Investigating Surface Heterogeneities of Polyacrylamide Hydrogels Result ing f rom Casting Conditions Hyd rogels are materials that have potential to be used as implants due to their biocompatibility, ease of polymerization, and water content. Further study of the behavior of these hydrogels provides insight into how these materials can be expected to perform in vivo Hydrogels are governe d by their mesh size, which drives their mechanical properties, transport properties, and recently has been shown to drive their friction behavior. Superlubricity has been observed in samples polymerized with a free surface, which has been hypothesized to be attributed to a very large mesh size. To investigate this result, confocal microscopy is used in conjunction with fluorescent microparticles of varying diameters to measure penetration into the hydrogel. Results show t hat beads larg er than the previous ly accepted mesh size penetrate into the hydrogel bulk. Penetration depth was inversely proportional to the polymer concentration. Samples cast between polystyrene surfaces accommodated a lesser number of microparticles than their exposed surface counterpa rts. Results indicate that casting conditions can produce sub stantially different mesh sizes at the surface. Introduction The advent of hydrogels in soft matter engineering has produced unique challenges, many of which concern the characterization of the material itself. Effectively described as a network of hydrophilic polymer chains, hydrogels have been produced using a var iety of constituents and synthesis techniques To assist in categorization, hydrogels have been broadly organized according to their usage of either synthetic or natural polymers, and then further classified according to their polymerization methodology [2 ] The conditions and concentrations necessary during synthesis of eac h type of hydrogel are the source of the significant variations Soft Matter Tribology Lab Honors Thesis Ryan Smolchek Magna Cum Laude Spring 2018
in mesh size which in turn dictate behavior. For practicality, the scope of this report will be limited to polyacrylamide (pAAm) hydrogel with an emphasis on the 3.8 % 7.5% 10.0%, and 12.5% volume concentrations. These varieties were selected for their relevance to ongoing studies and recent superlubricity findings [4 ] Classified as a synthetic polymer, the polyacryl amide hydrogels used in this experiment were formed using varying concentrations of acrylamide (AAm) and methylenebisacrylamide (MBA), with the former being the contributor of the polymer chains and the latter being the attributing cross linker. Ammonium p ersulfate (APS) and t etramethylethylenediamine (TEMED) are introduced to catalyze the reaction, where APS is broken down into sulfate free radicals and TEMED acts as a stabilizer to form the polymer network. W ater is typically added prior to the introdu cti on of TEMED and APS, and is deoxygenated to promote a complete reaction. The mesh size of the polyacrylamide hydrogel used in this procedure is considered to be directly dependent on the relative concentration of MBA to AAm, as well as the concentrat ion of the polymer constituents to water as a whole. Larger quantities of MB A during polymerization produces a greater degree of crosslinking, which in turn produces a more interconnected polymer network. When compared against the diameters of the fluoresc ent beads used for later confocal microscopy, all gel compositions investigated in this procedure should behave as effective barriers preventing significant bead penetration. Three varieties of fluorescent beads were sourced for this procedure from the Life Technologies branch of ThermoFisher Scientific. Each was provided as a 1% solids solution, with the first being green beads with a diameter of 26 nm 3.6 nm, and the second vari ety being red beads with a diameter of 45 nm 7.5 nm. The third variety was used largely for
surface measurements and measured 100 nm 16.3 nm. All batches of beads provided were of the carboxylate modified specification, indicating a surface coating of a polymer containing hydrophilic carboxylic acids. Differing diameter beads were selected for comparison during confocal microscopy, as the smaller diameter green beads were expected to migrate more readily into the hydrogel surface than their larger coun terparts. Of interest during this procedure is the effects of surface conditions during polymerization and the resultant effects on gel characteristics. Specifically, Pitenis et al 2014 found that hydrogels cast in open air produced friction coeffic ients nearing and within the range of superlubricity, defined as less than or equal to 0.001  These same gels when cast between polystyrene surfaces produced friction coefficients a full magnitude greater [UrueÂ–a et al.] 2018 indicating a significant deviation in hydrogel properties when allowed to polymerize against in an open environment [3 ] As hydrogel properties are largely dominated by the respective mesh size, this likely indicates that hydrogels cast in an open environment produce a drastically larger mesh size at the surface. By introducing a bead solution to hydrogels cast in both open air and bounded conditions, this procedure seeks to distinguish between the two resultant surfaces. Methods Before casting the sample hydrogel disks, seve ral stock solutions of constituents were produced using a combination of available powders and liquids. These stock solutions consisted of a 2.0 wt% MBA solution, a 30 wt% AAm solution, and 10.0 wt% solutions of both TEMED and APS. Each stock solution was prepared by adding the solute powder or liquid into a container and adding distilled water until reaching the desired concentration. Then, each
solution was taken to either a vortex or centrifuge mixer depending on solubility. Both TEMED and APS were remade for each batch of samples. The large quantity of samples produced during this procedure necessitated a more efficient method of calculati ng stock volumes, and to assist with this a MATLAB program was written which receives the desired hydrogel as well as both the stock solutions and desired sample volume. The program would then output stock volumes necessary to achieve the desired hydrogel and quantity. This approach greatly expedited sample production, and with a saved history of each sample set later results could be traced back to their respective mixes to ensure validity. This program was written with stock solution tables taken from Uru eÂ–a et al. 2015 to ensure applicable results  To establish a standardized procedure, 10.0% and 12.5% volume concentration pAAm samples were cast for the first imaging cycle due to their ease of handling For both concentrations there would be two additional sets of samples, one set cast with a surface exposed to air and another cast between polystyrene dishes. Five samples of 60 mm diameter Fig. 1 Illustration of the casting conditions and sample production methods.
