular point is often too long. Since the measured potential is the average of the potentials on that conductor with respect to the time required to make the measurement, this causes inaccuracy if the potential on the conductor changes significantly during the potential measurement. .

To avoid this second problem, the electron beam must be pulsed into very short pulses at a precisely defined time relative to the test signal that governs the operation of the circuit. This imposes design constraints that are not properly incorporated in conventional scheme configurations. For example, timing circuits used to control electron beam pulsing often limit the frequency of electrical signals that can be measured by these schemes.

Finally, the integrated circuit has a conductor covered with a layer of insulating material. There is a need for a means of measuring the potential on such buried conductors. The potential on these conductors produces an electrostatic potential on the surface of the insulating layer. Therefore, it should be possible to measure the potential of the lower conductor. However, conventional methods have not provided a satisfactory method for measuring these potentials. In the conventional method, the electron beam bombardment of the surface of the insulating material causes a change in the potential of the surface, which makes it difficult to infer the potential of the lower conductor.

Accordingly, it is an object of the present invention to provide an electron beam test probe apparatus that is less susceptible to electric fields at points on the sample than at points on the circuit being probed than in conventional approaches.

Another object of the present invention is to provide an electron beam test probe apparatus having an improved signal to noise ratio.

Still another object of the present invention is to provide a small-sized and easy-to-use electron beam test probe apparatus suitable for use in a production line.

Yet another object of the present invention is to determine the potential on a VLSI integrated circuit at any point within the physical boundaries of the integrated circuit, even if the physical boundaries of the integrated circuit exceed the field of view of the scanning electron microscope. It is to provide a test probe that can be measured.

Yet another object of the present invention is to provide an electron beam test probe device capable of measuring high frequency electrical signals.

Yet another object of the invention is to measure the potential of a conductor covered by an insulating layer.

Finally, it is a further object of the present invention to simultaneously measure the electrical potential at one or more of a plurality of selected points in the circuit under test while generating an image of the circuit under test. Is to provide a test probe device capable of

The present invention relates to an improved electron beam test probe device, which enables measurements to be made less sensitive to local electric fields in the vicinity of the point where the potential of a sample is measured than in conventional methods. There is. The apparatus of the present invention includes an electron beam that strikes the specimen at a point where the potential of the specimen is measured, a magnetic lens that collimates secondary electrons emitted from the specimen in response to the impact, so collimated. It has a detector system for measuring the energy distribution of secondary electrons. The limitations caused by the angular dependence of prior art sensing systems are significantly reduced by using magnetic lenses to collimate the secondary electrons before measuring the secondary electron energy distribution. Tubular electrodes are used in energy distribution sensing systems. Such electrodes have significantly higher field uniformity and intercept a smaller portion of the secondary electrons than the wire mesh electrodes used in conventional systems. The electrodes used in the present invention are supported on an insulator composed of a slightly conductive plastic that prevents the build-up of electrostatic charges that have generated unpredictable electric fields in conventional systems. The electron beam used to bombard the specimen is of much lower energy than that normally used in scanning electron microscopes, thus reducing the problems associated with high energy electron beam bombardment of conventional systems. There is. An improved electronic delay circuit has been developed that uses a tandem combination of digital and analog delay techniques to turn on the electron beam with short pulses at precise times in synchronization with the test signal pattern applied to the circuit under test. It is possible to let. This delay circuit allows the timing of these short pulses to be specified to an accuracy of 5 picoseconds relative to the trigger pulse applied to the delay circuit a few milliseconds earlier.

In addition, an improved signal averaging circuit has been developed which improves the signal-to-noise ratio of voltage measurements and which has a much faster response time than prior art signal averaging circuits. The circuit uses an operational amplifier with a gain set to zero during the period when only the noise signal is present. This makes for a better frequency response and lower noise.

The device can be used to measure the potential at the point of impact while forming an image of the specimen near that point. Furthermore, the method according to the invention makes it possible to measure the potential on a buried conductor located below the insulating layer. This method also prevents drift in the electron beam resulting from the changing surface electric field on the specimen.

Hereinafter, specific embodiments of the present invention will be described in detail with reference to the accompanying drawings.

A test probe constructed in accordance with the present invention is shown generally in FIG. 7 as 100, which comprises an electron beam source comprising a tungsten filament 102 which emits electrons when heated by passing an electric current through it. Have Factors affecting the configuration of the electron beam source will be described later. These electrons are the filament 10
It is accelerated by the potential difference maintained between 2 and the acceleration electrode 104. The electrons accelerated by the electrode 104 are formed into an electron beam by the orifice 106 in the electrode 104. This electron beam has two magnetic lenses, namely the magnetic lens 110.
And a collimating magnetic lens 112 focuses on a point within the specimen 108. A deflection coil 114 is used to correct for any slight misalignment between the electron beam column axis 116 and the filament 102. Secondary electrons generated by the electron beam in response to the impact of the sample 108 are formed into a second collimated beam by the collimating magnetic lens 112. This second collimated beam travels in a direction parallel to the direction of the electron beam used to impact sample 108 and in a direction opposite to the direction of the impacting beam. These secondary electrons, which have sufficient energy to pass through the filter electrode 118, are generated by the combination of the negative voltage applied to the suppression electrode 122 and the positive potential applied to the surface 124 of the secondary electron detector 120. It is deflected into the secondary electron detector 120.

Collimating Magnetic Lens The cross section of the collimating magnetic lens 112 will be described in more detail below.
It is shown in FIG. In the preferred embodiment, the collimating magnetic lens 112 has a slot 130 that is tilted downward toward a point 132 where the specimen is bombarded by the electron beam.
It has a coil 126 wound in an annular duct 127 formed in a pole piece 128 having a. The pole piece 128 is doughnut-shaped and has a cylindrical passage 134 with an axis that coincides with the axis 116 of the electron beam column. The pole piece is made of a ferromagnetic material having a high magnetic permeability.

When the current is lengthened through the coil 126, a magnetic field is generated as shown in FIG. 8 (b). The magnetic field thus generated forms a magnetic circuit in which the magnetic field lines lie in a plane perpendicular to the direction of the wires in coil 126. Since the pole pieces are constructed of a highly permeable material, most of this magnetic circuit is constrained to flow through the pole pieces 128 as indicated by the magnetic flux lines 136. However,
Since slot 130 does not thus contain a highly magnetically permeable material, the magnetic flux lines "bulge" at the slot and
At 132, a strong magnetic field region is formed. Pole piece 1
The small portion of the magnetic field not constrained by 28 forms a uniform, weaker magnetic field region within the bore 134 of the magnetic lens, as illustrated by the flux lines 138.

The collimating magnetic lens 112 acts on the secondary electrons in a manner similar to the action of the optical lens on the photon emitted from the point light source located at the focal point of the lens. This is shown in FIG. For example, photon 1 from light source 141
Suppose 40 is emitted at the focal point of optical lens 142 and travels toward lens 142 at an angle 144 with respect to the lens axis 146. When photon 140 strikes lens 142, its direction of travel is changed as it exits the lens, such that it travels parallel to lens axis 146, as indicated at 148. Similarly, an ideal magnetic lens would change the direction of travel of the secondary electrons emitted at the focal point, in which case the secondary electrons in question would differ from the angle at which the secondary electrons were emitted from the source. Are independently changed so as to travel in a direction parallel to the axis of the magnetic lens.

This secondary electron collimation significantly facilitates the measurement of secondary electron energy distribution and eliminates the "crosstalk" that was present in prior art electron beam test probe systems. Since a magnetic field that is constant with respect to time cannot give any energy to the electrons, only the traveling direction of the electrons is changed. Therefore, due to this collimating action, all the speeds are concentrated on the speed component parallel to the lens axis. Therefore, after the collimating action of the electrons by the magnetic lens, measuring the velocity component parallel to the lens axis is equivalent to measuring the total velocity of each electron, ie its energy. This measurement is independent of the angle at which the secondary electrons exit the sample and is also substantially independent of subsequent deflection by the adjacent electric field. Therefore, the "crosstalk" caused by the deflection of the secondary electrons by the adjacent conductor is substantially reduced. Therefore, the introduction of the collimating magnetic lens 112 corrects one of the major problems in the conventional electron beam test probe system.

