Superior Carbon Nanotube Field Emission for X‐Ray Source via Metal Silicide‐Induced Adhesion

Abstract

Carbon nanotube field emitters (CNT FEs) have obtained increasing attention for vacuum electronics devices such as cold cathode X‐ray tubes. However, the low adhesion of CNTs on the substrate thwarts their ability to achieve high field emission current density. To this end, the effects of Ni, Si, and Al₂O₃ fillers on their adhesion both in the paste and onto the Kovar (nickel‐cobalt ferrous alloy) substrate are investigated. Chemical reactions between Ni, Si, and the Kovar constituents lead to the formation of micrometer‐sized protruding particles. Si fillers are key in promoting their formation in the paste and on the substrate. During high‐temperature vacuum annealing, the Si fillers reacted with the Ni fillers and the Kovar constituents, forming Ni₂Si in the paste and Fe₂NiSi on the substrate, both of which strengthened adhesion. The adhesion of CNT FEs with both Ni and Si fillers is better compared to those containing Ni or Si fillers alone. With the resulting retention of more CNTs on the substrate after tape activation, a current density of 30.9 A cm⁻² and stable field emission for 14 h at 500 mA cm⁻² are achieved, indicating the commercial potential of CNT FEs in vacuum electronics.

The adhesion of CNT field emitters (FEs) is enhanced using Ni and Si fillers, which formed strong bonding phases (Ni₂Si and Fe₂NiSi) in the paste and Kovar substrate during high‐temperature annealing. CNT FEs containing both fillers exhibit superior adhesion and outperform those with a single filler, achieving a high emission current density (30.9 A cm^{− 2}) and stable performance, highlighting their commercial potential in X‐ray tube applications.

1. Introduction

Carbon nanotubes (CNTs) have emerged as prime candidates for field emitter materials owing to their unique properties, including a sharp tip geometry, large aspect ratio, excellent mechanical strength, and a high melting point.^{[ 1 ]} The suitability of CNTs as electron field emitters (FEs) in applications such as X‐ray tubes,^{[ 2 ]} high microwave power amplifiers,^{[ 3 ]} traveling wave tubes,^{[ 4 ]} ionization gauges,^{[ 5 ]} micro‐radiotherapy,^{[ 6 ]} microcomputed tomography,^{[ 7 ]} and ultraviolet‐light‐source tube^{[ 8 ]} has been demonstrated. For the aforementioned applications, however, CNT FEs must produce a high current density and exhibit long‐term emission stability. Meeting these requirements has been challenging because CNT FEs may be damaged during operation at high current densities, due to factors such as severe Joule heating,^{[ 9 ]} structural degradation such as CNT shortening^{[ 10 ]} and damage to nanotube walls,^{[ 11 ]} field evaporation,^{[ 12 ]} ion bombardment by residual gases,^{[ 13 ]} and physical detachment from a substrate by field‐induced stresses.^{[ 14 ]} The detachment of CNT FEs from the substrate due to poor adhesion can result in catastrophic arcing, leading to a sudden drop in emission current and a significant reduction in the lifetime of the CNT FEs.^{[ 15 ]} For example, CNTs rapidly detach from the cathode substrate at electric fields >7 V µm⁻¹.^{[ 16 ]} Consequently, the poor adhesion of CNTs to the substrate is the main obstacle preventing the long‐term use of CNTs in field emission devices working at high current densities or in strong electric fields.

CNT FEs have been fabricated using several different techniques, including electrophoretic deposition,^{[ 17 ]} chemical vapor deposition (CVD),^{[ 18 ]} mechanical rubbing,^{[ 19 ]} dip‐coating,^{[ 20 ]} spray coating,^{[ 21 ]} and screen‐printing.^{[ 22 ]} Of these, screen printing has advantages such as low cost, simplicity, easy control of emitter shape and size, scalability, as well as environmental friendliness^{[ 8 ]} and thus is widely used in the fabrication of CNT FEs. In this process, the CNT FEs initially buried in the paste but subsequently need to be exposed to the surface, typically by surface activation using either adhesive tape^{[ 23 ]} or a roller.^{[ 24 ]} These activation processes play a significant role in improving emission uniformity due to the reduction of shielding effects.^{[ 25 ]} The uniform formation of CNT FEs on the surface depends on the adhesion of the paste ingredients to each other and the substrate. When substrate adhesion is weak, CNT FEs will be removed by the activation process, leaving only a few that are vertically aligned on the surface. Moreover, CNT FEs loosely bonded with the paste ingredients may detach under high electrostatic stresses during field emission,^{[ 26 ]} potentially leading to catastrophic electric arcing. Conversely, if the adhesion is too strong, it may hinder exposing the buried CNT FEs over the surface by the activation process. Therefore, the excellent field emission performance of the CNT FEs requires optimizing adhesion between the paste ingredients and the substrate.

The adhesion of CNT FEs to a substrate is crucial for achieving a high current density in field emission.^{[ 27 ]} The fillers in a CNT paste play a significant role in controlling adhesion, as they affect both the performance and the durability of CNT FEs. Numerous studies have sought to enhance the performance of CNT FEs through the use of metallic and ceramic fillers. For example, Lei et al.^{[ 28 ]} improved adhesion and performance of CNT FEs by incorporating bismuth (Bi) fillers into the CNT paste and sintering them at 400 °C. This resulted in the formation of Bi blocks from the melted Bi fillers, which created strong bonds between the CNTs and the Bi blocks. Similarly, Qin et al.^{[ 29 ]} developed a CNT paste with silver (Ag) nanoparticles printed on a silicon substrate and sintered at 250 °C. The FEs adhered strongly to the substrate due to the melting of the Ag nanoparticles. Both Bi and Ag nanofillers improved adhesion by melting at low temperatures, which enhanced field emission performance. However, low‐melting‐point fillers such as Ag and Bi may evaporate during high‐temperature vacuum annealing in the brazing process used to seal the X‐ray tubes, thus degrading field emission performance. To address this issue, Kim et al.^{[ 30 ]} fabricated a high‐temperature‐resistant paste using nickel‐copper alloy and alumina (Al₂O₃) fillers. The resulting FEs exhibited high evaporation resistance during thermal annealing at 950 °C, and a substantial number of vertically aligned CNTs were observed after surface tape activation (TA), indicating good adhesion. In another study, Kim et al.^{[ 31 ]} optimized the adhesion of CNT FEs to achieve both a high emission current and uniformity, by adjusting the mixing ratios of the µm TiO₂ and nano SnO₂ fillers in the CNT paste. However, while these fillers serve as heat‐resistant bonding agents for FEs, their high melting point prevents strong adhesion. Thus, despite the achievements obtained using metallic and ceramic fillers, a better understanding of the role played by the filler is needed to identify the materials that can enhance adhesion and optimize CNT FE performance.

