Compact Sub Micro-resolution X-ray Microscope Based on Carbon Nanotube FE-SEM∗

Generally, nano-resolution X-ray imaging has been achieved using synchrotron radiation facility. The spatial resolution (below 100 nm) achieved with a Fresnel zone plate for focusing the X-ray in SPring-8 lately. We have developed a compact composite of field emission-scanning electron microscope (FE-SEM) with a single isolated multi-walled carbon nanotube (CNT) electron source in our previous study. The spatial resolution of SEM image was estimated as ≃ 10 nm. This result indicates that the electron source we developed is suitable for a pointlike X-ray source. We developed two kinds of X-ray microscope; projection-type X-ray microscope (PXM) and transmission-type X-ray microscope (TXM) in this study. The X-ray microscope based on the CNT-FE-SEM was constructed with a metal target for the X-ray source. Developing our X-ray microscope has been miniaturized to a compact size so that it can be placed on a desk. Moreover, the spatial resolution of 400 nm achieved the theoretical resolution limit. Because there is resolution limit by the acceleration voltage of 17 kV for the focused electron beam spot, the spatial resolution is an order of the electron penetration depth. The present study shows a potential that the CNT-based X-ray microscope would be applied to various field studies including the scene of an archaeological site excavation and future planet search. [DOI: 10.1380/ejssnt.2018.84]


I. INTRODUCTION
Carbon nanotube (CNT) possesses various benefits as a field electron emitter [1]. Multi-walled CNT (MWNT) exhibits high brightness and sustains stable field emission (FE) even under poor ultrahigh-vacuum condition 10 −6 -10 −7 Pa without ion pumping [2]. Furthermore, CNT field emitters possess excellent properties such as a small virtual source size of 2-3 nm and high brightness of 10 9 -10 10 A/cm 2 sr [3,4]. These properties are unique advantages of CNT electron source in comparison to conventional electron sources. However, the application of CNT emitter has not been in practical use yet in 2017. In the field of electron source, thermionic electron sources have been used for electron microscopes. In some of high-resolution electron microscopes, FE is used. Tungsten (W) field emitters are commonly used. Yabushita et al. tried to develop a compact scanning electron microscope (SEM) equipped with a MWNT bundle FE electron source and a ballast resistance [5]. They demonstrated the performance of MWNT bundle FE-SEM and showed SEM images taken under the poor vacuum condition at room temperature. Next, Yabushita et al. developed a transmission-type X-ray microscope with a MWNT bundle FE electron source [6]. An X-ray generated from a target in vacuum chamber and is ejected into air through a beryllium window. A sample is set in air and is observed by their transmission-type X-ray microscope. They estimated the resolution of SEM images as 50 nm and ob- * This paper was presented at the 11th International Symposium on Atomic Level Characterizations for New Materials and Devices '17, Aqua Kauai Beach Resort, Kauai, Hawaii, USA, December 3-8, 2017. † Corresponding author: irita.masaru@d.mbox.nagoya-u.ac.jp served X-ray images with resolution higher than 700 nm. CNT field emitter has been intensively studied in application to microscopes.
In our previous study, we had experimentally manufactured a compact composite of FE-SEM and X-ray microscope with a single isolated MWNT FE electron source [7]. The CNT emitter operated stably without any ballast resistance. Insertion of a ballast resistance is not desirable, because the electron beam spot becomes broad. The spatial resolution of SEM image was estimated as ≃ 10 nm [7]. This result indicates that the electron source we developed is suitable for a point-like X-ray source.
Generally, nano-resolution transmission-type X-ray imaging has been achieved using synchrotron radiation facility [8,9]. The spatial resolution was estimated to be 70 nm from a line profile at an edge of a gate pattern using 3D nano-ESCA in SPring-8 [9]. The 3D nano-ESCA is constructed with a Fresnel zone plate for focusing the X-ray. Synchrotron radiation facility has an advantage of wavelength tunable light source. In this study, we tried to develop a portable compact nano-resolution of X-ray microscope.

