PAPER www.rsc.org/softmatter | Soft Matter Mechanism of attraction between like-charged particles in aqueous solution Ekaterina Nagornyak, Hyok Yoo and Gerald H. Pollack Received 13th March 2009, Accepted 2nd July 2009 First published as an Advance Article on the web 21st July 2009 DOI: 10.1039/b905080a Although it has been long known that like-charged particles attract one another in aqueous media, the mechanism underlying this counter-intuitive phenomenon has remained controversial. We tested the hypothesis put forth long ago by Langmuir and again by Feynman and by Ise, that the attraction between like-charged entities lies in an intermediate of unlike charges. Tests were facilitated by the observation that the attractive forces could be confirmed between widely separated particles of macroscopic size. Two approaches showed comparable results. In the first, pH-sensitive dyes showed intermediate zones of opposite charge: an accumulation of protons was found between negatively charged spheres, whereas between positively charged spheres the intermediate zone contained OH groups. In the second and complementary approach, microelectrode measurements showed that in between negatively charged spheres, the electrical potential was relatively positive, whereas between positively charged spheres it was relatively negative. Hence, both approaches confirm theoretical expectations. The large number of unlike charges lying in between the like-charged spheres may come from the build-up of the recently reported ‘‘exclusion zone’’ surrounding each particle. Introduction Like-charged colloidal particles attract one another. In aqueous suspension, particles draw together to form a colloidal crystal, in which particle separation is on the order of particle diameter.1–6 Although counter-intuitive, attraction between like-charged particles has been confirmed time and again over more than half a century.7–12 Given the classical expectation that attraction should occur only between particles that are oppositely charged, Langmuir suggested early on that the required opposites might come from counter-ions situated in between the like-charged particles; these counter-ions would create the attractive force.13 Feynman likewise emphasized the role of intermediate counter-charges, coining the phrase, ‘‘like-likes-like through an intermediate of unlikes’’.14 Taking up the Langmuir–Feynman ideas, Ise and Sogami derived a formal theory based on those suggestions, demonstrating excellent agreement between theoretical expectations and experimental observations.15 On the other hand, the presence of counter-ions between the like-charged particles has never, to our knowledge, been experimentally validated. A reason may be the demands on spatial resolution: micron-scale separations typical of colloidal systems demand nanometre-scale spatial resolution, which is not readily achievable for charge-distribution measurements. Furthermore, the source of unlikes has never been entirely clear—possible candidates including various ions that might be present, and/or protons and hydroxyl groups coming from hydrolyzed water. Two recent findings have opened the possibility of testing the like-likes-like mechanism. The first is the identification of a potential source or abundant free charge. Adjacent to hydrophilic surfaces, extensive zones of organized water have been Department of Bioengineering, University of Washington, Box 355061, Seattle, WA 98195, USA 3850 | Soft Matter, 2009, 5, 3850–3857 found, which retain charge.16,17 As this charge builds, charges of the opposite polarity are released to the region beyond, into the bulk solution.18 These released charges could constitute the suspected unlikes. In other words, if each colloidal particle is enveloped by a shell of more-ordered water, then many charge carriers would be released into the solution beyond. Such enveloping shells have been confirmed for larger spherical particles,19,20 and presumably exist for smaller particles as well, at least for those whose characteristic dimension greatly exceeds the size of the water molecule. The second relevant finding is that the attraction implied in colloid-crystal formation can also be observed with much larger particles. As shown here, sub-millimetre-sized particles of like charge are attracted to one another over separations on the order of hundreds of micrometres. Such large-scale separation made it feasible to explore whether indeed the suspected unlike charges are situated between the like-charged particles. Results Attraction experiments Preliminary experiments had shown hints of attraction between like-charged beads. When two similarly charged beads were dropped into chamber-1 filled with deionized water, spontaneous interaction could be seen from time to time, even when bead surfaces were separated by as much as several hundred micrometres. The attractive interaction was relatively weak, typically 100 mm displacement in 30 minutes. Such spontaneous attractions were noted in five out of 20 experiments. In 13 others the beads did not move at all, while in two experiments they moved apart. The phenomenon was present also when the chamber