Archive for the ‘Uncategorized’ Category

Multilayer photolithography with manual photomask alignment

Frank Benesch-Lee, Jose M. Lazaro Guevara, and Dirk R. Albrecht

Worcester Polytechnic Institute, Worcester, MA 01609 USA

Why is it useful?

Modern microfluidic devices can incorporate channels of different heights to fulfill their designed function. Examples include hydrodynamic focusing [1], cell traps [2], and chambers that isolate cellular components [3]. These devices are fabricated from a multilayer SU-8 photoresist master mold. Each layer height requires a separate set of photolithographic steps, including photoresist spin, photomask alignment, exposure, and bakes, followed by a development step at the end to reveal the 3D resist pattern.

Mask aligners have microscopes and stage micrometers for precise, micron-scale alignment of each layer’s photomask with visible marks on the substrate wafer.  They are indispensable tools for creating multilayer patterns with accurate registration, but while available in cleanrooms at many research universities, their substantial expense may place them out of reach of teaching institutions and individual laboratories.

In contrast, single-layer microfluidics can be prepared using an inexpensive UV light source, or even a self-made one [4]. In principle, manual photomask alignment could be made under a microscope, then brought to the UV source, yet this poses several complications. First, alignment features can be very difficult to see using inexpensive microscopes or stereoscopes, especially in thin SU8 layers, due to poor contrast between exposed and unexposed regions before development. Second, misalignment can occur during movement to the exposure system.

Here we present a manual photomask positioning method that yields a 50 µm accuracy, without the aid of a mask aligner.


What do I need?

  • Equipment and supplies for photolithography:
    • Spin coater, and UV exposure system
    • Substrate wafer and SU-8 photoresist
  • Small microscope (e.g. USB) or stereo microscope
  • Photomask transparencies for each layer
  • Scotch tape
  • Fine-tip permanent marker
  • Straight razor blade
  • Cutting mat
  • 4 small (3/4”) or mini (1/2”) binder clips
  • Glass plate, approx. 4 x 5”, compatible with exposure system


What do I do?

  • Cut the photomasks from the transparency sheet, leaving 4 corner tabs. Align the two masks relative to each other under the microscope (Figure 1a) and clip them together with a binder clip. Ensure correct mask orientation and check alignment accuracy at multiple alignment marks across the mask. (Note that horizontal alignment accuracy with a stereomicroscope is low, because each eye’s optical path is angled 5 – 8 degrees, whereas vertical alignment is unaffected. Align in the vertical direction first then rotate the masks 90 degrees to ensure accurate alignment in both horizontal and vertical directions.) Add binder clips to each corner (Figure 1b), and verify alignment. Next, remove one binder clip at a time and use a straight razor blade to cut a sharp V-notch into each tab, through both masks. Press the blade straight down to avoid shifting the alignment. Replace the binder clip, and proceed to the next corner until all 4 notches are cut (Figure 1c).





  1. Spin the first layer of SU-8 onto the wafer to the desired thickness and prebake. Attach 4 pieces of scotch tape onto the bottom of the wafer so that the sticky side faces up (Figure 2a). Position the first mask on the wafer, pressing gently to adhere it to the tape tabs. Use a fine-tip marker to trace the alignment notches (Figure 2b) onto the scotch tape (Figure 2c).  Transfer to the UV exposure system and expose.  Carefully remove the mask without detaching the scotch tape from the wafer and postbake.  Scotch tape is compatible with 95 °C baking.  Apply an additional piece of tape to cover the sticky tape tabs to protect the marker from smearing and allow smooth alignment of the next mask.




  1. Spin coat the next photoresist layer and prebake (Figure 3a). Mount the wafer onto a glass plate with a loop of scotch tape to keep it in place. Position the second mask onto the wafer, ensuring that alignment “V” markings are centered within each alignment notch and across all 4 corners (Figure 3b). Affix the mask to the glass plate with thin (2-3 mm wide) pieces of tape, and adjust alignment as necessary.  Carefully transfer the glass plate with wafer and aligned photomask for exposure (Figure 3c).



