Solvent Driven Motion of Lithographically Fabricated Gels - American

Sep 30, 2008 - Noy Bassik, Beza T. Abebe, and David H. Gracias*. Department of Chemical and Biomolecular Engineering, Johns Hopkins UniVersity,...
0 downloads 0 Views 1MB Size
12158

Langmuir 2008, 24, 12158-12163

Solvent Driven Motion of Lithographically Fabricated Gels Noy Bassik, Beza T. Abebe, and David H. Gracias* Department of Chemical and Biomolecular Engineering, Johns Hopkins UniVersity, 3400 North Charles Street, Baltimore, Maryland 21218 ReceiVed April 28, 2008. ReVised Manuscript ReceiVed July 9, 2008 We investigated the solvent driven motion of lithographically structured poly-N-isopropylacrylamide (PNIPAm) gels. The gels were soaked in ethanol and then transferred to water, where they moved spontaneously. This movement was driven by the expulsion of the ethanol from the gel and subsequent ethanol spreading at the air-water interface. We utilized lithographic patterning of the gels at the micron-millimeter length scales to investigate the effect of size, shape and symmetry. Lithographic patterning allowed the structures to be fabricated in an identical manner, thereby allowing a single variable (such as shape, size, or symmetry) to be altered while minimizing change in other variables such as thickness, roughness and swelling characteristics. The diverse range of motions including translation, precession and rotation could be controlled and were recorded using videography. Gels were lithographically patterned with features less than 100 µm, and exhibited remarkably high linear and rotational velocities of up to 31 cm/s and 3529 rpm over time spans of seconds to minutes. We observed a reciprocal dependence of maximum rotational velocity on linear dimension. The linear velocity for all types of motion, along a line or curve, was analyzed and found to be similar across different shapes and sizes. This velocity was in the range of 17-39 cm/s even though our sizes and shapes varied across orders of magnitude. We postulate that this velocity is related to the velocity of spreading of ethanol on water, which is approximately 53 cm/s. Additionally, since this solvent powered motion is a clean, quiet and reusable source of motive power, with no need for on-board wiring or batteries, we explored applications in moving lithographically integrated metallic payloads on top of the gels and utilized the gels to move larger floating objects.

Introduction In order to fabricate miniaturized dynamic devices such as rotors, mixers and locomotors it is necessary to design strategies to move structured objects and systems in fluidic media, in predefined ways. While it is straightforward to power these systems on the macroscale using batteries or wireless antennas, it is increasingly difficult and cost prohibitive to power analogous systems, especially tetherless ones, on the mm and submm scale. There are a few limited examples of chemically (as an alternative to electrically) driven autonomous movement;1 including camphor boats2 in water and platinum coated particles in peroxide. Here, we study a less frequently explored mechanism, solvent driven motion. Osada and colleagues first described the spontaneous motion of millimeter to centimeter scale pieces of amphoteric gels, composed of acrylic acid and its derivates, based on the intermixing of two solvents.3 The authors utilized structures that were shaped by hand in the size range of several mm to cm. They studied structures in the shape of regular polygons and observed rotational motion with a maximum of 350-400 rpm and translational motion with a maximum speed of 5 cm/s. In a later paper,4 Osada and colleagues also studied smaller submm sized gels fabricated by emulsion polymerization; these structures were spherical, rather than the disk shaped structures utilized in the earlier study. To date, controlled gel based motion with patterned submm features, allowing for diverse shapes, has not been demonstrated and compared across size scales. Another limitation * To whom correspondence should be addressed. E-mail: dgracias@ jhu.edu. (1) Paxton, W. F.; Sundararajan, S.; Mallouk, T. E.; Sen, A. Angew. Chem., Int. Ed. 2006, 45(33), 5420–5429. (2) Nakata, S.; Doi, Y.; Kitahata, H. J. Colloid Interface Sci. 2004, 279(2), 503–508. (3) Osada, Y.; Gong, J. P.; Uchida, M.; Isogai, N. Jpn. J. Appl. Phys 1995, 34, L511–L512. (4) Gong, J. P.; Matsumoto, S.; Uchida, M.; Isogai, N.; Osada, Y. J. Phys. Chem. 1996, 100(26), 11092–11097.

