Direct Measurement of the Binding Force between Microfabricated

Langmuir 2005, 21, 11251-11261

11251

Direct Measurement of the Binding Force between Microfabricated Particles and a Planar Surface in Aqueous Solution by Force-Sensing Piezoresistive Cantilevers Hiroaki Onoe,*,† Murat Gel,‡ Kazunori Hoshino,† Kiyoshi Matsumoto,† and Isao Shimoyama† Department of Mechano-Informatics, Graduate School of Information Science and Technology, The University of Tokyo, and Center for International Research on Micromechatronics (CIRMM), Institute of Industrial Science, The University of Tokyo Received June 21, 2005. In Final Form: August 26, 2005 We propose a force measurement method for evaluating the binding force between microscale flat surfaces in an aqueous solution. Using force-sensing piezoresistive cantilevers with sub-nanonewton force resolution, we have directly measured binding forces between SiO2-SiO2 microcontacts, which were created by gravitydriven random collision between microfabricated SiO2 cylindrical particles and a planar SiO2 substrate in a HCl solution. First, to examine our method we measured the pH dependence of the binding force. The binding forces were 12 and 5.8 nN at pH 1.0 and 2.0, respectively. As the pH increased, the binding force decreased and became zero at pH greater than 3.0. We confirmed that the bindings were based on the van der Waals’ (VDW) force at pH 2.0 or less whereas a repulsive double-layer force acted between the surfaces at pH 3.0 or more. Second, the binding forces were categorized into a friction force or an adhesion force between the particles and the substrate. In the measurement, the friction force between the particle and the substrate was measured in the case when the particle slid on the substrate. On the contrary, the adhesion force was measured when the particle came off the substrate. Whether the particle slid or came off depended on the aspect ratio of the particle. We fabricated cylindrical particles with an aspect ratio of 0.03-2.0 and distinguished the friction force from the adhesion force by changing the aspect ratio of the particles. As a result, the friction force per unit contact area between SiO2-SiO2 flat surfaces was found to be 330 pN/µm2 ( 20% when we used particles with a low aspect ratio (0.4). For fluidic self-assembly that utilizes microscale surface contact in a liquid, our measurement method is an effective tool for studying and developing systems.

Introduction Fluidic self-assembly is a technique for assembling microscale components in a liquid environment for microscale integration. In the assembly process, many microfabricated parts are stirred in liquids, and microstructures are fabricated spontaneously in two or three dimensions through random collisions of the microparts. Microparts usually have microscale flat surfaces designed to be binding or nonbinding sites. The interactions between binding sites cause microparts to assemble themselves. Many research groups have proposed fluidic self-assembly techniques using various binding forces driven by gravity force,1 capillary force,2-6 the hydrophobic effect,7,8 bridging flocculation,9 or biological events 10 to create new functional * Corresponding author. E-mail: [email protected]. Tel: +81-3-5841-6318. Fax: +81-3-3818-0835. † Department of Mechano-Informatics, The University of Tokyo. ‡ Center for International Research on Micromechatronics. (1) Talghader, J.; Tu, J.; Smith, J. IEEE Photonics Technol. Lett. 1995, 7, 1321-1323. (2) Jacobs, H.; Tao, A.; Schwartz, A.; Gracias, D.; Whitesides, G. Science 2002, 296, 323-325. (3) Xiong, X.; Hanein, Y.; Fang, J.; Wang, Y.; Wang, W.; Schwartz, D.; Bo¨hringer, K. J. Microelectromech. Syst. 2003, 12, 117-127. (4) Srinivasan, U.; Liepmann, D.; Howe, R. J. Microelectromech. Syst. 2001, 10, 17-24. (5) Gracias, D.; Tien, J.; Breen, T.; Hsu, C.; Whitesides, G. Science 2000, 289, 1170-1172. (6) Breen, T.; Tien, J.; Oliver, S.; Hadzic, T.; Whitesides, G. Science 1999, 284, 948-951. (7) Onoe, H.; Matsumoto, K.; Shimoyama, I. J. Microelectromech. Syst. 2004, 13, 603-611. (8) Onoe, H.; Matsumoto, K.; Shimoyama, I. Proceedings of the 17th Micro Electro Mechanical Systems; Maastricht, 2004.

devices and structures such as light-emitting diode (LED) arrays,1-3 micromirror arrays,4 and electric circuit networks.5 A current challenge in fluidic self-assembly is realizing precise control of the bindings of the microparts (e.g., binding sequence control8 and binding recognition with shape matching9,10) to assemble more complex and functional structures. To achieve precise control of the bindings between microparts, we need to know exact values of the binding forces. In other words, we need to find out how strong the binding between microparts is, how much force the assembled microparts require to maintain their structures, and how binding sites and nonbinding sites in microparts work properly as designed. Therefore, a measurement method that enables us to ascertain the interactions between microparts in a liquid is needed. For binding force measurement in liquid, atomic force microscopy (AFM)11 and the surface force apparatus (SFA)12 have been used to study the interactions between two surfaces in liquid for a variety of interfaces. To explore the mechanisms of interfacial phenomena between various kinds of surfaces in a liquid environment, attractive and repulsive interactions on a nanometer-separation scale, adhesion, friction, and wear forces have been investigated (9) Nakakubo, T.; Shimoyama, I. Sens. Actuators, A. 2000, 83, 161166. (10) McNally, H.; Pingle, M.; Lee, S.; Beregstrom, D.; Bashir, R. Appl. Surf. Sci. 2003, 214, 109-119. (11) Binning, G.; Quate, C.; Gerber, C. Phys. Rev. Lett. 1986, 56, 930-933. (12) Israelachvili, J.; Adams, G. J. Chem. Soc., Faraday Trans. 1 1978, 74, 975-1001.