  1. Repeat step 3 for any additional layers. Remove the tape tabs and develop the photoresist. Evaluate alignment accuracy under a microscope (Figure 4).




In this tip, we present a method for manual alignment of multiple transparency photomasks.  We achieved repeatable accuracy of <100 µm and as good as 50 µm (Figure 4a). These accuracies are within required tolerances of many multilayer designs (Figure 4b).  In many cases, minor design alternatives can relax alignment tolerances, such as in a trap design containing a thin horizontal channel that allows fluid bypass but captures larger objects (Figure 4c). In this example, a 100 µm wide bypass channel only partially covered the trap indentations, whereas widening the bypass channel to 400 µm enabled a functional device despite slight misalignment.  Overall, this simple method allows fabrication of microfluidic device molds containing multiple layer heights, without expensive mask alignment equipment, to an accuracy of at least 50 µm.  Furthermore, after alignment marks are cut, no microscope is needed at all during the photolithography process, speeding the fabrication of multiple masters.


Funding provided by NSF IGERT DGE 1144804 (FBL), Fulbright LASPAU (JMLG), University of San Carlos of Guatemala (JMLG), NSF CBET 1605679 (DRA), NIH R01DC016058 (DRA), and Burroughs Wellcome CASI (DRA).Acknowledgments:



  1. Chih-Chang, C., H. Zhi-Xiong, and Y. Ruey-Jen, Three-dimensional hydrodynamic focusing in two-layer polydimethylsiloxane (PDMS) microchannels. Journal of Micromechanics and Microengineering, 2007. 17(8): p. 1479.
  2. Erickson, J., et al., Caged neuron MEA: A system for long-term investigation of cultured neural network connectivity. Journal of Neuroscience Methods, 2008. 175(1): p. 1-16.
  3. Taylor, A. M., et al., A microfluidic culture platform for CNS axonal injury, regeneration and transport. Nature Methods, 2005. 2(8): p. 599-605.
  4. Erickstad, M., E. Gutierrez, and A. Groisman, A low-cost low-maintenance ultraviolet lithography light source based on light-emitting diodes. Lab on a Chip, 2015. 15(1): p. 57-61.


Digg This
Reddit This
Stumble Now!
Share on Facebook
Bookmark this on Delicious
Share on LinkedIn
Bookmark this on Technorati
Post on Twitter
Google Buzz (aka. Google Reader)

Electric drill/driver as centrifuge with 3D-printed custom holders for non-conventional containers

Minkyu Kima, Guanya Shib, Ming Panc, Lucas R. Blaucha, and Sindy K.Y. Tanga*

aDepartment of Mechanical Engineering, Stanford University, Stanford, CA 94305, USA

bUndergradute Visiting Research Program, School of Engineering, Stanford University, Stanford, CA 94305, USA

cDepartment of Materials Science and Engineering, Stanford University, Stanford, CA 94305, USA


Why Is This Useful?

Many processes in biological and chemical preparation require centrifugation steps. The transfer of samples between the original sample container and tubes required by commercial centrifuges increases the risk of sample contamination, and often leads to the loss of samples. Commercial centrifuges are also not readily available outside laboratory settings. Here we show the design of a simple 3D-printed holder for attaching to the chuck of an electric drill/driver which we use as a centrifuge. The advantages of this method include: 1) The holder can be designed to hold non-conventional containers (e.g., syringes, glass vials, capillaries). 2) Electric drill/drivers are more widely available than centrifuges. We show that a variety of samples (e.g., water-in-oil emulsions, cell suspensions, food and drinks, wet soil) in various containers can be centrifuged with our method. This method should be useful for field work outside of the laboratory, and for the wider DIY community interested in home-based applications that require centrifugation, such as blood separation and related diagnostics, separation of interstitial water from wet soil for pollution detection, extraction and identification of allergens in food samples, and fluid clarification (e.g., olive oil, wine) by accelerating sedimentation.

What do I need?

Figure 1. Photo of components needed.

