soft lithography replication of a bioinspired unidirectional wicking channel

Upload: joe

Post on 25-Feb-2018

220 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/25/2019 Soft Lithography Replication of a Bioinspired Unidirectional Wicking Channel

    1/14

    Soft Lithography Replication of a Bioinspired

    Unidirectional Wicking ChannelRyan Blumenstein

    [email protected]

    Ethan [email protected]

    Sara [email protected]

    Washington University in St. LouisMechanical Engineering and Materials Science

    Micro-Electro-Mechanical Systems IMEMS 5801

    Abstract Capillary filling methods, and directional

    wicking techniques in particular, are powerful and

    widely-used tools in passive microfluidic devices, as the

    surface adhesion experienced by a microflow becomes

    large relative to viscous and inertial body forces. Here, we

    replicate a previously described unidirectional wicking

    channel topography inspired by the rain-harvesting

    behavior of the Texas horned lizard. Previously

    fabricated by laser-engraving PMMA, wicking channels

    are here fabricated by a PDMS soft lithography processwell-suited to batch processing. In addition to replicating

    the previously studied channel geometry, we take

    advantage of the resolution capabilities of photo- and soft

    generally grouped as actuated and passive microfluidic

    devices according to whether they utilize an external energy

    source or spontaneous filling phenomena resulting from

    device design. The former group includes all manner of

    pumps, typically displacement- or centrifugal-style pumps, as

    well as electric- or magnetic-field actuation, while the latter

    includes devices that drive flow with chemical gradients,

    osmotic pressures, permeation, and capillary forces. Actuated

    microfluidic devices are capable of producing higher and

    more consistent flow rates with the greatest control, butpassive devices offer greater portability, scalability, and low

    power consumption [1]. For this reason, passive microfluidic

    devices are of particular interest in many applications where

  • 7/25/2019 Soft Lithography Replication of a Bioinspired Unidirectional Wicking Channel

    2/14

    d i b bl h h b

    molecular analysis, biodefense, molecular biology, and

    microelectronics [4]. One of the most common uses of

    capillary action in microfluidics devices is in capillary

    pumping. In a capillary pump, the surface attraction between

    the working fluid and device surface spontaneously draw thefluid through a working channel without the need for external

    pumping. An example of a device that uses capillary pumping

    is shown below inFigure 2.Much work, both numerical and

    experimental, has been done to model the dynamics of

    capillary pumps and to develop tunable designs to control

    filling rates with various capillary geometries [1,5].

    FIGURE 1

    MAGNITUDE OF INERTIAL,VISCOUS,AND GRAVITATIONAL FORCESRELATIVE TO INTERFACIAL FORCES AS FUNCTIONS OF CHANNEL SIZE AND

    VELOCITY [6]

    By controlling feature size, shape, spacing, and alignment,

    some wicking devices have been designed to produce

    anisotropic filling, allowing for directional control of fluid

    motion [7,8,9,10]. Most such devices use micropatterned

    posts or nanohairs [11,12] on a surface or channel to produceasymmetrical conditions that favor movement of a fluid

    surface in one direction. Some wicking topographies achieve

    sufficient directional selectivity to effectively halt fluid flow

    in the reverse direction. These devices are referred to as

    liquid diodes, as their selectivity resembles an electrical

    diodes ability to allow current in only one direction. An

    example of the wicking behavior of such a device is

    illustrated inFigure 4.Anisotropic wicking topographies and

    unidirectional liquid diodes are powerful tools for the

    design of microfluidic devices, providing greater flow control

    and even allowing for the construction of simple logic circuits

    [13].

    FIGURE 4

    THIS EXAMPLE OF A LIQUID DIODE CHANNEL TOPOGRAPHY ALLOWS FLOW

  • 7/25/2019 Soft Lithography Replication of a Bioinspired Unidirectional Wicking Channel

    3/14

    from damp sand [16]. The skin of these lizards produces

    capillary wicking with a strong directional selectivity

    towards the lizards mouth (seeFigure 3), making the lizards

    skin an excellent natural model for unidirectional wicking

    topographies. As a result, P. cornutum has inspired highlysuccessful attempts to replicate the anisotropic wicking

    behavior in microfluidic devices [17,18,19].Figure 5shows

    a schematic and photograph of a polymethyl methacrylate

    (PMMA) biomimetic wicking channel for unidirectional

    transport based on observations of the Texas horned lizards

    rain-harvesting behavior and morphology. This channel was

    found to allow fluid flow in the forward direction while

    halting fluid spread in the reverse direction [18].