10.1021/la051666f CCC: $30.25 © 2005 American Chemical Society Published on Web 10/08/2005

11252

Langmuir, Vol. 21, No. 24, 2005

by many groups. AFM has been used to measure nearmolecular contact between a planar surface and a sharp tip with a typical curvature radius of 5 to 50 nm13-16 or between a planar surface and a 2- to 10-µm-radius sphere attached to a cantilever tip.17-23 Contact between two orthogonal cylindrical surfaces, with radii of about 10 mm, has been measured with SFA.24-29 However, with only the measurement data obtained by AFM and SFA it is difficult to predict the actual binding force acting on the microscale flat surfaces of microparts because of the surface roughness of the parts, the surface history during parts fabrication, and individual differences in the surface characteristics of parts. In this article, we propose a method that enables measurement of the binding force between microscale flat surfaces in an aqueous solution. A feature of our method is that we use microparts themselves for the measurement so that we can determine the actual interactions between the microparts. Figure 1 shows a schematic illustration of the measurement method. Microfabricated cylindrical particles, which correspond to microparts in the selfassembly, were distributed over a planar substrate immersed in a solution. The particles gradually settled down onto the substrate by gravity, and then microscale contacts of two flat surfaces were made between the particles and the substrate. After that, a force-sensing piezoresistive microcantilever30 was used to push the particles in a horizontal direction to break the binding. The binding force was calculated from changes in the cantilever’s resistance caused by its deformation. This method enabled us to measure many bindings of the distributed particles directly and evaluate them statistically. We had two objectives in this work. The first objective was to appraise our measurement method. We used silicon dioxide (SiO2) surfaces in a hydrochloric acid (HCl) solution as the target of our measurement and examined the pH dependence of the interactions between the SiO2 microscale flat surfaces. According to the Derjaguin-LandauVerway-Overbeek (DLVO) theory,31 the interaction be(13) Hoh, J.; Cleveland, J.; Prater, C.; Revel, J.; Hansma, P. J. Am. Chem. Soc. 1992, 114, 4917-4918. (14) Weisenhorn, A.; Hansma, P.; Albercht, T.; Quate, C. Appl. Phys. Lett. 1989, 54, 2651-2653. (15) Tsukruk, V.; Bliznyuk, V. Langmuir 1998, 14, 446-455. (16) Dicke, C.; Hahner, G. J. Am. Chem. Soc. 2002, 124, 1261912625. (17) Ducker, W.; Senden, T.; Pashley, R. Nature 1991, 353, 239241. (18) Ducker, W.; Senden, T.; Pashley, R. Langmuir 1992, 8, 18311836. (19) Butt, H.; Jachke, M.; Ducker, W. Bioelectrochem. Bioenerg. 1995, 38, 191-201. (20) Kappl. M,; Butt, H. Part. Part. Syst. Charact. 2002, 19, 129143. (21) Li, Y.; Tao, N.; Pan, J.; Gracia, A.; Lindsay, S. Langmuir 1993, 9, 637-641. (22) Vezenov, D.; Noy, A.; Rozsnyai, L.; Lieber, C. J. Am. Chem. Soc. 1997, 119, 2006-2015. (23) Clear, S.; Nealey, P. J. Colloid Interface Sci. 1999, 213, 238250. (24) Israelachvili, J. Proc. Natl. Acad. Sci. U.S.A. 1987, 84, 47224724. (25) Horn, R.; Smith, D.; Haller, W. Chem. Phys. Lett. 1989, 162, 404-408. (26) Grabbe, A.; Horn, R. J. Colloid Interface Sci. 1993, 157, 375383. (27) Shubin, V.; Ke´kicheff, P. J. Colloid Interface Sci. 1993, 155, 108-123. (28) Chapel, J. Langmuir 1994, 10, 4237-4243. (29) Atkins, D.; Ke´kicheff, P.; Spalla, O. J. Colloid Interface Sci. 1997, 188, 234-237. (30) Gel, M.; Shimoyama, I. J. Micromech. Microeng. 2004, 14, 423428. (31) Verwey, E.; Overbeek, J. In Theory of the Stability of Lyophobic Colloids; Elsevier: Amsterdam, 1948.

Onoe et al.

Figure 1. Conceptual illustration of our proposed measurement method. (A, B) Cylindrical silicon microparticles were dispersed in an aqueous solution and settled onto a silicon substrate by gravity. Bindings between microscale flat surfaces in the solution were created between the particles and the substrate. (C, D) Binding strength was measured with a forcesensing piezoresistive cantilever. By pushing the particle horizontally with the cantilever, the binding between the particle and the substrate was broken. The force applied to the particle was measured by monitoring the change in the cantilever resistance.

