Supplementary material for
Fullerene as electrical hinge Neng WAN1, 3, Pascal PERRIAT2, Li-Tao SUN1,*, Qing-An HUANG1, Jun SUN1, Tao XU1 1SEU-FEI Nano-Pico Center, Key Laboratory of MEMS of Ministry of Education, School of Electrical Science and Engineering, Southeast University, 210096 Nanjing, China
2Materials Engineering and Science, UMR 5510 CNRS, Université de Lyon, INSA-Lyon, 69621 Villeurbanne Cedex, France
3Laboratoire de Physico-Chimie des Matériaux Luminescents, Université Claude Bernard Lyon 1, UMR 5620 CNRS–UCBL, 69622 Villeurbanne Cedex, France
* Corresponding author: firstname.lastname@example.org
Description of the experimental setup
Figure s1. (a) Experimental setup. The driving mechanism was used to manipulate the CNTs by attaching the Au tip to the fixed Au tip. The STM-TEM holder (Nanofactory Instruments) has a manipulate resolution of better more than 0.05 nm at the X, Y, and Z directions, and has a current measurement sensory of more than 1 nA. (b) Typical TEM micrograph of the CNT prepared using the CVD method, which was purchased from Shenzhen Nanotech Port Co., Ltd. I–V measurements were performed after the contact formation. The positive and negative I–V scans were performed following the schematic shown in (c). A complete I–V scan includes a negative scan automatically followed by a positive scan.
Different connections investigated
Figure s2 shows connections corresponding to the three types of fullerene hinges.
Figures s2(a)–s2(j). The “electrical hinge” established using a fullerene clamped between two CNTs. The left CNT has an outer diameter of 3.3 ± 0.1 nm, whereas the one on the right has a diameter of about 4.4 ± 0.1 nm, which was covered with tinny graphite flakes. The fullerene in (a) has an outer diameter of 2.5 ± 0.1 nm and a wall thickness of approximately 0.6 ± 0.1 nm. The shaded regions separate the left CNT, the fullerene, and the right CNT, for guiding the eyes. In each figure, the two CNTs contacted with the fullerene with different contact geometries. Mode I contact is shown in (a)–(d), where the left CNT changed position while the fullerene remained almost unchanged. Mode II contact is shown in (e)–(f) and (i)–(j), where the distance between the two CNTs and the fullerene position changed simultaneously. Mode III contact is shown in (f)–(i), where the fullerene changed position between the two CNTs. These behaviors can be better understood by referring to Figure 1. Mixed contact modes were also extrapolated in the figure. The images were taken after finishing a complete I–V scan. Thus one or several I–V measurements could be performed between the adjacent images. Arrows in figures indicate the position of the fullerene. All figures have the same magnification. Bar = 10 nm.
Movies showing the motion of the connections are also given in sm1–sm5.
sm1 shows a similar structure to that in Figure 2. The CNT changed position during the I–V measurements (mode I connection). A mode II connection can also be seen at 5 s and 15 s.
sm2 shows a similar structure to that in Figure 2. A mode II connection can be seen at 5 s.
sm3 shows a similar structure to that in Figure 2. A mode III connection can be seen at 7 s.
sm4 shows a similar structure to that in Figure s3. A mode I connection can be seen at 2 s.
sm5 shows a similar structure to that in Figure s3. The “pick-up” of the fullerene indicates the existence of electrical static force.
Shortening of the CNT and the diminishing of the fullerene caused by carbon evaporation
Figure s3 shows the evolution of a fullerene hinge with time. When the connection is subjected to electrical solicitation, the geometry changes, the fullerene diameter decreases, and the CNT shortens due to carbon evaporation.
Figures s3(a)–s3(p). TEM images showing other examples of using the fullerene as an “electrical hinge.” The left CNT is similar to that in Figure 2. The fullerene was located on a cluster of graphite flakes (G) on the right-hand side. Typically in (i), the left CNT, the fullerene, and the right graphene flakes are shaded separately, for guiding the eyes. The fullerene changed its position, shape, and diameter during the I–V sweeping process. The diameter decrease (from an initial 2.7 ± 0.1 nm to 1.9 ± 0.1 nm) was due to the high-temperature-induced evaporation of carbon atoms during the mode I, II, and III connections, respectively. Arrows in figures indicate the position of the fullerene. All figures have the same magnification. Bar = 10 nm.