    FIGURE 5

    ABIOMIMETIC CHANNEL FOR UNIDIRECTIONAL WICKING,INSPIRED BY THE

    RAIN-HARVESTING ADAPTATIONS OF THE TEXAS HORNED LIZARD [18]

    In this investigation, we attempt to replicate the

    unidirectional wicking topography shown inFigure 5 through

    a soft lithography fabrication process. In addition, we

    investigate the effects of channel width and feature size by

    creating similar channels at varying scales. Compared to

    laser-engraving, soft lithography has the advantage of being

    suitable to batch processing. By replacing laser-engraving

    with soft lithography and experimenting with varying

    channel scales, we aim to design an optimized unidirectionalwicking device potentially suited to production at scale.

    METHODS

    I

    reasons. A 5:1 ratio would dictate a channel depth of 1500 m

    for the geometry described above, which was not feasible for

    the photolithographic methods used to produce the channel

    molds. With the methods and materials available, 450 m

    was considered a reasonable maximum channel depth, givingan aspect ratio of only 1.5:1.

    FIGURE 6

    CHANNEL GEOMETRY AT 6:1SCALE,INCLUDING CHANNEL,WICKING SCALEFEATURES,LOADING PAD,AND 1AND 5MM MEASUREMENT FEATURES

    In addition to reproducing previously described results, wewished to investigate the effect upon wicking behavior of

    different feature sizes. This was of particular interest as prior

    biomimetic devices inspired by the wicking behavior of the

    Texas horned lizard have used features approximately six

    times the size of the lizards scales[18]. The

    photolithographic and soft lithographic are not bound by the

    same size limitations as the laser-engraving methods

    employed previously, allowing for the production of similar

    geometry at much smaller scales. We therefore designed

    similar wicking channels at one-half and one-sixth of the size

    shown inFigure 6 to produce wicking features at approximate

    ratios of 1:1 and 3:1 compared to the biological model. The

    characteristic widths for these channels were 50 and 150 m,

  • 7/25/2019 Soft Lithography Replication of a Bioinspired Unidirectional Wicking Channel

    4/14

    the channels labeled a-d are hereafter referred to as 1-4,

    respectively).

    FIGURE 7

    SECTION OF A NOVEL ARRAY OF WICKING SCALE FEATURES,WITH

    ALTERNATING ROWS INVERTED

    FIGURE 8

    THE PHOTOMASK DESIGN CONTAINS ALL CHANNELS,INCLUDING A)THE 3:1

    CHANNEL, B) THE 6:1, C) THE 1:1, AND D) THE 3:1 ARRAY, ARRANGED TO

    same spin coating procedure a second time to achieve a total

    photoresist thickness of 450 m. The thick wafer was

    subsequently soft baked again at 65Cfor 8 minutes and 95C

    for 105 minutes.

    Pattern transfer from mask to wafer was performed byexposure to 365nm light in a Karl Suss MJB3 Mask Aligner

    (Figure 9). According to the data sheet, the thin wafer

    required an exposure dosage of 350 mJ/cm2 and the thick

    wafer required 500 mJ/cm2. However, both wafers were

    intentionally overexposed (11.46 mW/cm for 45 s/80 s,

    respectively) to ensure complete pattern transfer. Both wafers

    underwent post exposure bake (PEB) at 65C for 5 minutes

    and then at 95C for 15 minutes for the thin wafer and 30

    minutes for the thick wafer. Following PEB, the wafers were

    developed in MicroChem SU-8 developer for approximately

    30 minutes (thin) and 45 minutes (thick).

    FIGURE 9

    KARL SUSS MJB3 MASK ALIGNER USED TO EXPOSURE PHOTORESIST

  • 7/25/2019 Soft Lithography Replication of a Bioinspired Unidirectional Wicking Channel

    5/14

    again degassed in the vacuum chamber until all bubbles were

    sufficiently removed: this was a longer process for the thicker

    photoresist master than the thinner one. The devices were

    cured for 1 hour at approximately 70C in a Quincy Lab

    Model 20 Lab Oven. Finally, each device was removed witha scalpel and bonded to a glass slide.

    FIGURE 10

    VACUUM CHAMBER USED TO DEGAS PDMS(VACUUM PUMP NOT

    PICTURED)

    IV.

