We have developed an instrument control and image acquisition system for use with scanning electron microscopes. By making the system flexible over a wide range of operating voltages, scan generation and image acquisition modes can be easily accommodated to a wide range of instruments. We show the implementation of this system for use with a custom-built low-voltage scanning electron microscope. We then explore the simple modifications that are required for control of two instruments intended for use as free electron lasers.
We are developing a low-voltage scanning electron microscope (SEM) that uses a permanent dipole magnet placed below the specimen as the objective lens (Crewe & Kapp, 2003). The design specifications indicate that a resolution of 1 nm can be obtained at an accelerating voltage of 1 keV, significantly better than those obtained from commercial instruments (Crewe & Kielpinski, 1996; Tsai & Crewe, 1996). Using a spherical rare earth neodymium–iron–boron magnet, we have been able to obtain fields in excess of 0.6 T at the surface of a 9.5-mm-diameter sphere, sufficient to focus electrons of 6-keV energy or less in this device. The microscope consists of an electrostatic triode gun, which provides a parallel beam of electrons that are focused by the dipole magnet. With the use of a subspecimen lens, the specimen chamber is essentially open, allowing the use of a tilting stage and providing space for large specimens such as the latest generation of 300-mm semiconductor wafers. The use of low-voltage electrons has advantages for biological specimens because of the reduction in radiation damage (for a given amount of image information) compared to high voltages (Pawley, 1995).
We originally developed an instrument control and data acquisition system based on a PC (Intel 486) using Windows Visual Basic Version 3 (Ruan & Kapp, 1992a,b, 1993, 1994, 1995; Kapp & Ruan, 1997). In addition, we used a data acquisition card that was purchased in the early 1990s. This system had become obsolete, and we wanted to take advantage of the dramatic improvements in the speed of current PCs, advances in electronics, and improvements in available commercial software for development projects.
Our goal in this work was to develop an in-house solution using common off-the-shelf components for instrument control and data acquisition. Commercial software for contemporary SEMs is typically proprietary and not customizable. Because we essentially developed our own software using Microsoft Visual Basic, the control and image acquisition system can be easily modified by the user for particular needs.
We are in the process of developing three instruments based upon SEM instrumentation but differing somewhat in their particular requirements. These are a low-voltage SEM (LV-SEM) intended to operate below 5 kV (Crewe & Kapp, 2003), a free electron laser (FEL) based on a modified commercial SEM (Kapp et al., 2004), and a custom-designed FEL based on the use of a flat electron beam.
The LV-SEM has been in operation for a number of years and has a resolution goal of 10 nm at an acceleration voltage of 1000 V. It was originally controlled by an antiquated 486-PC-based control and image acquisition system. Because the computer and the associated cards were well beyond their useful lifetime, it was necessary to replace this system.
We are also developing an FEL based on a Cambridge S-200 SEM that has been highly modified in terms of electron source and beam control. However, image acquisition was still achieved through the approximately 20-year-old digital section of this original instrument. The performance of the electronics of this instrument has become unreliable, and we needed to develop a new control and imaging system.
Finally, we are designing a new FEL using a flat beam source, and we need a control system that can be easily modified as this instrument evolves.
Our goal was, then, to develop a system that could serve the above instruments and be flexible enough to accommodate future instrumental designs as they occur.
For the electronics of the LV-SEM, absolute precision is not an issue, and it is not necessary to know the voltages or currents to any better than about 1%. On the other hand, both stability and noise are of great concern. This can be illustrated by looking at the high voltage supplies. There are four relevant supplies and they operate in the range 0–5000 V. The spread of energy in the electrons from the source we are using is about 0.2 V, and we would like to keep the effect of the instabilities in the supplies to as low a level as possible. We would like this variation to not exceed 0.1 V for each supply.
The LV-SEM requires several voltage and current sources to supply the electrostatic and magnetic components of the device. Also required are additional electronics to detect and display the resulting image. In the past, these were supplied by a modified system originally designed for a sextupole-corrected 200-kV microscope built in this laboratory several years ago (Ruan & Kapp, 1992a,b, 1993). Extensive modification was required, with many compromises to accommodate the much lower voltages (5 kV) and currents used in the current microscope. This system had reliability problems due to its advanced age, and there were additional difficulties in obtaining replacement parts, many of which are now obsolete. The electronics were contained in several large equipment racks with extensive interconnections that exacerbated the reliability problem in addition to consuming a large amount of laboratory space. The use of these antiquated electronics was becoming untenable, and a decision was made to construct a new system for the microscope with reliability and compactness as design goals.