disks approximately 4 mm in height were cast for each, and a punch was used to extract several smaller sampl es approximately 15 mm in diameter from each. This method produced 20 samples for each concentration and casting condition, resulting in a total of 80 samples. Once punched, each disk was placed in an assigned and labelled petri dish and soaked in ultrapur e millipore water for a minimum of 72 hours to ensure sufficient swelling. Each sample was rinsed at 36 hours to remove any remaining unreacted solutes and soaked in fresh water. After swelling, each sample container was drained of water an d assigned either the 26 nm green bead solution or the 46 nm red and 100 nm blue bead solution Green beads were prepared in solution separate from the red and blue to to prevent cross fluorescence, an unde sirable result where beads of a certain color fluoresce in an other given enough light intensity Each container of bead solution provided by ThermoFisher Scientific was sonicat ed to break up any grouped nano particles, and then mixed to produce a 0.02% solids bead solution. These solutions were then poured over the s ample disks of appropriate assigned color and all petri dishes were then stored in containers which prevented UV light from bleaching the microparticles. To investigate the long term effects of bead diffusion and penetration, several discrete time in tervals were assigned before imaging. A logarithmic scale was chosen, and early testing indicated that the 1 second, 10 second, and 100 second discrete time intervals showed little to no bead permeation. Samples were then imaged using the confocal microsco pe at 1000, 10,000, and 314,000 seconds. The latter was chosen as it represents the halfway point between the 100,000 and 1,000,000 second interval and provides insight into long term behavior. When each sample was ready for imaging, the bead solution was rinsed from the sample and petri
dish using distilled water. The sample was then flipped over onto a clean petri dish, locating the surface of interest closer to the lens, and then placed into the confocal imaging tray. Calibrating the confocal for effective imaging was initially difficult due to the unknown conce ntration of beads within the surface of the hydrogel, and it was anticipated that with greater concentrations at higher time scales the settings of early imaging would produce oversaturated images. To circumvent this issue, a solution of beads of each colo r and concentration were imaged and the most desirable settings were recorded. The goal for this calibration step was to produce images that were approaching oversaturation for the 0.02% concentration, but not exceeding it, as the samples surfaces should n ever contain a greater concentration of beads than the solutions they were exposed to. Red and blue bead samples were imaged together as their respective response wavelengths would prevent cross fluorescence, while green bead samples were imaged separately Imaging was completed using the 10x magnification and 4.850 m step size for purposes of time, and imaging heights were kept consistent at each time interval to simplify later Fiji analysis. "#$%!! & %!'#()*#+#,-!#**./0120#34!3+!05,!634+362*!#(2$#4$!)136,//7!8,$#44#4$!9#05!05,!826:$13.4-!/.801260#34!)136,-.1,!3+!#(2$#4$! 2!/3*.0#34!9#05!2!:4394!6346,40120#34!3+!8,2-/%!;51,,!8,2-!/#<,/!3+!=%=&>!/3*#-/!6346,40120#34!9,1,!./,-!2*34$/#-,!2!/,1#,/!3 +! -#/61,0 ,!0#(,!#40,1?2*/%!@265!/2()*,!92/!+*#)),-!03!*3620,!05,!/.1+26,!3+!#40,1,/0!4,21!05,!*,4/!03!4,$20,!05,!)3//#8#*#0A!3+!05,! 5A-13$,*!8.*:!2++,60#4$!#(2$,/%!