    Surface functionalization

    In order to achieve the recommended contact angle range

    between 60-80 [18], the devices were plasma treated in aplasma cleaner (Harrick Plasma PDC-001-HP) with a

    maximum power of 30 W. According to Figure 11, it was

    determined that a plasma treatment of 3 minutes at maximum

    FIGURE 11PDMS/DICONTACT ANGLE AS A FUNCTION OF PLASMA TREATMENT DOSE

    (TIME AT 70 W)AND AIR EXPOSURE TIME [22]

    RESULTS AND DISCUSSION

    I. Photolithography

    The SU-8 2075 photoresist was highly viscous, which

    presented challenges in producing an even coating. Spin

    coating produced a sizable edge bead on both wafers. After

    pre-exposure baking, wrinkling defects became apparent in

    the edge bead of the thick wafer, as seen inFigure 12.This

    did not appear to impact the exposure, since the middle of the

    wafer was free of defects.

  • 7/25/2019 Soft Lithography Replication of a Bioinspired Unidirectional Wicking Channel

    6/14

    appeared to be fine, but after the pre-exposure bake there

    were several elliptical depressions on the wafer. These can be

    seen on the left side of Figure 13. These patterns would

    interfere with the exposure, because they cover a significant

    portion of the wafer.

    FIGURE 13

    THIN WAFER AFTER PRE-EXPOSURE BAKE SHOWING DEPRESSIONS IN

    PHOTORESIST LAYER AFTER SOFT BAKING

    The pre exposure bake for the two wafers took longer than

    expected. Based on the manufacturers guidelines, the SU-8

    for the thin wafer should have been 45 minutes at 95C, and

    the thick wafer should have been 45 minutes after each spin.

    However, after the prescribed time, both wafers were still

    very tacky. The thin wafer still had the defects described

    above. In order to be sure the wafer would not stick to the

    photomask in the photolithography stage, the surface should

    be only slightly tacky. The thicker wafer was soft baked for

    a total of 150 minutes and the thinner wafer was soft bakedfor a total of 145 minutes. After this extended bake the wafers

    were less tacky, and after cooling overnight both wafers were

    significantly less tacky.

    thin-coated wafer (design thickness 225 um) ranged from

    300-400 um (seeappendix A: Photoresist Thickness Data).

    Channel 1 (6:1) had significant issues due to it overlapping

    with the defects on the photoresist. The height of this channel

    was consistently below 100 um, so the device created withthis master would not be effective. The thicker wafer had a

    design thickness of 450 um, but our measurements ranged

    from 500-650 um. The higher-than-expected thickness of the

    channels increased the capillary aspect ratios, which may

    have improved the devices wicking capabilities.

    Overall, the developed features appeared well defined with

    few defects. There are a few notable exceptions, most notably

    in thin Channels 1 (3:1) and 3 (1:1). Channel 1 was

    significantly thinner than expected due to the mask

    overlapping with the defect on the wafer, but the pattern

    seemed to transfer well. As seen inFigure 14,the features are

    crisp but there is a rough texture in the channel.

    FIGURE 14

    PRMASTER CHANNEL 1THIN (3:1),SHOWING ROUGHNESS IN THE CHANNEL

    Channel 3 thin (1:1) showed precise pattern transfer for most

    of the length (Figure 15), but had significant defects on

    portions of the device (Figure 16). The severity of defects on

  • 7/25/2019 Soft Lithography Replication of a Bioinspired Unidirectional Wicking Channel

    7/14

    FIGURE 16PRMASTER CHANNEL 3THIN (1:1),SHOWING SIGNIFICANT DEFECTS ON THECHANNEL

    For the thick wafers, all of the masters seemed to be defect-

    free enough to produce usable devices. There were issues

    with Channel 2 thick (6:1) where the corners of the

    trapezoidal features were darker than the center, as seen in

    Figure 17.This is most likely due to incomplete etching of

    the photoresist for the entire height of the feature. The middle

    of the feature is completely etched, but the corners were lessexposed and therefore rounded off instead of etching to a

    point.

    Channel 3 thick (1:1) showed wavy surface distortion across

    the width of the channel, as seen inFigure 18.The features

    appear to be mostly intact, so was unclear how this defect

    would affect the performance of the resulting PDMS device.

    For additional images of the photoresist-patterned masters,

    seeAppendix B: Supplementary Images.

    properties (i.e. Youngs modulus), but should not have a

    significant effect on the surface chemistry, contact angle, or

    other fluidic behavior.

