Mimicking a gecko’s foot with strong adhesive strength based on a spinnable vertically aligned carbon nanotube array

Abstract

To mimic gecko foot adhesion, spinnable vertically aligned carbon nanotube (VACNT) arrays, which have a higher density and cleaner surface than ordinary VACNT arrays, were prepared by a normal chemical vapor deposition (CVD) process that is simple and easy to operate for large-scale fabrication, particularly compared with the low-pressure CVD process. The height of the spinnable VACNT array was tuned by varying the reaction time. The shear adhesion strength of the spinnable VACNT array (0.16 cm²) was increased from 21.4 ± 1.7 to 85.8 ± 8.7 N cm⁻² when the length of the spinnable VACNT array increased from 35 to 110 μm. Based on the enhanced van der Waals force induced by the large number of contact points on the high-density spinnable VACNT array, the maximum shear adhesion strength of the spinnable VACNT array (0.16 cm²) is 91.8 N cm⁻², which is comparable to that of the CNT-based adhesive (∼100 N cm⁻²) prepared by the low-pressure CVD process. Moreover, a spinnable VACNT array adhesive was prepared over a large area, and a maximum weight of 3.0 Kg was supported successfully by a spinnable VACNT array adhesive with a contact area of 0.96 cm².

---

1. Introduction

The amazing climbing ability of geckos has inspired scientific interest because geckos can make strong and reversible attachments against surfaces of varying roughness and orientation.^1,2 A gecko’s foot has countless specialized keratinous aligned microscopic elastic hairs (130 ± 30 mm in length) called setae. Each seta contains hundreds of projections terminating in 0.5 ± 0.2 mm spatula-shaped structures.³ It is now known that the gecko owes the extreme reversible adherence of its feet to the structure of its hierarchically arranged fibrils, which enable it to exploit van der Waals and capillary forces with great efficiency.⁴ Each spatula produces a miniscule force ≈10⁻⁷ N, but millions of spatulas acting together create a formidable adhesion of ≈10 N cm⁻², which is sufficient to keep geckos firmly on their feet, even when upside down on a glass ceiling.⁵ Since discovering the hierarchical structure of a gecko’s foot, extensive studies have been undertaken to design synthetic dry adhesives by mimicking gecko foot hairs.

Polymer-based and carbon nanotube (CNT)-based dry adhesives are the two main routes for fabricating gecko-inspired artificial adhesives.² The main reason for using polymers and CNTs as attachment tip materials is the simple and reproducible manner in which they can be prepared to obtain robust, high aspect-ratio (AR) structures. Another advantage of the polymer-based method is that the polymer's geometry (angle, radius, height, tip shape and hierarchy) can be tailored to a simple and scalable approach for fabricating gecko-mimicking nanohairs.⁶ Meanwhile, the properties (modulus, surface energy, etc.) of polymer-based gecko-mimicking nanohairs can be tuned in a rapid and cost-effective manner. However, the adhesion strength of polymer-based gecko nanohairs is usually lower than that of CNT-based adhesives because the resolution and AR of polymer nanostructures are restricted by the low mechanical strength of polymer materials. According to the splitting theory,⁷ nanostructures with smaller features have much higher adhesion forces. CNTs have outstanding structural properties, such as extremely small radii (∼10 nm) and a controllable, elevated height (micrometer- or centimeter-scale), as well as strong mechanical properties (elastic modulus ∼1000 GPa). Moreover, the unique thermal and electrical properties of CNTs can be exploited as additional assets in developing CNT-based adhesives with electrically switchable and high temperature-tolerant properties.^8,9 Therefore, CNTs are one of the most suitable materials for gecko-inspired dry adhesives. For example, Qu et al. prepared ordinary vertically aligned CNT (VACNT) array dry adhesives using low-pressure (10 mTorr) chemical vapor deposition (CVD) and obtained a shear adhesion (∼100 N cm⁻²)³ that was 10 times higher than the gecko's adhesion strength. However, these researchers found that the adhesion forces of ordinary VACNT arrays prepared by normal (1 atm) CVD are generally less than 1 N cm⁻² because of the absence of a nonaligned nanotube top layer and/or the poor quality of the CNTs.³ It has been noted that low-pressure CVD is a complicated process, which may limit the widespread use of VACNT array dry adhesives. Therefore, it is necessary to develop practical applications of CNT-based adhesives to prepare VACNT array dry adhesives with high shear adhesion strengths using normal CVD, which is a simple, easy and large-scale process that can operate at normal pressures.