of the prior studies was that the structures were fabricated by hand; hence, it was difficult to study or apply this phenomenon to precisely shaped and mass producible mm scale objects. In order to explore the utility of this gel based motion on the mm scale, we developed a system that allows for autonomous motion in a photopatternable PNIPAm gel (as opposed to the acrylic acid gel used by Osada). Photopatterning of polymers, e.g., photoresists, is widely utilized to fabricate microscale structures. Patterning of hydrogels, however, is challenging since these materials tend to be soft and wet (limiting processing time and precluding contact lithography). Additionally, hydrogel size can change during processing, based on the degree of hydration of the gel. Drawing inspiration from published strategies to lithographically pattern PNIPAm,5-7 we developed our own lithographic process. Our recipe utilized prepolymerized high molecular weight PNIPAm along with the N-isopropylacrylamide monomer to allow for photopatterning with high resolution and production of a patterned gel with sufficient strength necessary for rapid motion and handling. Using lithography, it was possible to create arbitrarily shaped PNIPAm structures with features as small as 100 µm. Thus, we were able to systematically study, for the first time, the motion of precisely structured gels and the dependence of their motion on size, shape and symmetry. After photopatterning, the gels were soaked in ethanol and then immersed in water; the ensuing motion was recorded with a video camera. In this paper, we focus on the motion of shapes with interesting, reproducible, and novel behavior. We discuss the effect of size on the spin speed for symmetric Y shaped gels. Additionally, in order to explore applications of this solvent powered motion, we integrated metallic structures with the gels and demonstrated autonomous motion of the integrated structures. (5) Ito, Y. J. Intell. Mater. Syst. Struct. 1999, 10(7), 541. (6) Hoffmann, J.; Plo¨tner, M.; Kuckling, D.; Fischer, W. J. Sens. Actuators A: Phys. 1999, 77(2), 139–144. (7) Ito, Y.; Chen, G.; Guan, Y. Q.; Imanishi, Y. Langmuir 1997, 13(10), 2756– 2759.

10.1021/la801329g CCC: $40.75  2008 American Chemical Society Published on Web 09/30/2008

Motion of Lithographically Fabricated Gels

Langmuir, Vol. 24, No. 21, 2008 12159

We also show that the small lithographically patterned gel shapes could move much larger objects.