    FIGURE 18

    PRMASTER CHANNEL 3THICK (1:1),SHOWING DISTORTION IN AREAS OF

    THE CHANNEL

    After degassing the PDMS in the beaker, the PDMS was still

    filled with air bubbles after being poured onto the wafers.

    After another round of degassing, the PDMS appeared to be

    free of air bubbles. After removing the PDMS from thewafers, the devices were examined for defects with a

    microscope. Most of the defects observed resulted from flaws

    in the PR master; however, several additional defects

    occurred during the casting process. We observed warping of

    the PDMS on the edge of Array 4 thin (3:1) (Figure 19). Most

    of the warping occurred near the scale bars and loading pads,

    so this defect was not believed to significantly influence fluid

    flow in the device.

  • 7/25/2019 Soft Lithography Replication of a Bioinspired Unidirectional Wicking Channel

    8/14

    leaving a dark hole behind. This was the only defect on the

    device, so it should not have a significant impact on fluid

    flow.

    FIGURE 20

    PDMSCHANNEL 3THICK (1:1),SHOWING DEFECT ON FEATURE AND

    SURFACE WAVINESS

    FIGURE 21

    PDMSARRAY 4THICK (3:1),SHOWING MISSING FEATURE DEFECT

    we used a diluted ethanol solution. A 35% ethanol solution

    produced a contact angle of 635. However, the channel was

    still too hydrophobic, as no channel filling was observed. A

    50% ethanol solution resulted in a contact angle of 515,

    which induced fluid filling on most devices.After preliminary testing, we discovered the fluid did not

    flow well in the thin channels. There was either no

    movement, or the fluid flowed over the features rather than

    in the channels. The thicker channels exhibited better wicking

    behavior, which is unsurprising. The thick channels have a

    greater aspect ratio (depth:width), which means that filling

    will be influenced more by the fluid interactions with the

    channel walls than with the floor. Since our interest is in

    measuring how the shape of the features determines the fluid

    flow, this is preferable. Channel 2 (6:1) exhibited slower

    filling speed than the other channels, while Channel 3 (1:1)

    tended to be the quickest. This result is expected due to the

    nature of capillary action: at small channel sizes and flow

    velocities, surface forces are dominant over body forces.

    Smaller features have a higher surface area to volume ratio,

    so the surface force pulls the liquid more effectively.

    Taking all of this into account, we decided to focus on three

    devices: Channel 1 thick (3:1), Channel 3 thick (1:1), and

    Array 4 thick (3:1). Two trials were performed in bothdirections on each device. The speed was found by measuring

    the time the fluid took to travel 7.1mm, the distance between

    two lines of college ruled lined paper underneath the device.

    We planned to use the distance markers on the edge of the

    channel, but they were difficult to see on the videos of the

    trials. Measured filling speeds are recorded inTable 1; full

    data are presented inAppendix C: Microfluidic Testing Data.

    TABLE 1MEASURED DEVICE FILLING SPEEDS

    Channel 1 Thick (3:1) Array 4 Thick (3:1)

    Speed (mm/s) Speed (mm/s)

  • 7/25/2019 Soft Lithography Replication of a Bioinspired Unidirectional Wicking Channel

    9/14

    the 7.1mm mark in the forward direction. To increase the

    speed so it was easier to measure, a higher concentration of

    ethanol was used to decrease the contact angle.Most of our data indicates the fluid flows faster in the reverse

    wicking direction. This may be due to many factors,including contact angle of the fluid, definition of the features,

    or material property differences between PDMS and PMMA.

    We had considerable difficulty controlling the contact angle

    of the fluid, and we found the flow through the devices was

    sensitive to the contact angle. When the device was too

    hydrophobic, the fluid would bead up on the loading pads

    instead of entering the channels. When the device was too

    hydrophilic, the fluid would flow quickly through the entire

    channel without being pinned as designed. With the 50%

    ethanol solution, we observed the fluid being pinned in the

    same manner as the researchers who created the scale design,

    as seen inFigure 22.In part a) the filling front approaches the

    capillary at the top, while it is pinned on the bottom. Then in

    b) and c), the filling front has advanced through the capillary

    while still being pinned at the top and bottom of the channel.

    Finally in d) the fluid from the capillary joins with the fluid

    pinned at the bottom, joining the two streams and advancing

    the filling front. In the reverse wicking direction, the fluid

    should not have had the same pinning and capillary fillingaction. The filling front did appear to be pinned, as seen in

    Figure 23.

