Acquisition of field ion microscope image using deflector during atom probe analysis

The combination of a field ion microscope and time of flight (ToF) mass analyzer was achieved as a field ion microscope-atom probe (FIM-AP) by Barofsky¹ in 1968 to obtain the structural information and the compositional analysis in a single instrument. The next breakthrough was made by the development 3D-atom probe (3D-AP) equipped with a two-dimensional position-sensitive ion detection system in the 1990s. The tomographic analysis on an atomic scale by 3D-AP has come to an essential technique in microtechnology today.² 

The pulse field evaporation performed at cryo-temperature enables AP to analyze samples on an atom-by-atom scale; therefore, AP has been assumed to be the most sensitive analysis technique on a nanometer scale because the ionization possibility is very high, close to 100%.³ Although the region of analysis is restricted to several hundred nanometers along one axis of the bulk sample, the high detection efficiency (DE) of AP is thought to be much higher than other analysis techniques.² 

We report the synchronization of AP analysis and the FIM image acquisition in conventional FIM-AP systems in this work by installing an ion deflector in front of the FIM screen and image-capturing system. The evaluation of the ion detection efficiency of AP was performed by the comparison between the ion hit count in the AP data set and the structural change in FIM in an atom-by-atom manner.

It should be noted that “the field of analysis” (FOA) in conventional AP in which the mass analysis is performed using a one-dimensional mass spectrometer is restricted to the small area corresponding to the probe hole located on the center of the phosphor screen of FIM. The probe hole also works as the aperture of ion optics of the mass spectrometer of AP. When we hope to “see” the FOA in common FIM-AP, we had to change the sample orientation so that the FOA pointing the neighboring area of the probe hole. By using the deflector before the screen, we can see FOA on the screen and the structural information can be acquired only by the application of voltage controlled by the ion detector system of AP.

The introduction of image gas of 10⁻³ Pa order in conventional protocol gives a bright FIM image that can be captured in less than 1 s exposure time. At the same time, the image gas ion must be blocked out from reaching the ion detector of the mass spectrometer because the amount of ion of the image gas ion surpasses that of the sample ion and the image gas ion is emitted continuously during the AP analysis. The toroidal electrostatic lens, which served as an energy filter, was utilized in this work to block the image gas ion from going into the ion detector.

The DE of field evaporated ions is an important factor in AP analysis.⁴ Dealing with the 3D-AP data set, several approaches had been proposed, such as Fourier transforms and lattice rectification procedures.^5–7 The measurement of the angle between two crystallographic directions in the reconstructed data is also proposed to offer the detector efficiency.⁸ The DE obtained by these methods on a 3D-AP system ranges from 37% to 80%, which is mainly limited by the open area ratio of the MCP (60%–90%).^4,9

In the analysis of the AP sample that contains multiple elements, the chemical composition is influenced by detection efficiency, which depends on the bonding nature and the evaporation field of each element. When the multi-hit event is the cause of the compositional error, it is reported that making the detection efficiency uniform by restricting the transparency of ions within a flight path gives better compositional results.⁴ 

Using FIM image spots as a basis for counting atoms, it is favored that the image contrast is homogeneous and the field evaporation sequence is uniform; therefore, a single element metal, tungsten, was chosen in this work.

To synchronize FIM observation and AP mass analysis without changing the setting of sample orientation, several points must be solved, (1) suppression of imaging gas (typically, He) ions reaching to ion detector, (2) viewing “the blind area” corresponding to the probe hole in the FIM image, and (3) dynamic control of ion detection and FIM image acquisition.

Home-built FIM-Poschönrieder-AP reported by Nishikawa et al.¹⁰ elsewhere was modified for this work. Toroidal electrostatic lens proposed by Poschönrieder,^11,12 which selects ion energy, is placed in a ToF mass spectrometer. This ion optic eliminates the incident image gas ions (He⁺) from the FIM chamber to the AP detector. The flight path of this ToF system is 2500 mm measured along the zero-field line.