On the other hand, the CNTs in spinnable VACNT arrays have a better alignment, cleaner surfaces, higher densities and a narrower diameter distribution compared to ordinary VACNT arrays.¹⁰ The higher densities and cleaner surfaces of CNTs enable spinnable VACNT arrays to have more contact points per unit contact area and a stronger van der Waals force for each of the contacts. As a result, enhanced adhesion forces can be expected for spinnable VACNT arrays. However, to the best of our knowledge, there have been no published studies using spinnable VACNT arrays to mimic gecko feet. In addition, most of the reported CNT-based adhesives are small (∼0.16 cm²).³ For practical applications, CNT-based adhesives must have larger contact areas.

In the present work, spinnable VACNT arrays prepared by a normal CVD method were first used to mimic gecko feet adhesion, and a large adhesive area (∼0.96 cm²) was realized. The maximum shear adhesion strength of the spinnable VACNT array reached 91.8 N cm⁻², comparable to that reported for ordinary VACNT adhesives (∼100 N cm⁻²) prepared using low-pressure CVD.³ A weight of 3.0 kg was supported on a glass slide using a spinnable VACNT array dry adhesive with a contact area of 0.96 cm².

2. Experimental methods

2.1. Synthesis of the spinnable VACNT array

The synthesis of spinnable VACNT arrays has been reported in detail elsewhere.¹¹ In brief, double layers of Fe (1 nm)/Al (10 nm) were deposited on a SiO₂/Si substrate as a catalyst using E-beam evaporation (ULVAC Electron Beam Evaporator). The Al layer was coated on the SiO₂/Si substrate before the deposition of the Fe film to enhance the attachment of grown CNTs on the Si wafer. The deposition rate was maintained at 0.03 nm s⁻¹. The spinnable VACNT arrays were synthesized in a one-zone quartz-tube furnace. The reaction chamber was a quartz tube of 1 inch diameter. The substrates were placed in the middle of the quartz tube, and the pressure in the quartz tube was reduced to 5 mTorr to remove any ambient gas. The total pressure was maintained at 1 atm for each experiment. The furnace was heated to the growth temperature (750 °C) at 20 °C min⁻¹ and under 100 sccm Ar flow. Next, C₂H₂ (99.999%) and/or H₂ (99.999%) was introduced into the reactor, and the spinnable VACNT arrays were synthesized at the growth temperature for a specified time period. Finally, the acetylene and/or hydrogen gas flow was turned off, and the furnace temperature was reduced to room temperature while the Ar gas continued flowing. Scanning electron microscopy (SEM, JEOL-6470, 10 KV) was used to characterize the morphology of each spinnable VACNT array.

2.2. Test of adhesion strength

Measurements were conducted by first pressing the spinnable VACNT array, grown on the SiO₂/Si substrate, against the surface of a glass slide (root-mean-square roughness 0.98 nm) with a normal preload force. The spinnable VACNT array was placed, topside down on the glass slide. Subsequently, different weights (0.5, 1.0, 1.5 and 2.0 kg) were placed on the rear surface of the silicon wafer to press the VACNT array onto the glass slide. After the preload force was completely removed, the normal and shear adhesion strength of the spinnable VACNT array were assessed by measuring the pull-off forces in the shear and normal directions (Fig. S1†). Weights ranging from 1 g to 3.0 kg were used to complete the test. To eliminate the effects of the attachment cycles and avoid the pre-alignment effect for subsequent adhesion measurements, the adhesion forces obtained from the first attachment of each sample were adopted. Adhesion tests for each reported condition were completed for 4 samples, and the average values were used. During the measurement, the temperature and humidity were ∼26 °C and ∼70%, respectively. In our experiments, the effects of substrate surface roughness, temperature and humidity on the adhesion strength of the CNT-based adhesive were not considered.