Figure 1 shows an overall block diagram of the LV-SEM. The upper chamber of the microscope is maintained at 1 × 10−10 torr using a 20 L s−1 Varian ion pump and a 400 L s−1 nonevaporable getter (model ST 172, SAES Getters). This chamber contains the electron gun, which uses a thermally assisted field emission Schottky source (FEI Corp., Hillsboro, OR, U.S.A.) and provides a parallel exit beam. In the gun, the emitter is placed in a suppressor fixture, which behaves in a fashion similar to the common Wehnelt cap in a conventional SEM or transmission electron microscope. The suppressor, extraction and focus electrodes are separated by sapphire balls (for electrical insulation), and a final anode at ground contains the limiting aperture. Beam deflection coils are wrapped around the mu-metal gun frame.
The upper chamber is separated from the lower specimen chamber by a miniature value which allows a pressure differential of 103 between the upper chamber and the 1 × 10−7 torr lower chamber. The lower chamber contains the lower scan assembly, consisting of the stigmator coils and the two raster scan coils, A and B.
The LV-SEM uses an yttrium aluminium perovskite (YAlO3) (YAP) scintillator, a quartz lightpipe and a photomultiplier (PMT) to detect secondary electrons from the sample. As part of upgrading the microscope, we redesigned the YAP–lightpipe–PMT assembly to improve electron and light collection for a better signal-to-noise ratio. The new lightpipe is 18 inches in length, with a diameter of 1 inch. The beam passes through a hole located at the end of the lightpipe (Fig. 1). A grounded stainless steel tube is fitted to the hole and exits out the bottom of the lightpipe below the YAP (#P47, SPI Supplies, West Chester, PA, U.S.A.) scintillator assembly, which is in turn attached to the flattened surface of the underside of the lightpipe. The 20-mm-diameter scintillator is coated with 20 nm of aluminium and is an annulus to allow passage of the incident beam. This configuration presents a symmetrical target for secondary electrons and minimizes distortion of the low-voltage beam, due to voltage biases applied to the scintillator.
The permanent magnet objective lens is placed below the specimen (Fig. 1, bottom) and provides a focusing field just above the specimen. Because the dipole field drops off with the inverse cube of the distance, the field is localized near the specimen, helping to prevent spurious effects further up the column.
Figure 2 shows a general schematic of the microscope and control and image acquisition system. With the exception of the computer and displays, all the electronics are housed in a one half-height equipment rack, as opposed to the several full-height racks needed for the old system. The new system is designed around a generic PC-type computer using a high-speed RS485 serial communication link between the computer and the peripheral equipment rather than the bulky, somewhat unreliable parallel link employed in the old system.
Below we discuss the control and image acquisition properties of the LV-SEM, starting at the top of the column and following through to the formation of an image. We begin by describing the hardware and then the use of the GUI (Graphical User Interface) for control of the various microscope parameters.
Tip heater supply
Heating power for the field emission tip is supplied by an isolated current source capable of supplying up to 3 A with a compliance of 5 V (Fig. 3). The power requirements are rather modest for this application, and 2–2.5 A of current are typically sufficient to produce enough heat to operate the Schottky electron source. The desired tip voltage (acceleration voltage) from the high-voltage supply passes through a current-sensing circuit before being applied to the tip. The circuit measures the electron emission from the tip in three ranges of 20, 200 and 2000 µA. The emission current and heating currents are displayed on separate 3.5-digit digital panel meters. The current supply and emission measurement are isolated to sustain the maximum 5000-V differential between the tip and circuit ground.
The heater supply is of straightforward design. However, one feature of note is the use of a low-cost commercial transformer for isolation. High-voltage isolation transformers are usually custom designed for their output characteristics and isolation voltage. We have found that commercially available transformers, at least for the moderately high voltages employed in the microscope, provide for long leakage paths and adequate voltage isolation. In our case, a low-cost transformer (part # 14A-56-20, Signal Transformer Inc., Inwood, NY, U.S.A.) conforming to universal electrical standards provides low leakage (< 100 nA) at primary to secondary voltage differentials as high as 6 kV.
Adjustment of the heating current to form the tip and establish the desired emission current is a delicate procedure that also requires close monitoring of the vacuum in the region of the tip. Although it may be possible to automate the current adjustment using computer control, doing so would require somewhat complicated software, as well as interfacing of the heater supply and vacuum gauges to the computer. At this stage of development, it was decided to use manual adjustment of the heating current to establish the emission current, with direct observation of the currents and vacuum conditions.
High-voltage settings (suppressor, acceleration, extraction and focus)
The suppressor, acceleration and focus supplies (Bertan, PMT, Hauppauge, New York, U.S.A.) are unipolar, have a range of 0 to −5000 V, and are controlled by 16-bit digital–analog converters (DACs) (Fig. 4). The extraction supply is unipolar and has a range of 0 to +5000 V.