Once a standardized procedure was developed and validated for the 12.5% and 10.0% samples, testing was expanded to the lower concentrations of 7.5% and 3.8%. Two samples of each concentration were cast in an open air petri dish and then soaked in water to swell for 72 hours. To reduce the number of samples teste d, each disk would be punched and then sliced in half with one semicircle flipped over, simulating a surface cast against polystyrene. Additionally, these samples were cast with blue 100 nm beads to assist with locating the surface when imaging. These samp les were then soaked in green 26 nm beads for 36,000 seconds and prepared for imaging in a procedure identical to the 12.5% and 10.0% samples. Of interest during this imaging cycle was the potential for the rinsing procedure to cause green beads to evacuat e the surface. To investigate, each sample was imaged before flushing to compare against results collected afterwards. Confocal settings for this image series were identical to those used in the 314,000 second time interval to permit direct comparison. Image analysis was conducted through Fiji, and consisted of a series of image manipulations with the intended result being an intensity profile of the surface. First, a TransformJ rotation of 90 Â¡ about the x axis produced a cross section of the image stack. Then, a correcting factor for the refraction during imaging was applied to produce an accurate length scale for measurements. A Z Project was then applied to produce a single image, and a plot profile function was applied to generate an intensity cu rve through the surface of the hydrogel. Several Z Project settings were assessed to investigate the effects on data retention and representation, and it was determined that the median setting produced a curve more authentic to the original data than the o thers. The plot profile function also provided a method
for exporting to Excel, and charts were constructed to compare the effects of both time and surface conditions. In order to produce more impactful charts and retrieve more immediately useful inf ormation, a relationship needed to be devised between the intensity reported through confocal imaging and the concentration of beads present. To construct this relationship, a 12 well plate was filled with incrementally increasing concentrations of green b ead solutions and imaged at settings identical to those used during the 10,000 and 314,000 second cycle. This relationship was found to be linear, and when applied to the intensity plots produced through Fiji pro duced contours plotting the concentration of the beads against the surface depth Results "#$%! B %!C13+#*,!)*30/!3+!1,-!DE!4(!24-!F==!4(!8,2-!#40,4/#0A!1,/)34/,/!+31!F=%=>!G 27! 8H!24-!F&%I>!G 67 -H!?3*.(,!6346,40120#34! /2()*,/%!J(2$,/!9,1,!02:,4!20!BFD7===!/,634-/!24-!63()*,0,-!+31!8305!,K)3/,-!/.1+26,!24-!)3*A/0A1,4,!/.1+26,!/2()*,/%!
Surface profiles of the samples exposed to the 46 nm red beads and 100 nm blue beads illustrated larger in tensities in accordance with several trends. Samples of the lower 10.0% polymer concentration illustrated both peaks of greater magnitude and thicker tails indicating substantially more bead diffusion (a, b) In addition, samples cast with an exposed surfa ce illustrated significantly greater responses in the 10.0% pAAm samples. The 12.5% samples illustrated much lower responses overall, with slight reductions when cast between polystyrene surfaces (c, d) Sub surface concentration maximums were present duri ng analysis, however in conjunction with the peak concentrations exceeding those of the bead solutions it is probable that confocal imaging resolution produced a response misrepresenting the peak concentration. It is far more likely that responses were pic ked up by the sensor in z steps prior to the polystyrene surface, however this necessitates confirmation through further imaging at greater magnifications. Fig. 4 Profile plots of green 26 nm bead concentration for varying time intervals, concentrations, and surface conditions. Distance is plotted beginning at the estimated surface, however recent findings have suggested that the true surface is locat ed at the pea k concentration. A Riemann sum integration was applied for the 10.0% and 12.5% volume concentration exposed surface samples to investigate the reduced peak.
Analysis of the discrete time interval testing for the 10.0% and 12.5% samples soaked in g reen beads illustrated distinguishing features resulting from elapsed time, sample concentration, and casting surface conditions. Samples exposed to beads for longer time scales produced substantially greater responses, with peak intensities of 314,000 sec ond samples often quintupling those of the lesser 1,000 second samples. In addition, samples with greater exposure periods also presented thicker tails when interpreted as a normal distribution. This effect was also present when comparing constituent conce ntrations, as the 10.0% pAAm samples (a, b) were much broader in their response when compared to the 12.5% samples (c, d) An in teresting result came in the for m of the conflicting trend exhibited here, as while the red and blue bead samples illustrated su bstantially more bead permeation into the exposed surface, here the green beads appeared to behave in the opposite manner. Intensity peaks and profiles were reduced across the full distribution for both concentrations, indicating a reduced number of beads entering the surface. Sub surface concentration maximums were present here as well, likely confirming this behavior as a procedural error. "#$%!! I %!C13+#*,!)*30/!3+!6346,40120#34!?/%!-,)05!+31!05,!B%L>!24-!M%I> ?3*.(,!6346,40120#34!/2()*,/!02:,4!20!BE7===! /,634-/%!'2()*,/!9,1,!1#4/,-!8,09,,4!#(2$,/!03!#4?,/0#$20,!05,!)30,40#2*!*3//!3+!8,2-/!#4!,K)3/,-!/.1+26,!/2()*,/%