ToF measurement systems using voltage pulse for triggering field evaporation are renewed to synchronize with the FIM image capture function. The schematic is shown in Fig. 1. The field evaporation is triggered by applying a voltage pulse of 22% of the total applied voltage, and the flight time of ions is recorded by a time-to-digital converter (Texas Instruments TDC7200 chip). The TDC can accept up to five hit signals in one trigger (multi-stop). The dead time of the TDC coupled with the timing discriminator (Phillips Scientific, #715) is 30 ns. The TDC box stops triggering and generates a hit signal when a field evaporated ion is detected. The deflector shifts the He⁺ ions trajectory during the CCD camera (FLIR) capturing FIM image. When the capture has finished and the deflection has been released, the TDC box sends triggers to the HV pulser again until the next field evaporation ion gets detected.

The pulse rate between the desorption and detection is around 200 shots/s. In a typical condition, the exposure time for image capture is shorter than 500 ms including the set-release time of the deflector.

An electropolished tungsten tip was analyzed in this work. The tip was cooled down below 50 K in the FIM chamber and image gas (helium) flowed at 2 × 10⁻³ Pa. The gas pressure was monitored at the FIM section of the chamber, and the pressure at the toroidal lens section was almost the same as that of the FIM section. To reduce the dose amount of He in the chamber, a gas dose tube was attached to the variable leak valve mounted on the chamber wall of the FIM section. The tube's opening was set within 10 mm of the specimen. Without the dose tube, a pressure of 4 × 10⁻³ Pa was required to obtain the same spot brightness of the FIM image keeping exposure time the same (∼400 ms).

Figure 2 shows the He⁺ ion trajectory displaced by a pair of deflector plates. A pair of deflector plates are placed between the specimen and the phosphor screen assembly in the FIM chamber. Two deflector plates are set to spread so as not to hinder the screen viewed from the tip. The simulation by simion7.0 (Ref. 13) shows that an ion flux of Ep = 5 keV is shifted laterally about 8 mm by the deflector biased at ±500 V when the screen is placed at a distance of 10 cm from the specimen. The displacement is almost the same as the probe hole size.

The deflector is driven by a high voltage operational amplifier (APEX PA97) and the logic signal from the ToF measurement system of AP. The logic signal from the ToF system also triggered the FIM image acquisition by the CCD camera. FIM images were captured by a monochrome CCD camera (FLIR BFS-U3-23S3), and the image files were stored on a PC. The image data were processed (contrast enhancement and subtraction of two sequential images) by using the software imagemagick 7.1.0-53.

Although the modeling of the deflector function is based on numerical calculation, the trajectory varies to the kinetic energy of imaging gas ion He⁺ and voltage applied on the deflector. The probed region in AP analysis, which is masked by the probe hole on the screen, must be assigned in the practical condition determined by the tip, which has a different apex radius. Figure 3(a) shows the FIM image around the probe hole without shifting. The gray circle corresponds to the probe hole. Figure 3(b) shows the shifted FIM image using a deflector. The dashed circle shows the probe hole visible in the image. The probe hole appears at the same position in sequentially captured two images (a) and (b) because all the components are not changed except the application of deflector voltage. The atomic spots in FIM images (a) and (b) are consistent, and the composite of two images of (a) and (b) [Fig. 3(c)] shows the displacement of the image by the deflector. The encircled region in a solid line in Fig. 3(d) is assigned as the analyzed region in AP analysis. The region of a relatively rough, stepped region is selected to see the individual atoms.

The longer the tip-screen distance is set, the wider the shift width of He⁺ gets. In Fig. 3, the shift width is ca. 2/3 of the probe hole diameter here. Although the two circles are not fully separated but overlapped, to keep the acceptance angle of the ToF spectrometer large enough, the configuration in Fig. 3 was maintained in the following experiment.

The change in the FIM image of tungsten was captured during AP analysis. The FIM image was shifted to observe the probe hole region in AP analysis. The sample configuration was adjusted to probe the stepped region where the packing density of atoms was lower than that of the close-packed terrace. Figure 4 shows the sequential change of the FIM image recorded with a superposing pulse voltage of 28% of DC bias voltage (22% of total voltage) at the same region shown in Fig. 3.