contained several beads of the same polarity instead of two: when the intervening distance between any two was several hundred micrometres or less, the beads showed a tendency to move This journal is ª The Royal Society of Chemistry 2009 spontaneously toward one another. This was observed three out of 17 times, while in the rest of the observations the beads did not move at all. These preliminary observations provided a reason to believe that attractive forces might be present. Hence, more systematic experiments were carried out to explore the possibility of attraction. In these experiments the strategy was to free the beads from any adhesion to the chamber floor, by gently tapping from below. A representative result is shown in Fig. 1. The graph shows the surface-to-surface distance between the two beads during the course of a series of taps. Above the graph are representative images of bead pairs from which measurements were made. Overall, the beads moved toward one another, although some net repulsive events were recorded as well. In this particular experiment, from an initial surface-to-surface separation of approximately 230 mm, the beads wound up touching one another. A series of experiments were carried out similar to those depicted in Fig. 1. Initial distances between bead surfaces ranged from 70 mm to 400 mm; the variation in initial separation distance was a mere consequence of inability to place beads exactly at the same position every time. Observations were made until the moment when either of the following events occurred: the beads touched, or the beads had separated by more than 1200 mm. A summary of 21 experiments similar to the one depicted in Fig 1 is shown in Fig 2. Initial separation (t ¼ 0) was normalized to a value of 1.0 in each experiment, and all subsequent spacings were normalized relative to that. To reduce clutter, spacings were averaged among five successive taps and placed in bins, average values for which correspond to time intervals of 10 s (t ¼ 0–10 s, 10–20 s, etc.). For the time interval t ¼ 0–10 s, the graph contains data from all 21 experimental runs, while successive bins have fewer and fewer experimental values because one of the two termination criteria had been satisfied. The final bin contained data from only eight experiments. In 18 of the 21 experiments, the beads wound up touching, while in the remaining three experiments, the beads apparently separated enough that the attractive force never took control. Because those large separation values were averaged into the Fig. 2 Average distance change between two negatively charged bead surfaces. Number of observations in successive bins: 21, 18, 15, 12, 12, 11, 8. data set of Fig. 2, the graph shows less dramatic attraction than was actually observed in the vast majority of bead pairs. The experiments above confirm that attractive forces exist between large, negatively charged beads separated by up to several hundred micrometres. Thus, attraction apparently occurs on a macroscopic scale in much the same way as it occurs on a microscopic scale. pH-dye experiments The next series of experiments was designed to test whether indeed counter-ions postulated to be responsible for the attraction exist in abundance in the span between the like-charged beads. To test for positive counter-ions between negatively charged beads, pH-sensitive dyes were used. A representative result is shown in Fig. 3. Panel a shows a low pH region (red; pH 3–4) surrounding each bead, which could be seen after the dye had been present for 10–20 seconds. The lighter color between the bead surface and the red region corresponds to the exclusion zone, where the dye molecules are excluded; its slightly red appearance is likely the result of color contributions from above and below the focal plane. The low pH region beyond the exclusion zone grew with time (panel b). After approximately five minutes the low pH clouds merged, creating a low-pH linkage between beads (panel c). Subsequent to that time the low pH region continued to spread, with clouds growing toward the periphery (panel d). This phenomenon was observed consistently in 27 experiments. The only notable variation was in the speed of growth of the low pH regions, presumably as a result of uncontrolled factors. Nevertheless, the low-pH zone between beads was seen in every experiment, implying a consistently high concentration of protons lying between the negatively charged beads. Electrical potential Fig. 1 Surface-to-surface distance between two beads following a series of taps, imparted every 2–3 seconds. This journal is ª The Royal Society of Chemistry 2009 While the high concentration of protons between beads is consistent with the proposed hypothesis, it does not necessarily Soft Matter, 2009, 5, 3850–3857 | 3851 Fig. 3 Growth and propagation of low pH zone formed around negatively charged beads. pH scale shown in left panel. prove that this region is more positive than areas immediately adjacent to the bead surface: some negative-charge carrier might neutralize the protons. To evaluate the charge distribution between beads, electrical potential was measured. Fig. 4A shows an