II. EXPERIMENTAL
We considered that developing two kinds of X-ray microscope: transmission-type X-ray microscope (TXM) and projection-type X-ray microscope (PXM). The X-ray microscope based on the CNT-FE-SEM was constructed with a metal target for the X-ray source. Our compact X-ray microscope has been miniaturized to a compact size so that it can be placed on a desk. The MWNT electron source was installed in the gun chamber, which was evacuated with a turbo-molecular pump down to 10 −7 Pa. The  sample chamber is equipped with a secondary electron detector and X-ray detector (XRI-UNO, XRAY IMATEK Co.). The detector is based on a silicon sensor being effective for X-ray photons with energy in a range between 4 and 25 keV. A more detailed structure was described in previous report [7]. For measuring X-ray micrograph, the target is placed at the position where the e-beam is focused and a sample is set between X-ray source and the X-ray detector. The distance from X-ray source to sample is defined as a and the distance from X-ray source to X-ray detector is b for explanation. Therefore, the magnification is (b/a).
First, we explain about PXM. Figure 1 shows a schematic diagram of the PXM apparatus used in this study. The electron beam axis is fixed by the PXM design and thus, the distance b is a constant value of 135 mm. The distance a could be directory observed by SEM as shown in Fig. 1(b). Therefore, the magnification is determined precisely. PXM is one of X-ray imaging methods, the principle is similar to shadow play. The SEM and PXM images of sample are different in an angle of observation.
Next, we explain about TXM. Figure 2 shows a schematic diagram of the TXM apparatus. A SiN film (50 nm thick) was used as a base of target for X-ray source. The base was deposited with an Au film (10 or 50 nm thick) by electron beam physical vapor deposition. An X-ray generate from controlled thickness of Au metal film. Because the X-ray generated along the electron beam axis as shown in Fig. 2(a), the distances a and b could not be directory observed in TXM method. Furthermore, the distances a and b are dependent on the electron beam focusing point. Therefore, it is necessary to observe a specific sample for the calibration of magnification. The SEM and TXM images of sample could be observed in a same angle. TXM method is preferably used in previous research. Our TXM is characterized as setting X-ray target and sample in the vacuum chamber.