The left row series is the raw deflected images that were captured just after the detection of desorbed ions. The region in the white circle in each image corresponds to the probe hole region. Ten tungsten species (nine tungsten atomic ions and one cluster ion) are detected during the sequence from top to bottom. The cluster ion was assigned to WO₄²⁺ by its m/Z value (124). The species is imaged as a bright single spot in the probed region of the FIM image captured before desorption. In this case, the number of spot changes after the evaporation event is 3. One helium ion is also detected, but it should be neglected here.

The subtracted image of two continuous images is shown in the right row. The bright spots that appear in each subtracted image correspond to the desorbed atoms from the former image by pulse desorption in AP. The desorbed ions must go through the probe hole, which is depicted as a white circle to be detected by the ToF system. The spots that appear outside the circle also correspond to desorption, but they are not blocked by the screen. The number of spots that appear in the probed region is marked on the right side of the images. The sum 33 is the number of desorbed atoms from the probed region. Here, we can evaluate the detection efficiency of AP analysis shown in Fig. 4 to be 10/33 ≈ 0.30.

The detection efficiency of the pulse counting system used in the ToF mass spectrometer is mainly restricted by that of the detector device, which converts the ion impact to an electric signal. Due to the opening ratio (≤60%) of the common microchannel plate (MCP) used in this experiment, more than 40% of incident ions are missed and not detected. Other experimental factors of signal loss might be ≈0.50 if we assume the signal conversion efficiency of the MCP to be 0.60.

To get rid of the MCP limitation, the replacement of the MCP with a conventional channel electron multiplier was proposed by utilizing the point focusing capability of the Poschönrieder-lens by Sakurai.¹⁴ As the channel electron multiplier has a single opening of ca. 5 mm to accept the incident ion, the detection efficiency is not restricted when the spot size of the focused ion is smaller than its opening. Although the detection is not restricted by the MCP in its setup, the acceptance angle of ion optics set by probe hole must be kept small (typically <1°) due to the aberration of the toroidal sector lens. Therefore, the synchronization of FIM observation and AP analysis remains difficult keeping a wide field of view/analysis in both FIM and AP.

As the number of desorbed ions is evaluated by the spots that appeared in the subtracted sequential images, the uniformity of the spots is expected for this method. In this experiment, the stepped region is selected to demonstrate spot counting because it is expected to be difficult to count individual atoms inside a densely packed terrace in a low-index plane. In addition to the counting problem, the disappearance of the flat terrace is accompanied by the evaporation of multiple atoms causing multi-hit in a single trigger. Therefore, the improvement of the FIM image manipulation procedure is required for this method for application to the other region of the sample.

On the reproducibility and the accuracy of detection efficiency measurement in this experiment, the overlap of the probed region shifted by the deflector to the probe hole should be discussed. It is clear in Figs. 3 and 4 that almost 1/4 of the probed region is invisible due to the insufficient shift width by the deflector. If there are some invisible spots hidden in Fig. 4 and they are considered, the true detection efficiency must be smaller than the value (0.30) estimated here. Except for the signal count loss (50% is lost), the ambiguity of atom counting in FIM image should be eliminated by the boost of deflector voltage and the improvement of image processing.

Several mechanisms of ion loss are caused by instrumentation and experimental conditions, such as deviation of machine setup from an optimal condition of ion-optics and signal processing, the collision loss of flying ion with He gas in the mass spectrometer, and the counting loss due to multi-hit of ions on the detector.

The signal yield may be improved by changing the threshold of the signal processing system lower although the noise count might get higher. In the above result, one He⁺ was found among W³⁺ and WO₄⁺² ions. These “regular” ions were pulse ionized at the timing of application of the trigger pulse. The He⁺ ions, which were continuously field ionized by DC bias voltage and had 22% lower kinetic energy, were blocked by the toroidal-sector lens. Although the misfit of the signal processing must be the decisive factor of the signal loss and it should be optimized for better signal yield, the random noise mainly caused by He⁺ between the triggers was perfectly suppressed during the ToF window of the analysis.

