A Portable Scanning Electron Microscope Column Design Based on the Use of Permanent Magnets

Transcription

1 SCANNING VOL. 20, (1998) Received October 8, 1997 FAMS, Inc. Accepted with revision November 9, 1997 A Portable Scanning Electron Microscope Column Design Based on the Use of Permanent Magnets A. KHURSHEED, J. C. PHANG, J. T. L. THONG The Centre for Integrated Circuit Failure Analysis and Reliability (CICFAR), Electrical Engineering Department, National University of Singapore, Singapore Summary: A portable scanning electron microscope (SEM) design is presented which makes use of permanent magnets. Simulation results predict that such an SEM is feasible and that it can be compact. The height is typically less than 12 cm. The is designed to be modular, so that it can fit onto a wide range of different specimen chamber types, and can also be readily replaced. Key words: s, miniature, permanent magnets, scanning electron microscope Introduction The microminiaturization of electron optical systems has received considerable attention over the last decade (Chang et al. 1990, Chisholm et al. 1997). At present, the scanning electron microscope (SEM) is a bulky device which, when installed at a specific location, is not designed to be moved again. There are several advantages to making the SEM portable, where the overall height from the specimen chamber to the electron gun is less than 12 cm. First, more compact SEM s give rise to the possibility of making the SEM modular. The bulky items such as vacuum pumps, drive, and display electronics are fixed, and the SEM s electron optics section, that is, the part comprising the, can be easily removed and replaced. Figure 1a depicts a schematic diagram of this concept. In principle, the user will be able to choose from a range of different SEM s, each suited to different applications, while employing the Address for reprints: A. Khursheed The Centre for Integrated Circuit Failure Analysis and Reliability (CICFAR) Electrical Engineering Department National University of Singapore 10 Kent Ridge Crescent Singapore eleka@nus.sg same vacuum and electronics system. By simply replacing the, the SEM can, in principle, be converted into a different analytical tool. At present for instance, quantitative voltage contrast or critical dimension topographic measurements, or Auger spectrometer analysis require the use of different SEMs, each designed to meet different requirements. The portable SEM provides the opportunity of having an SEM cluster, where each is designed for a different application. The cluster of SEM s will be compact enough to fit on a normal shelf or table top. The second advantage of a portable SEM is that it can be used on a variety of different specimen chambers. This possibility is illustrated in Figure 1b. Since specimen chambers greatly vary in their size, shape, and vacuum conditions, the ability to use the same SEM for them all greatly reduces the cost and space requirements for SEM systems in general. At present, if a different specimen chamber is required, it necessitates the use of a different SEM. The third advantage of the portable SEM comes from its reduced volume. The smaller volume leads to faster pump down times and simplifies vacuum requirements. Fourth, since the on-axis aberrations scale with the size of the electron lenses, the spatial resolution of the final primary beam is expected to be significantly better than that of conventional SEMs. Fifth, the shorter primary beam path length means that it will experience less variation from stray fields. A shorter path length is also expected to result in smaller probe sizes, since it reduces the energy spread in the primary beam due to the effect of electron electron interactions. Finally, a light-weight can be scanned or moved over large distances. This might be useful where the specimen is large and cannot be easily moved, or where its movement is restricted by a bulky item, such as another piece of test or measurement equipment. The main difficulty in scaling down the size of present SEM s relates to the current coil density in the magnetic electron lenses. The increase in current density is proportional to the square of the factor of reduction. Already at their present size, special considerations of cooling need to be taken into account. The use of permanent magnets is a way of overcoming this problem. Permanent magnets, unlike conventional SEM electromagnetic magnets, do not operate by employing current carrying coils and so avoid potential heating problems.