The high-voltage supply and scan generator are controlled by the computer over a high-speed RS485 serial data link (Fig. 4). The computer uses a commercially available plug-in card to send data over the link to receiving microprocessors in each device. An identical microprocessor circuit is used in both devices. Different software is programmed into the processors to establish a unique address and to format the data for each device. A common data format, which includes a checksum for data verification, is used for both the high-voltage supply and the scan generator. To minimize interference from the computer, the microprocessor circuit board and power supply are physically separated from the remainder of the device, and all signals are transferred to the main circuit through digital isolators.
The scan unit contains two 12 bit DACS that generate the X and Y ramps. Additional 12 DACs rotate, scale and offset these ramps (Fig. 5). The logic for controlling the 14 DACs is contained in a Xilinx XC95108 programmable logic chip. A phase-lock loop locked to the 60-Hz power frequency drives counters in the logic chip that step the X and Y ramp DACs. The ramps are fed to LTC 1590 dual 12-bit DACs, which first rotate the ramps and then scale them. Finally, a DC offset is added. All analog voltages are proportioned for ±10 V full scale. There are eight analog processing channels, four for the X ramp and four for the Y ramp. The processed ramps are fed to eight current amplifiers with ±2-A output range. The ramp currents and DC offset currents range to ±1 A, providing a total range of ±2 A. A full image is scanned in 12.8 s (768 lines/60 Hz).
PMT and YAG
The PMT supply is unipolar and has a range of 0 to −1000 V. The YAP supply is unipolar and has a range of 0 to −5000 V.
LV-SEM control program
The microscope is controlled through a GUI, written in Microsoft Visual Basic 6, and displayed on two monitors. The left monitor displays the microscope operating parameters while the right monitor displays the acquired image (Fig. 6). Adjustment of the control parameters is visually interactive, and the effects are immediately apparent in the image when the acquisition program is in continuous scan mode (see below); changes in focus, stigmatism correction, etc. are easy to view, because the image is continuously updated in a line-by-line fashion. Parameters are displayed on the control screen in both graphical and text format. Communication with the scan and high-voltage units is accomplished by a 115 000-Baud RS-485 serial link. Future plans include the replacement of this link with a higher-speed (USB or Firewire) link.
User manipulation of high voltages
The operator sets the gun, PMT and YAG (Yttrium aluminium garnet) voltages by either entering the value in the appropriate text box or using the up/down buttons. Step size can be from 100 to 0.01 V. The gun voltages that are displayed on the GUI interface (Fig. 6, left monitor) are referenced to the accelerating voltage for the convenience of the operator. The program calculates the actual voltage (referenced to ground) that has to be provided to each element. For example, if the program reports an accelerating voltage of 1000 V and a suppressor voltage of −1000 V, then the tip is at −1000 V relative to ground and the suppressor is at −2000 V (its supply is generating −2000 V). The extractor voltage is also reported by the computer display as the difference between the extractor and the tip. However, because the extractor output is positive, the difference is never less than the accelerating voltage. The PMT and YAG voltages displayed on the control GUI (Fig. 6, left monitor) are referenced to ground. The voltage applied to the PMT provides the brightness control for the signal. Contrast control can be achieved by adding a bias voltage.
User manipulation of scan parameters (amplitude, rotation and DC offset)
The operator controls the amplitude of the upper and lower (scan A and scan B, respectively) coils. The control program displays a scan screen, representing the area scanned by the microscope, which uses standard Cartesian coordinates. The scan screen features a visual representation, in the form of a grid icon, of the amplitude, rotation and DC offset values of the scans (Fig. 6, left screen). The size of the scan icon is scaled to the scan amplitude. For large scan amplitudes, the grid covers approximately one-quarter of the area allocated for the scan display. As the scan decreases in amplitude, the icon becomes progressively smaller, until a preset level is reached at which the icon changes in appearance to indicate that this scan (A or B) is off. Of course, if both A and B scans are off, the beam is stationary, striking the specimen without being swept in a raster. An image can be formed by scanning with A or B separately or by using both.
Scan A and B rotation covers the range from −180 to +180. The rotation value is visualized with the scan amplitude icon. Different fields of view can be obtained by using various combinations of rotation for scans A and B. To obtain the lowest magnification (largest scan size), the B scan is rotated 180° in relation to scan A.
A bipolar DC offset can be added to scan A or B. The relative position of the scan is indicated by the position of the scan icon on the scan screen. Positive values of X move the scan right, and positive values of Y move the scan up. This gives the operator a way of scanning different areas of interest on the sample.