Experimental Section We first prepared a PNIPAm solution which consisted of 3 g NIPAm monomer (Scientific Polymer Products Inc.), 0.4 g polyN-isopropylacrylamide (PNIPAm, 300k molecular weight, Scientific Polymer Products Inc.) and 0.18 g BIS-Acrylamide (N,N′-Methylenebis-Acrylamide) (Aldrich), dissolved in 7.5 mL of n-butanol (Sigma). The solution was vortexed to aid dissolution of the PNIPAm and then decanted (after settling) to remove any insoluble crystals. Five microliters of 0.17 g/mL crystal violet/ethanol (Sigma) were added to the NIPAm/PNIPAm solution, for visualization during handling. One hundred microliters of the photoinitiator, Irgacure 2100 (Ciba), was added to the solution just prior to photolithographic patterning. Thin films of PNIPAm were prepared by spin coating the solution onto cleaned silicon or glass substrates at 75-200 rpm. Thicker samples were fabricated by placing 0.75-1.2 mL of NIPAm/PNIPAm solution on the substrate and gently rotating the substrate by hand to level the liquid layer. The photopolymerization was performed in noncontact mode, using a transparency mask (Fineline Imaging, Colorado Springs, CO) and a commercial mask aligner (Quintel). The solution was exposed at energies of approximately 70 mJ/cm2 at 365 nm; the optimum exposure varied depending on sample thickness and photoinitiator concentration. After exposure, the films were rinsed in ethanol to remove any unexposed NIPAm/PNIPAm. The photopatterned shapes were then lifted off in a swollen state using either ethanol or water. In our experiments comparing the rate of motion of different sized gels, all the structures were fabricated simultaneously on a single wafer. For samples spun at 75 rpm, the thickness of the shapes was measured by profilometry and was approximately 40 µm when dry. We patterned shapes with metallic features by first creating the shapes as described above. After drying the gels, a 100 nm layer of chromium (Cr) and 300 nm layer of copper (Cu) were thermally evaporated onto them at a pressure of ∼10-5 Torr. An approximately 3 µm thick layer of a novolac resin based photoresist, Shipley 1827 (Microchem), was spun on top of the metal, and was patterned using lithography. After developing the photoresist, the Cu and Cr layers were etched using commercial wet etchants APS-100 and CRE-473 (Transene), respectively. We minimized exposure of the gels to aqueous media during lithographic patterning of the metal, and no swelling was observed during this period. The PNIPAm shapes with metallic patterns were lifted off in ethanol after metal etching. We also embedded lithographically fabricated, 500 µm cubes into the photopatterned PNIPAm gel. Assembly of the 500 µm cubes (metallic containers) is described elsewhere.8 We achieved the integration by first placing the cubes on a bare glass or silicon substrate and subsequently wetting them with the liquid photopatternable NIPAm/PNIPAm solution. The solution was then patterned as described earlier. For larger payloads we utilized nylon mesh (Small Parts) and 24 mm of steel wire. In all our studies of spontaneous motion described here, PNIPAm gels were soaked in ethanol for a minimum of 10 min before introduction into water to ensure proper solvent exchange. However, in general motion can be observed in water with gels soaked in ethanol for time periods as brief as one minute. A thin nylon mesh was used to lift the gels out of organic solution and place them into water, this is visible in supplementary video 1. Absorbent wipes were used to carefully blot away the excess organic solvent. For visualization purposes, the optical contrast was increased by soaking the gel in 0.17 g/mL crystal violet/ethanol prior to imaging. Low speed videography was performed on a commercial digital camera [Canon S3] at 60 frames/s; high speed images were obtained with a CMOS camera at rates of 500-2000 frames/s. We recorded the motion of the gels and then measured the angular and linear velocity by analyzing individual successive video frames. Figure S1 shows (8) Leong, T. G.; Lester, P. A.; Koh, T. L.; Call, E. K.; Gracias, D. H. Langmuir 2007, 23(17), 8747–8751.

Figure 1. Optical microscopy images of lithographically patterned PNIPAm gels. (A) The smallest resolvable features were less than 100 µm. We were also able to pattern a variety of shapes including (A) rectangles and (B) symmetric Y shapes in a highly parallel manner. Using lithographic patterning, one variable, such as size, could be altered while minimizing the change in other variables, such as thickness. All structures of a particular shape could be fabricated at once on a single wafer, using identical processing conditions.

captured images from high speed imaging of a spinning symmetric Y gel used to calculate rotational velocity.

Results and Discussion Shown in Figure 1 are PNIPAm structures patterned lithographically at the wafer scale. The main limiter of resolution was the requirement of noncontact lithography, since the gels had to remain wet. Our lithographic recipe required the polymer to remain in the wet state during photopolymerization, because the polymer crystallized on drying and was not photopatternable. Since the NIPAm/PNIPAm film was wet, noncontact lithography was necessary to avoid mechanical distortion and tearing of the film during lithography. In our system, we utilized mechanical spacers (roughly 1 mm thick) to separate the mask and the film during exposure. We were able to resolve features of approximately 100 µm using our process. After patterning, exposure to solvent (either water or organic) resulted in swelling and buckling of the films. Compressive forces resulting from this volume change due to swelling lifted off the patterned gels from the substrates. After lift-off, linear dimensions (length and width for a rectangular shaped gel) increased approximately 38% in room temperature water and 36% in ethanol. After soaking in ethanol and upon transferring to water, the gels moved spontaneously. This motion can be rationalized as follows:9 the gel has both a hydrophobic side chain (isopropyl) and a hydrophilic core (acrylamide). In ethanol, both the hydrophobic side chains and hydrophilic core can interact with the organic solvent. In water, on the other hand, the preferred molecular orientation leaves side chains hidden and the core exposed. Hence, when the ethanol soaked gel is placed in water (Figure 2A), and the water and ethanol mix readily, the polymer molecules reorient to shield their hydrophobic side chains. This reorientation begins at the surface and moves toward the center of the gel, resulting in contraction of the gel and expulsion of ethanol from the interior of the gel (Figure 2B).10,11 We observed shrinkage of the gel, indicating expulsion of ethanol, before it swelled again with water. Since ethanol is less dense and has a lower surface energy as compared to water, it spreads at the (9) Mitsumata, T.; Gong, J. P.; Osada, Y. Polym. AdV. Technol. 2001, 12(1-2), 136–150. (10) Li, Y.; Tanaka, T. Annu. ReV. Mater. Sci. 1992, 22(1), 243–277. (11) Dagani, R. Chem. Eng. News 1997, 75, 26–28.