example of the measured electrical potential as a function of position between two beads. The traces represent the time course of voltage scans made from left to right. At the left-bead surface, the potential ranged between 90 mV and 110 mV. It grew more positive with increasing distance from the bead, reaching the highest values when the tip of the electrode was roughly midway between the beads. With time, the midway potential became more negative (c.f. 10 and 30 minutes), and during scans made early on (t # 10 min) some measurements showed slightly positive midpoint values. The potential profile of Fig. 4A is slightly asymmetrical and we wondered whether the asymmetry might be a result of the microelectrode tip passing through the water and disturbing the system. To examine the source of spatial asymmetry the electrode was moved in the opposite direction, and the results are shown in Fig. 4B. A similarly asymmetrical pattern can be seen, practically the mirror image of panel A. Hence, the slight asymmetry most likely arises from the measurement method itself. More detailed analysis of the midway-potential time course is shown in Fig. 5. The microelectrode was positioned midway between the beads, and measurements were made every ten minutes following initial bead placement in the chamber. Data from twelve pairs of beads are summarized. The results show that the midway potential was most positive during the first few minutes after insertion, and became progressively less positive with time. This result is in accord with the results of the pH-dye experiments, which show the red color between beads most Fig. 5 Time dependence of electrical potential midway between beads. Standard deviations are shown as vertical lines. n ¼ 3 experiments. intense early after bead immersion, then diffusing in all directions with time. Presumably, the protons lodged between (and around) the beads progressively diffuse throughout the rest of the solution, diminishing the potential gradients. In the voltage-measurement experiments, the distance between bead surfaces varied appreciably, from 350 mm to 1200 mm. To determine how the potential profile varied with separation between beads, observations were categorized into three groups depending on bead separation: close (333 24 mm); medium (610 19 mm); and far apart (1104 14 mm). Potential measurements were made between 40 and 50 minutes following bead immersion. To reduce clutter, collected voltage values were averaged among ten successive measurements of the same potential profile. Each group—close, medium, and far—contains three experiments. The results are shown in Fig. 6. They show that the potential profile differs somewhat, depending on bead separation. When beads were in relatively close proximity, the profile was shifted to more positive values than when the beads were farther apart. One possible interpretation is that when beads were closer together the released protons were relatively more confined, giving the higher potential. Multiple-bead experiments Fig. 4 Representative voltage scans between two negatively charged beads. (A) The microelectrode moves from the surface of left bead to the surface of right bead. Different traces correspond to different times after bead immersion in water. (B) Similar to (A), but the microelectrode moves from right to left. 3852 | Soft Matter, 2009, 5, 3850–3857 To check the influence of the presence of multiple beads on beadpair interaction, experiments were carried out with eight to 20 beads in the chamber. First, proton distribution was measured using the pH-dye indicator. For these experiments, chamber-2 was used. The results of eight experiments were similar to those This journal is ª The Royal Society of Chemistry 2009 Fig. 6 Voltage-profile dependence on bead separation. Each curve represents average of three experiments. obtained with only two beads. At low pH (pH z 3) clouds formed around each bead. The clouds grew toward one another, merging midway, and then spread peripherally. The only notable difference from the two-bead results was that propagation was faster: it took approximately half the time to turn the chamber water, initially with pH z 6, into lower pH of 3. Presumably, the rapidity is a result of both the larger number of beads and the smaller water volume in the chamber. Electrical potential experiments were also carried out in the multiple-bead configuration. Approximately 20 negatively charged beads were dropped into chamber-3. Measurements were made between beads of a pair that were midway along the line of multiple beads. Voltage profiles were, again, similar to those obtained in the case of only two beads, with a voltage swing of approximately 100 mV. Positively charged beads To determine whether the attractive behavior might be polarity sensitive, a limited number of experiments were performed with beads that were positively charged. In the chamber-tapping experiment, the positively charged beads seemed to exhibit even stronger attraction than the negatively charged beads (c.f., Fig. 1). In nine out of ten experiments they wound up touching one another, and they reached the touch point with typically