However, it is much more difficult to form an ideal magnetic lens than to form an ideal optical lens. However, Kruit and Read, "Journal of Physics"
(J.Physics, E: Sci.Instrum., Vol.16, p.313,1983, UK) forms a magnetic field having a high intensity region of magnetic flux at the secondary electron emission point, and the magnetic field is uniform. It is shown that it is possible to construct a good approximation to an ideal magnetic lens by diverging to a magnetic flux region and making the magnetic flux line parallel to the lens axis in it. This is shown in FIG. The electrons emitted from the point 150 at an angle 152 with respect to the direction of the magnetic lens axis 153 are trapped by the magnetic field and
Constrained to travel along the path determined by the magnetic flux lines as shown at 54. When the electrons enter the constant magnetic field region, the angle of their velocity vector is rotated to a new angle 156, which is significantly smaller than the original angle 152. Kruit and Read show the following equations.

sin (A) / sin (B) = (Bf / Bi) ^0.5 where A and B are the final and first angles of the electron velocity vector with respect to the lens axis, and Bf and Bi are uniform magnetic fields. The respective field strengths in the region and in the high intensity diverging magnetic field. Therefore, if the magnetic field in the area of the electron emission point 150 is large compared to the magnetic field in the uniform magnetic field area, the electrons will be collimated into a beam having a direction substantially parallel to the axis of the magnetic lens.

In the device according to the present invention, the effect of this collimating action is that the secondary electrons emitted from the sample 108 are applied to the uniform magnetic field region of the magnetic lens in a direction basically independent of the angle emitted from the sample. enter. As a result of this collimating action, essentially all secondary electrons emitted from the specimen are collected and their velocity vectors are redirected such that they are parallel to the axis of the lens. Therefore, the probability of whether or not a given secondary electron is counted depends only on its energy, not on the emission angle at the point where the sample is bombarded by the electron beam. Therefore, artifacts such as crosstalk between adjacent conductors, which have arisen from the angular dependence of the energy filter, are substantially reduced.

Furthermore, essentially all the electrons emitted from the specimen reach the energy filter. A conventional electron beam test probe only collects a small portion of the secondary electrons emitted from the specimen. The time required to measure the potential on the sample with a given accuracy is inversely proportional to the fraction or proportion of electrons collected by the energy distribution sensing means,
By introducing a collimating magnetic lens, the time required to measure the voltage at a point on the specimen is significantly reduced.

Since the collimating effect is not perfect, each electron generally still has a small radial velocity component in the direction perpendicular to the axis of the lens 112 which coincides with the axis of the electron beam column 116. This velocity component in combination with the magnetic flux in this portion of the magnetic lens 112 causes the electrons to draw a spiral path around the magnetic field lines parallel to the lens axis. As long as this radial velocity component is small compared to the velocity component parallel to the axis of the lens, the potential barrier will still distinguish electrons of different energies based on their total energy. This can be understood as follows. The electron energy E is related to its velocity v as follows.

E = 0.5 mv ² where m is the mass of the electron.

The velocity v is related to the velocity components vp and vr as follows.

v ² = vp ² + vr ² where vp is the velocity component parallel to the axis of the magnetic lens,
vr is the velocity component perpendicular to the axis. If vp is much larger than vr, then v is approximately vp. For example, vr is 10% of vp
Then v differs from vp by only about 1%.
Therefore, a perfect collimating action is not necessary to form a scheme that allows the secondary electrons to be separated according to their total energy as they emerge from the sample.

The uniqueness of the magnetic lens 112 used in the device of the present invention is that the high-intensity magnetic flux region generated by this lens lies outside the physical boundary of the pole piece or pole member 128. . As mentioned above, this is achieved by tilting the gap 130 in the pole piece downwards towards the point where the secondary electrons are emitted. In the conventional structure for generating the high-intensity magnetic flux region of the magnetic lens, this region is located inside the limit of the pole piece. This tests integrated circuits where these configurations may be located on wafers up to 6 inches in diameter, and thus it is not possible to locate them in the high intensity flux region in the conventional manner. It is unsuitable for use in a test probe constructed accordingly. In addition to the space required to accommodate the specimen, the test probe must have space for circuits and connections on one side of the specimen and move the specimen in a plane perpendicular to the direction of the electron beam. It is also necessary to provide a space for it.
This makes conventional magnetic lenses for collimating less practical for measuring test points on such constructions.

In addition to removing the artifacts generated by the electrostatic field near the impact point of the sample by the electron beam,
The collimating action of secondary electrons improves the energy resolution of the test probe and improves the accuracy of determining the sample potential. As pointed out above, conventional test probes
Only the distribution of velocity components of secondary electrons parallel to the direction of the electron beam was measured. Since this distribution depends on both the energy distribution of the secondary electrons and the velocity vector of the secondary electrons, it is considerably broader than the energy distribution. Therefore, the accuracy with which the peaks (or other parameters characterizing the distribution) can be determined is less than the accuracy with which the peaks of the energy distribution can be determined. The accuracy with which the sample potential can be determined is directly related to the accuracy with which the peak of the measured distribution is determined. Therefore, the conventional test probe has a potential measurement capability significantly inferior to that of the test probe of the present invention.

In addition to improving the energy resolution of the test probe and eliminating the problem that conventional methods depended on the emission angle of secondary electrons, the collimating magnetic lens according to the present invention has several additional improvements. Is giving. First, a much larger part of the secondary electrons emitted from the specimen reaches the secondary electron detector 120. In the conventional method, the secondary electron needs to have two characteristics in order to reach the detector. First, it must have sufficient velocity in the direction parallel to the direction of the electron beam to overcome the potential barrier. Secondly, if its radial velocity component, which is perpendicular to the direction mentioned above, is too large, it will not be counted. This radial velocity component causes the secondary electrons to drift away from the axis of the electron beam column towards the wall of the vacuum vessel. If a secondary electron drifts to one of these walls before overcoming the potential barrier, that electron will be lost. The collimating magnetic lens used in the present invention overcomes this difficulty because secondary electrons are collected by the magnetic lens while they drift to the walls of the vacuum vessel or other structures in the device. is doing. When the secondary electrons are collimated into a uniform magnetic field region, the remaining radial velocity component simply causes the secondary electrons to draw a spiral path around the magnetic flux lines, and thus the secondary electrons at this point. It does not drift freely on the wall of the collimating magnetic lens.

The second improvement is that the collimating magnetic lens also collimates the electrons backscattered by the specimen. When so collimated, these secondary electrons are constrained to follow a spiral path around the magnetic field lines within the uniform magnetic field region of the magnetic lens. Therefore, these secondary electrons are prevented from colliding with the wall of the confinement container or the magnetic lens. In conventional methods, these backscattered electrons often collide with the walls of the vacuum vessel. Since backscattered electrons are much higher in energy than the secondary electrons generated in the sample, collisions with such vessel walls or other objects in the device are often a few low-energy background electrons. At this point, electrons with sufficient energy to overcome the potential barrier are generated, and such electrons are detected as if they were emitted from the sample as secondary electrons. These background electrons place significant limits on the signal-to-noise ratio of conventional test probe schemes.

A third improvement provided by the collimating lens of the present invention is that the collimating magnetic lens eliminates the need to place an extraction electrode between the sample and the electrode used to form the potential barrier. ing. The conventional method is
Such extraction electrodes were used to partially overcome the effects of the electric field on the surface of the specimen. These electric fields will bombard the area of the specimen with the electron beam during the process of scanning or moving the electron beam from one conductor to another. The insulating regions are charged by the electron beam interfering with these insulating regions. This charge produces an electric field near the surface of the specimen. When a conductor located in one of these charged insulating regions is bombarded with an electron beam, the secondary electrons emitted at a large angle are, as mentioned above, secondary electrons due to the potential on the adjacent conductor. Are deflected by these electric field regions in a manner similar to that of. Since these regions have high resistivity, this trapped charge only leaks slowly. In order to basically eliminate the influence of these electrostatic fields, the conventional electronic test probe method has used an extraction electrode positioned between the surface of the sample and the filter electrode. This electrode was charged to a positive potential of hundreds of volts. As a result, the secondary electrons emitted from the conductor bombarded with the electron beam are accelerated by the large positive electric field in the direction away from the surface of the sample. This acceleration reduces the amount of time secondary electrons spend in the vicinity of these electrically charged regions. Since the deflection of secondary electrons is proportional to the time they dwell in the electric field generated by this surface charge, the extraction electrodes reduce the magnitude of this deflection and partially compensate for the adverse effects of these surface electric fields. There is.