III. RESULTS AND DISCUSSION
We observed PXM images as shown in Figs. 1(c) and 3(a). Figures 1(c) and 3(a) are same PXM image. After PXM observation, the sample of the W needle was observed by conventional SEM, and the shape was investigated. Observed PXM and conventional SEM images of the W needle are same magnification of ≃ x300 as shown in Fig. 3. The PXM image of the W needle apex [ Fig. 3(a)] is transparent, as revealed by comparison with the corresponding SEM image [ Fig. 3(b)]. X-ray is transmitted through the thin part of the needle around less than 10 µm thick. Thus, it is necessary a thickness of more than 10 µm of the sample for X-ray observation. ≃ 250 nm from the W needle apex. A best PXM image is shown in Fig. 4(b). The sample was an Au 400-mesh and the magnification was ≃ x790. The magnification of Fig. 4(b) is higher than Fig. 3(a). The thickness of Au 400-mesh is ≃ 10 µm. From the PXM image, it can be seen that the Au 400-mesh becomes thinner toward the edge. Figure 4(c) shows a line profile across the edge of a Au bar. The intensity profile in Fig. 4(c) was fitted using a step function convoluted with a Gaussian function. The spatial resolution was estimated to be about 80 nm from the width at 80-20% height of a fitting profile at the edge. Here, we should not select good resolution because the edge is not sufficiently sharp. The average of spatial resolution was estimated to be ≃ 400 nm from the width at 80-20% height of each fitting profiles at all edge of Fig. 4(b). The spatial resolution varies widely because of signal-noise.
The X-ray target mode of Zr and Au are different in a critical excitation voltage. The critical excitation voltage is V c ≃ 15.7 kV (K α1,2 ) for Zr and a wavelength of X-ray is λ ≃ 0.08 nm. On other hand, the critical excitation voltage is V c ≃ 9.7 kV (L α1 ) for Au and a wavelength of X-ray is λ ≃ 0.13 nm. In this study, the acceleration voltage of 16 and 17 kV were used for the critical excitation voltage, respectively. The electron penetration depth in the energy range 5 to 15 keV is expressed as R z = kV 1.5 acc (µm), where k is a parameter dependent on material [10]. The X-ray source size is expressed as δ X = 0.033(V 1.7 is critical excitation voltage, A is atomic weight, Z is atomic number and ρ (g/cm 3 ) is density of material [11]. This is called Castaing's equation. The electron penetration depth and X-ray source size depended on the density of material. The calculated quantities for each material are listed in Table I. We found δ X ≃ 35 nm for Zr was obviously small as compared with δ X ≃ 270 nm for Au at V acc = 16 kV. We investigated the possibility of using Zr as X-ray target. Figure 5 shows EDX spectrum of the X-ray target mode of Zr and Au. The EDX spectrum of Au-L α1 was detected, although the EDX spectrum of Zr-K α1,2 was not detected. The X-ray almost did not generate from the X-ray target made of Zr. Thus, the Zr is not suitable for the X-ray source. Though the PXM FIG. 5. EDX spectrum of the X-ray target mode of Zr and Au. The X-ray target was the same as shown in Fig. 1(b). The EDX spectrum was observed by conventional SEM equipment. The X-ray target mode of Zr and Au were observed at the same time. The acceleration voltages was 17 kV and a collecting time was ≃ 4 min. The red arrows show the characteristic X-ray peak.
image was observed using the X-ray target made of Zr as shown in Figs. 1(c) and 3(a), the focussed beam current I foc ≃ 1×10 −8 A was higher than other measurements. In this case, the electron source suffered damage and the operation was unstable. The best PXM image was observed using the X-ray target made of Au with I foc ≃ 7×10 −10 A as shown in Fig. 4(b).
Let us compare the spatial resolution with theoretical predictions. The theoretical spatial resolution is expressed as δ = √ δ 2 e + δ 2 F + δ 2 X , where δ e is an electron beam spot size and δ F is Smearing due to Fresnel diffraction. In previous study, the electron beam spot size δ e  was estimated to be ≃ 10 nm [7]. Smearing due to Fresnel diffraction is expressed as δ F ≃ √ aλ, where λ is wavelength of X-ray. In the case of Fig. 4, the distance a ≃ 0.17 mm was determined, then δ F ≃ 147 nm. The theoretical spatial resolution is δ ≃ 360 nm for the Au target at V acc = 17 kV. Therefore, the theoretical result agrees well the experimental result.
We assumed that the X-ray source size might be controlled by a thickness of metal film as a target in the TXM. For better resolution below 100 nm, we tried to develop the TXM. We observed TXM images of sample with the Au film thickness of 10 and 50 nm as shown in Fig. 6. The magnification was calibrated with the specific sample. In the case of an Au film thickness of 10 nm, a bright spot appears at the center as shown in Fig. 6(a). We could not find the sample because the electrons are almost transmitted through the Au film. With an Au film thickness of 50 nm, W needle sample was observed, although the image was affected by transmitted electron as shown in Fig. 6(b). Observed all TXM images showed a bright spot at the center. We assumed that the electron beam penetrated through the thin Au film because of R z ≃ 320 nm for Au at V acc = 17 kV. The X-ray source size could be controlled by the deposition, although the transmitted electron could not be restricted by TXM method. To solve this problem, the X-ray detector should be installed tilting from the electron beam axis. As one of solutions of this problem, an X-ray generated from the target in vacuum chamber and is ejected into air in previous research [6].

IV. CONCLUSION
The present study has demonstrated the best performance of PXM housed in the CNT based compact FE-SEM. Because there is resolution limit by the acceleration voltage of 17 kV for the focused electron beam spot, the spatial resolution is an order of the electron penetration depth. The spatial resolution of 400 nm achieved the theoretical resolution limit. In the TXM, the X-ray source size may be controlled by the deposition, although there are still problems regarding the transmitted electron. Developing our X-ray microscope is portable and compact. Moreover, the spatial resolution is about synchrotron radiation facility. The technique of this research would be beneficial for the application in the near future.