Highly conductive vertically aligned molybdenum nanowalls and their field emission property

We report that vertically aligned molybdenum (Mo) nanowalls can grow on various substrates by simple thermal vapor deposition. Individual nanowalls have a typical thickness of about 50 nm and very good conductivity with a typical average value of about 1.97 × 104 Ω−1 cm−1, i.e., only an order of magnitude less than the value of bulk Mo. The formation process is characterized in detail, and it is found that Mo nanowalls grow from nanorods through nanotrees. The atomic arrangement, lattice mismatch relationship, and competition growth are all believed to contribute to the growth mechanism. The field emission performance is attractive, typically with a very low fluctuation of about approximately 1.18% at a high current density level of 10 mA/cm2, and a sustainably stable very large current density of approximately 57.5 mA/cm2 was recorded. These indicate that the Mo nanowall is a potential candidate as a cold cathode for application in vacuum electron devices, which demand both a high current and high current density.


Background
Field emission cold cathode has many important applications [1]. There are typically two areas which have attracted the attention of researchers, i.e., the field emission flat panel display (FED) and the vacuum electron devices, such as X-ray sources [2,3] and microwave sources [4,5]. FED requires large-area cold cathode arrays, which are addressable and thus need low-field emitters. The vacuum electron devices generally require both a high current and high current density. Now, it has been shown that nanostructured materials can give low-field cold electron emission [1]. However, one has come across a number of problems in pursuing a highcurrent and high-current density cold cathode. The major problem is that there are high-resistance regions in the individual nanostructure using nanowire as an emitter, which can lead to vacuum breakdown when it undergoes a high-current density operation [6]. Even more, so far, the conductivity of many nanotubes and nanowires is not high enough, except that for carbon nanotubes. Carbon nanotubes, however, are easily oxidized during the vacuum sealing process and thus lose its good properties.
We have been trying two other approaches aimed at overcoming the above difficulties. One is to grow nanowires or nanowalls with a high electrical conductivity [7][8][9][10][11][12][13][14][15][16]. In the recent years, one has seen a large number of studies focused on the preparation of one-dimensional molybdenum oxide nanostructures, such as MoO 2 nanorods [7], pinaster-like MoO 2 nanoarrays [8], and MoO 2 nanostars [9], with a few cases on twodimensional (2D) molybdenum oxide nanostructures such as MoO 3 nanobelts [10]. Our group has been developing a thermal vapor deposition method for synthesizing vertically well-aligned molybdenum oxide nanowires [11][12][13][14] and for a large-area preparation of 2D MoO 3 microbelts [15,16]. In this paper, we report that uniform Mo nanowalls can be grown on various substrates by simple thermal vapor deposition. The morphology and crystalline structure of the nanowalls are characterized by scanning electron microscopy, X-ray diffraction, and transmission electron microscopy techniques. The growth mechanism is discussed. The field emission performance and electrical transport property of the nanowalls are shown to be promising in meeting the requirements of high current and high current density of some vacuum electron devices.

Methods
The basic thermal vapor deposition method we used in the present study has been described in our previous report [14]. The method is developed based on the chemical reactions observed in the formation of Mo nanowires, which is similar to the formation of tungsten (W) [17]: Mo atoms of the Mo boat react with residual oxygen in the vacuum chamber spontaneously to give rise to MoO 2 (Equation 1). Then, Mo atoms coming from the decomposition of MoO 2 deposit during a condensation process (Equation 2). Here, we should note that when the substrate temperature is over 1,428 K, the condensation process of MoO 2 will be difficult. So, the key to synthesize the molybdenum nanostructures rather than oxide ones is the high-enough reaction temperature. Silicon, stainless steel, or silicon carbide was used as substrates in this study. They were first washed with acetone and then with alcohol, respectively, for more than 15 min in an ultrasonic bath. The substrate and Mo boat were then placed in the center of vacuum chamber, keeping them about 5 mm away from each other. The Mo boat as evaporation source was heated by electric current. The vacuum chamber was evacuated using a mechanical pump. When the vacuum reached below approximately 6 × 10 −2 Torr, high-purity argon (Ar) gas (99.99%) was introduced into the system at a flow rate of approximately 200 sccm, accompanied by high-purity hydrogen (H 2 ) gas (99.99%) at a flow rate of approximately 80 sccm. The boat temperature was then increased gradually at a rate of 50 K/min to above 1,623 K, in order to ensure the growth of the metallic molybdenum nanowall rather than the oxide one. After deposition for more than 10 min, the preparation of molybdenum nanowalls was completed. No catalysts were used in the whole process of growth.