Although the extraction electrodes used in conventional electronic test probe schemes reduce the effects of these surface fields, they often introduce other artifacts that introduce additional problems. As mentioned above, electron beam test probes used to analyze integrated circuits require the use of low energy electron beams to prevent circuit breakdown. As described in more detail below, when such a low energy electron beam strikes an insulating surface of the type present in a typical VLSI integrated circuit, more than one secondary electron is trapped on the insulating surface. The emitted electrons are emitted from the surface for each electron beam. In the absence of an extraction electrode, the surface is in equilibrium with a positive potential of about 4V. At this potential, as described in more detail below, some of the secondary electrons emitted from the surface in response to electron beam bombardment are pulled back toward the surface by this potential. At this surface potential, this equilibrium state occurs because only one of the secondary electrons emitted from the surface in response to electron beam bombardment is emitted for each electron beam electron captured by that surface. can get. When an extraction electrode is used, essentially all of the secondary electrons emitted from the surface in response to electron beam bombardment are wiped by the extraction field. Since more electrons are removed from the surface by the extraction electrode than are captured from the electron beam, the surface becomes progressively more positively charged as the impact continues. An equilibrium state is established when the surface is finally charged to a potential that balances the extraction electrode potential (ie, hundreds of volts). However, a positive potential of several hundred volts on the surface of the sample interferes with circuit operation. Therefore, the use of the collimating magnetic lens of the present invention significantly improves the performance of the electron beam test probe.

A fourth improvement is that the effective focal length of the magnetic lens used in the preferred embodiment is essentially zero. This offers a significant improvement in the ability to focus the electron beam on a small spot on the surface of the specimen.
In the conventional method, the distance between the sample and the bottom of the last magnetic lens used to focus the electron beam on the sample accommodates both the extraction electrode and the filter electrode forming the potential barrier. Had to be long enough.
As a result, the conventional method suffers from the problem of significant chromatic aberration,
It limited the range in which the electron beam could be focused on a small spot on the specimen. This chromatic aberration increases as the focal length increases. Chromatic aberrations result in electrons with different energies being focused at different distances from the lens. Chromatic aberration is a significant problem in conventional systems because the electron beam has a range of electron energies. By using a lens with a very short focal length, the present invention avoids this degradation in performance.

Electron Source Two factors affect the choice of electron source 102. First, the electron source should have the smallest possible spread in electron energy. The chromatic aberration of the electron beam optical system, that is, the magnetic lens 110 and the collimating magnetic lens 112 is proportional to the ratio of the spread in the energy of the electrons forming the electron beam and the average energy of the electron beam. Therefore, the electron beam energy can be reduced to 1,000 from the 20,000V level used in conventional test probes and scanning electron microscopes.
By reducing to V, the chromatic aberration is increased by 20 times. Fortunately, the improvement in chromatic aberration obtained by using a collimating magnetic lens with an essentially zero focal length compensates for this increase in the present invention.

Second, the vacuum conditions of the electron source should be as generous as possible. The lifetime of many electron sources is dramatically reduced when they are operated in poor vacuum conditions. However, operating in inferior vacuum conditions is highly desirable as it reduces the cost of vacuum pumps and pump down time. Pump down time can be an important factor in the time required to complete testing on a circuit. The tungsten source is a relatively inferior vacuum,
For example, it can be run at 10 ^-4 Torr. However, the spread in electron energy is relatively wide for tungsten. On the other hand, lanthanum
hexaboride) is half the energy spread of tungsten, but requires a vacuum of 10 ^-7 Torr to function properly. The present invention is suitable because it can withstand the large energy spread of a tungsten cathode and therefore can withstand its inferior vacuum conditions.

In the test probe method, which requires higher spatial resolution, barium calcium aluminate is used for the chromatic aberration generated from the electron source.
A further improvement is obtained by using a dispenser cathode composed of porous tungsten impregnated with. Such cathodes are known to those skilled in the short-wave circuit art and have never been used in electronic test probes. It is probably available from Spectra Mat, Inc. of Watsonville, Calif., USA. This cathode has an electron beam energy spread of about half that of pure tungsten.
Moreover, it can be operated in relatively poor vacuum.

Secondary Electron Detector As described above, the potential on the sample can be inferred from the energy distribution of secondary electrons. However, a detailed measurement of energy distribution is not necessary. It is sufficient to measure the number of secondary electrons with energies greater than the given energy. This can be accomplished by measuring the number of electrons with velocities greater than the first predetermined value but less than the second predetermined value. The upper limit is used to distinguish secondary electrons from backscattered electrons from the electron beam.

The velocity of secondary electrons traveling parallel to the axis of the lens can be measured by any of a number of methods. In the preferred embodiment, a potential barrier is used. If the potential barrier is given by an electrostatic field with a direction parallel to the lens axis, then electrons with a velocity parallel to the lens axis greater than a given value depending on the strength of the electrostatic field will impinge on this barrier. overcome. The remaining electrons are returned to the specimen. By measuring the fraction of electrons that pass through the potential barrier as a function of the strength of the electrostatic field, it is possible to obtain a distribution of secondary electron energy.

Referring to FIG. 7, the potential barrier used to detect the energy distribution of secondary electrons is formed by applying a potential to the filter electrode 118. In the preferred embodiment, the filter electrode 118 is a collimator magnetic lens 112.
Is a tubular electrode positioned within the bore of the. As will be appreciated by those skilled in the art, if the diameter of the tubular electrode is small compared to its length, a uniform potential barrier will be formed at the inlet 157 of the tubular electrode when a potential is applied to the tubular electrode. . In the present invention, the length of the tubular electrode is preferably at least twice its diameter. The optimum arrangement of the filter electrode 118 is in the uniform magnetic field region formed by the collimating magnetic lens 112, because the secondary electrons are collimated in this region. When the filter electrode 118 is located in the region where the magnetic flux diverges, the secondary electrons also have velocity vectors that diverge, thus reducing the improvement in energy resolution resulting from the collimating effect.

Conventional test probes use filter electrodes composed of a conductive grid. The grid electrode is at most
It is just a compromise between two competing requirements. In order to form a uniform potential barrier, the spacing between the wires that make up the grid must be small compared to the diameter of the wires. However, if the wire occupies a significant portion of the grid area, a significant portion of the secondary electrons with sufficient energy to pass the potential barrier will be lost due to the collision with the wire. The present invention circumvents these limitations because there is no material in the bore of the tubular electrode capable of intercepting secondary electrons.

Secondary electrons with sufficient energy to overcome the potential barrier generated by the filter electrode 118 will be absorbed by the suppression electrode 12
2 is "swept" or wiped into the secondary electron detector 120 by an electrostatic sweeping electric field formed by the combination of a negative potential applied to 2 and a positive potential applied to the surface 124 of the secondary electron detector 120. . As will be appreciated by those skilled in the art, the positive potential applied to surface 124 may also be applied to the electrode located between secondary electron detector 120 and suppression electrode 122. In the preferred embodiment, the suppression electrode is a tubular electrode having an axis parallel to that of the electron beam column. Section 158 has been removed from the side closest to the secondary electron detector 120, providing a path for secondary electrons.