tween small silica particles in an aqueous solution is explained by the interplay between the attractive van der Waals’ (VDW) force and the repulsive electrostatic doublelayer force. Because silicon dioxide resembles silica in its surface chemical properties, the interactive forces between SiO2 surfaces and their pH dependence in our measurement were expected to be similar to those between silica surfaces, which have been well examined with AFM and SFA.17,18,25,26,28,29 The second objective was to determine if the measured force should be categorized as a friction force or an adhesion force. In the case when a particle slid on a substrate in the measurement, the friction force between the particle and the substrate was measured. However, the adhesion force was measured when the particle came off the substrate. Whether the particle slid or came off depends on measurement conditions such as the particle size, the particle aspect ratio, and the height of the contact point between the particle and the cantilever. We propose a model of particle behavior when the particle is pushed by the cantilever in the measurement and have investigated appropriate experimental conditions that allow us to distinguish the friction force from the adhesion force. Materials and Methods Chemicals and Materials. We obtained hydrochloric acid (HCl, 35%), sulfuric acid (H2SO4, 96%), hydrogen peroxide (H2O2, 30%), acetone (99.5%), 2-propanol (IPA, 99.5%), and ethanol (99.5%) from Kanto Chemical and hydrofluoric acid (HF, 46%) from Morita Chemical Industries. All chemicals were used without further purification. Water was deionized to 18 MΩ cm with a Millipore purification system. Silicon-on-insulator (SOI) wafers were used for the fabrication of the cantilever and the cylindrical particles: a 1 µm SOI/1.5 µm buried oxide/450 µm handle silicon wafer (Soitec), a 3 µm/2 µm/450 µm wafer (ShinEtsu), a 5 µm/2 µm/450 µm wafer (ShinEtsu), and a 20 µm/1 µm/525 µm wafer (Mitsubishi)for the particle fabrication

Measurement between Microparticle and Substrate

Langmuir, Vol. 21, No. 24, 2005 11253 k)

Figure 2. (A) Whole view of the cantilever chip. (B) SEM photograph of the 300-nm-thick silicon cantilever (type B). An ultrathin piezoresistor (∼100 nm) was fabricated on the cantilever surface. Gold wiring was used for electrical connection. (C, D) SEM photographs of long (type-A) and short (typeB) cantilevers.

Figure 3. Design parameters for two-legged cantilevers. and 300 nm/400 nm/525 µm (Soitec) and 1500 nm/1000 nm/675 µm (Soitec) wafers for the cantilever fabrication. The 1500-nmthick SOI layer was thinned to 900 nm by ELM-D (20% tetramethylammonium-hydroxide (TMAH), Mitsubishi Gas Chemical) isotropic etching to form 900-nm-thick cantilevers. For microfabrication, OFPR800-20cp (Tokyo Ohka Kogyo) was used as a photoresist, OAP (Tokyo Ohka Kogyo) was used as a primer to improve adhesion between the photoresist and wafer, and NMD-3 (Tokyo Ohka Kogyo) was used as a developer. Force-Sensing Cantilevers. Force-sensing piezoresistive silicon cantilevers were two-legged shaped, and ultrathin piezoresistors (∼100 nm) were formed on the cantilever surfaces. Figure 2 shows photographs of the fabricated cantilever. Details of the fabrication process were reported in our previous work.30 We fabricated four types of cantilevers for our measurement. The design parameters are shown in Figure 3, and the dimensions and the characteristics are summarized in Table 1. By changing the length and thickness of the cantilever, we could easily change the spring constant from 1.0 to 140 pN/nm, which allowed us to measure a wide range of forces from the sub-nanonewton to the sub-micronewton level. The force applied to the cantilever tip, F, was calculated as

F ) k∆x )

k ∆R Sd R

(1)

where k is the spring constant of the cantilever, ∆x is the displacement of the cantilever tip, Sd is the displacement sensitivity, R is the cantilever resistance, and ∆R is the amount of change in the cantilever resistance when ∆x is applied. Here, Sd is defined as the ratio of resistance change against unit displacement, (∆R/R)/∆x. The spring constant, k, was calculated as32

(L1

3

Et3wb - L23)b + 3L23w

(2)