This study investigated the impact of different fillers on the adhesion of CNT FEs, both within the paste and on a Kovar substrate, thereby enhancing field emission performance. In our ternary filler system, consisting of nickel (Ni), silicon (Si), and aluminum oxide (Al₂O₃), where Si chemically reacted with the Ni in the paste and with the constituents of the Kovar substrate forming micrometer‐sized particle agglomerates consisting of Ni₂Si and Fe₂NiSi phases. The Si‐based filler played a key role in the formation of protruding particles on the paste and substrate, resulting in the reinforcing the adhesion of the CNT FEs. The CNT FEs strongly adhered to the substrate and demonstrated an excellent field emission performance, characterized by both a high current density and long‐term stability.

2. Experimental Section

High‐purity Ni (40 nm), Si (50 nm), Al₂O₃ (<50 nm), and multi‐walled CNTs (XNM‐HP) were purchased from Sigma–Aldrich, Avention, and Xinnano Inc., respectively, and used as received. Scanning electron microscopy (SEM, Hitachi, SU‐8010) images of the Ni, Si, Al₂O₃, CNTs, and Kovar (nickel‐cobalt ferrous alloy) images of the substrates used in fabricating the CNT FEs are shown in Figure S1 (Supporting Information). The CNTs used in this study have an average diameter of 4 nm, an average length greater than 10 µm, and an average I_D/I_G ratio of 0.26, as provided by the vendor (Figure S1h, Supporting Information). High‐resolution transmission electron microscopy (HR‐TEM) and Raman spectroscopy of CNTs are shown in Figures S1f,g (Supporting Information), respectively. Ethyl cellulose (EC) and 2,2,4‐trimethyl‐1,3‐pentanediol monoisobutyrate (Texanol) were purchased from Sigma–Aldrich and used as a binder and solvent, respectively.

Three types of CNT pastes were prepared by mixing various fillers with CNTs in a paste mixer. CNT Paste 1 was composed of nickel (Ni) and alumina (Al₂O₃), CNT Paste 2 incorporated silicon (Si) and Al₂O₃, and CNT Paste 3 contained Ni, Si, and Al₂O₃. To achieve a stable and homogeneous dispersion, a combined mechanical and noncovalent functionalization strategy was employed. The optimized formulation (CNT paste‐3) contained Ni, Si, and Al₂O₃ in a volume ratio of 1:5:1. The CNT content remained consistent across the three paste types. Initially, the fillers were dispersed in Texanol with 3 mm zirconia beads for 90 min. CNTs (0.66 g) were then added and mixed for an additional 20 min to form a well‐homogenized CNT‐filler composite. Ethyl cellulose (EC) binder was subsequently added and mixed for 5 min; EC noncovalently adsorbed onto the CNT and filler surfaces, providing steric stabilization and increasing viscosity to suppress aggregation. Finally, the paste was further homogenized using a three‐roll mill, producing a uniformly dispersed printable CNT paste, which was screen‐printed onto a Kovar substrate through a dot‐patterned mask. The CNT FEs consisted of 25 dots on the substrate, with each dot measuring 250 µm in diameter. They were fabricated by firing the screen‐printed CNT pastes at 400 °C in the air to remove the EC binder, followed by annealing at 865 °C under a high vacuum (10⁻⁷ torr). The topmost layer of the CNT paste was peeled off using adhesive tape to expose the underlying CNTs, which served as CNT FEs.

The field emission properties of CNT FEs were measured using a diode configuration in a high vacuum (10⁻⁷ torr). A high‐voltage power source (Fug HCN 1400–12500) was used to supply DC voltage to the anode, with the current between the cathode and anode recorded using a digital multimeter (Keysight 34461A). The current was measurable up to 100 mA, the maximum limit of our DC power supply. A tungsten plate served as the anode electrode, with a 500 µm gap between the cathode and anode. Before the measurement of the current (I) versus voltage (V) characteristics, the CNT FEs were electrically aged at stepwise currents of 0.6, 3, 6, and 12 mA, to remove weakly bonded CNTs and vertically align them to achieve a uniform height.^{[ 8 ]} Limit current tests of the CNT‐FEs were conducted up to 100 mA, the maximum current limit of the DC power supply. The lifetime of the CNT‐FEs was tested at a current of ≈6 mA, corresponding to a current density (J) of 500 mA cm⁻², for 14 h in a diode configuration. In addition, the long‐term emission stability of single dot CNT FEs was measured at a constant current of 0.26 mA, corresponding to 500 mA cm⁻² in a triode configuration. Because high currents generate heat that could damage both the gate and the CNT FEs, a lifetime test of the single‐dot CNT FE in the triode configuration was conducted at a current of 0.26 mA to mitigate heating issues.

The surface morphology of the CNT FEs after firing and vacuum annealing was examined via SEM (Hitachi, SU‐8010). High‐resolution transmission electron microscopy (HR‐TEM, Jeol, JEM‐ARM200F) was employed to investigate the structure and composition of the CNT FEs after high‐temperature vacuum annealing. Cross‐sectional TEM samples were prepared using a focused ion beam (FIB), and the elemental composition in the cross‐sectional samples was analyzed via energy dispersive spectroscopy (EDS). The crystalline structure was characterized using X‐ray diffraction (XRD, PANalytical/Empyrean/PC) with a Cu‐Kα1 radiation (λ = 1.5406 Å) source and a scan speed of 0.5° min⁻¹. X‐ray photoelectron spectroscopy (XPS, Theta Probe; Thermo Fisher Scientific) with monochromatic Al Kα radiation was used to measure the chemical composition of the samples.