  1. Electric drill/driver (DeWalt DC742KA Cordless Compact Drill/Driver Kit [1]).
  2. 3D-printed custom holder.
  3. 3D-printed custom support for the drill/driver.
  4. Four 1 mL-syringes (NORM-JECT, Part No.: 4010-200V0) as an example of non-conventional sample containers.
  5. One long bolt (Pan Head Machine Screw, Zinc, #8 x 1-1/4”) and one matching nut (Hex Nut, Zinc, #8-32). The dimensions of the bolt should match the chuck of the drill/driver.
  6. Four short bolts (Pan Head Machine Screw, Zinc, #6 x 1/2”) to secure the 1 mL-syringes to the 3D-printed custom holder.

What do I do?

  1. Design a custom holder using Solidworks or other CAD software.
    1. Measure the outer diameter (w1 = 6.5 mm) of the 1 mL-syringe (Fig. 2a). We included a 0.5 mm-tolerance in deciding the width of the slot (w2 = 7 mm) into which the 1 mL-syringe will be secured (Fig. 2b).
    2. Decide the angle (q = 60°) at which the 1 mL-syringe will be tilted relative to the plane of rotation.
    3. Decide the length (w3 = 80 mm) and the thickness (w4 = 15 mm) of the holder.
    4. Measure or identify the outer diameters of the short and long bolts, and use these values for f3 and f5 respectively.
  2. Design a custom support for the drill/driver (Fig. 2c). The dimensions of this support are not critical so long as the drill/driver is stable and does not topple during operation.
  3. 3D-print the holder and the support. We used a 3D-printer by ROBO 3D [2]. The resolution of the 3D-printer in xyz direction is 100 mm. The material we used was polylactic acid (PLA).
  4. Assemble the centrifuge (Fig. 2d).
    1. Put one long bolt through the center of the holder and tighten with the nut.
    2. Insert the bolt into the chuck of the drill/driver and tighten the bolt by pushing the trigger of the drill/driver a few times.
    3. Make sure the bolt is fixed in the chuck and aligned to the drill/driver.
    4. Place the drill/driver in the 3D-printed support.
    5. Secure four 1 mL-syringes to the holder using the four short bolts.
  5. Start the centrifuge by pushing the trigger of the drill/driver for 5-10 minutes. The plane of rotation should be parallel to the floor.
  6. Unscrew the short bolts to remove the syringes.
  7. If desired, measure the rotational speed of the drill/driver before inserting the real samples. We used the SLO-MO mode in iPhone to calibrate the rotational speed of the drill/driver [3] (Fig. 2e).
  8. Results (Fig. 3).
    1. Water-in-oil emulsion: Micron-sized uniform water-in-oil droplets were collected in a 1 mL-syringe. After a needle was connected to the syringe, the syringe was secured to the 3D-printed holder with needle pointing up. The holder was balanced before centrifugation by adding another syringe containing an equal weight of fluids to the opposite side of the holder. We centrifuged the sample at a speed of 374 rpm for 10 minutes. Droplets were then injected into a microchannel to measure the change of volume fraction before and after centrifugation. The volume fraction was defined as the ratio of the total volume of water droplets to the total volume of fluids filling up the channel. After centrifugation, the volume fraction of the emulsion increased from 60% to 86% without a change in the size of the droplets. Neither break-up nor coalescence of the droplets was observed.
    2. Stentor coeruleus: To demonstrate the concentration of cell suspensions, we used Stentor coeruleus ( as a model. We filled a 3 mL-syringe with about 20 Stentor cells suspended in 2 mL of aqueous culture media (concentration ~ 10 cells/mL). A needle was connected to the syringe which was then secured in the 3D-printed holder with the needle pointing down. The cell suspension was centrifuged at a speed of 374 rpm for 5 min. The cells were concentrated at the bottom, close to the entrance into the needle. It was then possible to inject this concentrated cell suspension through the needle into a polyethylene tubing. Fig. 2-i) shows the microscopic image of the cells in about 15 mL of aqueous culture media in the tubing (concentration ~ 400 cells/mL).
    3. Korean rice wine (Makgeolli): A separate holder was designed to hold a 20 mL-glass vial. The vial was filled with 12 mL of Korean rice wine and centrifuged at a speed of 1252 rpm for 10 minutes. The sediment was clearly observed after centrifugation. On the other hand, the sediment was not observed for more than 20 minutes without centrifugation.
    4. Wet soil: 10 mL of wet soil in a 20 mL-glass vial was centrifuged at a speed of 1252 rpm for 10 minutes. Interstitial water was separated from the soil.