It is significant that to obtain molybdenum nanowalls, the substrates should be placed below the Mo boat with a certain distance away, instead of being directly placed inside the boat, as we did when growing Mo nanowires [14]. In the present study, the growth orientation and the driving force of nucleation can be controlled by adjusting the spacing between the Mo boat and substrates.
Field emission scanning electron microscopy (XL-SFEG SEM, FEI Co., Hillsboro, OR, USA) was employed to investigate the morphology of molybdenum nanowalls. Transmission electron microscopy (Tecnai-20 TEM, Philips, FEI Co., Hillsboro, OR, USA), and X-ray diffraction (XRD; Rigaku RINT 2400, Rigaku Corporation, Tokyo, Japan) were applied to study their crystalline structures. The field emission properties of molybdenum nanowall films were measured in the field emission analysis system. In addition, the electrical transport measurement of the individual nanowalls was carried out using a micro-point anode in situ in a modified SEM system (JEOL-6380, JEOL Ltd., Akishima, Tokyo, Japan); the details of which are described in our previous work [6]. Figure 1a shows the top-view SEM image of molybdenum nanowalls grown on the silicon substrate with an artificially scratched area on the bottom left corner. The molybdenum nanowalls are seen to be well distributed all over the substrate, which helps to ensure the uniformity of field emission. The density of the nanowalls is about 3 × 10 8 /cm 2 . Moreover, we can see from the scratched area ( Figure 1b) that there is a layer of nanorods existing between the molybdenum nanowalls and underneath the substrate; here, we call it mid-layer. Figure 1c shows that, typically, the width of the nanowall ranges from 0.5 to 1 μm, with their thickness being about 50 nm. Also, Figure 1d reveals that some of the   Figure 2e shows the corresponding three-dimensional (3D) geometrical model of the individual Mo nanowall, as one may intuitively see the structure of such material. Figure 3a gives a low-magnification TEM image of the individual molybdenum nanowall. One can see that the surface of the nanowall is covered with some nanoprotrusions and folded regions and that the thickness is not uniform. We selected five areas from the nanowall to have electron diffraction analysis. Firstly, the area B in the top end of the nanowall is of a single-crystalline structure with little defect, since the selected area diffraction (SAED) patterns are sharp and ordered ( Figure 3b). A high-resolution TEM (HRTEM) image of area B shown in Figure 3g reveals that in this area, the nanowall is crystalline, with the mean lattice spacing  being 0.22 nm between two adjacent crystalline planes, which is consistent with the {110} planes of the bodycentered cubic (bcc) Mo. However, not the whole top end is of a single-crystalline structure. The area C is of a quasi-single-crystalline structure, as one may see from its SAED patterns (Figure 3c). Figure 3h shows the HRTEM image of area C, where the arrowheads point out the existing nanoinclusions. The SAED patterns (Figure 3d,e) of the folded regions of the middle part (areas D and E) show that they are polycrystalline with textured structures, which contain a lot of dislocations and stacking faults. Finally, the SAED pattern ( Figure 3f ) of area F (the bottom of the nanowall) shows ring patterns of a typical polycrystalline structure. Therefore, the molybdenum nanowalls have a polycrystalline part in the bottom and gradually grow into a crystalline structure at the top end. In addition, the energy-dispersive X-ray spectroscopy (EDX) spectrum of the individual molybdenum nanowall is displayed in Figure 3i. There are Mo, Cu, O, and Cr elements having the atom percentage of about 71.24%, 22.75%, 4.80%, and 1.22%, respectively. Here, the element Cu and Cr may come from the TEM copper grid and the background signal of the lens cone. One may notice that the element O is present in small amounts in the molybdenum nanowall (only 4.80 atom percentage), which probably originates from the amorphous layer on the surface of the individual nanowall (top left in Figure 3g and top right in Figure 3h). Because only very little oxygen exists, we believe that the prepared nanowalls are molybdenum nanowalls rather than molybdenum oxide ones. However, they may have a surface layer of amorphous molybdenum oxide.