3. Results and Discussion

3.1. Morphological and Structure Characterization

The surface morphologies of the three different CNT FEs with various filler combinations were analyzed via SEM (Figure 1 ). CNT FE‐1 and CNT FE‐2 consisted of binary fillers of Ni‐Al₂O₃ and Si‐Al₂O₃, respectively, and CNT FE‐3 had a ternary filler of Ni‐Si‐Al₂O₃. All CNT FEs were prepared using screen printing followed by firing, vacuum annealing, and a first TA. The topmost layer of the CNT paste was peeled off by the first TA, which exposed the buried CNTs over the surface, allowing them to function as FEs. The interface between the Kovar substrate and CNT paste was observed by completely removing the CNT paste from the substrate surface during seven, eight, and nine rounds of TA for CNT FE‐1, CNT FE‐2, and CNT FE‐3, respectively (Figure S10, Supporting Information). Figure 1a,b shows top‐ and tilted‐view SEM images of CNT FE‐1, respectively, after vacuum annealing and the first TA. The particle agglomerates of the Ni and Al₂O₃ fillers on the surface of CNT FE‐1 were formed by physical bonding with randomly oriented CNTs. After repeated TAs, the CNT paste was eliminated from the substrate, leaving a sub‐micrometer‐ to micrometer‐sized paste debris that flatly adhered to the Kovar substrate surface (Figure 1c,d). For CNT FE‐2, with Si and Al₂O₃ fillers, a similar surface morphology was observed after the first TA (Figure 1e,f). However, unlike CNT FE‐1, in CNT FE‐2, micrometer‐sized particles of different morphology formed that protruded from the Kovar substrate surface (Figure 1g,h), most likely due to the chemical reaction of the Si filler with the metal constituents of the Kovar substrate during vacuum annealing at 865 °C, as few such particles were observed after firing at 400 °C (Figure S2d, Supporting Information). The formation of a larger number of protruding particles on the substrate surface, evidenced by the presence of numerous pits (see arrow), was probably the result of their detachment after repeated TA (Figure 1g). These particles likely enhanced the adhesion of the CNT paste to the substrate by mechanical anchoring. Figure 1i,j shows the SEM images of CNT FE‐3, with Ni, Si, and Al₂O₃ fillers, after vacuum annealing and the first TA. The surface morphology of CNT FE‐3 was completely distinct from that of CNT FE‐2. White, micrometer‐sized protruding particles appeared on the CNT paste surface in CNT FE‐3 even after the first TA while they were absent in CNT FE‐2. The particles likely formed from the chemical reaction between the Ni and Si fillers during vacuum annealing at 865 °C, as they were not observed after the 400 °C firing (Figure S2e, Supporting Information). Thus, the Ni and Si fillers did not chemically react during firing. After removing the CNT paste by repeated TAs, the surface morphology of CNT FE‐3 on the Kovar substrate (Figure 1k,l) closely resembled that of CNT FE‐2 (Figure 1g,h) but with larger protruding particles. The SEM results suggested that the Si filler played a significant role in the formation of the protruding particles, not only in the paste but also on the substrate. When the Si filler chemically reacted with the Ni filler in the paste and with the constituents of the Kovar substrate, micrometer‐sized protruding particles formed in the paste and on the substrate, enhancing the adhesion of the CNT paste to the substrate and improving the electrical connection between CNTs and substrate (Figure S3, Supporting Information).

Figure 1.

Top‐view (the 1st and 3rd rows) and tilted view (the 2nd and 4th rows) SEM images showing the surface morphologies of carbon nanotube field emitters (CNT FEs) subject to an initial tape activation (TA) to remove the topmost layer (the 1st and 2nd rows) or subsequent TAs performed to fully detach the CNT paste from the substrate (the 3rd and 4th rows): a–d) Ni and Al₂O₃ fillers (CNT FE‐1), e–h) Si and Al₂O₃ fillers (CNT FE‐2), and i–l) Ni, Si, and Al₂O₃ fillers (CNT FE‐3). All CNT FEs underwent vacuum annealing at 865 °C before TA.

The protruding particles present in the paste and on the substrate were characterized using XRD. After vacuum annealing, the CNT paste was collected from the substrate in powder form for analysis. The XRD spectra of the raw materials used for the CNT paste and CNT FEs, including Ni, Si, Al₂O_3, and the CNTs, and of the Kovar substrate were also obtained for comparison (Figure S4, Supporting Information). Figure 2 presents the XRD spectra of CNT FE‐1, ‐2, and ‐3 before and after vacuum annealing at 865 °C. ZrO₂ peaks were consistently observed at 30.2° and 35.2°, attributed to ZrO₂ debris from the ball milling process used to prepare the CNT paste. The XRD pattern for CNT FE‐1, containing Ni and Al₂O₃ fillers, was unchanged before and after vacuum annealing, except for the appearance of nickel oxide (NiO) peaks after vacuum annealing (Figure 2a). The NiO was formed due to oxidation of the Ni filler in the paste during firing at 400 °C in air^{[ 32 ]} (Figure S5a,c, Supporting Information), performed to remove the organic binder in the CNT paste. The absence of other new peaks after vacuum annealing (Figure 2a) indicated that there was no chemical reaction between the Ni and Al₂O₃ fillers in the paste or between the fillers and constituents of the Kovar substrate during the annealing process. Figure 2b shows the XRD pattern for CNT FE‐2, containing Si and Al₂O₃ fillers. After vacuum annealing, new XRD peaks appeared at 45.4° and 83.8°. In the control Si paste without Al₂O₃ filler and CNTs, the new XRD peaks seen in Figure S6a (Supporting Information) were detected at the same positions as those in Figure 2b, confirming them as a Fe₂NiSi phase.^{[ 33 ]} This indicates that Si had diffused into the Kovar substrate in the CNT FE‐2 during vacuum annealing, forming iron‐rich silicide phases. Thereafter, the Ni component in the Kovar substrate may have diffused to the iron silicide, due to its high solubility in the binary iron silicide phase.^{[ 34 ]} The micrometer‐sized protruding particles observed on the substrate (Figure 1g,h) corresponded to a Fe₂NiSi phase, confirmed by the XRD results shown in Figure 2b and Figure S6a (Supporting Information). This phase had a highly ordered Heusler structure, with the Si component forming strong covalent bonds with other elements in the Fe₂NiSi phase to enhance the adhesion of the protruding particles to the Kovar substrate.^{[ 33a ]} In the XRD pattern of CNT FE‐3, containing Ni, Si, and Al₂O₃ fillers, the peaks that formed after vacuum annealing (Figure 2c) were more complex than those of CNT FE‐1 and CNT FE‐2, due to the formation of a Ni₂Si phase in addition to NiO and Fe₂NiSi phases. The new XRD peaks for the control sample (Figure S7, Supporting Information) a powder prepared by collecting the Ni and Si paste from the Kovar substrate after vacuum annealing were confirmed as a Ni₂Si phase (ICDD PDF 00‐048‐1334).^{[ 35 ]} The XRD results indicated that the Si filler chemically reacted with the Ni filler in the paste during vacuum annealing, leading to the formation of Ni₂Si‐phase, micrometer‐sized protruding particles in the paste (Figure 1i,j). The remaining Si in the paste reacted with the Kovar substrate, resulting in the formation of micrometer‐sized protruding Fe₂NiSi particles (Figure 1k,l). The combination of the Ni and Si fillers in CNT FE‐3 thus acted as a bonding agent that significantly enhanced adhesion both in the paste and for the CNT FEs, to the substrate via the formation of Ni₂Si and Fe₂NiSi phases.

Figure 2.