2 88 Scanning Vol. 20, 2 (1998) Permanent Magnets Efforts to introduce permanent magnet lenses into electron microscopy date back from the very beginnings of electron microscope design. At that time, they were advocated as a way of overcoming power supply instability problems (Rieke 1982), but it soon became apparent that the field strength of permanent magnets was not strong enough for many common applications in electron optics. It was not until the 1960s, when new permanent magnet materials based upon rare earth metals became available, that permanent magnet technology rapidly improved. At present, the coercive force (H c ) of a neodymium-ironboron permanent magnet can be as high as A/m (Magnet Sales and Manufacturing, Inc.), which, for a cylindrical tube shaped magnet 2.5 cm high, corresponds to an equivalent coil strength of 22,500 AT. This excitation strength is much higher than that required by conventional SEMs in their normal mode of operation. The coil excitation in a standard objective magnetic lens required to focus a 25 kev beam at a working distance of 5 mm is around 1,000 AT, a factor of 22.5 times smaller than that provided by the permanent magnet. Recently, the possible use of permanent magnets to make miniature electron lenses has been proposed (Adamec et al. 1995). Replacement Specimen chamber SEM Display Permanent magnets are very stable; after 100,000 hours of operation, the change in the field strength is only a few fractions of a percent. Another attractive feature is that they can be made in a variety of different shapes and sizes and with different orientations of magnetisation. Since magnetic electron lenses have cylindrical symmetry, the two possible directions of magnetisation are in the longitudinal direction, parallel to the electron optic axis, or perpendicular to it, in the radial direction. Both possibilities have been advocated for electron optics applications (Adamec et al. 1995, Rieke 1982). In this study, the direction of magnetisation is chosen to be parallel to the electron beam. Although it is possible to manufacture the permanent magnet directly in the required cylindrical tube shape, this is not usually done in many different sizes for high coercive force materials; however, this does not present a major difficulty, since the permanent magnet can be made of curvilinear blocks. Typically, four blocks can be used, one for each quadrant. Furthermore, it is quite common for permanent magnet manufacturers to make custom designs. The permanent magnet lens designs presented in this paper were made through the use of the KEOS finite element programs (Khursheed 1992). These programs are able to include the effect of saturation in the iron circuit and calculate thirdorder deflection aberrations on the primary beam. The incorporation of permanent magnets in KEOS was done by modelling the coercive force in terms of two thin current-carrying coils at the iron-air interface, as shown in Figure 2. The effective relative permeability of a permanent magnet is around unity, and this value is input into the simulation program. It is relatively straightforward to show from Maxwell s magnetostatic equation for the field intensity H and current density J in cylindrical coordinates that the magnitude of the excitation current at each interface NI is given by H c L, where L is the Table Vacuum system Drive electronics Z (optic axis) SEM NI North NI H C L Specimen chamber 1 Specimen chamber 2 South FIG. 1 Scanning electron microscope (SEM) replacement. The portable SEM and different specimen chambers. FIG. 2 Simulation model of the permanent magnet. R (radius)

3 A. Khursheed et al.: Portable SEM design using permanent magnets 89 magnet height. The directions of these interface currents are in the azimuthal direction and are of opposite sign. Note that this formula is only valid if the thickness of each coil is small relative to the magnet s height. A suitable way of choosing the thickness of each coil is to make it equal to the local numerical mesh spacing of the finite element program. The accuracy of the finite element programs was investigated for a two-gap permanent magnet test lens. The axial field distribution for this lens was measured experimentally by the use of a Hall probe. The experimental results agreed to within 10% of the simulation predictions (Khursheed et al. 1997). The Basic Scanning Electron Microscope Column Design Figure 3 shows a schematic of the portable SEM. The SEM has an overall height of around 12 cm. Both the objective lens and condenser lens were designed to avoid saturation problems by increasing the thickness of the iron in the magnetic circuit and placing the position of the permanent magnets far from the electron optic axis. The condenser lens uses two gaps in the same magnetic circuit, and so, in effect, is 80 mm 14 FIG. 3 Portable scanning electron microscope schematic. 