where E is the Young’s modulus of silicon, 190 × 109 N/m2, and L1, L2, b, and w are the design parameters of the cantilever shown in Figure 3. The displacement sensitivity, Sd, was measured by pushing the cantilever tip with a silicon sharp edge controlled by a piezostage.30 The values of k and Sd are also listed in Table 1. Figure 4 shows the responses of the type-B cantilever when a sharp needle controlled with a piezostage pushed the cantilever tip. A triangular wave was used to drive the piezostage. Peakto-peak displacements of 12 µm, 1.2 µm, and 120 nm were applied to confirm the response linearity and minimum force resolution. A semiconductor parameter analyzer (Hewlett-Packard, 4145B) monitored the cantilever resistance at the sampling frequency of 10 Hz. The force, F, was calculated using eq 1. As shown in Figure 4A and B, we obtained good linearity from 1.2- to 12-µm displacement. Because the maximum displacement in our measurement was adjusted to 10 µm, as we explain below in Measurement Method, the linearity of the cantilever was sufficient for our measurement. The force resolution was confirmed as shown in Figure 4C. The 120-nm displacement was equivalent to about 630 pN for the type-B cantilever. Thus, the force resolution was found to be on the sub-nanonewton scale. Microparticle Preparation. The microparticles were fabricated by etching a silicon layer of an SOI wafer and released from the wafer by etching a silicon dioxide sacrificial layer. Table 2 summarizes the design of the cylindrical silicon microparticles. We fabricated 11 types of particles, with the height, h, ranging from 1 to 20 µm and the diameter, d, ranging from 10 to 30 µm. The height of the particle was determined by the thickness of the SOI layer, and the diameter depended on the design of the photolithography mask. The aspect ratio of the particle, defined as h/d, was designed to be from 0.03 to 2.0. Figure 5A shows examples of particles with h/d values of 0.3, 1.0, and 2.0. The particle fabrication process was as follows. The SOI wafers (with a 1- to 20-µm SOI layer) were diced into 25 mm × 25 mm squares. A photoresist (OFPR800-20cp) was spin coated onto each wafer and patterned to form a masking layer. The silicon layer was then etched by reactive ion etching (A601E ICP-RIE etching system, Alcatel) to form the cylindrical particles. Figure 5B shows an SEM image of particles fabricated on the wafer after the photoresist was removed by IPA, acetone, and an ethanol rinse. The wafer chip on which the silicon particles were fabricated was dipped into hydrochloric acid for 5 to 15 min in a 1.5-µL centrifuged tube to release the particles. After the wafer chip was removed, the released particles were centrifuged at 12 000 rpm (Iwaki, CFM-200) and deposited at the bottom of the tube. After that, the HF was replaced by water: we removed the HF with a pipet, taking care to keep the particles at the bottom, and then added water to the tube. Agitation was carried out by soaking the tube in an ultrasonic cleaner (Iuchi, VS-150) for 3 to 5 min, and then the tube was put into the centrifuge to redeposit the particles. We repeated these procedures six times to completely replace the HF with water. After that, we replaced the water in the tube with piranha solution (3:1 H2SO4/H2O2) to modify the silicon surface of the particles to silicon dioxide (SiO2). The replacement procedure was the same as that for the HF-water replacement. The particles were immersed in the piranha solution for 12 to 24 h to modify their surfaces sufficiently. Subsequently, the piranha solution was replaced with water through the replacement

Table 1. Design and Characteristics of the Fabricated Force-Sensing Cantilevers type

L1 [µm]

L2 [µm]

b [µm]

w [µm]

t [nm]

R [kohm]a

Sd [µm-1]b

k [pN/nm]c

A B C D

350 200 350 200

280 170 280 170

30 30 30 30

7 7 7 7

300 300 900 900

13.5 16.9 21.0 9.14

2.2 × 10-4 8.1 × 10-4 8.3 × 10-4 22.4 × 10-4

1.0 5.3 27 140

a R: electrical resistance of the cantilever. b S : displacement sensitivity defined by (∆R/R)∆x. c k: spring constant obtained from a d theoretical calculation.

11254


Onoe et al.

Figure 4. Responses of the type-B cantilever. The tip of the cantilever was pushed by a small silicon fragment controlled with a piezostage. The piezostage motion was a triangular wave at 1 Hz. Responses for 12- and 1.2-µm peak-to-peak displacements are shown in A and B, respectively. The responses show good linearity in the range from 1.2 to 12 µm. The response for 120-nm peak-to-peak displacement is shown in C. The minimum force resolution was estimated to be on the sub-nanonewton level.

Figure 5. SEM images of the fabricated silicon microparticles. (A) Example of particles with different aspect ratios. The aspect ratio, h/d, of the left, the center, and the right particles were 0.3, 1.0, and 2.0, respectively. (B) Particles before release from a wafer. The diameter and height of the particles were 10 and 5 µm, respectively. (C) Particles after being released with hydrofluoric acid (HF).

Table 2. Design of the Cylindrical Silicon Microparticles type

d [µm]a

h [µm]a

h/da

S [µm2]b

a b c d e f g h i j k

30 30 20 20 10 10 30 20 15 12 10

1 3 3 5 3 5 20 20 20 20 20

0.03 0.10 0.15 0.25 0.30 0.50 0.67 1.00 1.33 1.67 2.00

707 707 314 314 78.5 78.5 707 314 177 113 78.5

a d, h, h/d: diameter, height, and aspect ratio of the cylindrical particle, respectively. b S: contact area between the particle and the substrate.

procedure again. Figure 5C shows the released particles after the surface modification. We stored the particles in water until we used them for the measurement. The surface roughness at the surface of the cylindrical particles was measured by AFM (Pacific Nanotechnology, Nano-R). Figure 6A shows an AFM image of a typical area (800 nm)2 on the top surface of the particle after the surface modification. The roughness average was 0.24 nm, and the highest asperity was 4.3 nm. Substrate Preparation. We used a diced 25 mm × 25 mm SOI wafer (5-µm SOI layer) as the substrate in our measurement. The substrate was rinsed with acetone and ethanol and dipped in piranha solution for more than 1 h to modify the surface to SiO2. The advancing water contact angle on the substrate surface was less than 10°, which ensured a high density of surface silanol groups. The surface roughness of the substrate was also measured by AFM. Figure 6B shows an AFM image of the substrate. In a typical area (800 nm)2, the average roughness was 0.14 nm, and the highest asperity was 3.0 nm. Measurement Setup. Figure 7A shows a schematic illustration of the measurement system. The substrate was clamped onto an XYZ stage, and a cell was made on the substrate for pooling an aqueous solution. The cell comprised 2-mm-thick rubber spacers and a 0.15-mm-thick glass cover. Water or HCl solution, whose pH was 2.0 to 6.0, was injected into the cell. The microparticles were dispersed in the solution to create bindings to the substrate. The cantilever chip was attached to an XYZ (32) Tortonese, M.; Barrett, R.; Quate, C. Appl. Phys. Lett. 1993, 62, 834-836.