X‐ray diffraction data of a) CNT FE‐1 with Ni and Al₂O₃ fillers, b) CNT FE‐2 with Si and Al₂O₃ fillers, and c) CNT FE‐3 with Ni, Si, and Al₂O₃ fillers before (blue curves) and after vacuum annealing at 865 °C (red curves).

The composition and structure of the protruding particles formed in the paste and on the Kovar substrate were further examined by TEM (Figure 3 ). Cross‐sectional samples of CNT FE‐3 after vacuum annealing were prepared for TEM using the FIB technique. Figure 3a shows a cross‐sectional high‐angle annular dark‐field scanning transmission electron microscopy (HAADF‐STEM) image of the CNT paste, with particles of varying sizes distributed across the paste layer. EDS mapping (Figure 3b–h) revealed that these particles were mostly composed of nickel silicide. Fe, Co, and Ni components from the Kovar substrate were not detected, but carbon components were uniformly distributed around the particles and connected, forming an electrically conductive network (Figure 3f). The nickel silicide particles (Figure 3b,c) were analyzed using HR‐TEM and based on their selected area electron diffraction (SAED) patterns, as displayed in (Figure 3i,j). The latter revealed the crystalline nature of the Ni₂Si phase, corresponding to the (1 ®01 ®), (1 ®21 ®), and (020) planes along the [101 ®] zone axis. The lattice spacing of 0.195 nm, measured from the fast Fourier transform profile, was consistent with the (1 ®21 ®) plane of Ni₂Si. Further investigation of the particles formed on the Kovar substrate (Figure 3k) using EDS mapping (Figures 3l–r) clearly showed the Fe, Ni, and Si elements in the particles (Figure 3l–n) and their presence as an iron‐nickel silicide phase. In addition, the particle and substrate surface were coated with a thin layer of silicon oxide (Figure 3n,o), which may have resulted from the surface oxidation in the firing process but was not detectable by XRD. HR‐TEM imaging of the protruding particles (Figure 3s) and analysis of the SAED patterns (Figure 3t) showed that the latter were related to the (22 ®0), (12 ®0) and (110) planes along the [001] zone axis. The lattice spacing of the (22 ®0) plane was 0.204 nm, corresponding to the crystalline nature of a Fe₂NiSi phase. The d‐spacing values in the HR‐TEM images of Ni₂Si and Fe₂NiSi were obtained through the phase profile spectra (Figure S8, Supporting Information) and matched the XRD data (Tables S1 and S2, Supporting Information).

Figure 3.

Electron microscopic images of cross‐sectional (a–j) paste and (k–t) substrate for CNT FE‐3 after vacuum annealing. a) High‐angle annular dark‐field scanning transmission electron microscopy (HAADF‐STEM) image, b–h) energy dispersive X‐ray spectroscopy (EDS) chemical mapping, i) high‐resolution (HR) TEM image, and j) selected area electron diffraction (SAED) pattern of Ni₂Si phase in the cross‐sectional paste region. k) HAADF‐STEM image, i–r) EDS chemical mapping, s) HR‐TEM image, and t) SAED pattern of Fe₂NiSi phase in the cross‐sectional substrate region. The cross‐sectional TEM sample of CNT FE‐3 was prepared using the FIB technique after the first and repeated TA of the paste and substrate, respectively.

To investigate the surface chemical states of the protruding particles in the pastes and on the substrate surface after vacuum annealing, XPS analysis was conducted. Figure 4 shows the XPS spectra of CNT FEs after the first and repeated TA. Deconvolution of the XPS peaks for the Ni, Si, and Fe elements revealed the chemical bonds and interactions. Figure 4a–c shows the XPS peaks of the paste surface for CNT FE‐1, CNT FE‐2, and CNT FE‐3, respectively, after the first TA. In the Ni2p region (Figure 4a), CNT FE‐1 and CNT FE‐3 exhibited peaks at binding energies of 852.8, 854.6, 856.5, and 861.4 eV, assigned to Ni⁰, Ni²⁺, and Ni³⁺ oxidized states and satellite peaks, respectively, while in CNT FE‐2 there was no signal of the Ni element due to the absence of Ni filler. In CNT FE‐3, an additional peak was deconvoluted at a binding energy of 853.5 eV, with a positive chemical shift of 0.7 eV from Ni⁰. This can be attributed to the electron transfer from the Ni to the Si atoms resulting from the covalent interactions between Ni and Si due to Ni₂Si phase formation.^{[ 36 ]} This was similarly identified in the Si2p XPS spectra (Figure 4b). The XPS peak of CNT FE‐3 could be resolved into five deconvoluted peaks representing Ni‐Si bonds, Si2p_1/2, Si2p_3/2, Si³⁺, and Si⁴⁺ oxidized states at binding energies of 98.7, 99.2, 99.8, 102.6, and 103.5 eV, respectively.^{[ 37 ]} No Si‐related XPS peaks were observed for CNT FE‐1 due to the absence of Si filler. The deconvoluted peak at a binding energy of 98.7 eV, compared to those of Si and CNT FE‐2 (without Ni filler), corresponded to a Ni₂Si phase. The new peak appearance suggested a chemical interaction between Si and Ni elements that led to the formation of a Ni₂Si phase during vacuum annealing at 865 °C. There were no Fe and Co elements in the paste surface, evidenced by the absence of Fe2p and Co2p peaks for all CNT FEs (Figure 4c; Figure S9a, Supporting Information). This implied that neither Fe nor Co elements diffused from the Kovar substrate into the paste. No changes were observed in the Al2p peaks for all CNT FEs (Figure S9b, Supporting Information), indicating that the Al₂O₃ filler did not undergo a chemical reaction with the Ni and Si fillers in the paste during vacuum annealing. The C1s spectra of all CNT FEs (Figure S9c, Supporting Information) contained two deconvolution peaks at essentially the same binding energies, 284.6 and 286.1 eV, corresponding to sp²‐hybridized carbon and C─O, respectively.^{[ 38 ]}

Figure 4.

X‐ray photoelectron spectroscopy (XPS) profiles of a) Ni2p, b) Si2p, and c) Fe2p regions of CNT paste surface after the first TA, and d) Ni2p, e) Si2p, and f) Fe2p regions of substrate surface after repeated TAs for CNT FE‐1, CNT FE‐2, and CNT FE‐3.