12160 Langmuir, Vol. 24, No. 21, 2008

Bassik et al.

Figure 3. Different motions observed with phopatterned PNIPAm gels of different shapes. Still images from a video are shown. (A) 2:1 rectangular gel (convex) shape translating in a straight line with velocity of 31 cm/s. (B) A trapezoidal (convex) shape with aspect ratio of 1:1 precessing with angular velocity of 225 rpm and radius of rotation ∼15 mm. (C) A 3:1 oval (convex) shape spinning with angular velocity of approximately 900 rpm. (D) A symmetric Y (concave) shape spinner with an angular velocity of approximately 1400 rpm. Measurements for parts C and D were limited by the speed of the camera at 60 frames/sec; high speed camera studies were used to confirm velocities. Still images shown here are captured from supplemental video 1. Figure 2. Conceptual schematic diagram illustrating the organic solvent driven motion of an oval shaped PNIPAm gel. Dark red circles represent organic solvents (ethanol) and the light blue is water. (A) A PNIPAm shape soaked in ethanol is placed into water. (B) Contraction of the gel begins from the surface toward the interior due to unfavorable hydrophobic interactions with water; ethanol is expelled from the gel. (C) The ethanol spreads on the surface of the water due to its lower density, lower surface tension, and miscibility with water. The spreading of the ethanol causes a reactive pushing force on the polymer shape in the opposite direction to the expulsion of ethanol. The variation in forces across the gel cause it to move in a specific direction. In this schematic, more ethanol appeared on the right side of the gel due to asymmetric initial conditions. This resulted in a net force pushing the gel to the left. The gel contracts as it undergoes a phase change and ethanol is expelled. (D) Water diffuses into the contracted gel, while the supply of ethanol is exhausted; the gel swells again with water and the motion ceases.

air-water interface.12,13 This spreading of ethanol at the airwater interface pushes the gel and causes it to move (Figure 2C). We also soaked the gel in several other commonly used organic solvents and studied the resulting motion in water. It is essential that the solvent swells the gel appreciably in order for the gel to soak up sufficient quantities of the solvent and allow for prolonged motion. While doing so, it is also important that the solvent leaves the gel undamaged. The solvent also needs to be less dense than water and have a low surface energy to spread at the air-water interface. Alcohols and small ketones had the highest levels of motion,13 while solvents such as octanol, chloroform, benzyl alcohol and xylene produced no motion. We observed that solvents such as ethanol and tetrahydrofuran that are both oil and water soluble worked best. We were able to control the type of PNIPAm gel motion by photopatterning specific symmetry, shape and size. These motions were observed for time periods as long as a minute for subcm scale shapes, but they were often limited by collisions with the walls of the dish. The important factors that were observed to (12) Suciu, D. G.; Smigelschi, O.; Ruckenstein, E. J. Colloid Interface Sci. 1970, 33, 520–528. (13) Davies, J. T.; Rideal, E. K. Interfacial Phenomena; Academic: New York, 1963; p 344.