Figure 6. Surface topographical AFM images of a particle and a substrate. The observation area was 800 nm × 800 nm for both images. The average roughness and the highest asperity were, respectively, 0.24 and 4.3 nm for the particle and 0.14 and 3.0 nm for the substrate. The height of the images was magnified 5 times. manipulator. The cantilever was carefully inserted into the solution as shown in Figure 7B. The setup was built on a vibration isolation system. The XYZ manipulator was equipped with a Z-direction piezostage to control the cantilever height from the substrate with a resolution of several nanometers. We could precisely adjust the height of the contact position between the cantilever tip and the microparticle. The XYZ stage was also equipped with a Y-direction piezostage to control the motion of the substrate. The collision velocity between the cantilever and the particle was adjusted with the piezostage. We used an optical microscope (Keyence, VH-5910) to observe the particles and the cantilever in the solution. The cantilever resistance was monitored with a semiconductor parameter analyzer (Hewlett-Packard, 4145B) at a sampling frequency of 10 Hz. Measurement Method. After inserting the cantilever into the cell, we had to align the cantilever with a target microparticle. Alignment in the X and Y directions was achieved by moving the


Langmuir, Vol. 21, No. 24, 2005 11255

Figure 9. Height control of the cantilever. (A) The cantilever was controlled with a piezostage so that it approached the substrate. (B) Contact between the cantilever tip and the substrate was detected by monitoring the cantilever resistance. (C) The height of the cantilever tip was adjusted accurately by measuring the moving distance of the cantilever with a position sensor of the piezostage.

Figure 7. (A) Schematic illustration of the experimental setup. An aqueous solution was pooled in a cell composed of rubber spacers and a glass cover. Particles were dispersed into the solution to create bindings between the particles and the substrate. The cantilever was then carefully inserted into the cell. An optical microscope was used to observe the particles and the cantilever in the solution. The cantilever height, which determined the contact position of the cantilever tip and the particle wall, was precisely controlled with a piezostage. The velocity of the collision between the cantilever tip and the particle was kept at 2 µm/s by another piezostage. The cantilever resistance was monitored with a semiconductor parameter analyzer during the measurement. (B) Photograph of the cantilever inserted into the cell.

Figure 8. Optical image (obtained with the optical microscope) of particles and a cantilever in a HCl solution. Two microparticles bound to the substrate were observed at the center and the right of the photograph. Black objects seen on the left were also microparticles laid sideways on the substrate. The substrate moved in the Y direction at 2 µm/s so that the particle would make contact with the cantilever for the measurement. The moving distance was 10 µm. The particle diameter and height were 10 and 5 µm, respectively (particle type f). The cantilever was type B. XYZ stage of the substrate. Figure 8 shows a photograph (observed with an optical microscope) of a cantilever aligned with a microparticle in a solution. After the XY alignment, we aligned the height of the cantilever as shown in Figure 9. We approached

the substrate with the cantilever by using the Z-direction piezostage. Contact between the cantilever and the substrate was detected by monitoring the cantilever resistance. After contact, the cantilever was lifted from the substrate to adjust the cantilever’s height, hcantilever. The height was measured with a position sensor of the piezostage. After finishing the alignment, we measured the binding force between the particle and the substrate. The substrate moved along the Y direction together with the particle bound to the substrate so that the particle came into contact with the cantilever. The velocity of the substrate motion was 2 µm/s, and the moving distance of the substrate was 10 µm. The particle and substrate moved toward the cantilever for 5 s and then returned to the original position. During the movement, the cantilever resistance was recorded.

Results and Discussion Typical Measurement Result. Figure 10 shows a typical measurement result when the particle is 10 µm in diameter and 5 µm in height. We used a type-B cantilever, and the solution was 0.01 M HCl (pH 2). The substrate velocity, which was equal to that of the particle, was 2 µm/s. The height of the cantilever from the substrate was set to 3.5 µm. The substrate started to move at 0 s. The particle touched the cantilever at a and then started to push the cantilever. The cantilever resistance decreased gradually as it was deformed. The binding between the particle and the substrate broke at b. The particle then came off the substrate, and the cantilever sprang back to its original shape. The particle bonded to the substrate again at c, and then the cantilever started to deform again. These breaking and rebinding processes happened a few times during the measurement. In the case of Figure 10, two breakings (at b and d) and two rebindings (at c and e) were observed. The substrate started to return to its original position at f. The binding force could be evaluated when the particle came off the substrate. In the case of Figure 10, particles came off at two points, b and d. Though the binding forces at these two points were almost the same, the binding strength usually differed for each breaking point in a measurement. We attributed the different binding strengths to differences in the conditions when each binding was created. The binding at b in Figure 10, which was measured primarily, was created when the particle randomly settled on the substrate in the solution by gravity. The value of the binding force at d was affected by other factors, though, such as the breaking and rebinding history and the loaded force of the cantilever. Therefore, we defined a primarily obtained force, which

11256


Onoe et al.

Figure 10. Typical measurement data for the type-f particle (d ) 10 µm and h ) 5 µm). A type-B cantilever was used. The upper trace shows the substrate motion that was controlled with a piezostage. The velocity was 2 µm/s, and the moving distance was 10 µm. The lower trace shows the cantilever resistance. The substrate started to move upward at 0 s. The particles bound to the substrate also moved together. The particle made contact with the tip of the cantilever at a, and the resistance started to decrease as the cantilever deformed. The binding was broken at b, and the particle was pushed back with the cantilever springing back into its original shape. The particle was bound to the substrate again at c. The processes observed at a-c occurred again at c-e. The substrate started to return at f. Measured force was defined as the force between a and b, which was found to be 5.2 nN.