The potential applied to the suppression electrode 122 is selected to distinguish secondary electrons from the sample and backscattered electrons. Backscattered electrons have much more energy than those of secondary electrons. Therefore, a larger electrostatic field is required to sweep backscattered electrons into the secondary electron detector 120. The potential applied to the suppression electrode 122 is sufficiently high to sweep secondary electrons into the secondary electron detector 120 and continuously suppresses all backscattered electrons present from the filter electrode. Bore 1 of electrode 122
It is chosen low enough to be able to get into the 60's. Many of these backscattered electrons ultimately strike some structure above the suppression electrode 122. The low-energy electrons generated by these collisions do not affect the potential measurement of the sample. This is because such electrons cannot enter the secondary electron detector 120. The exact value of the suppressor electrode potential depends on the physical dimensions of the suppressor electrode 122 and its positioning relative to the other components of the electron beam column.

Secondary electron detector 120 is a scintillation detector in the preferred embodiment. This counter is composed of scintillation material that emits light when charged particles collide. Such materials are known to those skilled in the art of charged particle detectors. In the preferred embodiment, the scintillation material is P47, which can be purchased from PELCO, Inc., Tustin, Calif., USA. The light generated by the scintillating material is transmitted to a photomultiplier tube, which converts this optical signal into an electrical signal having a magnitude representative of the number of secondary electrons striking the detector.

Other secondary electron detectors will be apparent to those skilled in the art of charged particle detectors. For example, the scintillation counter used in the preferred embodiment can be replaced with a solid state or solid state charged particle detector or a charged detector such as a Faraday cup. A suitable electron multiplier may be placed in front of the charge detector to increase the number of electrons detected by the detector. Any detector that produces a signal related to the number of impacting electrons and does not limit the rate at which measurements are made is suitable. P47 phosphorus was chosen because it has a long lifetime under high-intensity electron impact.

For any given filter electrode potential, the electrical signal from the secondary electron detector is related to the potential on the specimen. However, the relationship is not linear. To provide a signal that is linearly related to the potential of the sample, the test probe of the preferred embodiment uses a feedback circuit which regulates the potential applied to the filter electrode 118. As for the filter electrode potential, a certain part of the secondary electrons is detected by the secondary electron detector.
It is located at a value that allows it to reach 120. The filter electrode potential determined in this way is linearly related to the potential on the sample at the probed point. When the sample potential increases by a fixed voltage, the energy distribution of the emitted secondary electrons is shifted by the same fixed voltage. This causes the filter electrode potential to be increased by the same fixed voltage, maintaining a condition that allows the same proportion of secondary electrons to reach the secondary electron detector 120. Thus, the filter potential so adjusted is a signal that is linearly related to the potential on the surface of the specimen.

As will be explained below, the device of the present invention is normally used in a mode in which the electron beam is pulsed for a short period. To reduce noise, the filter electrode potential is adjusted to allow a fixed portion or percentage of secondary electrons to reach the detector during the fixed time window to be detected. In principle, secondary electrons should not reach the detector when the electron beam is off, so averaging the secondary electrons arriving at the secondary electron detector 120 seems good. However, electronics and other noise sources continue to generate noise events, which are indistinguishable from electrons hitting the secondary electron detector 120. These events continue even when the electron beam is off. By limiting the measurement to the time period during which secondary electrons generated from the electron beam pulse arrive at the secondary electron detector 120, the effects of noise generated outside this time window are eliminated. This type of averaging is often referred to as "boxcar" averaging.

A typical conventional boxcar averaging circuit is shown in FIG.
It is shown in 2. It has a current-to-voltage converter 164, which matches the impedance of the secondary electron detector with the rest of the circuit, and also has an integrator 166 that averages the signal from the current-to-voltage converter 164. Further, it has a buffer / level shifter 168 for adjusting the filter electrode potential in response to an input signal. This input signal is basically an error signal representing the difference between the desired number of secondary electrons reaching the secondary electron detector 120 and the actual number of secondary electrons reaching said detector. When more than the desired number of secondary electrons reach the secondary electron detector 120, the signal present at the input to the buffer / level shifter 168 is positive and the buffer / level shifter 168 raises the filter electrode potential, which causes the secondary The number of secondary electrons reaching the electron detector 120 is reduced. If too few secondary electrons reach the secondary electron detector 120, the signal at the input to the buffer / level shifter 168 will be negative and the buffer / level shifter 168 will decrease the filter electrode potential, which will Increase the number of secondary electrons that reach the secondary electron detector 120. In addition to matching the impedance of the secondary electron detector with the rest of the circuit, the current-voltage converter 164 also includes a reference current used to identify the portion or percentage of secondary electrons that should reach the detector. 17
It forms the difference between 0 and the current from the detector 120. Therefore, the output of the current-to-voltage converter 164 is the error signal present at 172, with the desired portion of the secondary electrons reaching the detector 120 and the actual portion of the secondary electrons reaching the detector 120. Represent the discrepancy between. This error signal is averaged in integrator 166 over several electron beam pulses.
Switch 174 prevents integrator 166 from averaging error signal 172 at times outside the time window corresponding to the secondary electrons arriving at detector 120 from sample 108. This time window is line 1
Specified by an enable pulse on 76. The averaged error signal is provided to the buffer / level shifter 168, which increases or decreases the filter electrode potential in the direction of decreasing the error signal.

The output of the current-voltage converter 164 is high when the detector 120 is not counting secondary electrons generated from electron beams that interfere with the sample. This is because this circuit is the secondary electron detector 120
It is possible to discriminate between an "error" due to too few electrons arriving at, that is the result of improperly setting the filter electrode potential, and an error as a result of the electron beam being turned off. Because there is no.
Therefore, as the electron beam is turned on and off during each pulse, the output of the current-voltage converter 164 must swing a large voltage range. This voltage range is the output of the current-voltage converter when there is no secondary electron detector signal at the input and the secondary electron detection when the detector is counting a substantially desired number of secondary electrons. Corresponding to the difference between the output of the detector when the input signal corresponds to that generated by the detector.

Differential amplifiers or op amps are usually used to construct a typical current-voltage converter. The slew problem in such amplifiers limits the speed at which this conversion can be done, and thus the minimum time window that can be used. The device of the present invention uses a unique current-voltage converter configuration, which avoids these slew limits.

Referring to FIG. 11, the current-voltage converter 164 used in the device of the present invention has a standard operational amplifier 178,
Its gain is set by a resistor 180 in the feedback path and also has a FET switch 182, which
When connected across 180 and switch 182 is on, it shorts the resistive path through resistor 180. In the first state, the resistance of resistor 180 is shorted by switch 182 to provide a zero resistance feedback path. In this first state, the gain of operational amplifier 178 is substantially zero. This will result in the error signal voltage at output 172 being set to zero, even though the input current will normally produce a large signal at this output. The current-voltage converter 164 of the present invention uses the switch 182 to short the gain resistor 180 at all times except when secondary electrons are present in the secondary electron detector 120. This FET switch is driven by the same signal that drives switch 174 which gates the error signal to integrator 166. In the second state, the switch 182 turns on the current-voltage converter 164 by inserting the resistance value specified by the resistor 180 into the feedback path. During the time window when a non-zero resistance value is inserted in the feedback path, the input of the current-voltage converter 164 is near the correct input, so the error signal at the output of the current-voltage converter 164 has a large swing. No need. This significantly reduces the problems from slewing operations that were inherent in conventional boxcar averaging circuits.

In principle, by inserting the switch 182, the switch used in the conventional boxcar averaging circuit
Removing the need for 174. If switch 182 completely shorts resistor 180, the operational amplifier gain is zero and the signal from the operational amplifier input does not reach averaging circuit 166. However, commercially available switches and operational amplifiers are not ideal. Therefore, if switch 182 shorts resistor 180, a small portion of the signal present on the input to operational amplifier 178 will reach point 172. The second switch 174 ensures that this small residual signal does not affect the averaging circuit 166.

Although the preferred embodiment of the present invention uses a potential barrier to detect the energy distribution of the secondary electrons, it will be appreciated by those skilled in the art of charged particle detectors that they are emitted in response to an electron beam. Any of a number of energy sensing schemes that can be used in conjunction with the collimating magnetic lens 112 to sense the secondary electron energy distribution can be used. For example, the time required for secondary electrons to travel from the specimen 108 to the secondary electron detector 120 can be measured. Since this time is inversely proportional to the velocity of the secondary electrons, such a measurement gives the same information as a direct measurement of the velocity of the secondary electrons. In such a system, the electron beam is pulsed, and then each such pulse and the secondary electron detector 12
It can be constructed by observing the interval between the arrival of secondary electrons at 0.