control the type of motion observed (symmetry, aspect ratio, and concavity) are discussed below. Effect of Symmetry. We observed that symmetry was important in determining the type of motion observed. Rotationally symmetric shapes tended to precess or rotate, while asymmetric ones translated (until a critical aspect ratio was reached). Symmetric Y shaped gels (Figure 1B) spun (Figures 3D and S1), but when symmetry was broken on such a gel, chaotic motion resulted. This feature was demonstrated by adding a small bulge to only one of the ends of a Y shape (via an alternate photomask design, not shown) to render it asymmetrical. All symmetrical Y’s across size scales spun while all asymmetric Y’s experienced chaotic translational and orbital motion. Effect of Aspect Ratio. Another important factor that determined the type of motion for simple shapes was the length: width aspect ratio (in the plane of the liquid surface). Aspect ratio was varied by patterning polygons such as 1:1 squares, 2:1 rectangles and 3:1 oval shapes. A 2:1 rectangular shape translated along its longer side while a roughly 1:1 trapezoidal shape (Figure 3B) and square (not shown in Figure 3) precessed around a central point. The motion of the 2:1 rectangle in the direction perpendicular to its longer side can be rationalized by noting that the surface of the rectangle with the longer side experiences a larger pushing force (as compared to the side with the shorter length) due to its larger surface area. When the aspect ratio was increased further, rotary motion was observed, e.g. a 3:1 oval shape rotated (Figure 3C). The four types of motion are demonstrated in supplemental video 1, which features the gels from Figure 3. Effect of Concavity. A convex planar polygon is defined as a polygon that contains all lines connecting any two of its vertices; hence a rectangle is considered to be a convex polygon.14 Alternatively, polygons that do not contain all lines that join any two of its vertices are concave; these polygons tend to be indented, (14) Weisstein, E. W. “Conxex Polygon” From Math World-A Wolfram Web Resource. http://mathworld.wolfram.com/ConvexPolygon.html (accessed Apr 2008).

Motion of Lithographically Fabricated Gels

Langmuir, Vol. 24, No. 21, 2008 12161

Figure 4. Variation of velocity with gel size for symmetric Y (concave) shaped gels. (A) Angular velocity as a function of time across four sizes, 1.6-12.7 mm. The shapes were imaged with a high speed camera at 500 or 2000 frames/s for 4 or 8 s. Each line shows the linear fit for one gel placed into water. (B) The maximum velocity observed from any experiment plotted vs size of each spinner gel. Supplemental video 2 shows clips from the high speed videography of spinning gels used to calculate this data.

such as the Y shaped polygons in Figure 1B. Concave shapes such as symmetric Y shapes, tended to produce rotary motion, with observed speeds of 3500 rpm for fractions of a second, while convex shapes produced translational or precessional motion. As described in earlier studies with camphor boats,15 initial conditions are also important and dictate transitions from translational to rotational motion. Hence, if slightly more ethanol appears on one side of the shape due to initial conditions, that side experiences more of a reactive force pushing back from the spreading ethanol. As the shape accelerates, positive feedback occurs since more ethanol is released and rises on that side. This process is illustrated in Figure 2B-C. By photopatterning a symmetric Y shape, this positive feedback occurs at three rotor wings linked to a common point of rotation. Each wing experiences a pushing force from the rising and spreading organic solvent. The spreading is channeled into high rotational velocities, and spinning occurs around the fixed center of mass (Figure 3D). This spinning is evident in supplementary video 2, taken with a high speed camera. By patterning convex shaped objects with a roughly 1:1 aspect ratio, we produced a precessing motion or rotation around a fixed point (that did not pass through the center of mass of the object), with a reproducible orbital radius on the order of the linear size of the gel shape itself (Figure 3B). Effect of Size. We chose to study the effect of size with the symmetric Y shaped gels as they spun in place. These gels could be imaged at high resolution with a microscope; this imaging is difficult to achieve for rapidly translating objects. We were also able to image these symmetric Y shaped gels in the center of a relatively large dish, thereby minimizing any edge effects from the walls of the dish. We fabricated several photopatterned symmetric Y shaped gels on a single wafer with edge lengths ranging from 413-3317 µm and effective diameters ranging from 1.6 to 12.7 mm (Figure 1B). In general, larger sizes moved more slowly than similarly shaped smaller structures, but for longer duration. Larger shapes absorbed more organic solvent and released it over longer periods, allowing for motion on the order of 10 min (for a 2 cm shape). For shapes that would nominally translate, we found that when the shape was larger (cm scale) the motion became more rotational. As we kept the thickness constant, we observed that large shapes buckled when rapidly translating or rotating; this buckling can be attributed to the large ratio between their width/length (2 cm)