3. Results and discussion

Fig. 1 shows typical SEM images of spinnable VACNT arrays on Si wafers grown under the following conditions: C₂H₂ = 100 sccm, H₂ = 50 sccm and 750 °C. The heights of the spinnable VACNT arrays in panel (a) are approximately 110 μm after 8 min of growth. As shown in Fig. 1a, the as-synthesized CNTs aligned almost normal to the substrate surface and had uniform lengths. It should be noted that the heights of the spinnable VACNT arrays grown in the experiment were extremely uniform and varied only by 1–2% over the sample, which is important to the CNT adhesion force. The uniform height of the VACNT array enabled close contact between the CNT array and the glass slide. At the top layer of the VACNT array, the CNTs were tangled and not aligned (Fig. 1b), which is similar to what was observed with ordinary VACNT arrays prepared by low-pressure CVD.² The surface roughness of the spinnable VACNT array was measured using an atomic force microscope (AFM), and the root-mean-square roughness was 100.81 nm, with a scanning area of 10 × 10 μm². At the start of the growth of the spinnable VACNT array, the initially formed CNTs grew randomly and formed a randomly entangled top layer under which the underlying aligned CNT arrays then emerged.^3,11 The curly entangled segment at the top layer of the spinnable VACNT array was able to enhance side contact with flat surfaces.¹² As a result, the adhesion strength of the spinnable VACNT array was improved significantly. As shown in Fig. 1c, the CNTs in spinnable VACNT arrays are very clean and bundled in the vertical direction. We obtained packing densities for the spinnable VACNT arrays that were as high as ∼5 × 10¹¹ CNTs per cm², as estimated from Fig. 1c. The CNT bundles were parallel to one another, and the wavy CNTs switched between different straight CNT bundles. Generally, the spinnable VACNT arrays had higher packing densities compared to ordinary VACNT arrays.¹⁰ The higher packing densities and clean surfaces of CNTs are intrinsic characteristics of spinnable VACNT arrays that make them suitable for mimicking gecko foot adhesion.

Fig. 1 Typical SEM images of spinnable VACNT arrays. (a) Cross-sectional SEM image of a VACNT array (length ∼112 μm). (b) Top view of a VACNT array showing entanglement of the nanotubes at the surface. (c) Super-aligned CNTs in a spinnable VACNT array.

To characterize the performance of the spinnable VACNT array dry adhesive, we systematically measured the macroscopic shear and normal adhesion strengths as a function of the length of the spinnable VACNT arrays. As shown in Fig. 2a, the adhesion strengths of the spinnable VACNT arrays increased almost linearly as the lengths of the spinnable VACNT arrays were increased. For example, the shear adhesion strengths of the spinnable VACNT arrays increased from 21.4 ± 1.7 to 85.8 ± 8.7 N cm⁻² when the lengths of the spinnable VACNT arrays increased from 35 to 110 μm. Generally, the longer VACNT arrays supported longer randomly entangled nanotube segments on the top surface of the VACNT arrays,³ which is beneficial in that it increases side contact between the CNTs and their target surfaces. Meanwhile, there was more flexibility with regard to tilting the long VACNT arrays, enabling the closest contact with the target substrate to be obtained. The corresponding normal adhesion strength also increased from 13.8 ± 2.2 to 33.7 ± 4.3 N cm⁻² over the same range of CNT lengths. The maximum shear adhesion strength of spinnable VACNT arrays reached 91.8 N cm⁻², which is comparable with measurements reported for CNT-based adhesives (∼100 N cm⁻²) prepared by low-pressure CVD.³ Compared with the low-pressure CVD process, the normal CVD process is simple and critical for the practical application of CNT-based adhesives.