The chemical states of the protruding particles on the surface of the Kovar substrate were examined by completely removing the CNT paste by repeated TA. Figure 4d–f shows the resulting XPS peaks of the substrate surface for CNT FE‐1, CNT FE‐2, and CNT FE‐3. In the Ni2p regions (Figure 4d), Ni⁰ and its oxidized XPS peaks were present in CNT FE‐1 and CNT FE‐3, both of which included Ni fillers. A new peak in the Ni2p regions, at a binding energy of 853.5 eV, was generated for CNT FE‐2, with Si filler, and CNT FE‐3, with Ni and Si fillers. This peak may have corresponded to a Fe₂NiSi phase, with Ni and Fe likely originating from the Kovar substrate, as indicated by CNT FE‐2, which lacked Ni fillers in the paste composition. Similarly, XPS peaks corresponding to Si⁰ and its oxidized states (Figure 4e) were observed in CNT FE‐2 and CNT FE‐3, both of which included Si filler. For these CNT FEs, a new peak in the Si2p regions at a binding energy of 100.0 eV and a positive chemical shift of 0.2 eV from elemental Si were observed, attributable to a Fe₂NiSi phase.^{[ 39 ]} To understand the role of the Fe component within the Kovar substrate, the Fe2p XPS peaks were analyzed (Figure 4f). The Fe2p region in the Kovar substrate exhibited XPS peaks representing Fe⁰, Fe²⁺ and Fe³⁺ oxidized states at binding energies of 706.4, 710, and 713.9 eV, respectively.^{[ 40 ]} The similar XPS peaks in CNT FE‐1 and the Kovar substrate suggested that the Ni filler did not react with the Fe elements in the substrate. However, a new peak at 707.1 eV was detected for both CNT FE‐2 and CNT FE‐3, with Si fillers, along with a positive chemical shift of 0.7 eV from elemental Fe, indicating a Fe₂NiSi phase.^{[ 39} , ^{41 ]} This peak appearance in the Fe2p region was a result of a change in the binding energy of the core element. On the other hand, in the Co2p spectra for all CNT FEs (Figure S9d, Supporting Information), consistent XPS peaks for Co° and Co²⁺, at binding energies of 777.6 and 780.6 eV, respectively, were detected.^{[ 42 ]} Thus, in contrast to the Fe and Ni elements, the Co element in the Kovar substrate did not chemically react with the Si fillers. Al2p spectra were not detected on the substrate for any of the CNT FEs (Figure S9e, Supporting Information), given that Al₂O₃ is chemically stable and Al₂O₃ filler, incapable of substrate adhesion, was removed completely from the substrate through repeated TA. The C1s spectra of all CNT FEs (Figure S9f, Supporting Information) contained three deconvolution peaks, at binding energies of 284.5, 285.8, 288.1 eV, corresponding to sp²‐hybridized carbon, C─O, and C═O, respectively.^{[ 38 ]} The position and shape of those peaks remained unchanged after vacuum annealing at 865 °C but underwent slight oxidation. The emergence of new peaks in the Fe2p, Ni2p, and Si2p regions of the XPS spectra confirmed that the Si filler in the paste played a crucial role in Fe₂NiSi phase formation, through a chemical reaction with the Fe and Ni elements in the Kovar substrate.

The chemical reactions of the Ni and Si fillers in the paste and with the Kovar substrate are schematically represented in Figure 5 . The CNT paste was screen printed onto a Kovar substrate, fired at 400 °C in air, and annealed at 865 °C under high vacuum. During high‐temperature vacuum annealing, the chemical reaction between Ni and Si nanoparticles in the paste and between Si nanoparticles and the components of the Kovar substrate resulted in the formation of micrometer‐sized protruding particles in the paste and on the substrate surface. These particles were strongly attached to the substrate and acted as an anchoring structure for the CNT FEs. The Ni and Si in the paste reacted to form a Ni₂Si phase, which improved the adhesion between the paste ingredients. In addition, Si elements in the paste diffused into the Kovar substrate to form an iron silicide phase while the Ni elements diffused into the iron silicide phase, forming a Fe₂NiSi phase. The Fe₂NiSi phase strongly attached to the Kovar substrate and thus improved the adhesion of the CNT FEs to the substrate. The use of Al₂O₃ as a chemically stable filler to adjust the adhesion in the paste facilitated the uniform formation of vertically aligned CNT FEs on the substrate through TA, thereby preventing excessively strong adhesion between the paste ingredients, which would have hindered TA and in turn, the formation of vertically aligned CNT FEs. Instead, the sufficiently strong adhesion layer of Ni₂Si in the paste and Fe₂NiSi on the substrate was expected to significantly enhance the performance of CNT FEs during field emission.

Figure 5.

Schematic of a chemical reaction between nickel (Ni) and silicon (Si) within a paste, and chemical reaction of Ni and Si with Kovar substrate in high‐temperature vacuum annealing.

3.2. Evaluation of Adhesion Strength for CNT Field Emitters

The effect of the fillers on the adhesion of CNT FEs was investigated by examining the morphological changes in the 25‐dot pattern of CNT FEs formed on the Kovar substrate as a function of the number of TAs (Figure S10, Supporting Information). As the number of TAs increased, the dot patterns of the CNT FEs became progressively blurred, indicating that the CNT FEs were being peeled off layer by layer with each TA. Complete removal of the dot patterns from the substrate occurred after the seventh, eighth, and ninth TAs for CNT FE‐1, CNT FE‐2, and CNT FE‐3, respectively. The more extensive peeling of CNT FE‐1 and CNT FE‐2 than of CNT FE‐3 suggested weaker adhesion between the fillers in the paste and less adhesion to the substrate. This has significant implications for the field emission performance because electrical‐field‐induced stress often leads to the detachment of CNT FEs from the substrate, resulting in electrical arcing and emitter failure.^{[ 26 ]} The curves obtained in plots of the field emission current versus the electric field for the three different types of CNT FEs after the first, third, and fifth TA are shown in Figure 6a. The current was measured up to the maximum emission current (I_max) achievable by the CNT FEs without electrical arcing. The I–E curves gradually flattened out as the number of TAs increased, but even after the fifth TA the field emission performance of CNT FE‐3 was still superior to that of either CNT FE‐1 or CNT FE‐2. The I_max values of CNT FE‐1 and ‐2 were 60.6 and 82.1 mA, respectively, whereas after the first TA the I_max value of CNT FE‐3 was higher, 99.2 mA (Figure 6c). Figure S11 (Supporting Information) shows the optimization of Ni, Si, and Al₂O₃ fillers for CNT FE‐3. The results indicate that reducing Al₂O₃ content significantly enhances the emission current in CNT FE‐3. This suggests that a high Al₂O₃ content in the paste negatively affects the field emission due to reduced electrical conductivity and weakened adhesion, attributed to its insulating and chemically inert nature. The optimal filler ratio of Ni:Si:Al₂O₃ that yielded the highest field emission current was 1:5:1. In this composition, the reactions between Ni and Si formed conductive phases, such as Ni₂Si and Fe₂NiSi, while the moderate Al₂O₃ content helped control adhesion, ensuring adequate CNT exposure on the substrate. Here, Al₂O₃ plays a beneficial role by fine‐tuning adhesion without significantly compromising conductivity. However, when the Al₂O₃ filler was completely eliminated, the field emission current drastically declined. The paste without Al₂O₃ filler showed excessive adhesion due to strong reactions between Ni/Si and the Kovar substrate, which hindered CNT exposure and degraded emission performance. In this case, strong adhesion adversely impacted field emission despite electrically conductive phases. These results indicate that the fluctuation in the emission current caused by varying Si/Al₂O₃ ratios is due to adhesion strength. We also investigated the electrical characteristics of each filler using literature data, which further supports our findings.^{[ 43 ]} The electrical properties of each filler are presented in Table S3 (Supporting Information). According to the literature, Ni, Ni₂Si, and Fe₂NiSi phases are electrically conductive, Si exhibits moderate conductivity, and Al₂O₃ is an insulator filler.