to their thickness (40 µm), a ratio of 500:1. This buckling caused instabilities that led to asymmetric ethanol spreading and chaotic motion. Figure 4A shows the measured angular velocity plotted as a function of time for symmetric Y shaped gels across four sizes. The angular velocity decreased linearly with time; the explanation for this observation is discussed later in the paper. The rate of decrease (slope of Figure 4A, given in rpm/s) was roughly the same for large gel shapes at 15 rpm/s. For smaller gels (such as 1.6 mm diameter) the rate of rotation decreased by 315 rpm/s, much more quickly than for the 6 or 12 mm sized gels. Figure 4B shows maximum observed velocities of each symmetric Y shaped gel as a function of its size. We observed that the maximum angular velocity varied as the reciprocal of the diameter of the gel which implies that the maximum linear velocity is constant across sizes. We observed that in addition to being constant the maximum linear tip velocity of the shapes along a line or circle was comparable to the spreading velocity of ethanol on water, and that this was the common parameter among different shapes, motions, and experiments. Comparison to Ethanol Spreading Experiments. Several investigations have explored the spreading of organic solvents on an aqueous surface; in these experiments organic solvent was introduced via continuous feed from a capillary.12,16 The solvent spreading occurs as a result of Marangoni convection, driven by a surface tension gradients due to concentration differences. In our system, ethanol expulsion from the gel is required to observe the unique types of motion. For example, when ethanol was pipetted near a water swollen gel shape, the gel always moved away from the pipet, but no complex motion such as rotation was ever observed. We observed one remarkable similarity between ethanol spreading experiments and the movement of our gels. It is known from ethanol spreading experiments in water,13 that the ethanol has an initial velocity of approximately 53 cm/s close to a concentrated source. This value is slightly greater than that of the initial linear velocities measured in all our moving gel shapes (in the range of 17-39 cm/s) for widely varying shapes, as well as sizes of rotating gel shapes ranging from 1.6 to 12.7 mm in diameter (Table 1). Figure S2 shows the geometry used to calculate initial linear velocities for comparison to ethanol spreading. Since by definition the angular velocity ω ) V/r, it then implies that the angular velocity should scale inversely with gel diameter, which is what we have observed (Figure 4B). Regarding the linear deceleration of the rotating gels, the motion of the gel is dictated by a balance between the torque on the gel

(15) Kitahata, H.; Hiromatsu, S.; Doi, Y.; Nakata, S.; Islam, M. R. Phys. Chem. Chem. Phys. 2004, 6(9), 2409–2414.

(16) Suciu, D. G.; Smigelschi, O.; Ruckenstein, E. AIChE J. 1967, 13(6), 1120–1124.

12162 Langmuir, Vol. 24, No. 21, 2008

Bassik et al.

Table 1. Angular and Linear Tip Velocities of Various Photopatterned PNIPAm Gelsa

gel shape

motion

dimensions/diameter on mask (mm)

RPM (ω)

diameter of rotation (mm)

time at which velocity measured (s)

velocity (VT) (cm/s)

225

15

2.7

17.7

trapezoidal [Figure 3B]

precess

4.0 × 2.35

circle w/ Cu stripes [Figure 5B]

precess

1.4

360

9.2

3.3

17.3

2:1 rectangle [Figure 3A]

translate

3.0 × 6.0

n/a

n/a

1.3

31

symmetric Y-shaped gels [Figures 1B, 3D, and 4]

spin in place

1.6 3.2 6.3 12.7

3529 1500 1081 475

spin in place

initial velocity after placement

31.5 26.9 38.6 33.9

a For translating or precessing motions velocities were measured after stable motion was achieved. This time at which the velocity was measured is also noted. The linear tip velocity is similar across different sizes and shapes and correlates well with the velocity of spreading of ethanol on water.