The collision of flying ions in the drift path of the ToF spectrometer with introduced helium gas should cause count loss of sample ions. The pressure of helium in this study was 2 × 10⁻³ Pa measured at the chamber wall of the FIM compartment. At this pressure, the mean free path of He–He is 7 m at 300 K and that of He⁺ ion with neutral He gas at 300 K is 10 m when the kinetic energy of He⁺ ion is 7 keV. These path lengths are assumed to be larger than the dimension of the system used in this work (drift path length = 2.5 m), but the factor of safety for avoiding collision is not large enough; therefore, the reduction of image gas flow should be important in future work.

The third factor, the decrease in the ion count due to the multi-hit of the same ions in the detector, is discussed by De Geuser et al.¹⁵ and Meisenkothen et al.¹⁶ De Geuser reports that the multi-hit event shows some spatial correlation on the specimen surface. The simultaneous evaporation of closely related sites like the probed region in this experiment is a potential mechanism for signal loss. On the other hand, the influence of multi-hit events on a composition analysis of a sample that consists of multiple elements is discussed by Thuvander et al.⁴ and Meisenkothen et al.¹⁶ The multi-hit event of species causes a signal discount due to the dead time of TDC. A material that consists of multiple species should require a careful approach in the evaluation of atoms desorbed from the surface. Because the sample used in this study is pure tungsten that consists of only one element, the miscounting of the minor element by the multi-hit event is not the most serious factor of signal loss in this experiment.

The recent development of AP has been focused on the acquisition of a three-dimensional data set and the improvement of throughput. The former point was realized by the utilization of position-sensitive detectors such as delay-line (3D-AP), and the latter was achieved by using a short pulsed laser of high repetition rate for triggering pulsed evaporation.

Although the throughput of laser-assisted 3D-AP is far better than that of conventional FIM-AP, FIM-AP has the advantage of spatial resolution (FIM) and mass/resolution (AP with energy focusing facility). Introducing the low-pressure imaging gas in the order of 10⁻⁸ Pa or utilization of the residual gas in an AP chamber, Chen et al.¹⁷ and Katnagallu et al.¹⁸ showed that the FIM information obtained directly using a two-dimensional ion detector in 3D-AP is useful for morphological investigation during AP analysis. The discrimination of field ionized gas ion and field evaporated sample ion has been also examined, and the merit of FIM has been reevaluated in comparison with 3D-AP. In both approaches, the partial pressure for image information must be much lower than 10⁻³ Pa in order not to interfere with normal AP ions. The reduction in the gas pressure degrades image quality (image brightness and increase in the statistical error); therefore, the accumulation time is required.

The experimental setup to obtain the FIM image of FOA presented in this study can apply to the common one-dimensional FIM-AP system equipped with an energy filter between the FIM partition and the ToF mass spectrometer. Contrary to this, in the modern 3D-AP system with a 2D detector that is directly facing the sample with no obstacle, the acquisition of an FIM image is not practical because the introduction of image gas sufficient for FIM image capture raises the background noise level due to the image gas ions asynchronously ionized to the ToF trigger.

FIM image acquisition of probed region at every step of the ion detection of AP analysis was enabled by the ion deflector in FIM driven by the AP system. AP's detection efficiency was evaluated by the comparison between the ion count and the structural change observed by FIM at each ion detection step. The obtained detection ratio was 0.30. Taking the efficiency of the MCP conversion rate (<60%) into consideration, this value suggests that the half of emitted ions were lost in the drift region or rejected in the signal processing. This method provides the experimental value of detection efficiency, which can be the basis of one-dimensional FIM-AP system tuning for AP analysis.

This work was supported by JSPS KAKENHI Grant-in-Aid for Scientific Research (C) (No. 19K0518800). The authors are grateful also to Owari of The University of Tokyo for his kind encouragement and advice on the construction ion deflector and drive circuit.

The authors have no conflicts to disclose.

Masahiro Taniguchi: Funding acquisition (lead); Project administration (lead); Writing – original draft (lead). Yasuo Yamauchi: Data curation (equal); Resources (equal). Kenji Yoshikawa: Data curation (equal); Resources (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.