only 50 taps, compared to 100 taps in the case of the negatively charged beads. One possible reason for the difference is their relatively small mass (33% lighter than the negatively charged beads), which allows them to translate more readily. Another contributing factor might be weaker adherence to the chamber floor. Tapping was sometimes unnecessary for generating the attraction. In three of four experiments with these positively charged beads, the beads moved together immediately after they were added to the bath, and it was difficult to separate them either by tapping the chamber or by jarring one of the beads with a pipette. With fairly brutal force they could be separated, only to move back together within a few seconds. When the pH indicator solution was added, a high-pH region (purple) was formed around each bead (pH 9–10). The high pH clouds were initially (10–20 s) relatively small and better defined than the corresponding low pH clouds seen with the negatively This journal is ª The Royal Society of Chemistry 2009 Fig. 7 Growth and propagation of high-pH zone formed around positively charged beads. pH scale shown in left panel. charged beads. Within one minute a clear high-pH bridge formed between the two positively charged beads—much the same as with the other polarity (compare Fig. 3 and Fig. 7). One notable difference, however, is that the link tended to diffuse in the direction normal to the axis connecting the bead centers. After about 5 to 10 minutes, the high pH region formed a boomeranglike figure (Fig. 7). In such instances the attractive bead movements were seen to be directed along the color lines rather than directly toward the companion bead. Perhaps, the most notable observation is the complementary nature of the results: with the positively charged beads the pH change was the opposite of that seen with the negatively charged beads. Proton-rich bridges formed between negatively charged beads, while hydroxyl-rich bridges formed between positively charged beads. In both instances, therefore, the linkage between ‘‘likes’’ is indeed formed by ‘‘unlikes.’’ The electrical potential profile was measured as well. The potential was 90 mV positive at the bead surface, and near zero (10 mV to +10 mV) in the mid region. These measurements may not be quantitatively accurate because of the abovementioned dye-distribution asymmetry. Nevertheless, the shape did correspond roughly to that measured with the negatively charged beads, except that it was inverted. In summary, the results obtained with positively charged beads are qualitatively similar to those obtained with negatively charged beads. In both instances the beads attracted one another. And, both pH and voltage measurements indicated that the attraction results from oppositely charged species lying in between the beads. Uncharged beads As a final control, a series of experiments were carried out with bead pairs that had lost their charge in order to determine whether inter-bead attraction was absent. In these experiments positively charged beads were used, and the beads examined were those that had lost their color and hence their ion-exchange capacity. Such beads no longer showed exclusion zones (solutefree zones), and hence should not have contained the large charge that lies within those zones. Ten trials were carried out. The standard tapping strategy was used to free the beads from any Soft Matter, 2009, 5, 3850–3857 | 3853 adhesion to the floor of the chamber. For each bead pair, 30 to 100 taps were made. Overall, the exhausted beads did not move toward one another, although some net attractive events were recorded and in one case the two beads finally touched one another (possibly because of some residual charge; electrical potential measurements showed that charge on the exhausted beads had diminished appreciably, but not necessarily to zero). For statistical analysis, initial separation in each experiment was normalized to a value of 1.0, and relative separation was then measured following the 30th tap. Among the ten experiments the mean value was 1.22 0.61. Hence, the exhausted beads showed no attraction. Discussion Many experiments have demonstrated that like-charged particles in aqueous solution attract one another.1,21–26 The attraction is counter-intuitive, for we are schooled to think that like charges must repel, and not attract. It is perhaps for this reason that some have questioned whether the seemingly attractive movements might actually not require attractive forces,27,28 although these challenges have been vigorously rebutted.4,11 The results obtained here confirm that like-charged particles in suspension attract one another, and provide evidence for the mechanism. Attractive forces are observed over unexpectedly long distances, and the results show that it is indeed the unlike charges lying in between that are responsible for these attractive forces. Forces between beads We found that like-charged beads in solution are drawn toward one another at separations up to several hundred micrometres. Attraction was found between positively charged beads and also between negatively charged beads, the former being