Results and discussion
Further, Figure 4 shows the typical XRD patterns of the samples of the following growth times: 1 min, 2 min, 3 min, 4 min, 6 min, and 10 min, respectively. At the initial stage of growth, there are two peaks indexed to MoO 2 and Mo 8 O 23 , respectively. With the increase of growth time, new peaks indexed to Mo appear. For 3 min, the peaks of both molybdenum and molybdenum oxides appear. When the growth time is more than 10 min, all the peaks can be indexed to metallic Mo with of a pure bcc structure with a cell constant of a = b = c = 3.15 Å, while the original two peaks indexed to molybdenum oxides disappear. This sequential XRD spectra indicate that from the substrate to the Mo nanowalls, there are nanostructures of Mo oxide in the mid-layer. Now, we shall explain how Mo nanowalls are formed. Our group discussed previously how the molybdenum  nanowires were formed [14], and this is the base of our present discussion. To explore the growth mechanism of molybdenum nanowalls, we first need to know the growth process. Figure 5 shows the side-view SEM images of the samples of the following growth times: 3 min, 4 min, 6 min, and 8 min, respectively. One may see that the formation of molybdenum nanowalls may be divided into four stages: uniform and closely aligned nanorods (Figure 5a), nanotrees (Figure 5b and 5c), nanoflakes (Figure 5d), and nanowalls shown in Figures 1 and 2. This is the result of the combined effects of atomic arrangement [18], lattice mismatch relationship [19], and competition growth mechanism [14], as explained below.
We first discuss why in the first growth stage, only nanorods grow. From the HRTEM image (Figure 6c) of the rectangular enclosed area in Figure 6a, the fringe spacings of 0.240 nm, 0.241 nm, and 0.244 nm may be found, corresponding to the lattice spacings of the (200), (002), and (202) planes of the monoclinic-structured (Tugarinovite) MoO 2 , respectively. The axial growth plane is (202), and the growth direction is [202]. The SAED patterns of the nanorod were shown in Figure 6d. Both the patterns of monoclinic MoO 2 and bcc Mo were observed, indicating that the quasi-one-dimensional nanorods are a mixture of Mo and MoO 2 . Some groups [8,9,11,20] have reported that MoO 2 can act as a seed for the growth of nanorods. The related chemical mechanism has also been reported [11]; some unstable MoO x solid phase that formed may react with the residual oxygen in vapor and then become a stable MoO 2 crystal.
It is important to know that a set of the SAED patterns of Figure 6d can be indexed to the [010] zone axis of the monoclinic-structured MoO 2 crystal. With reference to Figure 7, that means that the normal direction of the wide surface (facing us) of the nanorod corresponds to [010], and the surface plane is (010). As indicated in Figure 7, the (010) plane corresponds to the close-packing plane, and it is not for fast growth. Thus, initially, the [202] direction grows fast to form nanorods.
In the second stage of the growth process, the nanorod grows into a nanotree. Typical TEM images and the corresponding SAED patterns are shown in Figure 8. From With reference to Figure 7, theoretically, there are two narrow side surfaces existing on both sides of the (202) axial growth plane, and they correspond to potential  ½ MoO z have a small lattice mismatch relationship. The natural misfit m between the two lattices is [19]:  Figure 8a. Here, it should be pointed out that in the bcc structure of Mo, <211 > shows six geometrically equivalent directions (Figure 8d), and in theory, nanowires may grow along all these directions as Zhou et al. [14] suggested. However, we did not observe these in the present work, and this is attributed to the limitation due to the spatial confinement [21].
In the third stage, Mo nanowalls start to form from a flake shape. Figure 9 shows TEM images of the samples in different moments of the growth process from the nanotrees to the nanoflakes. When the growth of main branches along the [121] and [112] directions is terminated by the adjacent nanotrees, the energies of growth will be redistributed into the following three parts: While the trunks continue to grow, more and more main branches (primary dendrites) can grow out from the sides of the trunk. Furthermore, as emphasized by the red arrowheads, new mini-branches can grow out from the existing branches. As these growth processes continue, two-dimensional flake structures will finally form. The competition growth process [14] plays an important role in this stage as the growth of some branches terminates due to limitation of growth space. However, how the competition growth process here leads to the only two-dimensional growth needs further detailed investigation. The illustration given in Figure 10 may help one to understand the whole growth process.