All scan parameter values can be either entered directly by the user into a designated text box, or modified by using slide bars (see Fig. 7 for detail of the right side of the first display in Fig. 6). In addition, the DC offset, as well as the current through the gun and stigmator coils, can be modified by moving the corresponding scan icons on the scan screen with the mouse. For the finest positioning (at high magnifications), the operator should use either the text entry boxes or the slide bars; the mouse drag capability should only be used for coarse positioning at low magnifications.
In addition to modifying scan parameters, the control program allows the user to switch from a normal scanning mode to a line scan mode in which, instead of scanning in a square raster, both the A and B scans hold their Y position and scan continuously across a single line of the specimen. As a result, an image acquired in line scan mode consists of this single line of grey scale pixels repeated vertically. Under many conditions, this feature allows for quick and accurate adjustment of focus. A line scan is initiated by the image acquisition program and not the control program; the user selects a line to be scanned, which the former then communicates to the latter using standard Windows protocols.
User diagnostic utility
The control program features several menus that display diagnostics to aid the user if an unexpected condition is experienced during operation of the microscope. For instance, one diagnostic screen provides a way of adjusting the gun and scan parameters outside of ‘normal’ operation. By use of the diagnostic utility scan coils A and B, gun and stigmator parameters can be verified for proper operation. Enhancements to the hardware and software can be checked for proper operation before program operation takes place. Caution should be exercised when using the diagnostic mode, as the regular error checking performed by the program is not in place. For example, if scan A and scan B amplitudes are set to zero, the beam will be stationary, possibly causing damage to the sample or to the internal structures of the microscope.
Saving and recovering microscope settings
For convenience, reproducibility, and image annotation purposes, the control program includes the ability to save the operating parameters of the microscope to an ASCII file. The control program is automatically prompted to do so by the image acquisition program whenever the user saves an image using the latter; the file is then used to annotate the saved image (see below). The user can also directly prompt the creation of the file. The control program also has the ability to read an image data file (see below) and recover from it the operating parameters at the time the image was saved. This feature provides the operator with the ability to resume a session days or weeks later, using the same settings and returning to the same field of view.
Image acquisition − EFI-Acquire
In order to acquire images generated by the LV-SEM, the PC through which the microscope is controlled was equipped with a DT-3152 card, a PCI frame grabber manufactured by Data Translation, Inc. (Marlboro, MA, U.S.A.). The card is accompanied by a library of Visual Basic functions that allow the user to control and operate the card in the framework of a Visual Basic interface.
Such an interface was provided by the card manufacturer in the form of DT-Acquire, a sample image acquisition program written in Visual Basic. However, this program as written was somewhat inflexible and lacked key functions, i.e. storage and display of settings, image saving, image annotation, and many others. A more sophisticated interface was needed. This interface, EFI-Acquire, was created by heavily modifying and building upon the source code for DT-Acquire. EFI-Acquire incorporated the features necessary for the operation of the microscope, in particular the ability to save images displayed by the card to disk. This was done in a user-friendly and economical fashion, so as to minimize the learning curve for both users of the program and programmers who might need to modify or supplement the code in the future.
One important feature implemented in EFI-Acquire was a means for the program to automatically set the card parameters to default values. This was lacking in the DT-Acquire template, which meant that all card parameters had to be re-entered every time the program was launched.
The most important feature of EFI-Acquire is the ability to save acquired images to disk. The low-level functions that make this possible are included in the library that accompanies the DT-3152 card. In addition, each image stored by the microscope operator has a corresponding set of data that must be recorded and stored. These data consist of the values of microscope parameters at the time of save, imported from the ASCII text file generated by the LV-SEM control program (see above), as well as any comments that the user wishes to make. All of this information is stored in an ASCII text file, separate from the image file.
Along with implementing the above features, it was necessary to create an organizational scheme that efficiently archives saved images and allows the user to easily browse through them and their accompanying data files. After implementing this scheme, it was possible to equip the program with the ability to display an image file and to simultaneously find and display its associated data file, as well as to delete an image file along with its data file. Because these features were implemented in EFI-Acquire, the user does not have to resort to a second program in order to manage the archive of saved images. A second program is needed only when the user decides to process or manipulate a particular image.
During use of EFI-Acquire, image acquisition runs independently of the control program on the second video monitor. This allows the operator to make adjustments to the scan and see the effect in real time. Image acquisition is performed by taking the output from the PMT and converting it into a 256 grey level raster-scanned video display (see Fig. 6, right monitor). The scan signals consist of a pixel clock, frame sync, and line sync, and are generated by the scan unit. The raster consists of a 1024 × 768 pixel array. These signals are fed into the DT-3152 board, where they are processed.