Fig. 2 (a) Shear and normal adhesion forces of spinnable VACNT array dry adhesives as a function of the length of the VACNT array. The preload pressure used to attach the fastener was 122.5 N cm⁻². (b) Preload pressure-dependent adhesion forces of the spinnable VACNT array (length ∼35 μm) adhesive. The errors in panels (a) and (b) represent standard errors calculated from 4 measurements.

In addition to the height of the spinnable VACNT array, the adhesion strength of the spinnable VACNT array was also affected by the applied preload pressure in that it determined the strength with which the spinnable VACNT array was pressed against the surface of the glass slide. As shown in Fig. 2b, the adhesion strengths of the spinnable VACNT arrays (length ∼35 μm) increased with applied preload pressure. The shear adhesion strength of the spinnable VACNT arrays increased from 3.1 ± 0.9 to 21.4 ± 1.7 N cm⁻² when the preload pressure was increased from 30.6 to 122.5 N cm⁻². Van der Waals forces are mainly responsible for the adhesive force between the VACNT array and the glass slide.³ Van der Waals forces are extremely weak at greater-than-atomic-distance gaps, and close contact between the adhesive and the surface is necessary to obtain a large adhesive strength.¹³ Thus, increasing the preload pressure enables more close contact between the spinnable VACNT array and the glass surface, thereby strengthening the van der Waals and adhesion forces involved.

The maximum shear adhesive force was estimated to better understand the adhesive properties between the spinnable VACNT array and its target surface. The contact between the CNT and its target surface was assumed to be a line contact with an effective length L after preloading. The maximum van der Waals force between a single CNT and the target surface per unit length is ,¹⁴ where A is the Hamaker constant, R is the radius of the CNT and D₀ is the “cut-off” distance between the CNT and the glass slide.¹⁴ The maximum shear adhesive force per unit area of the spinnable VACNT array is P^max_f = μF^max_vdwLρ, where μ is the friction coefficient, L is the contact length and ρ is the effective contact density per unit area.³ Under conditions where A = 6 × 10⁻²⁰ J,³ R = 5 nm, D₀ = 0.34 nm,³ μ = 0.09,¹⁵ L = 50 nm and ρ = 5 × 10¹¹ tubes per cm², P^max_f was 395.88 N cm⁻², which means the adhesive force of the spinnable VACNT array dry adhesive could be improved significantly by increasing the effective contact length between the CNT and the target surface.

Fig. 3 shows typical SEM images obtained for spinnable VACNT arrays after the adhesion-strength measurements were completed. Consistent with published reports,³ the top layer of the randomly entangled nanotube segments became horizontally aligned after the shear adhesion-strength test (Fig. 3a). Meanwhile, the vertically aligned spinnable VACNT array was tilted along the shear direction after the test (Fig. 3b). Upon shear pulling, the external shear force caused the nonaligned nanotube segments to align and the vertically aligned nanotube trunks to tilt along the shear direction. However, the morphology of the randomly entangled nanotube segments did not necessarily change in an obvious manner after normal adhesion-strength measurements (Fig. 3c). This is because the CNTs made point contact with the target surface during the normal adhesion-force test.² In contrast, the CNTs made side contact with the target surface during the shear adhesion-force measurement.¹⁶ Side contact of the CNTs with the target surface could provide a stronger adhesion force than tip contact, significantly increasing the shear adhesion strength of the VACNT array beyond its normal adhesion strength. As shown in Fig. 3d, a few zigzag buckles formed along the VACNT array after several cycles of normal adhesion-force measurements. As demonstrated by Cao et al.,¹⁷ the VACNT array could be compressed to form ordered wavelike folds when the strain was less than 15% and returned to the original state upon stress removal. With strains higher than 15%, irreversible deformation occurred, and the VACNT array could not recover its original morphology upon load release. In fact, only a few buckles formed along the spinnable VACNT array after several attachment-detachment cycles. The main reason for this observation is that the spinnable VACNT array has a higher density than that of an ordinary VACNT array. It is difficult for nanotubes to buckle in a densely aligned nanotube array because of the proximity of the neighboring tubes.¹⁷ The force-induced buckling of the VACNT array lowered the adhesion strength of the VACNT array after several loading cycles.⁸ In our case, the spinnable VACNT array could be repeatedly attached to and detached from the glass slide without decreasing the supported weight for 10 cycles. However, the supportable weight of the spinnable VACNT array decreases after 10 cycles of attachment–detachment tests (data not shown).