Electron Beam Energy Selection In the preferred embodiment, the electron beam energy is significantly lower than that used in the prior art. Many conventional systems are basically scanning electron microscope adaptations.
As such, it is optimized for electron beam energies in the 10-20 KeV range. The test probe of the present invention uses electron beam energies in the 500-1500V range. As mentioned above, high energy electron beams can cause a number of problems such as circuit damage, high background of secondary electrons from backscattered electrons, and unwanted surface charging of non-conductive parts of the sample surface. Is possible. Such high energy is guaranteed in the scanning electron microscope. Because these devices are optimized to inspect very small features on the surface of the specimen to be scanned. The resolution of the electron microscope is determined in part by the energy of the electron beam. For example, as described above, the energy of the electron beam affects the chromatic aberration of the optical system. Generally, increasing the energy of the electron beam reduces the chromatic aberration of the optical system. This is because chromatic aberration is proportional to the spread in the energy of the electron beam, expressed as a percentage of the average electron beam energy.
The spread in electron beam energy is determined by such factors as the type of electron source used to generate the electron beam. These factors are unaffected by the accelerating voltage used to accelerate the electrons. Therefore, the chromatic aberration is halved by doubling the electron beam energy. Furthermore, the resolution of electron microscopes is ultimately limited by diffraction effects similar to those that limit the resolution of light microscopes. The smallest observable object size is determined by the wavelength of the electron beam. Objects smaller than about one wavelength are unobservable. The wavelength of the electron beam is inversely proportional to its velocity, so a high energy electron beam must be used to observe very small details.

However, since the conductors being probed are on the order of 1 micron wide, the test probe need only generate spots that are a fraction of 1 micron in diameter.
Electron beam energy is not a limiting factor in observing objects within this range. Therefore, the test probe can operate satisfactorily at electron beam energies substantially lower than standard scanning electron microscopes.

The use of a low energy electron beam allows for a number of simplifications in test probe construction, making the present invention more suitable for use as a production line device as compared to conventional approaches. For example, in accordance with the present invention, a much smaller electromagnet is needed to focus and steer the electron beam. As a result, the magnetic lens does not have to be water cooled. This simplifies the preparation of the physical location for mounting the test probe. Furthermore, the distance between the various magnetic lenses is reduced, thus allowing a test probe with a much smaller geometry. This allows the test probe to be in a system form that can be physically moved relative to the specimen. Finally, as will be explained in more detail below, the use of a low energy electron beam allows the device of the invention to be used for probing the potential on a conductor covered with an insulating layer.

Electron Beam Positioning Conventional test probe methods have had an insufficient "field of view" to probe the entire VLSI circuit. In the present invention, this problem is solved by providing both coarse and fine adjustment means for positioning the electron beam relative to the sample. The fine adjustment means includes a steering magnet 186 shown in FIG. 7, which deflects the electron beam relative to the sample. The steering magnet 186 can deflect the electron beam in either one of two orthogonal directions perpendicular to the direction of the electron beam. Therefore, they can be used to scan an area on the specimen and to finely position the electron beam on the specimen. This scanning motion can be used to form an image of the surface of the specimen that has been enhanced to show the different electrostatic potentials present on the specimen. Other uses for this scanning movement are described below with respect to methods of using the test probe of the present invention.

In a preferred embodiment, the steering magnet has steering coils wound on a somewhat conductive plastic support for improved frequency performance, as described in more detail below.
As will be appreciated by those skilled in the art, the steering magnet 186 can be replaced with an electrostatic deflection plate that also deflects the electron beam when an appropriate potential is applied.

The coarse adjustment means is shown in FIG. Electron beam column 18
7 is mounted in a vacuum container having boundaries indicated by 188, 190 and 192. This boundary also surrounds the stage 194, which has a platform 196 that holds the specimen 108 under exploration. The electron beam column has means for positioning the beam column shown at 198 and 200 by translational movement in two orthogonal directions, x and y, which are perpendicular to the electron beam direction. Electron beam column is a positioning means
It can be moved using micrometers 202 and 204 which control 198 and 200. In addition, stage 194
Provides a means for translating in any of three directions, x, y, z, and rotating about an axis A about an axis parallel to the electron beam direction to position the specimen. The various translational movements performed by stage 194 are controlled by micrometers 206, 208, 210. The rotary movement is controlled by the micrometer 212. In the preferred embodiment, these micrometers are controlled by a stepper motor under computer control.

Flexible bellows 190 provide a means for maintaining a vacuum in the container when the electron beam column is moved with respect to the vacuum chamber. Bellows 190 allow the column positioning means to remain outside the vacuum vessel. The positioning means preferably uses conventional bearings or other lubricated surfaces to support the column load. Outgassing from such lubrication during the pump down process increases the length of time required to pump down the vacuum chamber when a new specimen is introduced into the test probe. Furthermore, the volume of the vacuum vessel, which must be pumped down each time a sample is introduced into the device, is also minimized in this configuration.

The probe card 214 has a platform or table 196.
Means are provided for making an electrical connection to the specimen 108 mounted above. This probe card 214 has a hole through which an electron beam passes.

One of the intended uses of the device of the present invention is to make potential measurements on integrated circuits. As is known to those skilled in the integrated circuit art, integrated circuits are typically manufactured on wafers containing hundreds of such circuits. The apparatus of the present invention can be used to test individual integrated circuits before separating them from other integrated circuits on the wafer. Such measurements use the present invention to position an integrated circuit, referred to as the desired die, within the field of view of the electron beam and to move the electron beam to various points on the die where electrometry is desired. Done by. The wafer is mounted at 108 on table 196. The position controls 206, 208, 210, 212 are used to position the desired die relative to the probe card 214. The probe card has mechanical probes 216 for connecting power and test signals to the die under test.

Once the desired die is accurately positioned relative to the probe card 214, the die is moved by translating the electron beam column 187 relative to the die using coarse adjustment xy position controls 198 and 200. Select the area of interest above. The steering magnet 186 is then used to inspect the area of interest.

A substantial use contemplated for the present invention is to measure the potential of a node in a circuit provided on the surface of specimen 108. However, in order to search for a particular node, the image of the sample within the node region must be compared to the circuit layout on the sample, usually in the shape of a mask, showing the locations of the various components of the circuit. Thus, in effect, one scans the area of the specimen that has the node in question, this area is displayed to the operator, and the operator selects the exact point within the area where the potential is to be measured. In the conventional scheme, the electron beam is directed at this point and the potential is measured as a function of time. This procedure is often an artifact resulting from the electrostatic potential on the surface. These potentials are formed by the scanning motion used to search for the point of interest.

The surface of the specimen has insulating areas between the various conductors which have a very low conductivity. As the electron beam scans these areas, these areas become charged due to the interference of the low-conductivity material with the electron beam. This surface charge is
As the surface charge of interest slowly dissipates, it creates an electrostatic field whose intensity decreases slowly. This changing electrostatic field interferes with the electron beam and its position drifts as the electric field dissipates.

The method of the present invention avoids this problem by periodically refreshing the surface charge in the area surrounding the node being measured. This is accomplished by periodically scanning the area around the node during the potential measurement. In the proposed method of the present invention, the node of interest is first searched by scanning the area in the vicinity of that node. The electron beam is then moved to the node in question and the potential measurement is started. Periodically during this potential measurement, the electron beam scans the area around the node in question, refreshing the surface charge in these areas. This maintains the electrostatic field around the node at a substantially constant value during the potential measurement. Therefore, drifts in electron beam position during these potential measurements resulting from changes in the local electrostatic field are eliminated or at least minimized.

The method of maintaining the insulating region at a constant potential is possible only in the electron beam test probe in which the electron beam has an energy of about 2 KeV or less. The number of secondary electrons generated as a result of electron beam electron bombardment is shown in FIG. For electron beam energies above about 2 KeV, less than one secondary electron is generated for each electron beam electron striking the insulating surface. As a result, electrons are accumulated on the surface. Since the surface in question is an insulator, this accumulated charge cannot leak. Therefore, the surface becomes progressively more negatively charged, which continues until it is at a potential high enough to effectively change the energy of the electrons striking it.