Figure 6.

a) Field emission current versus electric field curves, b) F‐N plots, c) maximum emission currents, and d) field enhancement factor (β) values for CNT FE‐1, CNT FE‐2, and CNT FE‐3 after the first, third, and fifth TAs.

Furthermore, CNT FE‐3 had higher I_max values than CNT FE‐1 or ‐2 even after the third and fifth TAs. The decrease rates of I_max for the CNT FEs after the third and fifth TAs relative to the first TA were calculated. For CNT FE‐1, CNT FE‐2, and CNT FE‐3, the rates after the third TA were −28.8%, −24.6%, and −13.3%, and after the fifth TA −72.2%, −59.4%, and −39.6%, respectively. The significantly lower decrease rate in the I_max values of CNT FE‐3 than of CNT FE‐1 or CNT FE‐2 underscores the close relationship between the I_max values and the black‐color intensity of the dot images after the third and fifth TAs (Figure S10, Supporting Information). It was, therefore, likely that the superior adhesion of CNT FE‐3 resulted in more CNTs being left on the substrate surface after the first, third, and fifth TAs (Figure S12, Supporting Information). Although the number of standing CNTs on the substrate decreased with increasing TAs, a higher number remained in CNT FE‐3 than in CNT FE‐1 and CNT FE‐2, which would explain its higher I_max values in the field emission. Figure 6b presents the Fowler–Nordheim (F–N) plots^{[ 44 ]} obtained using Equation (1) and the I–E curve in Figure 6a.

$$\ln \left(\frac{I}{V^2}\right)=\ln \left(A\alpha {\mathrm{\Phi }}^{-1}\frac{{\beta }^2}{d^2}\right)-\frac{B{\varnothing }^{\frac{3}{2}}d}{\beta }\left(\frac{1}{V}\right)$$ (1)

$$\beta =\frac{Bd{\varnothing }^{3/2}}{S}$$ (2)

The equation relates the emission current (I) to the applied voltage (V), which depends on several factors, including an effective emission area (α), a field enhancement factor (β), the work function of the emitter (Φ), cathode‐to‐anode distance (d), and slope (S). The constants A and B have values of 1.54 × 10⁻⁶ A eV V⁻² and 6.83 × 10⁷ eV^−3/2 V cm⁻¹, respectively. Assuming a work function of 4.61 eV for CNTs,^{[ 45 ]} the β values for CNT FEs in the low electric field region after the first, third, and fifth TAs were determined: 679, 362, and 90 for CNT FE‐1; 867, 517, and 232 for CNT FE‐2; and 1328, 994, and 582 for CNT FE‐3, respectively. Thus, the β values of CNT FE‐3 were consistently higher than those of CNT FE‐1 and CNT FE‐2. The β values in the F‐N plots decreased with repeating TAs relative to the first TA; for CNT FE‐1, CNT FE‐2, and CNT FE‐3 they were −53.6%, −45.2%, and −25.2% after the third TA and −86.7%, −73.2%, and −43.8% after the fifth TA, respectively (Figure 6d). The β value depends on the emitter geometry β $\approx \frac{\mathrm{l}}{\mathrm{r}}$,^{[ 22} , ^{46 ]} where l and r are the length and apex curvature radius of the CNTs, respectively. This relationship clearly shows that β is inversely proportional to the CNT radius (r), meaning that as the radius decreases, the β value increases. Conversely, β is directly proportional to the CNT length. According to this relationship, the observed decrease in the β value in our emitter is due to the elimination of long CNT FEs from the substrate during TA (Figures S10 and S12, Supporting Information). Since the same CNTs were used for all three types of emitters in our study, the effect of the apex curvature radius remains consistent. Accordingly, the difference in the β value between CNT FEs may be closely related to the number and length of CNTs in the emitting area. For CNT FE‐3, the larger number and longer length of the CNTs on the substrate led to β values higher than those of CNT FE‐1 and CNT FE‐2. As mentioned above, the superior field emission performance of CNT FE‐3 can be attributed to two factors. First, the chemical reaction between Ni and Si fillers in the paste enriched the Ni₂Si phase, increasing the adhesion between the CNTs and the fillers in the paste. Second, the chemical reaction of the Si filler with the Kovar substrate resulted in the formation of a Fe₂NiSi phase, facilitating good adhesion and electrical contact between the CNTs and the substrate. Hence, the strong adhesion of CNT FEs plays a critical role in enhancing the field emission performance. Adhesion strength was experimentally evaluated by quantifying the retention of CNTs on the emitter surface after repeated tape activation. To further support these findings, we calculated the stresses required to detach the CNT FEs from the substrate under a strong electrical field using Maxwell's stress equations:^{[ 27 ]}

$$\overrightarrow{{\sigma }_s}=\frac{1}{2}{\in }_{\mathrm{o}}\left(\mathrm{E}\mathrm{local}\right)2\vec{\mathrm{n}}$$ (3)

$$\overrightarrow{{\sigma }_s}=\frac{1}{2}{\in }_{\mathrm{o}}\left(\beta \frac{\mathrm{V}}{\mathrm{d}}\right)2\vec{\mathrm{n}}$$ (4)

Here, $\overrightarrow{{\sigma }_s}$ represents the stress on the conductor, ϵ_o is the permittivity of free space (8.854  ×  10⁻¹²  F/m), E_local is the local electrical field (β V/d), and $\vec{n}$ is the unit normal vector to the surface. The stresses required to detach CNT FE‐1, CNT FE‐2, and CNT FE‐3 from the substrate were 2.42 × 10⁴, 3.00 × 10⁴, and 5.51 × 10⁴ N cm⁻², respectively. The stresses required to remove CNT FE‐3 were ≈127.7% and 83.7% higher than those of CNT FE‐1 and CNT FE‐2, respectively. The calculated Maxwell stresses for each CNT field emitter are also listed in Table 1 . These results indicate that CNT FE‐3 required the highest detachment stress, demonstrating the strongest adhesion to the substrate.