Operating the program
EFI-Acquire is operated as follows. At launch, the user is asked to either choose or create a directory that will contain saved files. After a save directory is defined, the main program form is displayed (the user may change the save directory at any future point).
In order to acquire images, the user must load the DT-3152 card, at which point the program sets card parameters to their default values (card parameters may be changed, or the default values restored, at any point). After doing so, the user may perform a single frame acquisition, multiple frame acquisition, or continuous image acquisition. In a single frame acquisition, a single image frame is acquired but not saved. In a multiple frame acquisition, the user specifies how many frames are to be acquired; these are then consecutively acquired and saved to disk. When performing a continuous acquisition, the user may save to disk an instantaneous snapshot of the image on the screen.
When an image is saved to disk, EFI-Acquire also creates and saves an associated data file. The image and accompanying data file are named and saved to disk in accordance with the organizational scheme. At any point during which images are not being acquired, the user may open an image file for viewing, and subsequently close or delete this file. When opening an image file, the program searches for the accompanying text file and, if it exists, displays it alongside the image.
Figure 6 shows the display arrangement of the two monitors used to operate the microscope and display the microscope images. The right screen shows the 1024 × 768 image obtained from the microscope. The slow scan is 12.8 s per frame (768 lines/60 Hz). The displayed image can be immediately stored and later displayed in the second window for review, as described above. The image in Fig. 6 shows a detail (the letter W) of a standard metal specimen displaying letters of the alphabet and numerals. Because the specimen consists of readable letters, it is convenient for use in testing the behaviour of the scans, particularly magnification and rotation. At higher magnification it is much easier to keep track of scan position, due to the readily identifiable markers. The image shown is the sum and average of 50 images taken in sequence with the new system over a time period of 10.7 min (50 × 12.8 s). Details can be seen in the 20-µm-thick arms of the letter. Full scale is 170 µm.
Application to similar instrumentation
Implementation for the Cambridge S-200-based FEL
The control and image acquisition programs can conveniently be used with some simple modifications to operate our FEL (Kapp et al., 2004) which is based on a modified commercial SEM (Cambridge S-200). This instrument was modified to produce Smith–Purcell far-infrared radiation by the insertion of a metal grating with an appropriate period into the path of the beam. Photons are emitted from the grating surface and pass through a polyethylene window, where they are detected by a helium-cooled bolometer. This instrument operates at higher voltages (20–40 kV) than the LV-SEM and uses either a tungsten filament or an LaB6 source. Conditions of geometry, Wehnelt voltage and filament/tip current are different for the two sources, and the latter two may be easily set by the operator, using the controls described above. The source current may be analog (as we chose for the LV-SEM) or digital. As mentioned earlier, the digital section of the S-200 is quite antiquated and we can bypass this by connecting to the scan circuit of this instrument and coupling this to the output from the YAG–lightpipe–PMT secondary electron source, as was done with the LV-SEM. Again, a two-monitor setup is used and the image acquisition is accomplished with the modified DT-Acquire software, using an identical DT-3152 card.
Implementation for a flat-beam FEL
The flat-beam FEL requires some additional control parameters. Specifically, the electron gun is enclosed in a solenoid so that electrons are born with angular momentum. In addition, production of the flat beam requires the presence of several quadrupoles, each with identical supplies. The control of these elements will be added to the control screen using the same GUI features as described above. In addition, we would like to digitally collect the far-infrared output from the bolometer as well as oscilloscope traces and integrate these as part of the daily documentation of the instrument operation.
We have developed an instrument control and image acquisition system for use with three imaging instruments with somewhat different operating requirements. The control program, written in Visual Basic 6.0, was designed to be easily modified to accommodate these systems, and the image acquisition software was designed to address the different requirements for image output. These two programs run independently and can be easily integrated from the operational standpoint by the use of a single computer with two displays. We have demonstrated the use of the system on our LV-SEM and are in the process of modifying the program to control the FEL. As we settle on the design parameters for the new flat-beam FEL, we will be able to concurrently modify the code so that the development cycle for this part of the program is minimized.
This work was supported in part by NSF grant #5-27483 (to OHK and AVC) and a seed grant from Argonne National Laboratories (to KJK and OHK). Additional support was obtained from the Illinois Consortium for Accelerator Research #6-34022 (to KJK), NSF grant #PHY 0104619 (to KJK) and through the Block Fund of the University of Chicago (to KJK and OHK). Several undergraduate summer students participated in this work through support from the NSF-sponsored REU programme.