1 = Magnetic specimen holder; 2 = nonmagnetic specimen chamber; 3 = magnetic circuit of the objective lens; 4 = deflection coils; 5 = permanent magnet of the objective lens; 6 = magnetic slip ring of the objective lens; 7 = electron detector; 8 = predeflection coils; 9 = magnetic circuit of the condenser lens; 10 = permanent magnet for second air gap of the condenser lens; 11 = permanent magnet for the first air gap of the condenser lens; 12 = upper magnetic slip ring of the condenser lens; 13 = lower magnetic slip ring of the condenser lens; 14 = electron gun equivalent to two condenser lenses. Magnetic slip rings are used to short the magnetic circuit and so control the strength of the magnetic field on axis. These slip rings provide a way of varying the demagnification of the primary beam spot size. The outer magnetic slip ring on the objective lens can be used to focus the primary beam onto the specimen. Focusing can also be achieved by varying the specimen height. The objective lens design is based upon a single pole magnetic immersion lens. The peak magnetic field typically lies around 2 mm below the lower pole piece tip. Deflection coils are set above and within the objective lens and operate together to scan the primary beam in a moving objective lens configuration. The electron detector is placed above the objective lens and collects the secondary and backscattered electrons. The detector may be either a microchannel plate or a scintillator detector. The trajectory paths of the secondary electrons will be collimated due to the sharply decreasing magnetic field of the objective lens, and their transport efficiency from the specimen to the electron detector is expected to be high, better than the transport efficiency for secondary electrons in conventional SEMs. The kind of collection system described here is known as a through-the-lens secondary detection system and is commonly used in high resolution SEM systems (Electron Beam Testing Technology, 1993). The proposed permanent magnet is NdFeB grade 35 (H c = A/m) (Magnet Sales and Manufacturing, Inc.). Various types of iron with different saturation characteristics were considered for use in the magnetic circuit of the lenses. Obviously, high saturation material such as Permendur is preferable, but the simulations showed that the results were not critical on the material type. For the size considered, ordinary mild steel produced results that were only marginally worse than the Permendur material. The flux lines for both the condenser lens and objective lens are shown in Figure 4a and b, respectively. As a first approximation, each lens was analysed separately. The open boundaries around the magnets were substituted with a zero vector potential boundary, which has the effect of lowering the field strength on axis by around 5 to 10%, so the on-axis field strengths presented in this paper are slightly underestimated. It should also be noted that the flux lines at the top of the objective lens and at the bottom of the condenser lens unit are inaccurate since, in practice, the objective lens is connected to the objective lens. This error does not significantly affect the field distribution inside the SEM. In practice, flux lines will connect the top of the condenser lens with the bottom of the objective lens. The flux lines in Figure 4a and b show that the field distribution for the portable SEM is quite different from conventional SEM s. For conventional SEMs, almost all the field distribution is contained within the microscope, but for the portable SEM, the field distribution extends into the region around the. For most applications, this should not affect the operation of the SEM. The equivalent excitation current for the permanent magnets of the condenser lens is 7,200 AT. The permanent mag-

4 90 Scanning Vol. 20, 2 (1998) nets produce a maximum axial field strength of 0.71 Tesla at the centre of each gap in the condenser lens, which is able to produce a focal length of 1.9 mm and a demagnification of 20 for a 25 kev beam. The total demagnification of the condenser lens is thus typically 400. When the slip rings are pushed to the positions which short circuit the permanent magnets, the maximum field strength falls to 0.14 Tesla, giving a focal length of 1.9 mm and a demagnification of 20 at each gap for a 1 kev beam. The axial flux density for the condenser lens is shown in Figure 5. The two narrow peaks correspond to the two gaps in the magnetic circuit. Each gap spans 1 mm. In principle, the condenser lens can be made at least two to three times smaller in size since the height of the magnets is only 8 mm. Figure 6a shows a schematic representation of the pole tip of the objective lens. It is mainly the pole tip dimensions that determine the primary beam optics of the lens. These dimensions include the effective on-axis gap size G in the magnetic circuit, the radius of the lower pole tip R, and the working distance W. The allowable thickness of the specimen is obviously given by G-W. In the following design, R=2.4 mm, which allows sufficient space for in-lens deflection coils to be mounted on the lower pole tip of the objective lens (Zhao and Khursheed et al. 1997). The use of a magnetic specimen holder helps to increase the field strength at the specimen. As the gap size G decreases, the peak magnetic field strength on the axis increases, which in turn lowers the on-axis aberrations. Significant improvement occurs for gap sizes which are < 15 mm; beyond this amount, the results are similar to those obtained for a nonmagnetic specimen holder. For a gap size of mm, a 1 kev primary beam focuses at a working distance of W=1.16 Axial magnetic field strength (Tesla) Gap 1 Gap FIG Distance along the axis (mm) Axial flux density for the condenser lens. Pole tip R W G Axis 0.80 Axial magnetic field strength (Tesla) Lens gap G = mm G = 2 mm FIG. 4 Flux lines for the condenser lens unit; flux lines for the objective lens Pole tip Distance along the axis (mm) FIG. 6 Pole tip and lens gap schematic of the immersion objective lens. Axial field variation around the pole tip for two different gap sizes in the immersion objective lens.