slightly stronger than the latter. Attractive forces are anticipated to fade at distances beyond several Debye lengths, which are typically less than 0.1 mm. However, the attractions observed here evidently took place over much larger spans. One potential artifact might arise from the method used. In order to free the beads from adhesion to the floor of the chamber and allow them to move as they may, the chamber was gently tapped from beneath. This approach was fairly crude; nevertheless, with some practice the tapping procedure could be carried out with a reasonable degree of consistency. However, a possibility is that the tapping induced some kind of vibrationinduced convection that drew the beads toward one another, and that the drawing together was not the result of a genuine attractive force but this kind of convection. Several arguments militate against such an explanation. First, the bead pairs were situated at different positions within the chamber from one experiment to the next; hence flow toward some particular point in the chamber should not necessarily have brought the beads together as consistently as they did. Second, attractions were often seen in the absence of tapping; they sometimes occurred spontaneously. When such beads were forcibly pulled apart (with difficulty), they would spontaneously return toward one another. For these reasons we believe that attractive force between beads is not the result of a convective artifact. 3854 | Soft Matter, 2009, 5, 3850–3857 A further reason to think the attractive forces are genuine is that the attractive potential has been identified. Two methods show that in between the like-charged beads are charge carriers of opposite polarity. The first method, the pH-dye indicator experiments, shows that between negatively charged beads there is a region of low pH. Protons emanating from the region around each bead eventually build a bridge between the negatively charged beads, thereby attracting the beads toward one another. The opposite pH change occurs between positively charged beads. Hence, a high concentration of unlike charge accumulates between the like-charged beads, creating the attractive potential that can draw the beads together. In the case of the positively charged beads, the bridge often bends for some reason that is not yet clear; yet the beads move together by following along the color lines, reinforcing the view that it is indeed the opposite charges of the bridge that attract the beads. The second confirmatory method was that of the potential distribution. Consistently, those results showed that for the negatively charged beads, regions midway had potentials on the order of 100 mV more positive than the zones immediately surrounding the beads. The negatively charged beads are inevitably drawn toward the more positive region and hence toward one another. For the positively charged beads the potential distribution was opposite—more negative in between. Thus, the electrical potential results are consistent with the results of the dye experiments: both show that the like-charged entities are attracted toward one another because of the unlike charges lying in between. A slight difference of behavior between bead types was that the positively charged beads tended to attract one another more strongly than the negatively charged beads. One reason might be the magnitude of the pH difference. The initial pH of the chamber solution was commonly 6; it was slightly acidic, with an abundance of protons. The negatively charged beads added additional protons to this positive charge. Hence, the difference between the final inter-bead pH and the bulk solution pH may not have been as large as with the positively charged beads, which released negatively charged OH groups. This explanation is evidently conjectural, and requires additional study. Source of unlikes A question that arises is the source of the unlike charges lying between the likes. One putative source is the beads themselves. Counter-ions diffusing from the bead may spread broadly within the fluid, accounting for the observed pH changes. Although this is anticipated, the rather extreme pH changes observed raise some question whether the counter-ion source can account for what is observed. The pH changes amounted to three pH units in the acidic direction, and up to four units for the positively charged beads. A pH change of three units implies a change of 1000 times and one wonders whether any such massive change can be accounted for by counter-ions alone. Another potential source of unlikes is the ordered water zone that grows around each bead. Such ordered zones are surprisingly extensive. They vary from tens of micrometrers up to hundreds of micrometres next to some polymeric surfaces such as Nafion and next to gel beads.16,17,29,30 A critical feature of these This journal is ª The Royal Society of Chemistry 2009 zones is that they are charged: negative next to negatively charged surfaces, and positive next to positively charged surfaces.31 Water, on the other hand, is neutral. In order for these zones to transition from neutral to charged, they must release carriers of opposite