Figure 11. Distribution of the measured force for particle f (d ) 10 µm and h ) 5 µm). The cantilever was type B. The number of measurements, N, was 78. The mean and standard deviation were 3.9 and 2.3 nN, respectively. The solid line shows the Poisson distribution calculated from the data. The data follow the solid line.

Figure 12. Dependence of the measured force on pH. The measured force vs the pH of the HCl solution is plotted. Bindings between the particles and the substrate were observed when the pH was 1 or 2, but the measured force became zero when the pH was 3, 4.3, or 6.0. The zero values mean that the particles did not bind to the substrate and just floated on the substrate.

was measured as 5.2 nN at b in the Figure 10 case, as the “measured force” to evaluate the binding between the particle and the substrate. We measured 78 particles from the particles dispersed in the solution. The measurement was done only once for each particle. The distribution of the measured force is shown in Figure 11. All of the measurement conditions were the same as described above. The mean value of the measured force was 3.9 nN with a standard deviation of 2.3 nN. The solid line shows a Poisson distribution curve calculated from the measured result. Although the Figure 11 distribution is wide, the measured force approximately followed a Poisson distribution. The values of the measured force were broadly distributed possibly because of individual differences in the surface conditions of the particles and the contact conditions between each particle and the substrate. From this point on, we will use the mean value and standard deviation of the measured force when discussing the measurement results. Dependence on pH. We examined the pH dependence of the bindings to verify our measurement method. We used type-b particles (d ) 10 µm and h ) 5 µm) and type-A and -B cantilevers. The substrate velocity was 2 µm/s, and the cantilever height was 2.5 µm. The pH was adjusted by diluting the HCl solution with water. We used a pH meter (Shindengen, KS723) to measure the solution pH. Measurements were made at pH values of 1.0, 2.0, 3.0, 4.3, and 6.0. The number of measured particles, N, was 25 for each measurement. Figure 12 shows the measured force versus the pH of the HCl solution. With the type-B cantilever, the binding

forces were 12 and 5.8 nN at pH 1.0 and 2.0, respectively. A greater measurement force was observed at pH 1.0 than at pH 2.0. Even when we used the type-A cantilever, which had the minimum force resolution, the measured force was zero at pH 3.0, 4.3, and 6.0. When the pH was 3.0 to 6.0, we observed that the particles on the substrate were slightly moved by a liquid flow generated by the evaporation of the HCl solution. Thus, in this pH range the bindings between the particles and the substrate were too weak to hold the particles at their positions against the liquid flow, or the bindings were not created. We think the change in the binding force that accompanied the change in pH was caused by the interplay of the attractive VDW force and the repulsive electrostatic double layer force, as was discussed with regard to DLVO theory31 for the behavior of colloid particles. Two closely approaching ( |F Badhesion|/ (2h/d), the particle rotated with the particle edge as the center of rotation and then came off the surface.

Figure 16. log-log graph of measured force per unit area, |F Bcantilever|/S, vs aspect ratio, h/d. The solid line in the graph was predicted by our model based on the particle behavior shown in Figure 15. For a low aspect ratio that satisfies eq 7, the value of |F Bcantilever|/S was expected to remain constant because |F Bcantilever|/S was not affected by h/d as shown in eq 3. For a high aspect ratio that satisfies eq 8, the value decreased as the aspect ratio increased, following eq 6. The slope of the decreasing line was -1. The intersection of the two solid lines indicates the border between the slide and jump phenomena.

equal to |F Bfriction| and starts to rotate when |F Bcantilever| reaches |F Badhesion|/(2h/d), as shown by eqs 3 and 6, respectively. Which phenomena will occur is determined by the relative values of |F Bfriction| and |F Badhesion|/(2h/d):

(slide) (rotate)

|F Bfriction|

1 |F B | 2h/d adhesion

(7) (8)

By using a low-aspect-ratio particle satisfying eq 7, we can measure the friction force, which equals the measured force as shown in eq 3. Alternatively, by using a highaspect-ratio particle satisfying eq 8, the adhesion force can be measured and expressed by the measured force and the aspect ratio of the particle as written in eq 6. Thus, if we use particles with various aspect ratios in the measurement, we would expect to find that the relationship between the measured force per unit area, |F Bcantilever|/S, and the aspect ratio, h/d, can be expressed in a log-log plot as shown in Figure 16. At a low aspect ratio, |F Bcantilever|/S remains constant because h/d does not influence the measured force. When the aspect ratio of the