Fig. 3 SEM images showing the morphology of a spinnable VACNT array after the adhesion-force measurement. (a) Top and (b) side views of the VACNT array after the shear adhesion-force measurement. (c) Top view of the VACNT array after the normal adhesion-force measurement. (d) Side view of the VACNT array after several cycles of the normal adhesion-force test.

It should be noted that the CNT-based adhesives reported in the literature are usually small (∼0.16 cm²).² The adhesion force per contact area can be enhanced significantly by reducing the contact area. However, the CNT-based adhesive should be prepared over a large area for practical applications. To investigate the adhesion performance of spinnable VACNT arrays, a large-area spinnable VACNT array was prepared. Fig. 4 shows the inverse dependence of the adhesive strength of the spinnable VACNT array on the contact area between the VACNT array and the glass surface. As shown in Fig. 4, the shear adhesion strength of the spinnable VACNT array decreased from 85.8 ± 8.7 to 38.3 ± 10.8 N cm⁻² when the contact area of the VACNT array increased from 0.16 to 0.64 cm². A possible explanation is that the Si substrates are relatively stiff, and the roughness of the target surfaces prevents uniform engagement of the nanotubes over large contact areas.¹⁸ The shear adhesion strength of the spinnable VACNT array decreased to 33.2 ± 3.6 N cm⁻² when the contact area of the VACNT array was further increased to 0.96 cm². As shown in Fig. S2,† a preload pressure of 122.5 N cm⁻² was able to attach a spinnable VACNT array (0.64 and 0.96 cm²) onto a vertically positioned glass slide and to enable weights of 2.0 and 3.0 kg, respectively, to be held in the shear direction. The maximum weight that could be held by a spinnable VACNT array dry adhesive with a surface area of 0.96 cm² (3.0 kg) was 3 times higher than that of a polymer-based adhesive with a contact area of 1.0 cm².^19,20 In addition, the shear force supported by a 0.96 cm² spinnable VACNT array dry adhesive was comparable to that of a 1 cm² patterned CNT array transferred on flexible tape.²¹ This result indicates that only 20 pieces of this small spinnable VACNT array dry adhesive (0.96 cm²) with a total contact area of approximately 20 cm², which is smaller than the palm of the hand (over 200 cm²), are needed to collectively hold a person of ca. 60 kg.

Fig. 4 Shear adhesion forces of a spinnable VACNT array (length ∼110 μm) with the adhesive as a function of the contact area of the VACNT array. The preload pressure used to attach the adhesive was 122.5 N cm⁻². The errors represent the standard errors calculated from 4 measurements.

4. Conclusions

In summary, spinnable VACNT arrays were prepared using a normal CVD process to mimic gecko foot adhesion. Compared with the low-pressure CVD process, the normal CVD process is simple and enables large-scale and practical applications of CNT-based adhesives. The heights of the spinnable VACNT arrays were tuned by varying the reaction times. The shear adhesion strengths of the spinnable VACNT arrays were increased from 21.4 ± 1.7 to 85.8 ± 8.7 N cm⁻² when the lengths of the spinnable VACNT arrays (0.16 cm²) increased from 35 to 110 μm. The maximum shear adhesion strength of the spinnable VACNT arrays (91.8 N cm⁻²) was comparable to that of CNT-based adhesives (∼100 N cm⁻²) prepared using low-pressure CVD. The spinnable VACNT array dry adhesive was able to support a weight of 3.0 Kg when the contact area of the spinnable VACNT array increased from 0.16 to 0.96 cm².