Figure 5. Lithographic integration of metallic structures with movable PNIPAm gels, and the motion of larger structures propelled by the PNIPAm gels. (A) Lithographically integrated PNIPAm gel with Cu metallic spirals, at rest. (B) The ethanol soaked metallic-gel structure consisting of a circular gel with Cu stripes moving autonomously in a controlled manner in water. The gel prescesses around a central point at 360 rpm in a circle with 9.2 mm diameter. Inset: gel at rest. (C) A metallic microcontainer embedded during lithographic patterning within photopatterned PNIPAm gel. The structure translates in water (not shown). (D) Photopatterned PNIPAm gels propelling a metallic wire in a larger nylon mesh in a circular manner. Images in panels B and D are composites from multiple video frames from supplementary video 3, combined with the “Darken” algorithm (Adobe Photoshop).

due to Marangoni convection and the viscous drag torque. We expect maximum linear velocities of all gel shapes to be lower than pure ethanol (Table 1) due to drag forces. The ethanol spreading velocity V is a function of local surface tension γ and ethanol concentration c.12

V∝

dγ dγ dc dγ ) ) ∆c dr dc dr dc

(1)

The ethanol concentration gradients which are necessary for gel motion are affected by the volume of ethanol released from the gel, the geometrically dependent release profile, and time. The type of motion, scaling effects and rotational deceleration are controlled to a large extent by the dynamics of the ethanol/water exchange within the gel, which determine the quantity of ethanol released as well as local concentration gradients. While expressions for the drag force on a spinning disk in a viscous fluid have

been developed,17 the dynamics of water/ethanol mixing with PNIPAm are complex and less well understood.18,19 However, based on the equation (1) we can see that the spreading velocity depends on the ethanol concentration and the surface tension gradient. The local concentration gradients are affected by molecular restructuring (visible as opacity changes) and shrinkage of the gel that occur during its motion. Moreover, it is known that dγ/dc is itself a complicated function of concentration20 and time,21 making it difficult to develop an exact analytical expression for the deceleration of the spinning gels. Effect of Surfactant. When surfactant (a drop of 1% TritonX-100) was added to the water, no motion was observed. Just as in situations with no surfactant, the gel’s opacity changed from transparent to opaque white and then back to transparent as the ethanol was released and replaced with water. This is a well-known phase transition that occurs in PNIPAm gels.10 However, we believe that the surfactant prevented gel motion, by impeding the surface spreading of ethanol. Nonetheless, ethanol diffusion from the gel into water was observed. Surfactants have known surface tension lowering effects, which remove the differential surface tension gradient between the water-rich regions far from the gel and ethanol-rich regions near the gel. Temperature of Water. PNIPAm in water is also known to collapse above the lower critical solution temperature temperature (LCST),10 typically in the range of 30-40 °C. The photopatterned gels undergo a phase transition in water at an LCST of ∼35 °C, as measured by heating gels on a hot plate. We investigated the effects of placing ethanol soaked gels into water at different temperatures. When placed into 40 °C water (above the LCST), vigorous motion was observed, and as the gel was then immediately transferred into cold 3 °C water (below the LCST), this motion ceased. In the opposite case, much less motion was observed when an ethanol soaked gel was placed in cold water below the LCST, but the rapid motion commenced if the shape was subsequently moved to warm water above the LCST. Therefore, solvent powered motion was enhanced in water above the LCST, since the gel collapsed at those temperatures, expelling ethanol more readily. On the contrary, below the LCST, ethanol exited more slowly resulting in slower gel motion. pH of Water. We did not observe any significant changes in the motion of PNIPAm structures when the pH of the aqueous solution was changed in the range of 4-10. This observation is consistent PNIPAm’s known temperature sensitivity but pH insensitivity.22 (17) Nowinski, J. L. Ing. ArchiV 1984, 54(4), 291–300. (18) Costa, R. O. R.; Freitas, R. F. S. Polymer 2002, 43(22), 5879–5885. (19) Lele, A. K.; Karode, S. K.; Badiger, M. V.; Mashelkar, R. A. J. Chem. Phys. 1997, 107(6), 2142–2148. (20) Bircumshaw, L. L. J. Chem. Soc. 1922, 121, 887–891. (21) Joos, P.; Serrien, G. J. Colloid Interface Sci. 1989, 127(1), 97–103.