The field emission properties of molybdenum nanowall films were investigated. Here, the molybdenum nanowall films with an area of 0.02 cm 2 on the stainless steel substrate were studied in an ultra-high vacuum (approximately 2 × 10 −7 Torr) system by using a transparent anode technique with vacuum gaps of 100, 200, and 350 μm, respectively. After a few cycles of applying an electric field in the gap, the field emission reached a repeatable and stable stage, and the current  and voltage data were recorded. Figure 11a shows the curves of field emission current density versus applied field, obtained with the different vacuum gap spacings, revealing that the turn-on field (E to ) and threshold field (E thr ), which are defined as the values of the applied electrical field to induce field emission current densities of 10 μA/cm 2 and 10 mA/cm 2 , respectively, decrease when the vacuum gap increases from 100 to 350 μm (Table 1). This indicates the significant role the geometrical field enhancement played by the Mo nanowalls. A larger vacuum gap leading to a lower turn-on field was also observed for molybdenum and its oxide nanostructures by others [10,20,22]. The turn-on field of our samples is higher than the best data from other metallic Mo one-dimensional nanostructures [11,14,22] and is due to their smaller aspect ratio (approximately 70). However, this is not hidden in their application in high-current, high-voltage vacuum devices, as may be seen below. Their typical emission current density can reach 25 mA/cm 2 , and the largest current density of 57.5 mA/cm 2 was recorded from the sample grown on the silicon carbide substrate before vacuum breakdown. Figure 11b shows the corresponding F-N plots. One can see that these curves may be seen to compose of some discrete regions. One may artificially define the regions so that the regions are nearly straight lines. However, they have different slopes.
The values of field emission (FE) enhancement factor β of the molybdenum nanowall films were calculated using the F-N formulation [23]: where Ф is the work function of the molybdenum nanowall, which adopts the work function (4.3 eV) of bulk molybdenum materials. S FN is the slope of the F-N plot. According to Equation 4, we obtained the values of FE enhancement factor β for the sample with different vacuum gaps. As shown in Figure 11b, the β values in low-field emission regions are calculated, and they are 181, 395, and 817 for different d values of 100, 200, and 350 μm, respectively. It is clear that as vacuum gap increases, the β value increases. On the other hand, in each curve, the absolute value of the slope at the high-field region is lower than that at the low-field region; thus, the corresponding value of β at the high-field region is higher than that at the low-field region. In order to understand what causes the non-linearity in the F-N plots, we consider the following: First, the whole emission process may be affected by the complexity of the material property of the nanowall films. We have known that the molybdenum nanowall films consist of a nonmetallic part. The individual nanowall is covered with a very thin layer of amorphous molybdenum oxide, and there is a layer of nanorods containing molybdenum oxides existing between the nanowalls and the substrates. Both of them may have an effect on the conductivity of the material. When the applied field is very high and, thus, the emission current is very high, the FE enhancement factor β would be suppressed because the increase of the voltage drop in the nanowall film itself can result in a decrease of the local field on the surface of emitters. However, the calculated value of β at the high-field region is higher, so the above reason does not suit our case.
In fact, we will show that the mid-layer and the molybdenum nanowalls both have very good conductivity, while the oxide layer is too thin to have such effect. The higher value of β at the high-field region may be explained by both the increasing effect of the thermionic emission and the increasing number of  emission sites. Firstly, thermionic emission will gradually increase its effect in the high fields when the current density becomes high. The temperature of the emitters will increase due to the Joule heating.
More and more hot electrons will take part in the emission, and thus, the emission will be enhanced. Secondly, the number of emission sites will increase in the high fields. We suggest that the sharp edges of the nanowalls serve as main electron emitters in the low fields, although it is not yet clear without observing the emission pattern of an individual nanowall. At higher fields, however, the nanoprotrusions adhere to the nanowall bodies, and/or the nanorods may be expected to contribute to the total emission current. The increasing emission sites with different values of β may enhance the emission, thus leading to the non-linearity in the F-N plots. The detailed explanation awaits to be given after more theoretical and experimental studies. Figure 11c shows the field emission stability curve, revealing that the molybdenum nanowall films have a relatively stable field emission at a current density of 10 mA/cm 2 with a fluctuation of about approximately 1.18% in an hour of testing time. The distribution of field electron emission sites was also recorded (bottom right inset in Figure 11c), showing that over 70.01% of the area has bright spots and that the distribution of emission sites is relatively homogeneous, with a value of 66.7%.