When the electron beam energy is lowered below about 2 KeV, it is possible to avoid this large negative storage potential. Thirteenth
As shown in the figure, if the electron beam energy is in the 1 KeV region, one for each electron beam electron that strikes the surface.
More than one secondary electron is generated. This results in the surface being positively charged because more electrons are emitted than are absorbed by the surface. When the surface becomes progressively positively charged, some of the secondary electrons exiting the surface will be recaptured by the surface because it is positively charged. The higher the positive charge, the larger the fraction of secondary electrons that are recaptured. This is because more than one secondary electron is emitted per electron beam electron, but low-energy secondary electrons are recaptured by the positive potential, so the number of secondary electrons emitted per incident electron is effective. Decrease. Ultimately, the surface reaches a stable positive potential and one secondary electron is trapped for each electron beam electron. This potential is about 4V for insulating materials used on silicon VLSI chips. Therefore, by refreshing the surface charge periodically, a stable potential of about 4V is maintained on the insulating surface. This potential remains substantially constant over the time between measuring potentials on nearby conductors, so it does not cause beam drift.

It should be noted that conventional approaches using extraction electrodes maintained at a positive potential near the surface of the specimen do not allow this method to be easily implemented. The large positive potential of the extraction electrode results in all secondary electrons escaping from the surface. Therefore,
When a low-energy electron beam is used, instead of reaching the stable positive potential of 4V mentioned above, the surface becomes progressively positively charged.

Slightly Conductive Plastic Support The support for the steering magnet 186 must have a path through which the electron beam passes. If the passage is formed from an insulator, stray electrons will strike it and electrostatic charges will accumulate on the surface of the passage. This electrostatic charge creates an electrostatic field in the passage, which adversely affects the electron beam. The magnitude of this charge is not unpredictable or even reproducible. To avoid this charge buildup, it is possible to use a conductive support. However, the fluctuating magnetic field generated by the steering magnet causes eddy currents in such conductors. These eddy currents reduce the frequency response of the steering coil by creating its own magnetic field. In order to avoid these eddy currents, the conventional method uses an insulating tube coated with a thin carbon layer. This thin carbon layer leaks electrostatic charge before it reaches a size that interferes with the electron beam without being sufficiently conductive to allow significant eddy currents to be generated by the steering coil. It has sufficient conductivity to make it possible. This coating system is expensive. This is because it requires depositing a very thin carbon layer on the surface after forming the insulating tube. The device of the preferred embodiment of the present invention avoids this cost.

The preferred embodiment uses a material that is a slightly conductive plastic for this purpose. The slightly conductive plastic is sufficiently conductive to allow the charge stored as a result of trapping stray electrons to escape. However, its conductivity is low enough to prevent eddy current generation by the steering coil. It is a more economically attractive solution to this problem because plastics are easily moldable and do not require any subsequent carbon deposition after molding.

A preferred slightly conductive plastic is formed by incorporating carbon black into a non-conductive plastic. The resulting preferred resistivity is 10 ⁶ Ω · cm. Satisfactory performance is obtained with a resistivity in the range of 10 ^{4 to} 10 ⁸ Ω · cm. Plastics with these properties are available from Wilson Fibefil, Inc., Evansville, IN 47732, USA.

Slightly conductive plastics can also be used to form the insulative support used to connect the various electrodes to the support structure in the device of the present invention. For example, as shown in FIG. 7, the tubular filter electrode 118 is
Insulator 218 formed from slightly conductive plastic
Is attached to the collimator magnetic lens 112. The slightly conductive plastic allows any electrostatic charge present on the support to escape. Therefore, the problem of electrostatic charge accumulation of the conventional method is avoided.

In the conventional manner, such a structure is mounted on a conventional insulator. Electrons that strike one of these insulators are usually trapped on the insulator. because,
This is because the insulator has a very high resistivity and trapped charges do not leak. This trapped charge creates an unpredictable electric field, which affects the trajectories of both electrons and secondary electrons in the electron beam. To avoid these adverse effects, conventional methods have used complex shields around such insulators. These shields intercept the stray electrons before they reach the insulator of interest and manage to pass through these shields and generate an electric field from the electrons trapped on the insulator of interest in the device. It is configured to shield the rest. The use of slightly conductive plastics in the present invention avoids an economically unattractive solution to this charge storage problem.

Strobe Behavior of Test Probes Improvements in integrated circuit technology provide circuits in which the potential at a given node changes in less than the time required to measure that node's potential using an electron beam test probe. is doing. The time required to make an accurate potential measurement at the node is the time required for the secondary electron detector 120 to collect a statistically significant number of secondary electrons. In the test probe of the present invention, since the collimating magnetic lens supplies a larger part of the secondary electrons leaving the node to the secondary electron detector, such a statistically significant number of secondary electrons is conventionally obtained. It is possible to accumulate in a shorter time than the method of. However, even with this significant improvement, it is not always possible to follow potential changes at a node simply by observing secondary electrons emitted from the node as a function of time.

This problem can be eliminated by using stroboscopic techniques. Here, several measurements with a limited number of secondary electrons are averaged to produce a single measurement with sufficient secondary electrons that is statistically accurate. The test signal pattern applied to the circuit under probing is periodic in nature. It consists of a series of signals that repeat in time. In stroboscopic systems, the electron beam is turned on for a short time at the same point in this test signal pattern. A point in the test signal pattern is identified by a delay time relative to some fixed point in the test signal pattern. The time during which the electron beam is turned on should be small compared to the period during which the test pattern potential changes. Because the measured potential is the average of the potentials on the conductor during the time the electron beam is turned on. Although the number of secondary electrons collected in each of these short periods is insufficient to obtain a statistically accurate measurement, secondary electrons obtained over many of these periods are The average of the numbers of s gives a statistically meaningful measure of the potential at the node in question.

Strobe techniques require the test probe scheme to apply a short pulse of electrons to the same point on the sample at exactly the same time as the trigger pulse derived from the test signal pattern. Fluctuations in the timing of the electron beam pulse relative to the trigger pulse degrades the potential measurement. Variations in the position of the electron beam on the specimen also degrade the potential measurement if the variation moves the electron beam off the conductor being probed.

The invention comprises means for pulsing the electron beam with short pulses. This is accomplished by applying a suitable potential across the blanking electrode 220 shown in FIG. When there is no potential across the blanking electrode, the electron beam is imaged by the magnetic lens 110 into the aperture 160. The collimating magnetic lens 112 then images the electron beam into a small spot on the specimen 108. When a potential is applied across the blanking electrodes, the electron beam is sufficiently deflected to prevent it from entering aperture 160.

Although the conventional method proposes a stroboscopic technique, the circuit provided in this conventional method is not sufficient for accurately observing the waveform in the 1000 MHz range on the conductor under investigation. To operate effectively in strobe mode, the test probe must bombard a circuit node on the sample with a burst of electrons, in which case
The duration of the burst is small compared to the time during which the signal being probed changes significantly. Any variation in the timing of these pulses relative to the test signal pattern will result in degradation of the potential measurement. The present invention samples the potential on the conductor in question for a minimum of 100 picoseconds.
The time that this pulse is applied to the sample relative to the test signal pattern is determined by a delay circuit that triggers the blanking electrode circuit at a specified time relative to the trigger pulse derived from the test signal pattern. If the delay time produced by this circuit changes by an amount that is a significant part of the electron beam pulse time, then a degradation in the performance of the strobe system will occur. Therefore,
During an electron beam pulse of 100 picoseconds, the delay circuit must generate a delay that varies by at most a small part of 100 picoseconds, eg 10 picoseconds, and gives a delay of zero to a few milliseconds.