To investigate the conflicting trends between the green bead response and that of the blue and red beads, further imaging was completed for lower concentrations in which the effects of rinsing samples would be recorded. 7.5% and 3.8% hydrogels of bot h surface conditions were imaged before and after rinsing with distilled water, with intensity profiles indicating a sizeable reduction in response for the exposed surface samples. This effect was diminished for the 3.8% concentration when cast against pol ystyrene (a) and absent for the 7.5% samples (c, d) Concentrations were lower for exposed surface samples prior to rinsing as well, however the difference was much greater for those at 7.5% volume concentration (c, d) Discussion Investigating the b ehavior and structure of hydrogel surfaces presents an interesting challenge due to both the minute scale of measurements required and current limitations in imaging and hardware. The intent of this procedure was to develop and test a method which circumve nts these limitations by imaging fluorescent nanoparticles and their movement patterns rather than the surface itself. Often several magnitudes greater in size than the resolution capable through confocal microscopy, these fluorescent nanoparticles can pro vide valuable insight into the macro behavior of the hydrogel surface. In addition, findings using this approach could potentially quantify and predict deviations in the surface which produce drastically different material properties exemplified by the sup erlubricity observed by UrueÂ–a et al. Before conducting this procedure, several conjectures were made in an attempt to both predict the effects of exposed surface casting and to explain the observed resultant
superlubricity. UrueÂ–a et al determined conclusively that the friction coefficient was driven by the mesh size, and a theorized model was developed that by casting against an exposed surface hydrogels would form a microscopic layer of substantially larger mesh size. This model aligns with the findings and trends previously established, while also taking into consideration the greater availability of reaction inhibiting oxygen at the surface during polymerization. To coincide with this model, hydrogels cast with an exposed surfac e should exhibit both greater bead permeation and a surface more readily accepting of larger bead sizes. Initial results indicated trends which appeared to both adhere and directly conflict with the proposed model. The larger red and blue beads were of substantially greater number in the exposed surface samples, however green beads were observed in significantly lower concentrations than in their polystyrene cast counterparts. No reasonable explanation was conceived which could explain this behavior c onclusively, which necessitated further investigation into the procedure of imaging itself. The effects of rinsing were of great interest, as while this step would rid the surface of nanoparticles suspended in the bordering water layer, there was a suspici on that the rinsing could be ejecting the green beads from within the hydrogel surface. To investigate, lower concentration samples were imaged before and after rinsing. These samples illustrated considerable bead loss in the second cycle of imaging. This result was significantly more apparent in the 7.5% samples when compared to the 3.8% set, suggesting that the larger bulk mesh size of the 3.8% hydrogel was less affected by the casting environment. These findings have led to a prediction that casting agai nst an open surface produces a mesh size much larger than present in the bulk, capable of both readily accepting relatively large beads and providing a means for smaller nanoparticles to be flushed out. This
prediction aligns with the model initially prese nted, and provides evidence to support the reasoning behind the superlubricity observed by Pitenis et al. 2016  Further imaging and measurements are necessary before drawing defendable conclusions concerning concentration gradient profiles, howeve r the trends established here provide a degree of insight into the surface behavior of hydrogels previously unavailable. Current models undergoing further investigation suggest that casting with an exposed surface produces mesh sizes large enough to produc e uncharacteristic behavior, with bead permeation and friction coefficients expected of hydrogels of considerably lesser polymer concentration. Only through imaging completed at greater resolution can further conclusions be drawn and validated, however the se results suggest that regulated bead permeation could provide a valuable tool for investigating hydrogel surfaces. References  Juan Manuel UrueÂ–a, Angela A. Pitenis, Ryan M. Nixon, Kyle D. Schulze, Thomas E. Angelini, W. Gregory Sawyer, Mesh Size Co ntrol of Polymer Fluctuation Lubrication in Gemini Hydrogels, Biotribology, Volumes 1 2, 2015, Pages 24 29, ISSN 2352 5738  Morteza Bahram, Naimeh Mohseni and Mehdi Moghtader (August 24th 2016). An Introduction to Hydrogels and Some Recent Applications, Emerging Concepts in Analysis and Applications of Hydrogels Sutapa Biswas Majee, IntechOpen, DOI: 10.5772/64301.  Juan Manuel UrueÂ–a, Eric O. McGhee, Thomas E. Angelini, Duncan Dowson, W. Gregory Sawyer, Angela A. Pitenis, Normal Load Scaling of Friction in Gemini Hydrogels, Biotribology, Volume 13, 2018, Pages 30 35, ISSN 2352 5738  Pitenis AA, Manual UrueÂ–a J, Cooper AC, Angelini TE, Gregory Sawyer WW. Superlubricity in Gemini Hydrogels. ASME. J. Tribol. 2016;138(4):042103 042103 3. Doi:10. 1115/1.4032890.