Figure 2. a) A 1-mL syringe as non-conventional container. b) Drawing of 3D-printed holder generated by SolidWorks. c) Drawing of 3D-printed support generated by SolidWorks. d) Photograph of experimental setup. The 3D-printed holder was connected to the drill/driver placed on the 3D-printed support. Four 1 mL-syringes were then tightened using the short bolts. e) Calibration plot of the rotational speed versus the trigger levels on different modes of the drill/driver. Mode 1 and Mode 2 indicate different gear settings in the transmission of the drill/driver. The numbers 1 to 3 in each mode indicate user-defined trigger levels. The rotational speed ranges from 120 rpm to 1252 rpm. The rotational speeds were measured using SLO-MO function in iPhone 6.

Figure 3. a) Photographs of emulsion in 1 mL-syringe and microscopic images of emulsion injected into a microchannel before and after centrifugation. The scale bar in the photographs is 5 mm. b) Photographs of Stentor cells in 3 mL-syringe before and after centrifugation. The red arrows indicate individual cells. The scale bar is 5 mm. i) Microscopic image of 6 Stentor cells in a polyethylene tubing after centrifugation. The scale bar is 300 mm. c) Photographs of Korean rice wine in 20 mL-glass vial before and after centrifugation. The red box indicates sediments. The scale bar is 10 mm. d) Photographs of wet soil in 20 mL-glass vial before and after centrifugation. The red box shows interstitial water separated from wet soil. The scale bar is 10 mm.

What else should i know?

  1. The centrifugal force can be increased by lengthening the arms (w3) holding the containers.
  2. The rotational speed can be measured using a high-speed camera or a smart phone with slow-motion videotaping capability, so long the frame rate is sufficient for the rotational speed used.
  3. The load in the centrifuge should be balanced.
  4. For safety purposes, safety goggles should be worn. The centrifuge should also be placed inside a safety barrier (e.g., a sturdy laundry basket). The safety instructions for the drill/driver should also be observed.
  5. After centrifugation for 10 minutes, we found that the drill/driver started to heat up. If centrifugation time longer than 10 minutes is needed, it should be possible to perform multiple rounds of 10-min centrifugation steps with breaks in between to cool down the drill/driver.

In this work, we demonstrated that 3D-printed holders attached to an electric drill/driver can be used for the centrifugation of samples in non-conventional containers. As 3D-printers and hand drills are easily accessible, we expect this tip to find immediate use in settings outside laboratories for field work, and also at home for DIY users.


    We acknowledge support from the Stanford Woods Institute for the Environment and the National Science Foundation (Award #1454542 and #1517089).





Digg This
Reddit This
Stumble Now!
Share on Facebook
Bookmark this on Delicious
Share on LinkedIn
Bookmark this on Technorati
Post on Twitter
Google Buzz (aka. Google Reader)

Simple and Low-cost Contact Angle Measurements Using a Smartphone with a PDMS-Lens

Jonas M. Ribea, Nils R. Skovb, Ole-Andreas K. Kavlia, Armend G. Håtia, Henrik Bruusb and Bjørn T. Stokkea

a Department of Physics, Norwegian University of Science and Technology, NO–7491 Trondheim, Norway

b Department of Physics, Technical University of Denmark, DK–2800 Kongens Lyngby, Denmark

Why Is This Useful?

Contact angle measurements are important for characterizing the wettability of a liquid to a solid surface. In microfluidics they are of special interest as they provide insight into the intermolecular interactions between the sample liquid and the microchannel surface. Contact angle measurements are also important when assembling polydimethylsiloxane (PDMS) devices using oxygen plasma bonding. For optimal bond strength the water contact angle of plasma treated PDMS should be minimized as shown by Bhattacharya et al. [1] A current hurdle in measuring contact angles is the requirement of a setup that is expensive and non-portable. Here we show a method for measuring contact angles using materials and equipment found in a typical microfluidics lab.