charge to the bulk, and such release has been confirmed.32 Because such ordered zones are fairly massive, with volume on the order of bead volume, large numbers of charges must be released and it seems likely that this may account for a major fraction of the unlikes. Application to smaller scale particles Another question that arises is whether the explanation of the attractions found here also applies to smaller scale particles, or even those on molecular scale. The reason for employing the larger scale particles here is that various probes could be easily applied. It was possible to measure the distribution of charge carriers using pH-sensitive dyes; and, it was possible to manoeuvre a microelectrode between beads to measure the potential distribution. With smaller particles this would not have been so easily feasible. On the other hand, the question whether the results apply on the smaller scale is not directly addressed. Attractive movements are evidently seen at these smaller-scale levels.4,12,20,33 And, there are arguments that indicate the likely presence of ordered water at these small scales: For example, ordered water zones are readily seen in one-micron thick slices of Nafion sandwiched between glass plates.34 Ordered water is also broadly confirmed within the living cell, the water molecules lying adjacent to nanometre-scale structures.35 If such zones are found adjacent to hydrophilic surfaces in general, then it seems reasonable that scale is relatively unimportant so long as it is appreciably larger than the size of the water molecule. Hence, we think it is likely that the mechanisms uncovered here also apply on smaller scales, possibly down to molecular scales. One notable difference between the present observations and those at the scale of microspheres is the ultimate particle separation relative to particle diameter. In the case of the microspheres, the attraction does not typically yield an end point of microsphere contact: microspheres move attractively to a separation on the order of particle diameter, at which point the attractive force is balanced by the repulsive force between likecharged particles. It is by such a mechanism that the colloid crystal is formed. In the case of the larger spheres studied here, the ultimate point, by contrast, is actual contact between spheres, although a very thin film of water cannot be ruled out. Several explanations seem possible. One reason could be the higher surface-charge density of ion exchange beads compared to that of microspheres. Higher surface charge might yield more counter-ions and therefore higher attractive force. Another possibility stems from the fact that the large spheres touch only at one point, leaving ample intervening space for unlikes to reside. Hence, the attractive force would derive from the unlikes residing in those substantial gaps. Applications of the like-likes-like mechanism The observations made here would seem to have potentially broad implications. The immediate implication is that similarly This journal is ª The Royal Society of Chemistry 2009 charged particles in aqueous solution attract one another even at distances very much greater than the Debye length. Attraction occurs even at macroscopic separations. Hence, even in dilute suspensions, where particles are greatly separated from one another, attractive forces may be present. Further, one needs to take into account the likely pH gradients that may be present throughout the solution domain. At present, these are not considered. Similar attractions may occur even between particles outside of aqueous solution. Consider, for example, water aerosols. One long-standing question is why evaporating water often clusters into discrete puffy clouds, instead of remaining uniformly dispersed over space. If aerosol droplets consist of the kind of organized water seen in other contexts, then an ample source of unlikes is present, which can bring the particles together to coalesce. Thus, the like-likes-like phenomenon may be significant in cloud formation. At the smaller end of the size scale is the issue of molecular selfassembly. Newly minted molecules assemble into mesostructures such as filaments and organelles, but the mechanism of assembly has remained a mystery. In such situations the like-likes-like mechanism may act at the sub-molecular level, creating distributions of unlikes that serve to bring like molecules together by the simple physical mechanism described herein. It appears, then, that the mechanism implied by Langmuir and advocated by Feynman and Ise is not only adequate, but may have broad application. Methods General Deionized water (type I HPLC grade (18.2 MU cm)) collected from a Barnstead D3750 Nanopure Diamond purification system was used for all experiments. In order to minimize contamination, experimental chamber materials were thoroughly washed with water from the above-mentioned source. The experiments involved two types of ion-exchange beads (Bio-Rex MSZ 501(D) resin): anionic and cationic. Prior to use, the beads were washed with spectroscopic grade methanol and then with deionized water. The negatively charged beads, which were used