Motion of Lithographically Fabricated Gels

Salt Concentration. When placed into water with increasing concentrations of salt, a gel such as a 2:1 rectangle that normally translated for under a minute with a smooth linear motion now had a jerky translational motion, which was prolonged and lasted for more than 2 min. Symmetric Y shaped gels showed a decrease in rotational velocity and then became jerky in a linear manner. PNIPAm gels would remain motionless for several seconds, and then rapidly translate a fixed distance with a small rebound. Initial tests were conducted with several salt solutions (NaCl, KI, KCl); differences were observed between the salt concentration required for oscillatory motion and the extent of change from normal motion in pure water. While it is expected that amplitude of motion would increase due to the increase in surface tension of water with salt,4,23 the jerky motion has not been described before. Based on previous reports that suggest that excess ions lower the LCST of gels in a concentration and species dependent manner, we believe there is an oscillating shift between collapse of the gel as salt enters the network, and rehydration with ethanol from other portions of the gel shape. In the future it may be interesting to study the correlation of this jerky motion to the type ions in water, such as the Hofmeister series.22,24,25 In these experiments, no significant difference in motion was observed between tap water and reverse osmosis water. Microlithographic Integration. A distinct advantage of this process is that the gels were processed by lithography, suggesting a straightforward strategy for integration with electronic and microelectromechanical systems. To demonstrate this feature we patterned metallic stripes on a photopatterned PNIPAm shape using evaporation, photolithography and etching (Figure 5A). The PNIPAm gels moved spontaneously in a predetermined manner with the attached metallic stripes when transferred from ethanol to water. As seen in Figure 5B, a circle with Cu stripes precessed in an orbit of 4.6 mm radius at 360 rpm. Since the gels were transparent to visible light, the structures were easily tracked by their opaque metallic strips; we believe these movable metallicgel structures could serve as oscillatory reflectors or electromagnetic modules. (22) Zhang, Y. J.; Furyk, S.; Bergbreiter, D. E.; Cremer, P. S. J. Am. Chem. Soc. 2005, 127(41), 14505–14510. (23) Baltes, T.; Garret-Flaudy, F.; Freitag, R. J. Polym. Sci., Part A: Polym. Chem. 1999, 37(15), 2977–2989. (24) Zhang, Y.; Furyk, S.; Sagle, L. B.; Cho, Y.; Bergbreiter, D. E.; Cremer, P. S. J. Phys. Chem. C 2007, 111(25), 8916–8924. (25) Lopez-Leon, T.; Elaissari, A.; Ortega-Vinuesa, J. L.; Bastos-Gonzalez, D. Chemphyschem 2007, 8(1), 148–156.

Langmuir, Vol. 24, No. 21, 2008 12163

The PNIPAm gels were also patterned around large microfabricated structures. We demonstrated a gel shape resembling a fish with an embedded microcontainer featuring 50 µm pores (Figure 5C). This ethanol soaked gel translates on immersion in water. The pores allowed PNIPAm to flow inside the container and selectively polymerize around it, resulting in excellent anchoring; thus, the gel was able to move the metallic structure without being torn. Finally, the gels were used to move larger structures such as the metal wire in a floating mesh in a circular manner (Figure 5D). Supplemental video 3 features the lithographically patterned gels shown in Figure 5B and the gel shown in Figure 5D. Hence, the gels can be utilized to move objects in a predefined manner (e.g., rotation, precession, translation).

Conclusions We have demonstrated control of solvent driven motion by photopatterning specific shapes. Parallel and precise fabrication via photolithography allowed an investigation into the size dependence of observed motion, permitted creation of new types of motion via shapes too difficult to manually construct, and revealed a direct relation between solvent spreading velocity and gel shape behavior as a key parameter. Additionally, photolithographic processing opens up the possibility for integration of these structures with semiconductor or polymer payloads. We believe that apart from being intellectually stimulating, these autonomously moving structures will find utility in laboratory-on-a-chip applications where predefined types of motion are required. Acknowledgment. The authors acknowledge funding support from the Beckman Foundation and the Camille and Henry Dreyfus Foundation. In addition, the authors thank Anum Azam and Aasiyeh Zarafshar for assistance with figures and Nathan Capallo for assistance with high speed imaging. Supporting Information Available: Still images from high speed imaging of a photopatterned PNIPam symmetric Y shaped gel, schematic of the geometry used to calculate linear and tip velocities, video showing several photopatterned PNIPam gels in different types of motion, video showing high speed imaging of spinning photopatterned PNIPam symmetric Y shaped gels, and video featuring a lithographically patterned gel and a gel interacting with a floating raft. This material is available free of charge via the Internet at http://pubs.acs.org. LA801329G