Table 1.

Adhesion strength of CNT field emitters.

Maxwell stress ($\overrightarrow{{\sigma }_s}$) = ½ ϵ0E2 local $\vec{\mathrm{n}}$	CNT FE‐1 [N cm−2]	CNT FE‐2 [N cm−2]	CNT FE‐3 [N cm−2]
First TA	2.42 × 104	3.00 × 104	5.51 × 104
Third TA	8.35 × 103	1.30 × 104	3.70 × 104
Fifth TA	6.16 × 102	3.38 × 103	1.65 × 104

3.3. Field Emission Performance of CNT Field Emitters

Figure 7a depicts the I and J curves of CNT FEs with 25 dots as a function of E. For CNT FE‐1 and CNT FE‐2, I_max values 60.6 mA at 10.7 and 82.1 mA at 9.3 V µm⁻¹, corresponding to a maximum current density (J_max) of 4.93 and 6.69 A cm⁻², respectively, were determined. However, CNT FE‐1 and CNT FE‐2 failed to achieve higher current levels, due to electrical arcing. By contrast, CNT FE‐3 reached an I_max of 99.2 mA at 8.1 V µm⁻¹, corresponding to a J_max of 8.1 A cm⁻², without electrical arcing. The excellent field emission performance can be attributed to the number of CNT FEs remaining on the surface after TA. The stronger the adhesion of the fillers to the substrate, the larger the number of CNT FEs remaining on the surface (Figure S12, Supporting Information). F–N plots (Figure 7b), obtained using Equation (1), from the J–E curve in Figure 7a, showed a linear inverse relationship between the logarithmic value of I/V² and 1/V. The β values calculated at the low field region were 679, 867, and 1328 for CNT FE‐1, CNT FE‐2, and CNT FE‐3, respectively. The I_max achieved by the 25‐dot CNT FE‐3 was imprecise due to the current limit (100 mA) of our DC power supply. To address this issue, one‐dot CNT FEs with a diameter of 250 µm were fabricated and their field emission properties were evaluated (Figure 7c). The I_max values for one‐dot CNT FE‐1, CNT FE‐2, and CNT FE‐3 were 5.1 mA at 17.4 V µm⁻¹, 9.7 mA at 16.7 V µm⁻¹, and 15.2 mA at 14.5 V µm⁻¹, corresponding to J_max values of 10.3, 19.8, and 30.9 A cm⁻², respectively. The field emission performance of one‐dot CNT FE‐3 was ≈1.9 times higher than that of one‐dot CNT FE‐2 and 3 times higher than that of one‐dot CNT FE‐1. The better performance of CNT FE‐3 reflected its better adhesion in the paste and to the substrate. In addition, the field emission performance of CNT FE‐3 was 2.8 times higher than that of a one‐dot CNT emitter with a diameter of ≈50 µm, prepared using Ni and SiC fillers, which produced a current density of 11.2 A cm⁻².^{[ 47 ]} A detailed comparison of the field emission characteristics for all CNT field emitters is listed in Table 2 . Digital images of CNT FEs taken after measurements of the I_max values of the one‐dot CNT FEs are shown in the inset in Figure 7c. CNT FE‐1 was almost entirely detached from the substrate during field emission, due to poor adhesion. Detached CNTs may cause severe electrical arcing, leading to catastrophic failure at low current^{[ 48 ]} in the case of CNT FE‐1 but with less damage to CNT FE‐2, due to the enhanced adhesion resulting from silicide formation with the substrate, despite weak adhesion between the CNTs and the fillers in the paste. However, there was no damage to CNT FE‐3, as it benefited from strong adhesion both in the paste and to the substrate via the chemical reactions of the Ni and Si fillers, respectively. Figure 7d presents a comparison of the field emission properties of screen‐printed CNT FEs according to their filler materials and illustrates the variation in current density among the different CNT FEs. The superior current density of CNT FE‐3 demonstrates that the choice of the filler material can enhance the emission performance.

Figure 7.

Field emission current (I) and their current density (J) as a function of electric field (E) in diode configuration for a) CNT FE‐1, CNT FE‐2, and CNT FE‐3 with 25 dots and b) their F‐N plots, c) I and J as a function of E for CNT FE‐1, CNT FE‐2, and CNT FE‐3 with one dot. The inset shows the dot images taken after the limit current test. d) Comparison of the field emission J performance of our CNT FEs with other CNT FEs in the literature that are dependent on the different fillers.

Table 2.

Field emission characteristics of CNT FEs measured in a diode configuration. The table compares the emission performance of 25‐dot and 1‐dot CNT emitter arrays. Here, E is the applied electric field (V µm⁻¹), I is the emission current (mA), and J is the current density (A cm^{− 2}).

25‐dot CNT Field emitter
CNT FEs	E [V µm−1]	I [mA]	J [A cm−2]
CNT FE‐1	10.71	60.61	4.93
CNT FE‐2	9.3	82.1	6.69
CNT FE‐3	8.1	99.2	8.1
1‐dot CNT Field emitter
CNT FE‐1	17.4	5.1	10.3
CNT FE‐2	16.7	9.7	19.8
CNT FE‐3	14.5	15.2	30.9

The emission stability of 25‐dot CNT FEs was evaluated by recording the variation in the anode voltage with time at a constant current of 6.1 mA, equivalent to 500 mA cm⁻², in a diode configuration (Figure 8a). For CNT FE‐1 and CNT FE‐2, the anode voltage continuously increased, from 3.62 to 4.56 kV and 2.91 kV to 3.30 kV, respectively, during 14 h of field emission. CNT FE‐3, however, had a highly stable anode voltage, increasing from 2.31 to 2.37 kV during 14 h, indicating a consistent performance. At the 14 h mark, the anode voltage for CNT FE‐1, CNT FE‐2, and CNT FE‐3 increased by 25.96%, 13.40%, and 2.59%, respectively. CNT FE‐3 demonstrated superior stability, and even after 100 h of continuous testing, it maintained stable emission without any performance degradation, highlighting its outstanding FE characteristics (Figure S13, Supporting Information). The field emission performance of CNT FEs may deteriorate due to factors such as physical detachment from the substrate, CNT evaporation, and ion bombardment caused by field‐induced stress, joule heating, and the presence of residual gases.^{[ 12} , ^{13 ]} At high current density, the excessive heat generated at higher dot numbers of the CNT field emitter cathode detrimentally affects the long‐term emission performance. The CCD images in Figure S14 (Supporting Information) show the severe damage in the dot patterns of the CNT FEs. The triode measurement was thus conducted using a one‐dot CNT field emitter instead of the 25‐dot array. Lifetime measurements of one‐dot CNT FE‐3 in a triode configuration during 14 h at a constant current of 0.26 mA, corresponding to 500 mA cm⁻², and at a 6 kV anode voltage are shown in Figure 8b. The minimal leakage current of 5.1% was due to the almost exact alignment of the gate hole over the CNT field emitter. The initial voltage (V_G) was 2.53 kV, gradually increasing to a final voltage after 14 h of 2.85 kV, 12.64% higher than the starting voltage. The higher percent increase in the voltage in the triode configuration was primarily due to the complexity and control mechanisms introduced by the gate electrode. Table 3 presents a comprehensive summary of the emission stability for all CNT field emitters in diode and triode configurations. Figure 8c shows the ripple effect of CNT FE‐3 on the percent increase in the triode measurements. Here, “ripple” refers to the variation in or fluctuation of the current. It can be expressed as shown in Equation (5):