for all experiments unless otherwise specified, were 600–650 mm in diameter. They contained an irreversibly bound yellow dye. The positively charged beads, which had an irreversibly bound blue dye, were 400–500 mm in diameter. Both types of bead turned clear after long-term use, indicating exhaustion of their ion-exchange capacity. To standardize experiments, only the beads with highest ion-exchange capacity, as shown by their color, were used. Attractive force between beads To observe bead displacements due to their attraction, a chamber, (chamber-1) was placed on the sample stage of a bright-field-inverted optical microscope (Zeiss Axiovert-35) with a 5 objective lens and a color CCD a camera attached (Color Digital Camera, model CFW-1310C, Scion Corporation). Chamber-1 was used to monitor bead displacement and later to visualize pH distribution around two beads of the same polarity. It was constructed from a rectangular plastic plate, Soft Matter, 2009, 5, 3850–3857 | 3855 1 mm thick, 6 cm long, and 3 cm wide. A 10 mm diameter hole was drilled through the center of the plate. The bottom of the hole was then sealed with a microscope coverslip (150 mm thick) using UV-glue. This arrangement provided a 1 mm high cylindrical chamber, 10 mm in diameter. The coverslip facilitated microscopic observation from below. For bead-displacement measurements, chamber-1 was first filled with deionized water. No dye was added. Then, two negatively charged beads were placed into the chamber. After five minutes, when the beads had settled for some time, the chamber was lightly tapped from beneath, which slightly lifted the beads from the chamber floor. This tapping procedure was repeated every 2–3 seconds. Each time, the separation distance was measured just after the beads had re-settled. Following this procedure we could construct a graph of bead separation vs. time. Visualization of pH distribution with pH-sensitive dye To detect local proton-concentration differences in the vicinity of bead surfaces, we used a universal pH indicator (Catalog #: SLU1051, Sciencelab.com, Inc., TX). The indicator changes color in response to pH change, and solution pH can then be determined from the color chart, acquired from acid–base titrations. The manufacturer’s instructions—1 ml indicator to 10 ml water—were followed when diluting the concentrated pH-dye solution for experimental use. Deionized water showed a yellow color upon addition of the pH-dye solution, indicating pH of 6. Chamber-2, created in order to observe pH distribution around a network of more than two beads, was constructed from a rectangular plastic plate the same as for chamber-1. Instead of a circular hole, a 2 cm long 1 mm wide groove of height 1 mm was cut from the plate to provide a long, thin box-like chamber. The bottom of this groove was sealed with a coverslip (150 mm thick) to facilitate optical microscopy. The groove configuration permitted a linear alignment of beads. To prevent shadow artifacts on the collected images, the top edges of the groove were smoothed. For experiments, two or more beads of the same polarity were dropped into the appropriate chamber filled with a solution of water and, when required, pH dye (50 ml for chamber-1 and 30 ml for chamber-2). Images were acquired using a stereo microscope (Nikon, model 75860, Japan) and a color CCD camera (model #: CFW-1310C, Scion Corporation). A white LED-ring array was used as a light source. The acquired images were then analyzed with ImageJ. then direct-filled using a flexible long syringe needle. Tip resistance was measured to be 10 MU, indicating tip diameters on the order of 1 mm. A micro Ag/AgCl electrode (World Precision Instruments, Driref-450) with tip diameter 450 mm was used as reference electrode. For the measurement of electrical potential distribution, chamber-3 was used. This chamber was constructed from 1 mm thick microscope slide pieces glued together such that the final chamber was an open box, 2 cm high, 5 mm wide and 5 cm long (see Fig. 8). To facilitate a linear arrangement of beads inside the chamber, a custom-made alignment piece, consisting of two glass slides, 4.5 cm 0.5 cm, glued together with a 300 mm gap, was inserted into the chamber. The alignment of beads simplified the potential-distribution measurement to a two-dimensional problem. Chamber-3 was first filled with deionized water. No dye was added. Then, the microelectrode and reference electrode were inserted into the chamber from the top. The reference electrode was situated 20 mm away from the beads in the bulk region. Two or more beads were immediately dropped onto the groove (Fig 8). Surface-to-surface separation was on the order of bead diameter, but never exceeding 1.5 mm. Microelectrode position was monitored from the side of the chamber with a stereo microscope (model #: Wild M7, Heerbrugg, Switzerland) coupled to a CCD camera (Scion Corporation, Model CFW-1310C). Potential distribution was measured while translating the microelectrode from the surface of one bead to the surface of the other. A step-motor (DC Motor Mike, Lot-Oriel, Germany) imposed translation at constant velocity. Before making