An analog circuit that generates a delay by charging or discharging a circuit element such as a capacitor to a predetermined voltage lacks the required accuracy for such a high speed delay circuit.
An accuracy of 10 picoseconds in millisecond delay requires a part of accuracy in 100,000,000, which is not possible with such analog circuits. It is not possible to achieve this degree of accuracy by using a simple digital delay technique in which a pulse is generated after a preset number from the oscillator circuit has been counted. Unless the oscillator always starts in the same phase in the oscillation cycle, there is an uncertainty of up to half a cycle of oscillation in the delay so generated. This uncertainty arises from the uncertainty in the phase of the oscillator when counting is started. Furthermore, the minimum delay that can be generated in this way is one oscillator cycle. Therefore, such a simple digital delay circuit requires an oscillator with a frequency exceeding 100 GHz. This type of circuit is economically unattractive.

The device of the present invention produces long delays with high precision by combining digital delay techniques with analog fine tuning circuits. Digital delay gives a coarse delay. The analog delay circuit gives a fine adjustment to this coarse delay time. This is
Allows a delay of less than one oscillator period generated. A restartable oscillator is used to remove the uncertainty created without knowing the point in the oscillation cycle at which counting was started. The delay circuit of the present invention is shown schematically at 222 in FIG. It has a restartable oscillator 224, a counter 226, and an analog delay circuit 228. The coarse delay is specified by the signal on bus 230, which specifies the number of oscillator pulses that must be counted by counter 226 before generating an output pulse.
The magnitude of the analog delay is specified by the signal on bus 231.

The restartable oscillator always starts oscillating in the same phase of its waveform. Therefore, the period does not have to be smaller than the period in which the electron beam burst is performed. Such oscillators are known to those skilled in the art of electronic circuits. Similarly, analog delay circuits are known to those skilled in the electronics circuit art. Significant improvements have been obtained by the combined use of two delay techniques as compared to conventional methods of producing a delay between a trigger pulse and an electron beam pulse.

The device of the present invention uses a 200 MHz restartable oscillator. Therefore, each count of counter 226 corresponds to a delay of 5 nanoseconds. Each of these 5 nanosecond steps is further divided by an analog delay circuit into 1000 5 picosecond steps.

The placement of the blanking electrode 220 has been optimized to further improve the strobe capability of the present invention. Three factors must be considered in order to properly optimize the positions of these electrodes. First, the positioning of the blanking electrodes affects the amount of beam movement that occurs when the electron beam is turned on and off during a given electron beam pulse. During the rise and fall of the potential applied to the blanking electrode, the electron beam tends to move on the sample. If the electron beam moves a small distance compared to the width of the conductor, which is hypothetically centered, there is no adverse effect since all of the conductors are at the same potential. However, as the electron beam travels off the conductor, the measured potential is a weighted average of the potentials encountered along its path. This can cause inaccuracies in the measured potential and can also cause crosstalk with the potential on the next conductor if the electron beam moves to the next conductor.

Second, the magnitude of the blanking potential that must be applied to the blanking electrode 220 to produce the desired electron beam blanking is a function of the position of the blanking. In general, large blanking potentials should be avoided. Large potentials require more expensive circuits.
Furthermore, the rise and fall times of the blanking potential that can be obtained at a reasonable cost are smaller for smaller potentials than for large potentials. It is these finite rise and fall times that make the movement of the electron beam an important factor during the pulsed operation of the electron beam. It is possible to turn on and off the blanking electrode potential in a short amount of time compared to the length of the electron beam pulse, and from these points on the electron beam path off the conductor that was initially centered. The contribution of the above-mentioned weighted potential to the average is negligible.

Third, the accuracy with which the blanking electrode 220 must be relatively aligned with the electron beam column axis 116 is also a function of the blanking electrode placement. Obviously, placing the blanking electrodes in locations that require precise alignment should be avoided for economic reasons.

In the conventional method, the blanking electrode is typically arranged between two magnetic lenses as shown in FIG.
The electron beam source 232 emits electrons over a cone angle 234.
These electrons are imaged in the second magnetic lens 236 by the first magnetic lens 238. The blanking electrode 240 is located at the point where the electrons imaged by the magnetic lens 238 intersect the axis 242 of the electron beam system. The aperture 244 intercepts the electron beam electrons when a blanking potential is applied. The magnitude of the blanking potential must be sufficient to deflect the electrons traveling along the extreme trajectory shown at the exit 246 of this aperture. Positioning the blanking electrode 240 produces minimal electron beam movement during the rise and fall of the potential.
However, it requires much higher potential and alignment accuracy than if the blanking electrode 240 were positioned between the first magnetic lens 238 and the source 232.

Referring to FIG. 16, the blanking electrode 248 of the present invention.
Are arranged between the electron source 250 and the first magnetic lens 252. This is because the electron beam on the specimen is pulsed when the electron beam is pulsed using the blanking electrode relative to the movement that would be obtained if the blanking electrode were positioned between the magnetic lenses as described above. Increase movement. To limit this increase in electron beam movement during electron beam pulse operation, a first aperture 254 is placed between the electron source 250 and the blanking electrode 248. The second aperture 256 intercepts the electrons deflected by the blanking electrode 248 when a potential is applied to the blanking electrode. When there is no potential on the blanking electrode, the first magnetic lens 252 images the electrons passing through the first aperture 248 into the second aperture 256.
Three parameters must be optimized, the dimensions of the first aperture 254, the placement of this aperture relative to the magnetic lens, and the placement of the blanking electrode relative to the magnetic lens. The minimum size of aperture 254 is determined by the physical size of source 250. The source is shown in Figure 16 as a point source, which actually has
It is on the order of 50 microns. If the aperture 254 is significantly smaller than the source size, it will significantly reduce the number of electrons focused on the specimen. Therefore, the aperture 254
Is selected for the source dimension order. The placement of aperture 254 relative to source 250 and magnetic lens 252 is selected to minimize the movement of the electron beam at point 258 on the sample during changes in the potential applied to blankin electrode 248. . The placement of blanking electrode 248 relative to source 250 and magnetic lens 252 is selected to minimize the potential required to deflect the electron beam from aperture 256. This position of the blanking electrode 248 is 1/4 and 1/2 the distance between the source 250 and the magnetic lens 252.
Is selected to be between and. The optimum position of the aperture 254 is as close as possible to the blanking electrode 248 within a mechanically and electrically feasible range. In the preferred embodiment shown in FIG. 7, this array of apertures and lenses is
When a voltage of 5V is applied to the blanking electrode, a movement of 0.2 micron or less occurs in the source. Conventional schemes typically required a blanking potential of 40V. Since the conductors on a typical integrated circuit are on the order of 2 microns, this amount of motion is acceptable.

Measuring the Potential on a Buried Conductor The sample to be measured in the device of the present invention often includes a conductor coated with a layer of insulating material, as shown in FIG. The voltage applied to the lower conductor 260 is
It is the potential generated on the surface 262 of the. Therefore, in principle, it should be possible to measure the waveform applied to the lower conductor by measuring the potential on the surface of the insulator covering the conductor. In practice, this is difficult to achieve due to variations in the surface potential generated by the electron beam itself during the measurement process.

As explained with reference to FIG. 13, the number of secondary electrons generated as a result of electron beam bombardment is a function of electron beam energy. For electron beam energies above about 2 KeV, less than one secondary electron is generated for each electron beam electron that strikes the insulating surface. This results in the accumulation of electrons on the surface. Since the surface in question is an insulator, this accumulated charge cannot escape. Therefore, the surface becomes progressively more negatively charged, which is
This is continued until the potential of the electrons striking the sample surface becomes sufficiently high to effectively change the energy. The surface potential at this point is so large that any changes in the sample potential resulting from the potential applied to the lower conductor are too small to be detected. Therefore, the conventional method uses a high-energy electron beam, and therefore it is impossible to measure the waveform on the conductor buried under the insulating layer.

Reducing the electron beam energy below about 2 KeV can avoid this large negative potential buildup. In this case, as mentioned above, the surface reaches a stable positive potential with one secondary electron escaping for each trapped electron beam electron. This potential is about 4V for the insulating materials used on silicon VLSI chips. This potential is small enough that it is possible to observe the difference generated therein by applying a typical 5V potential to the lower conductor.