Aside from the change in contact geometry, other features observed in the fullerene hinge are the shortening of the CNT and the decrease in the fullerene diameter. Figure s3 shows these effects. The CNT was shortened by ~10 nm from Figures s3 (a)–(p), and simultaneously, the fullerene was shrunk by about 30%. There are two possibilities for these features. The first possibility is the knocking off of the carbon atoms due to e-beam irradiation, which is frequently observed during TEM observation . However, TEM is a relatively slow process and the diameter and length of the CNT should simultaneously decrease. Considering that only a slight diameter decrease was observed, e-beam irradiation considered as a minor effect. Instead, the evaporation of the carbon atoms at the contact site was regarded as the main reason. Carbon evaporation is a result of the high temperature created during the I–V measurements. According to previous reports, a temperature of nearly 3000 K can be achieved using the current experimental setup, where carbon evaporation could be observed [30, 31].
Techniques for the characterization of CNT and fullerene dimensions
Figure s4. Determination of the diameter and wall thickness from the TEM images (top). A line was drawn across the CNT (middle) or the fullerene (bottom), and the dimensions were measured from the contrast profile. The contact area between CNT and fullerene is estimated from direct measurement based on TEM image as shown by red lines. The average contact diameter we measured is around 1.0 nm. An uncertainty of 0.1 nm resulted from the diffusion of the image contrast, which made it difficult to determine accurate edges.
Details about the electrical transportation
Figure s5 shows the I–V curves relative to the connections described in Section II.
Figure s5. (a) and (b) show the selected positively and negatively biased I–V curves of the CNT–fullerene–CNT contact, respectively. The labels correspond to the different contacting geometries in Figure s2. (c) and (d) show the log–log plot for the I–V curves, where the positive and negative parts are shown simultaneously.
Fluctuation observed at a higher bias voltage
In some of the I–V curves, small fluctuations and asymmetries were observed. The fluctuations of the curves were more evident under a high bias of 1.5 V to 2 V, due to the change in the contact geometry during the I–V measurement caused by the larger Faraday force or/and the high-temperature-induced interface bond breakage. In some cases, detachment of the contact was observed, accompanied by the current jumps to zero, such as in the I–V curve labeled as “IIa –” in Figure 3 (see also “b,” “e,” and “h” in Figure s5). All the breaking points in the curves were found to occur at a relatively lower current, and their I–V curves are relatively lower than the others (smaller current under the same bias). Considering that their corresponding TEM figures show almost the same contact area, the failure of the contact could be more probably due to the weaker covalent carbon bonds at the CNT–fullerene contact.
The I–V relationship based on different electrical transportation models was analyzed to understand the conducting mechanism better. Fitting with different kinds of transportation models failed to give an accurate description of the I–V curves. However, an acceptable fitting using the space-charge-limited current (SCLC) model gives a linear relationship of . [r1–r3]. The SCLC relationship shows a linear relationship in the log–log plot of the I–V relationship, as shown in the lower panel of Figure 3 (see also Figure s5). Curve fitting gives a slope of α = 1.0–3.5, but it does not show a clear dependence on the contact geometry. The observation of the SCLC process indicates that the Ohmic contact was formed between the CNT and the fullerene [r4]. Considering that the two CNTs have relatively large diameters and more than three walls, they should be treated as metallic-like. For the fullerene with a size the similar to that of C720, the previous calculation gives the band gap of around 0.5 eV for a single-shelled fullerene [37, 38]. For a double-layered fullerene, the band gap was assumed to be smaller but still semiconductor-like because of the unavailability of reference data [r5, r6]. Thus, the fullerene hinge forms a symmetrical metal–semiconductor–metal structure. The SCLC could be attributed to the semiconductor nature of the fullerene.