The electrical transport properties of the nanowall films were also investigated. Figure 12 gives the SEM images showing the tungsten (W) microprobe in touch with the sample surface for electrical conductivity measurements and the conduction I-V characteristics. First, the W microprobe contacted the top surfaces of the nanowalls and touched only the top ends of three nanowalls in the measurement (Figure 12a). Sweeping voltage was applied. From Figure 12b, the resistance for forward applied voltage (R forward 1 ) and the resistance for reverse applied voltage (R reverse 1 ) are 27.26 and 27.74 Ω, respectively, which were calculated by fitting the I-V plots. Second, the W microprobe contacted the surface of the mid-layer (Figure 12c). Similarly, the resistance for forward applied voltage (R forward 2 ) and the resistance for reverse applied voltage (R reverse 2 ) were obtained by fitting the I-V plots from Figure 12d. The corresponding values are 15.43 and 15.74 Ω, respectively. Considering the fact that the measured three nanowalls A, B, and C are side-by-side and have nearly the same size, we assume that the resistances R A , R B , and R c are parallel connected and approximately equal. The resistance of a typical nanowall may be estimated through the following: R forward ¼ 3 Â R forward 1 À R forward 2 ð Þ ¼ 35:49Ω; and R reverse ¼ 3 Â R reverse 1 À R reverse 2 ð Þ ¼ 36Ω , according to the schematic circuit diagram shown in Figure 12e. Here, the resistance of the mid-layer and the contact resistance between the mid-layer and the substrate have been eliminated. However, the contact resistance between the W microprobe and nanowall cannot be completely eliminated since the contact in each measurement is different. Because the resistance for forward  applied voltage is nearly equal to the resistance for reverse applied voltage, we believe that such difference is small. Furthermore, the shape of the I-V curves is almost linear, suggesting that there were good Ohmic contacts existing between the nanowalls and the mid-layer, between the mid-layer and the substrate, and between the W microprobe and the contacting surfaces.  Accordingly, the conductivity of the single molybdenum nanowall may be estimated using the equation below: where R is the resistance, while h, w, and t are the height, width and thickness of a single nanowall, respectively. Taking the average values of a height of 3.5 μm, width of 1 μm, thickness of 50 nm, and R equal to 35.49 and 36 Ω, we can obtain the conductivity of a single molybdenum nanowall: σ forward ¼ 1:97 Â 10 4 Ω À1 cm À1 ; and σ reverse ¼ 1:94 Â 10 4 Ω À1 cm À1 . Table 3 lists the calculated conductivities of molybdenum nanowall samples grown on different substrates, showing that these values, ranging from 0.88 × 10 4 Ω −1 cm −1 to 5.23 × 10 4 Ω −1 cm −1 , are usually about one to four orders of magnitude higher than those of other reported nanostructures [6,26,28,[31][32][33][34]. We should like to note that most of the conductivity values of the Mo nanowall on different substrates are about an order of magnitude less than the value of the molybdenum bulk material (approximately 19.46 × 10 4 Ω −1 cm −1 ) [35]. Such high conductivities explain why molybdenum nanowalls can sustain a stable high field emission current density. This outstanding electrical transport property of the molybdenum nanowall is mainly due to its metallic intrinsic property and could be related to the 2D wall-like structure. Beyond doubt, Mo nanowalls with their excellent electric conductivity is favorable to the high-current and high-current density application in vacuum electron devices.

Conclusions
Vertically aligned molybdenum nanowalls can be prepared on varied substrates by thermal vapor deposition. Their formation undergoes through a four-stage process associated with the production of molybdenum and its dioxide nanorods, molybdenum nanotrees, molybdenum nanoflakes, and molybdenum nanowalls. Atomic arrangement, lattice mismatch relationship, and competition growth all together contribute to the growth mechanism. The nanowalls can sustain a very much larger field emission current density and better emission stability as compared to most of other nanostructures. Moreover, the single nanowall is shown to have a remarkably high electrical conductivity. The results suggest that molybdenum nanowalls should have promising futures in the application of vacuum electron devices that demand both high current and high current density.