    FIGURE 23IMAGE TAKEN OF PINNING IN REVERSE WICKING DIRECTION AT 1:1SCALE

    CONCLUSIONS

    This investigation attempted to produce a biomimetic

    unidirectional wicking channel from previously published

    descriptions of similar such devices. This topography isinspired by observations of the rain-harvesting morphology

    and behavior of the Texas horned lizard, whose skin is able

    to transport water to its mouth for ingestion. Such

    unidirectional wicking behavior is of great interest in

    microfluidic applications where close control of filling

    direction and rate may be required.

    We recreated previously successful wicking topographies in

    a PDMS casting instead of engraved PMMA. This approach

    offers greater control over feature size and definition,particularly at smaller scales, and could potentially allow for

    production at scale; however, channel depth is more limited,

    which is problematic for larger geometries. In addition, the

    natural hydrophobicity of PDMS must be overcome by

  • 7/25/2019 Soft Lithography Replication of a Bioinspired Unidirectional Wicking Channel

    10/14

    in greater surface adhesions greater relative to viscous and

    inertial body forces.

    Unexpected, however, was the result that filling proceeded

    preferentially in the reverse direction, which was expected to

    halt the advance of a filling front. No conclusive explanationis offered for this result, which is inconsistent with previous

    results [17,18,19]. An excessively low contact angle may

    help to explain why the device failed to halt filling in the

    reverse direction, but does not account for the filling rate

    being greater in reverse than in the forward direction. Further

    investigation is necessary to explain the reverse filling

    observed and devise improved methods for more robust

    control of wicking directionality.

    REFERENCES

    1.Zimmermann, M., Schmid, H., Hunziker, P. &Delamarche, E., Capillary pumps for autonomous

    capillary systems.R. Soc. of Chemistry7, 119-125

    (2007).

    2. Wikisource contributors, 1911 Encylopaedia

    Britannica/Capillary Action, Available at

    https://en.wikisource.org(2015).3. Washburn, E. W., The dynamics of capillary flow.

    Phys. Rev17, 273-283 (1921).

    4. Whitesides, G. M., The origins and the future of

    microfluidics.Nature368-373, 05058 (2006).

    5. Saha, A. A., Mitra, S. K., Tweedie, M. & Roy, S.,

    Experimental and numerical investigation of capillary

    flow in SU8 and PDMS microchannels with integrated

    pillars.Microfluidics and Nanofluidics7(4), 451-465

    (2009).

    6. Gunther, A. & Kreutzer, M. T., in Micro Process

    Engineering, Vol. 1: Fundamentals, Operations and

    Catalysts edited by Hessel V Renken A Schouten

    12. Chu, K.-H., Xiao, R. & Wang, E. N., Uni-directional

    liquid spreading on asymmetric nanostructured

    surfaces.Nature Materials9, 413:417 (2010).

    13. Toepke, M. W., Abhyankar, V. V. & Beebe, D. J.,

    Microfluidic logic gates and timers.Lab on a Chip7,1449-1453 (2007).

    14. Feng, J. & Rothstein, J. P., One-way wikcing in open

    micro-channels controlled by channel topography.J.

    Colloid and Interface Science404, 169-178 (2013).

    15. Sherbrooke, W. C., Integumental water movement and

    rate of water ingestion during rain harvesting in the

    Texas horned lizard, Phrynosoma cornutum.Amphibia-

    Reptilia25, 29-39 (2004).

    16. Sherbrooke, W. C., Scardino, A. J., de Nys, R. &

    Schwarzkopf, L., Functional morphology of scale

    hinges used to transport water: convergent drinking

    adaptions in desert lizards (Moloch horridus and

    Phrynosoma cornutum).Zoomorphology126, 89-102

    (2007).

    17. Comanns, P. et al., Moisture harvesting and water

    transport through specialized micro-structures on the

    integument of lizards.Beilstein Journal of

    Nanotechnology2, 204-214 (2011).

    18. Comanns, P. et al., Directional, passive liquid

    transport: the Texas horned lizard as a model for a

    biomimetic 'liquid diode'.J. R. Soc. Interface12,

    20150415 (2015).