Note that although SCLC transportation was observed in the system, the bulk and contact effects are difficult to distinguish because of the small dimension of the fullerene compared with the size of the contact. Differentvalues of α = 1.0–3.5 were observed in the “fullerene hinge” with changing contact geometry, and the difference in the interface structures could be the most probable reason for the diversity in the α values. According to the SCLC model, different α values indicate different levels of trap occupation in the band gap or different relationships between the concentration of the injected carrier and the carrier in the semiconductor. α = 1 refers to the Ohmic behavior, indicating that there are enough free carriers for current transportation. α = 2 corresponds to the SCLC region, where all trap levels have been occupied. α = 1–2 is the transition region from Ohmic to SCLC conduction, and α > 2 refers to the SCLC in the trap-filling region [r1–r3]. In the current study, the SCLC transportation process is tentatively attributed to the formation of different interface bond structures between the CNT and the fullerene, because the interface bond structure, such as the type (single/double/triple) and number of bonds, significantly affects electron transportation [34–37]. The change in the interface structure has a similar effect with the different levels of trap-filling processes. The formation of the interface bond increases the electron density of state in the band gap of the fullerene, which serves as the channel for electron transportation [34–37]. The changes in the bond type and number will cause the change in the conducting channels. This change has a similar significance to that of the trap-filling process. More conducting channels tend to induce Ohmic conductivity, that is, smaller α values near 1 and lesser conducting channels tend to induce SCLC conduction, that is, larger α values [r7, r8]. Although the type and number of bonds under the current stage are difficult to determine, the movement of the fullerene at the contact and the change in the contact geometry are believed to be the causes of the change in the interface bond structure.
The I-V curves of the CNT-CNT connections for reference
Figure s6 gives the information of the electrical property in the CNT-CNT contact without fullerene. Comparing to situations with a fullerene between them, the CNT-CNT connections show increased current level (increased ~20% at corresponding voltages). As a matter of fact, although increased current level was normally observed in the CNT-CNT contact, the I-V curves of the CNT-CNT contact were also affected by the contact geometry. Simply, larger contact area shows accordingly higher current level and vise versa. The detailed discussion about this is out of the extent of this paper.
Figure s6 the I-V curves of the CNT-CNT connection without fullerene between them. (a) Corresponds to the situation shown in figure s3 and (b) corresponds to the situation shown in figure s2. I-V curves noted with different numbers correspond to different scans. Inset TEM figures present the structure of the CNT-CNT connections in different I-V scans, as indicated. Bar= 10 nm. Arrows point out the contact sites between the two CNTs.
method in construction the two kinds of contact in figure 4(a) and (b)
We constructed the different electrical setups by precisely controlling (not by chance) the movement of the tubes and verified by more than 10 times of similar experiments (considering both the images and I-V measurement). We make it sure following the below steps in construction the two kinds of connect: firstly, the CNT-fullerene-CNT contact is constructed, the I-V curve(s) is obtained; then the CNT on the movable probe side was pushed forward carefully, step by step (movement resolusion: 20 pm). The CNT-fullerene-CNT contact is maintained during whole manipulation process (following the mechanism described in figure 1(b)). Secondly, the fullerene is pushed aside and the two CNTs contacted with each other. This process also can be precisely performed under controlling the movable probe. This process can be performed under carefully controlled manner and can be prepared repeatedly, which insures the results given in figure 4.
Supplementary references [r1] A. A. Talin, F. Léonard, B. S. Swartzentruber, X. Wang, and S. D. Hersee, Phys. Rev. Lett., 101(2008)076802
[r2] A. Carbone, B. K. Kotowska, and D. Kotowski, Phys. Rev. Lett., 95(2005) 236601
[r3] Y. Gu and L. J. Lauhon, Appl. Phys. Lett., 89(2006)143102
[r4] A. Rose, Phys. Rev. 97(1955)1538
[r5] Y. L. Lin and F. Nori, Phys. Rev. B, 49(1994)5020
[r6] M. Pudlak and R. Pincak, Phys. Rev. A, 79(2009)033202
[r7] F. Börrnert, C. Börrnert, S. Gorantla, X. J. Liu, A. Bachmatiuk, J. O. Joswig, F. R. Wagner, F. Schäffel, J. H. Warner, R. Schönfelder, B. Rellinghaus, T. Gemming, J. Thomas, M. Knupfer, B. Büchner, and M. H. Rümmeli, Phys. Rev. B, 81(2010)085439
[r8] A. Nagataki, T. Kawai, Y. Miyamoto, O. Suekane, and Y. Nakayama, Phys. Rev. Lett., 102(2009)176808