particle is high, |F Bcantilever|/S decreases with an increase in the aspect ratio. The slope of the decreasing line is -1 because the dimension of h/d in eq 6 is -1. The intersection of these two lines in Figure 16 indicates the border between the slide and jump phenomena: in regions to the left and right of the border, the static friction force and adhesion force can be measured, respectively. Therefore, by plotting |F Bcantilever|/S against h/d, we can identify the border between sliding and jumping and separately examine the friction force and the adhesion force between the particle and the substrate. Discriminating between Friction and Adhesion. We experimentally confirmed our method of determining whether the measured force represents the friction or the adhesion force. The measured forces when we used particles with various aspect ratios were plotted in an |F Bcantilever|/S versus h/d graph and compared with the prediction shown in Figure 16. We prepared 11 types of microparticle whose aspect ratios ranged from 0.03 to 2.0 (Table 2). The measurement conditions were as follows. The substrate velocity was 2 µm/s, and the pH of the HCl solution was 2. To ensure proper contact between the cantilever tip and the particles, the height of the cantilever from the substrate, hcantilever, was adjusted to hcantilever ) h - 1.5 µm. For type-a particles only, hcantilever was adjusted to hcantilever ) h - 0.5 µm because the height of the type-a particles was only 1 µm. We defined the contact position between the cantilever tip and the particle, hcontact, as shown in Figure 17. Though we did not know the exact contact position because the cantilever tip and the particle wall overlapped in a certain region, the contact position was within the overlapped area shown in Figure 17. Thus, we defined the midpoint of the overlapped area as the contact position, hcontact:

1 hcontact ) (h + hcantilever) 2

(9)

In the measurement, we had to consider the ratio between the contact position and the particle diameter, hcontact/d, instead of the aspect ratio, h/d, used in the model. Figure 18 shows the log(|F Bcantilever|/S) versus log(hcontact/ d) graph for the 11 types of particle. Note that we used hcontact/d instead of h/d to plot this graph. The dots and the bars indicate the mean and standard deviation for N ) 25, respectively. We confirmed that the plotted data were divided into slide and jump phenomena: In Figure 18, the values of |F Bcantilever|/S were constant when hcontact/d was low and then decreased as hcontact/d rose, as we expected from Figure 16. The dashed line shows an approximate constant value, which is the mean value of the three dots for the lowest value of hcontact/d. The dashed line is written as

|F Bcantilever| ) 330 pN/µm2 S

(10)

11260


Figure 18. log-log graph of |F Bcantilever|/S vs hcontact/d for 11 types of particle. The dashed line was an average of three dots for the lowest hcontact/d values, written as |F Bcantilever|/S ) 330 pN/µm2. The solid line was a least-squares approximated line calculated using three dots for the highest hcontact/d values, expressed as |F Bcantilever|/S ) (hcontact/d)-1(49 pN/µm2). The value of hcontact/d at the intersection of the two lines was about 1.5. The plots agreed with our prediction (Figure 16): when hcontact/d was less than 0.1, the friction force was evaluated. When hcontact/d was greater than 0.4, the adhesion force was measured.

The solid line is an approximate decreasing line, which we calculated by the least-squares method using the three points for the highest hcontact/d values. Following eq 6, the solid line is expressed as

|F Bcantilever| ) S |F Badhesion| 1 1 ) 49 pN/µm2 (11) 2hcontact/d S hcontact/d Two plots for hcontact/d values of less than 0.075 agreed well with the dashed line. When hcontact/d was greater than 0.1, the plots gradually fell below the dashed line and then decreased along the solid line for hcontact/d of 0.4 to 2.0. The plots for hcontact/d greater than 0.4 agreed well with the approximated solid line. The value of hcontact/d at the intersection of the dashed and solid lines was 0.15. These experimental results were consistent with Figure 16, showing that our model explained the behavior of the particles in the measurement. Using the results, we can identify which force we measured, friction or adhesion. The intersection of the dashed line and the solid line at hcontact/d ) 0.15 in Figure 18 indicates the border between the slide and jump phenomena, which respectively correspond to friction and adhesion between the surfaces. The slide and jump phenomena were not clearly divided at hcontact/d ) 0.15, though, because either can occur near the border condition depending on differences in the individual surface properties of the particles. Thus, we consider the force being measured to be the friction force when hcontact/d is low (e.g., less than 0.1) and the adhesion force when hcontact/d is high (e.g., more than 0.4). The static friction force, |F Bfriction|, was about 230 nN between SiO2-SiO2 flat surfaces with a contact area of 707 µm2. Assuming that the static friction force was proportional to the contact area as mentioned above, we calculated that the static friction force per unit area, |F Bfriction|/S, was 330 pN/µm2 ( 20%. The adhesion force between the particles and the substrate was calculated using eq 6. Figure 19 shows the relationship between the adhesion force, |F Badhesion|, and the contact area, S. All of the data in Figure 19 were replotted using the data in Figure 18. The replotted data satisfied hcontact/d > 0.4, which is the condition used to measure the adhesion force. The adhesion force was proportional (R2 ) 0.94) to the contact area over the range of 78.5 to 707 µm2 as shown

Onoe et al.

Figure 19. Relationship between adhesion force, |F Badhesion|, and contact area, S. The adhesion force is proportional to the contact area.