Acknowledgements

This work was supported by the Japan Society for the Promotion of Science under a Grant-in-Aid for Scientific Research (A) 23246024.

References

1. B. Bhushan, Langmuir, 2012, 28, 1698–1714.
2. H. E. Jeong and K. Y. Suh, Nano Today, 2009, 4, 335–346.
3. L. Qu, L. Dai, M. Stone, Z. Xia and Z. L. Wang, Science, 2008, 322, 238–242.
4. L. F. Boesel, C. Greiner, E. Arzt and A. del Campo, Adv. Mater., 2010, 22, 2125–2135.
5. A. K. Feim, S. V. Dubonos, I. V. Grigorieva, K. S. Novoselov, A. A. Zhukov and S. Y. Shaproval, Nat. Mater., 2003, 2, 461–463.
6. M. K. Kwak, C. Pang, H. E. Jeong, H. N. Kim, H. Yoon, H. S. Jung and K. Y. Suh, Adv. Funct. Mater., 2011, 21, 3606–3616.
7. E. Arzt, S. Gorb and R. Spolenak, Proc. Natl. Acad. Sci. U. S. A., 2003, 100, 10603–10606.
8. S. Hu, Z. Xia and L. Dai, Nanoscale, 2013, 5, 475–486.
9. L. Qu and L. Dai, Adv. Mater., 2007, 19, 3844–3849.
10. K. Liu, Y. Sun, L. Chen, C. Feng, X. Feng, K. Jiang, Y. Zhao and S. Fan, Nano Lett., 2008, 8, 700–705.
11. Y. Cui, B. Wang and M. Zhang, J. Mater. Sci., 2013, 48, 7749–7759.
12. H. Ko, Z. Zhang, J. C. Ho, K. Takei, R. Kapadia, Y. L. Chueh, W. Cao, B. A. Cruden and A. Javey, Small, 2010, 6, 22–26.
13. K. Autumn, Y. A. Liang, S. T. Hsieh, W. Zesch, W. P. Chan, T. W. Kenny, R. Fearing and R. J. Full, Nature, 2000, 405, 681–685.
14. D. Leckband and J. Israelachvili, Q. Rev. Biophys., 2001, 34, 105–267.
15. P. L. Dickrell, S. B. Sinnott, D. W. Hahn, N. R. Raravikar, L. S. Schadler, P. M. Ajayan and W. G. Sawyer, Tribol. Lett., 2005, 18, 59–62.
16. L. Ge, L. Ci, A. Goyal, R. Shi, L. Mahadevan, P. M. Ajayan and A. Dhinojwala, Nano Lett., 2010, 10, 4509–4513.
17. A. Y. Cao, P. L. Dickrell, W. G. Sawyer, M. N. Ghasemi-Nejhad and P. M. Ajayan, Science, 2005, 310, 1307–1310.
18. Y. Zhao, T. Tong, L. Delzeit, A. Kashani, M. Meyyappan and A. Majumdar, J. Vac. Sci. Technol., B: Microelectron. Nanometer Struct.--Process., Meas., Phenom., 2006, 24, 331–335.
19. T. I. Kim, H. E. Jeong, K. Y. Suh and H. H. Lee, Adv. Mater., 2009, 21, 2276–2281.
20. M. P. Murphy, B. Aksak and M. Sitti, Small, 2009, 5, 170–175.
21. L. Ge, S. Sethi, L. Ci, P. M. Ajayan and A. Dhinojwala, Proc. Natl. Acad. Sci. U. S. A., 2007, 104, 10792–10795.

---

Footnote

† Electronic supplementary information (ESI) available: Additional figures as described in the text. This material is available free of charge via the Internet at http://pubs.rsc.org. See DOI: 10.1039/c3ra46113k

---