    19. Buchberger, G. et al., Bio-inspired microfluidic

    devices for passive, directional liquid transport: Model-

    based adaption for different materials.Procedia

    Engineering120, 106-111 (2015).20. MicroChem, SU-8 2000 Permanent Epoxy Negative

    Photoresist: Processing Guidelines for SU-8 2025, SU-

    8 2035, SU-8 2050 and SU-8 2075, Available at

    h // i h

    https://en.wikisource.org/https://en.wikisource.org/http://www.microchem.com/http://www.microchem.com/http://www.microchem.com/https://en.wikisource.org/
  • 7/25/2019 Soft Lithography Replication of a Bioinspired Unidirectional Wicking Channel

    11/14

    APPENDIX A:PHOTORESIST THICKNESS DATA

    TABLE 2

    WAFER 1MEASURED THICKNESSES (TARGET:225 M)

    Channel Top Middle Bottom Average

    1 - - - -

    2 265 318 326 303

    3 - 233 - -

    4

    412

    359

    389

    386.7

    Note: - denotes no measurement taken of visibly flawed regions

    TABLE 3

    WAFER 2MEASURED THICKNESSES (TARGET:450 M)

    Channel Top Middle Bottom Average

    1 614 660 673 649

    2 556 616 620 597.3

    3 540 549 599 562.7

    4 482 510 524 505.3

  • 7/25/2019 Soft Lithography Replication of a Bioinspired Unidirectional Wicking Channel

    12/14

    APPENDIX B:SUPPLEMENTARY IMAGES

    Photoresist Master

    PRMASTER CHANNEL 2THIN (6:1),SHOWINGSUCCESSFUL PATTERN TRANSFER

    PRMASTER ARRAY 4THICK (3:1),CLOSE UP OFSMALLEST FEATURE

  • 7/25/2019 Soft Lithography Replication of a Bioinspired Unidirectional Wicking Channel

    13/14

    PDMS Devices

    PDMSCHANNEL 1THIN (6:1),SHOWING RESULTSOF SUCCESSFUL CASTING

    PDMSCHANNEL 2THIN (3:1),SHOWING RESULTSOF SUCCESSFUL CASTING

  • 7/25/2019 Soft Lithography Replication of a Bioinspired Unidirectional Wicking Channel

    14/14

    14 December 7, 2015

    APPENDIX C:MICROFLUIDIC TESTING DATA

    Video # Channel

    #

    Ethanol

    Conc.

    (%)

    Direction

    (away/towards

    ID)

    Time (1

    segmt,

    s)

    Time (2

    segmt,

    s)

    Speed, seg.1

    (mm/s)

    Speed, seg.2

    (mm/s)

    Video Notes Testing notes

    2306 4 70 t 5.12 n/a 1.38671875 fill @ moderate rate ~ length

    2305 3 50 t 1.55 6.4 4.580645161 2.21875 filled ~50%, then stagnation

    2303 4 70 a 3.7 21.16 1.775 0.671077505 filling, decent rate (~ channel)

    2301 3 50 a 2.66 7.48 2.669172932 1.898395722 goes to 3 lines filled whole channel

    2299 1 50 t 12.8 n/a 0.5546875 preferential filling in reverse (slight advantage in distance, noticeablyfaster)

    2298 1 50 a 20.97 n/a 0.338578922

    2296 3 50 a 2.22 5.25 3.198198198 2.704761905 rapid filling whole length

    2294 1 50 a 19.09 n/a 0.371922472 n/a fill approx. length

    2292 4 50 t n/a n/a n/a n/a runs for more than 60s add 30 both sides; slight preference for rev direction

    2289 1 50 t 7.72 38 0.919689119 0.373684211 two channels run atdifferent speeds

    some filling

    2287 3 50 t 1.2 3.26 5.916666667 4.355828221 rapid filling whole length

    2280 4 50 a n/a n/a n/a n/a takes 30s to travel halfsegment, no progressfor another 30s

    photo shows end of filling front @ stasis

    2279 4 50 t 27 n/a 0.262962963 n/a only one portion ofarray reaches 1 seg

    2278 2 35 both n/a n/a n/a n/a some flow in towarddirection, drop in

    middle

    filling both dir; noticeably faster in rev dir

    2277 1 50 both drop in middle filled both directions, preerentially in reverse direction

    2276 thin 4 both

    2275 4 both better capillary filling - no clear directional preference

    2274 3 both spread in reverse direction, strongly preferential

    2273 2 both beading - no flow

    2272 1 both little flow; contact angle: ? (pic @ 4:33)

    2270 1 both