in Figure 19. The calculated adhesion force per unit area, |F Badhesion|/S, was approximately 90 pN/µm2, and the error of the five plots was within (20%. Because the adhesion force is proportional to the contact area, we can estimate the adhesion force of microscale flat surfaces in an aqueous solution. In a general friction force measurement, a certain force is vertically loaded onto a probe surface, and the frictionload relationship is measured. For example, the load is about a millinewton-order force when the friction force is measured with SFA,34 and a nanonewton-order force is applied to the probe surface in friction force microscopy (FFM).34-36 With our method, it is difficult to load the force vertically onto the particle-substrate interface, so we did not examine the friction-load relationship. A feature of our method is that the friction force is measured under an extremely low vertical load. According to recent friction measurements,34,37 the friction force depends on two separate components, the adhesion and the load components. In our measurement, the adhesion component was the VDW interaction, which was measured to be ∼90 pN/µm2, and the load component was only the particle mass calculated to be 0.01-0.1 pN/µm2. The contribution of the particle mass is negligible compared to that of the VDW interaction between the surfaces. Thus, we consider that the adhesion-component-based friction force was mainly evaluated with our measurement and the static friction coefficient for the particle of S ) 707 µm2 was calculated to be |Ffriction|/|Fadhesion| ) 3.6. For our measurements, the contact between two surfaces was created by gravity-driven random collision. The surface-contact process resembles that of fluidic selfassembly because the microparts in fluidic self-assembly also make contact and bind to each other through random collision without any load. Therefore, our force measurement method is suitable for evaluating the binding of selfassembled structures and microfabricated objects operating in a liquid. Conclusions We have proposed a measurement method for directly examining the friction force and adhesion force between microscale flat surfaces in an aqueous environment. Binding between flat surfaces was created between microfabricated cylindrical particles and a planar substrate through gravity-driven random collision of the particles and the substrate. Piezoresistive force-sensing (34) Ruths, M.; Alcantar, N.; Israelachvili, J. J. Phys. Chem. B 2003, 107, 11149-11157. (35) Brewer, N.; Beake, B.; Leggett, G. Langmuir 2001, 17, 19701974. (36) Ruths, M. Langmuir 2003, 19, 6788-6795. (37) Bernam, A.; Drummond, C.; Israelachivili, J. Tribol. Lett. 1998, 4, 95-101.


microcantilevers, which allow us to measure sub-nanonewton to sub-micronewton forces, were used to push the particles horizontally to break the bindings for the measurement. As a measurement target, the bindings between microscale SiO2-SiO2 surfaces in a HCl solution were examined with our method. When the pH was changed, we observed changes in the binding force due to the interplay of the VDW force and the double-layer force, which has been widely investigated using AFM and SFA. We confirmed that the VDW force became dominant when the pH of the HCl solution was less than 2 and the doublelayer force acted between the surfaces when the pH was greater than 3. Our results agree with the findings of studies on the zeta potential using AFM and SFA, and this indicates that our method can correctly evaluate the interactions between flat surfaces in the solution. We could distinguish between the friction force and the adhesion force by changing the particle aspect ratio, h/d. We proposed a model of particle behavior during the measurement: the slide and jump model. When h/d was low, the slide phenomenon occurred, and the friction force between the surfaces was evaluated. In contrast, the jump phenomenon occurred for particles with a high h/d, and the adhesion force was measured. By using 11 types of particles with aspect ratios of 0.03 to 2.0, we experimentally examined the border value of hcontact/d that separated these slide and jump phenomena. We found that this value was about 0.15. We thus regarded the measured force as being the friction force when hcontact/d was less than 0.1 and the adhesion force when hcontact/d was greater than 0.4. The friction force was approximately 230 nN for 708µm2 SiO2-SiO2 flat surfaces in a pH 2 HCl solution, so the friction force per unit area was 330 pN/µm2 ( 20%. The adhesion force was proportional to the contact area, and the adhesion force per unit area was found to be 90 pN/ µm2 ( 20%. Thus, we confirmed that our model appropriately represented the particle behavior during the measurement, and we were able to directly measure the friction force and adhesion force between microscale flat surfaces in an aqueous environment. Our measurement method, which enables us to measure friction and adhesion forces between microscale flat

Langmuir, Vol. 21, No. 24, 2005 11261

surfaces in an aqueous solution, will be an effective tool for studying and developing fluidic microsystems that utilize actual microscale surface contacts in a liquid, especially when a theoretical prediction of the binding force is difficult. Though our measurement was carried out mainly at pH 2, it is applicable to the measurement not only in the acidic pH range but also in the basic pH range, which are also interesting ranges in colloid and surface science. One of the potential applications of this method is to measure the interactions between surfaces modified with self-assembled monolayers (SAMs). SAMs were frequently used for the surface modification of microdevices to improve their chemical and tribological properties.38,39 The measurement of friction and adhesion between SAM-modified microscale surfaces using our method will certainly contribute to these fields. For measurements to support advanced studies using this method, such as those concerning the relationship between the measured force and surface roughness, the contact conditions between the particle and the substrate must be investigated more precisely. The contact conditions are particularly essential to the discussion of the link between the adhesion force and the thermodynamic free energy of adhesion that is expressed in mJ/m2. By combining our measurement method with surface-contact analysis techniques such as infrared radiation (IR) interference imaging, which is used to examine the bonding conditions of silicon wafers,40 we hope to uncover a more concrete relationship between surface conditions and the binding between microscale flat surfaces. Acknowledgment. This research was supported by the Program of Promotion of Basic Research Activities for Innovative Biosciences (PROBRAIN). The photolithography masks were fabricated using the EB lithography apparatus of the VLSI Design and Education Center (VDEC) of the University of Tokyo. LA051666F (38) Whitesides, G.; Ostuni, E.; Takayama, S.; Jiang, X.; Ingber, D. Annu. Rev. Biomed. Eng. 2001, 3, 335-373. (39) Tsukruk, V. Adv. Mater. 2001, 13, 95-108. (40) Miki, N.; Zhang, X.; Khanna, R.; Ayon, A.; Ward, D.; Spearing, S. Sens. Actuators, A 2003, 103, 194-201.

Direct Measurement of the Binding Force between Microfabricated

Recommend Documents