$$Ripplepercentage=\frac{I_{max}-I_{min}}{I_{avg}}\ast 100$$ (5)

Figure 8.

Emission stability of a) CNT FEs with 25 dots at 6.1 mA, equivalent to 500 mA cm⁻², for 14 h in diode configuration, b) CNT FE‐3 in triode configuration at constant anode current of 0.260 mA corresponding to 500 mA cm⁻², c) ripple percentage in the emission current indicating stability variation during operation, and d) comparison of the emission stability of CNT FEs at different current density.

Table 3.

Emission stability of CNT FEs in diode and triode configurations. The table shows the initial voltage required to maintain a 6.1 mA emission current and the final voltage after a 14 h stability test, with a percentage change in voltage over time indicating emitter degradation or stability.

Emission stability in diode configuration
CNT Field emitter	Initial voltage [kVi]	Final voltage [kVf]	Voltage increase (%) (kVf−kVi/kVi) × 100
CNT FE‐1	3.62	4.56	25.96
CNT FE‐2	2.91	3.30	13.40
CNT FE‐3	2.31	2.37	2.59
Emission stability in triode configuration
CNT FE‐3	2.53	2.85	12.64

Ripple can affect the stability and quality of the X‐ray beam. A high ripple percentage may lead to variation in X‐ray output that reduces image quality and consistency. Consistent electron emission, and thus a low ripple percentage, is crucial to generate stable, high‐quality X‐rays. A high ripple percentage indicates significant variation in the current, attributable to factors such as thermal effect variation in the electric field at the emitter tip and the structural integrity of the CNTs. In this study, CNT FE‐3 had a low ripple percentage (2.4%) in the triode measurements, due to the excellent contact between the CNTs and substrate achieved by improving the adhesion strength. A comparison of the emission stability of the paste‐based CNT FEs at the different current densities reported in the literature is shown in Figure 8d.^{[ 30} , ^{49 ]} Emission stability is a key parameter for evaluating the practical applicability of CNT FEs. Several critical factors influence emission stability, including vacuum level, the presence of reactive gases (especially oxygen), uniformity of CNT tips, CNT crystallinity, and the adhesion strength between CNTs and the substrate.^{[ 50 ]} A notable advancement in emission stability was reported by Shimoi et al., who demonstrated that highly crystalline single‐walled carbon nanotubes (SWCNTs) could maintain stable emission for 1300 h at a current density of 30 mA cm⁻².^{[ 51 ]} Their research highlighted that controlling the crystallinity of SWCNTs was key to achieving remarkable stability. In this study, we focus on the adhesion of CNTs to the substrate, which plays a pivotal role in maintaining emission stability. During field emission, strong electrostatic forces act on the CNT emitters, which can lead to their detachment from the substrate. Such detachment may lead to electrical arcing, ultimately resulting in catastrophic emitter failure. To address these challenges in CNT paste emitters, J. W. Kim et al. achieved enhanced emission stability for 130 h at a current density of 110 mA cm⁻² by improving CNT–substrate adhesion using Ni and SiC fillers.^{[ 49a ]} However, this issue becomes more severe at higher current densities due to increased Joule heating and electrostatic stress, which impose greater mechanical and thermal loads on the CNTs. Therefore, under such severe conditions, the selection of an appropriate filler system is required for the CNT paste emitter which significantly enhances the adhesion. Building on this strategy, we enhanced the adhesion of CNT FE‐3 by promoting the formation of Ni₂Si and Fe₂NiSi intermetallic phases in the paste and the Kovar substrate. As a result, CNT FE‐3 exhibited stable emission for 100 h at a significantly higher current density of 500 mA cm^{− 2} (Figure S13, Supporting Information). CNT FE‐3 had higher J values and more stable field emission than the CNT FEs in previous studies. In summary, the strong adhesion of CNT FEs not only with the fillers in the paste but also with the Kovar substrate is crucial to improve the emission current and ensure emission stability with a low ripple percentage.

4. Conclusion

The adhesion of CNT FEs is a crucial factor in efforts to increase their field emission performance. We enhanced the stability of CNT FEs by investigating the role of fillers, specifically Ni, Si, and Al₂O₃, in improving adhesion to the Kovar substrate. Chemical reactions between Ni, Si, and Kovar substrate components (Fe, Ni, and Co) led to the structural formation of micrometer‐sized protruding particles. Si filler played a significant role in the formation of these particles both in the paste and on the substrate. During vacuum annealing, the Si filler reacted chemically with the Ni filler in the paste, forming a Ni₂Si phase, and with the Kovar substrate, resulting in the formation of a Fe₂NiSi phase. Together, the Ni and Si fillers acted as strong bonding agents that significantly improved the adhesion of CNT FEs in the paste and to the substrate. The adhesion of CNT FE‐1 and CNT FE‐2, with only Ni or Si as filler, respectively, was weaker than that of CNT FE‐3, which incorporated both Ni and Si fillers. The superior adhesion of CNT FE‐3 was due to chemical reactions that enriched the Ni₂Si and Fe₂NiSi phases, resulting in more CNTs remaining on the substrate surface after TA. Consequently, CNT FE‐3 outperformed CNT FE‐2 and CNT FE‐1 in field emission performance, by ≈1.9‐ and 3‐fold, respectively. Overall, CNT FE‐3 had the highest current density (30.9 A cm⁻²), and field emission was stable for 14 h at a current density of 500 mA cm⁻². This development holds promise for commercializing CNT FEs in vacuum electronic applications.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Abbas S. Z., Lee J., Mehdi S. M. Z., Cho J., Raza A., Kim S., Goak J. C., Lee N., Superior Carbon Nanotube Field Emission for X‐Ray Source via Metal Silicide‐Induced Adhesion. Small 2025, 21, 2500242. 10.1002/smll.202500242

Data Availability Statement

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