Applying a positive potential of 5 V to the lower conductor, the surface of the sample
The equilibrium potential of 4V moves to a potential of about 9V, the exact potential depends on the thickness of the insulating layer. Since the surface is now positively charged to a potential greater than the 4V equilibrium potential, a larger portion of the secondary electrons will be recaptured by the surface. This large recapture results in less secondary electrons reaching the electron detector, ie the test probe measures a positive potential on the surface of the specimen. Therefore, a reduction of the electron beam energy below 2 KeV makes it possible to measure the potential on the surface of the insulator. In the preferred embodiment, the electron beam has an energy of 1 KeV.

However, the larger number of secondary electrons that are recaptured by the sample surface become progressively less positive until they reach equilibrium potential again. This is illustrated in FIG. As the conductor potential Vc changes from V- to V +, as shown at 266, the surface potential Vp is expected to rise from Veq and become more positively charged, which reduces the number of secondary electrons ejected. This is as the positive potential measured by the test probe is shown at 268. These additional recaptured secondary electrons make the surface less positively charged, so the measured surface potential is
Decrease until q is reached. The potential Vc on the lower conductor returns to zero, as shown at 270, and the surface potential suddenly decreases by V +. This results in a larger number of secondary electrons escaping from the surface, ie the measurements show a negative excursion at the surface potential as shown at 272. The surface becomes progressively more positively charged until the increased number of secondary electrons escaping from this surface reaches Veq again. In practice, the measured signal is differentiated. Therefore, simply reducing the electron beam energy in the region where more than one secondary electron is generated for each electron beam electron is not sufficient to accurately measure the waveform on the lower conductor. Absent.

This differentiation causes the surface to capture more secondary electrons than it loses during the V + portion of the lower waveform and more than it gets during the V− portion of the waveform. This is the result of losing the next electron. The apparatus of the present invention sequentially commands the potential measurements such that secondary electrons obtained during the V + portion of the waveform tend to compensate for secondary electrons lost during the V− portion of the waveform. This avoids this differentiation. In the method of the present invention, a repeating waveform is applied to the lower conductor. Two times are entered into the present invention that specify the period of the repeating test pattern during which a measurement should be made. Potential measurements are then taken at random times between these two times. Each measurement is stored with the time it was made. When enough data has been collected to plot the potential as a function of time, the potential result is displayed. If the lower waveform is V + for half of the time and V- for the other half of the time, a continuous measurement is made on average for the lower conductor at V + and then V-. Secondary electrons obtained on the surface when the lower conductor is at V + are compensated by secondary electrons lost by the surface when the lower conductor is at V-. Therefore,
There is no net long-term change in the surface potential of the conductor resulting from the electron beam impinging on the surface during the measurement process.
If the waveform applied to the lower conductor is half V + and half V-, the compensation is not perfect, but still significantly better than the conventional method.

The specific embodiments of the present invention have been described above in detail, but the present invention should not be limited to these specific examples, and various modifications can be made without departing from the technical scope of the present invention. Of course, it is possible.

[Brief description of drawings]

FIG. 1 is a schematic diagram showing the basic principle of operation of an electron beam test probe, FIG. 2 is an explanatory diagram showing energy distribution of electrons emitted from a conductor after being bombarded with an electron beam, and FIG. 3 is a conventional electron beam. FIG. 5 is a schematic diagram showing the basic method and apparatus used by the test probe method, FIG. 4 is an explanatory diagram showing decomposition of the secondary electron velocity vector into components parallel and orthogonal to the direction of the electron beam, FIG. Is an explanatory view showing a secondary electron flight path obtained from an electric field generated by an adjacent conductor, and FIG. 6 shows a secondary electron flight path between the sample and the focusing magnetic lens.
Explanatory drawing showing the problems of the prior art arising from the need to use a single grid, FIG. 7 is a schematic drawing showing a test probe according to the present invention, and FIG. 8 (a) is a preferred embodiment of the present invention FIG. 9 is a schematic cross-sectional view of a magnetic lens for collimation used in an embodiment, FIG. 8 (b) is an explanatory view showing a magnetic flux pattern generated by the magnetic lens for collimator shown in FIG. 8 (a), and FIG. Is an explanatory view for explaining the collimating action of photons by the optical system, FIG. 10 is an explanatory diagram showing the collimating action of secondary electrons by the magnetic lens for collimating used in the preferred embodiment of the present invention, and FIG. 11 is the present Schematic diagram of the "boxer" integrator circuit according to the invention, FIG. 12 is a schematic diagram showing the mechanical adjustment provided according to the invention for moving the electron beam column relative to the specimen, FIG. 13 is an impact electron On the beam FIG. 14 is a graph showing the relationship between the number of secondary electrons emitted from the insulator per electron and the energy of the impact electron beam, FIG. 14 is a schematic diagram of a delay circuit according to the present invention, and FIG. 15 is a typical diagram. FIG. 16 is a schematic view showing the arrangement of blanking electrodes in a conventional electron beam test probe system, FIG. 16 is a schematic view showing the arrangement of blanking electrodes in a test probe according to the present invention, and FIG. Explanatory diagram showing the charge distribution generated in response to the embedded conductor and the positive potential applied to the conductor, FIG. 18 is the potential measured when a rectangular wave is applied to the embedded conductor shown in FIG. There is an explanatory diagram showing. (Explanation of symbols) 100: Test probe 102: Filament 104: Accelerating electrode 106: Orifice 108: Specimen 110, 112: Magnetic lens 114: Deflection coil 116: Axis 120: Secondary electron detector

Claims (4)

[Claims] 1. An electron beam test probe apparatus for measuring an electric potential at one or more points on a specimen, comprising an electron beam column, a vacuum vessel, a probe card for providing electrical connection to the specimen, and a vertical axis perpendicular to a predetermined axis. Sample positioning means for positioning the sample with respect to the probe card by performing translational movement in two mutually orthogonal directions and rotating about an axis parallel to the predetermined axis, and the electron beam column The probe card and the sample positioning means are all housed inside the vacuum container, and the electron beam column is moved by translational movement in two mutually perpendicular directions perpendicular to the predetermined axis. A column positioning means for positioning with respect to the probe card is provided. The electron beam test probe apparatus, wherein the step is arranged outside the vacuum container. 2. The electron beam column according to claim 1, wherein the electron beam column generates an electron beam having an energy between 500 and 1,500 eV; and the electron beam with respect to a predetermined axis. A first aperture means for defining the divergence of the electron beam, a blanking signal generating means, a blanking electrode means for deflecting the electron beam when the blanking signal is generated, Second aperture means for intercepting the electron beam when the electron beam is deflected by the blanking electrode means, and second electrons for the electrons not removed by the first aperture means when no blanking signal is generated. A first magnetic lens means for forming an image on an aperture means; and the electron beam on the specimen. Electron beam steering means for deflecting the electron beam in one of two mutually orthogonal directions perpendicular to the predetermined axis so as to impact the selected point; and for focusing the electron beam on the sample, Collimating magnetic lens means for collimating secondary electrons generated in response to the electron beam impact of the sample, wherein the first magnetic lens means focuses the electron beam into the collimating magnetic lens means. A magnetic lens means and a second specified energy that is greater than the first specified energy and that is collimated by the collimating magnetic lens means and that is lower than the potential energy from the electron beam backscattered by the sample. An electron detection means for detecting secondary electrons having a small energy, Electron beam test probe apparatus according to claim. 3. The collimating magnetic lens means according to claim 2, wherein the magnetic field for strong collimation substantially coincides with the point where the electron beam strikes the sample, and the predetermined axis. A magnetic lens having a region of uniform magnetic flux having magnetic lines parallel to each other, and the electron detecting means has an electron counting means for counting electrons; A potential barrier means for preventing an electron having a small energy from reaching the electron counting means, and an electron deflecting means for deflecting an electron having an energy smaller than the second specified energy into the electron counting means. An electron beam test probe device comprising: 4. The specimen positioning means according to any one of claims 1 to 3, wherein the specimen positioning means has means for translating the specimen in a direction parallel to the predetermined axis. An electron beam test probe device characterized in that