5 A. Khursheed et al.: Portable SEM design using permanent magnets 91 mm. For these low-voltage conditions, the focal length is calculated to be 1.53 mm, the spherical aberration coefficient is mm, and the chromatic aberration coefficient is mm. Conventional magnetic lenses have chromatic aberration coefficients which are typically > 2 mm. Since, at low voltages, it is chromatic aberration which mainly determines the final primary beam spot size, the spatial resolution of the miniature is predicted to be typically a factor of two better than that of conventional SEMs, which is in keeping with the performance expected for magnetic immersion objective lens systems (Reimer 1993). Obviously, for thin specimens the lens gap size G can be reduced. Figure 6b shows that, for G=10.22 mm, the maximum axial flux density is 0.24 Tesla, while for G=2 mm it rises to Tesla and is able to focus a 20 kv beam at a working distance of 2.19 mm. For these operating conditions, the focal length is 1.76 mm, the spherical aberration coefficient is 1 mm, and the chromatic aberration coefficient is 1.24 mm, giving a spot resolution of 0.63 nm for a Schottky gun emitter (with an assumed energy spread of 0.8 ev). These results show that the portable SEM should be capable of operating over a wide range of beam-operating voltages. Further increases in the axial magnetic field strength can be obtained by using a longer permanent magnet. For low-beam voltages, such as 1 kv, the outer slip ring is used to short the magnetic circuit. There are, in principle, many ways in which a number of portable SEM s can function together. One possibility is to use several SEM s for simultaneous inspection of different parts of a large specimen, such as a 150 mm diameter integrated circuit wafer. In this case, the specimen is placed on a large magnetic table, and instead of moving the specimen to examine the circuits on it, each SEM is moved. The diameter of each SEM can be designed to be < 50 mm, making it possible to have an array of at least four SEM s simultaneously inspecting the specimen. For other applications, it may be possible to create an SEM cluster, where an array of SEM s collect information in parallel. Conclusion A portable SEM design has been presented which is based upon the use of permanent magnets. The SEM s height is designed to be below 12 cm. Computer simulations predict that the SEM should be able to provide high spatial resolution. The is designed to be modular, so that it can be fitted onto a wide variety of different specimen chambers, and can also be easily replaced. A patent application on the portable SEM design has been made, and a project involving the National University of Singapore and the Institute of Materials Research and Engineering in Singapore has been initiated to provide the experimental verification of the simulation results presented in this paper. References Adamec P, Delong A, Lencova B: Miniature magnetic electron lenses with permanent magnets. J Microsc 179, Pt. 2, (1995) Chang THP, Kern DP, Muray LP: Microminiaturization of electron optical systems. J Vac Sci Technol B8 (6) (1990) Chisholm T, Liu H, Munro E, Rouse J, Zhu X: A compact electrostatic lithography for nanoscale exposure. Proc SPIE, Charged Particle Optics III, 3155, (1997) Electron Beam Testing Technology (Ed. Thong JTL), Ch. 2, 3, and 5. Plenum Press, New York (1993) Khursheed A: KEOS: Khursheed Electron Optics Software. Department of Electrical Engineering, University of Edinburgh, Mayfield Road, Edinburgh EH9 3JL, Scotland, UK (1992) Khursheed A, Thong JTL, Phang JC, Ong IP: Miniature scanning electron microscope design based upon the use of permanent magnets. Proc SPIE, Charged Particle Optics III, 3155, (1997) Magnet Sales and Manufacturing, Inc., Playa Court, Culver City, CA Reimer L: Image formation in low voltage scanning electron microscopy. Ch. 1, Tutorial texts. SPIE (1993) Rieke WD: Practical lens design. In Magnetic Electron Lenses, section 4.1 and 4.6, (Ed. Hawkes PW). Springer-Verlag, New York (1982) Zhao Y, Khursheed A: In-lens deflectors for a LVSEM magnetic immersion objective lens. Proc SPIE, Charged Particle Optics III, 3155, (1997)