the measurement, the system was calibrated by placing the reference electrode close (<2 mm) to the probe electrode and verifying a near-zero potential difference. Then, by using micromanipulators (Daedal, Inc., PA), the reference electrode was placed several millimetres ([ distance between the beads) beyond the network of two (or more) beads, while the tip of the probe electrode was brought to the right surface of the left bead. The microelectrode tip was then moved horizontally, toward the surface of the neighboring bead, at a speed of 20 mm s1. A custom-written LabView program was used to monitor and record the electrical potentials. Measurements were taken at 100 Hz, giving voltage values every 0.2 mm. Electrical potential distribution Voltage measurements were made using tapered glass microelectrodes and a standard electrometer (Electro 705, World Precision Instruments, Fl). Thin-walled, single-barrel capillary tubes (o.d. ¼ 1.2 mm, World Precision Instruments, Model TW150F-6) were used as blanks to fabricate the electrodes using a standard electrode puller (Electrode Puller P-87, Sutter Instruments, CA). The pulled electrodes were immersed with their tips up in 3 M KCl solution. The solution climbed to the tips by capillary action, and microscopic examination ensured that the tip regions were bubble-free. The microelectrode shanks were 3856 | Soft Matter, 2009, 5, 3850–3857 Fig. 8 Experimental setup (chamber-3) used for electrical potential measurements. This journal is ª The Royal Society of Chemistry 2009 Conclusion Despite its central role in self-assembly, the mechanism of attraction between like-charged particles in aqueous solution has remained controversial. It is demonstrated here that abundant charge separation around each particle is the agent responsible for the attraction. This conclusion fits with earlier hypotheses. The mechanism should be valid not only on the colloidal scale explored here, but also at smaller scales possibly down to the nano and molecular levels, which would have important implications for self-assembly. Acknowledgements The authors thank Jeff Magula for fabrication of the experimental chambers and Ronnie Das for software development. This study was supported by NIH Grants (AT-002362 and AR-44813), and ONR Grant (N00014-05-1-0773). References 1 N. Ise and T. Okubo, Acc. Chem. Res., 1980, 13, 303. 2 S. Dosho, N. Ise, K. Ito, S. Iwai, Kitano, H. Matsuoka, H. Nakamura, H. Okumura, T. Ono, I. S. Sogami, Y. Ueno, H. Yoshida and T. Yoshiyama, Langmuir, 1992, 9(2), 394. 3 H. Matsuoka, T. Harada and H. Yamaoka, Langmuir, 1994, 10, 4423. 4 N. Ise, Proc. Jpn. Acad., Ser. B, 2002, 78, 129. 5 O. Gomez-Guzman and J. Ruiz-Garcia, J. Colloid Interface Sci., 2005, 291, 1. 6 Y. Liang, N. Hilal, P. Langston and V. Starov, Adv. Colloid Interface Sci., 2007, 134–135, 151. 7 J. D. Bernal and I. Fankuchen, J. Gen. Physiol., 1941, 25, 111. 8 N. Ise and T. Okubo, J. Phys. Chem., 1966, 70, 2407. 9 N. Ise, H. Matsuoka, K. Ito, H. Yoshida and J. Yamanaka, Langmuir, 1990, 6, # 2, 296. 10 J. C. Crocker and D. G. Grier, Phys. Rev. Lett., 1996, 77(9), 1897. This journal is ª The Royal Society of Chemistry 2009 11 N. Ise, Proc. Jpn. Acad., Ser. B, 2007, 83, 193. 12 B. V. R. Tata, P. S. Mohanty and M. C. Valsakumar, Solid State Commun., 2008, 147, 360. 13 I. Langmuir, J. Chem. Phys., 1938, 6, 873. 14 R. P. Feynman, R. B. Leighton and M. Sands, The Feynmsn Lecture on Physics, Addison-Wesley, Reading, MA, 1963, ch. 2, p. 2. 15 I. Sogami and N. Ise, J. Chem. Phys., 1984, 81, 6320. 16 J. Zheng and G. H. Pollack, Phys. Rev. E, 2003, 68, 031408. 17 J. Zheng, W. Chin, E. Khiijniak, E. Khijniak Jr. and G. H. Pollack, Adv. Colloid Interface Sci., 2006, 127, 19. 18 G. H. Pollack, 32nd Annual Faculty Lecture, University of Washington, Seattle, 2008, http://uwtv.org/programs/displayevent.aspx?rID ¼ 22222. 19 J. Zheng, A. Wexler and G. H. Pollack, J. Colloid Interface Sci., 2009, 332, 511. 20 Q. Zhao, J. Coult and G. H. Pollack. Phys. Rev. E, submitted. 21 N. Ise, T. Okuto, M. Sugimura and H. J. Nolte, J. Chem. Phys., 1983, 78, 536. 22 K. Ito, H. Yoshida and N. Ise, Science, 1994, 263, 66. 23 G. M. Kepler and S. Fraden, Phys. Rev. Lett., 1994, 73(2), 356. 24 A. E. Larsen and D. G. Grier, Nature, 1997, 385, 230. 25 J. Wu, D. Bratko and J. M. Prausnitz, Proc. Natl. Acad. Sci. U. S. A., 1998, 95, 15169. 26 Daoud and Williams, Soft Matter Physics, Springer, Heidelberg, Germany, 1999. 27 R. Fitzgerald, Phys. Today, 2001, 54, 18. 28 R. Zhang and B. I. Shklovskii, Phys. Rev. E, 2005, 72, 021405. 29 B. Chai, J. Zheng, Q. Zhao and G. H. Pollack, J. Phys. Chem., 2008, 112, 2242. 30 Q. Zhao, J. Zheng, B. Chai and G. H. Pollack, Langmuir, 2008, 24, 1750. 31 J. Zheng and G. H. Pollack, in Water and the Cell, ed. G. H. Pollack, I. L. Cameron and D. N. Wheatley, Springer, The Netherlands, 2006, p. 165. 32 B. Chai, H. Yoo and G. H. Pollack. Phys. Chem. Chem. Phys., submitted. 33 J. C. Crocker and D. G. Grier, Phys. Rev. Lett., 1994, 73(2), 352. 34 A. Wexler and G. H. Pollack., unpublished work. 35 G. H. Pollack., Cells, Gels and the Engines of Life, Ebner and Sons, Seattle, 2001. Soft Matter, 2009, 5, 3850–3857 | 3857