What do I need?



  • Smartphone
  • Digital scale
  • Desiccator with vacuum pump
  • Oven
  • Syringe pump (optional)
  • Light source

For measurements:

  • Pipette (0.5–3μL)
  • Sample (e.g. deionized (DI) water or other liquid sample)

What do I do?

Prepare PDMS:

  1. Weigh 10:1 PDMS (Sylgard 184) in a plastic cup on the digital scale
  2. Mix the PDMS by hand using a plastic spoon
  3. Degas the PDMS in a desiccator to remove the bubbles

Make PDMS-lens:

  1. Use the tip of the plastic spoon handle (or a pipette) to place a small droplet of uncured PDMS in the center of a glass cover slip. Repeat with various amounts of PDMS to obtain lenses with varying magnification.
  2. Mount the cover slips upside down (e.g. between two glass slides) and cure the PDMS hanging at 70 °C for 15 min. Longer curing times might be necessary, if the drop is relatively large.
  3. Center the PDMS-lens over the camera of your smartphone and fixate it using tape.
  4. Test the focus of your camera. For our camera setup the best images were captured with lenses that focus around 2 cm.

Contact angle measurements:

Smartphone contact-angle setup: (A) Focus test of a PDMS lens. (B-C) The smartphone mounted on a syringe pump. The PDMS-lens is mounted on the front facing camera of an iPhone 6S and the sample is centered in front of the lens. The sample is mounted on the pusher block of a syringe pump which can be moved to adjust the focus.

  1. Make a sample stage preferably using a syringe pump or some other system that you can move. We mounted the smartphone on the syringe holder block with the camera pointing towards the pusher block. Make a sample holder on the pusher block using glass slides or other consumables found in the lab. Align the center of the stage with the center of the camera. Tip: aligning is easier if done using the sample that you want to measure. Put the sample on the block and move it into focus by releasing the pusher block and sliding it away/towards the camera. Increase the height of the stage until the top of the sample is centered in the camera.
  2. Place the light source behind the sample and illuminate the stage evenly. Tip: put the sample stage in front of a white wall and light up the wall for a homogenous background and optimal contrast.
  3. Place a small drop (0.5–3 μL) of DI water on top of the sample using a pipette. Place the drop near the sample edge closest to the camera.
  4. Move the sample edge into focus. Block out ambient light in the room.
  5. Measure the contact angle of the drop in the image e.g. using ImageJ [2] software with a plugin for contact angle measurements [3] or get a rough estimate using an app on your smartphone.[4]

Contact angle measurements of water on PDMS: (A) Raw image from iPhone 6S front-facing camera with PDMS-lens. (B) Direct measurement using app on smartphone (based on θ/2 calculation) (C-E) ImageJ measurements using DropSnake plugin. Unmodified PDMS (C) and PDMS treated with oxygen plasma with increasing intensity (D-E).

What else should I know?

The focal length of the PDMS-lens is determined by the volume of PDMS used as described by Lee et al. [5]. However, it is difficult to control the volume of PDMS using a pipette due to the high viscosity of PDMS. We recommend making a range of lens sizes and testing them on your smartphone camera to see which gives the right focal length. If your digital scale has milligram precision you can measure the amount of PDMS used for each lens. The mass of each PDMS-lens is typically less than 10 mg. You can decrease the focal length further by adding PDMS to an already cured lens. Modern smartphones have both a rear-facing and a front-facing camera and in our experience the drop focusing was easier when using the front facing camera. The images taken here were captured with an iPhone 6S from Apple using the front-facing camera with a 5MP sensor. The weight of the cured PDMS lens was 7 mg.

Tip: you can also remove the PDMS-lens from the cover slip and place it directly on your camera. Although, it might be more difficult to center.

Calculating contact angles from images of sessile drops can be done using a range of techniques.[6] If the drop volume is small and the contact angles are not extreme, we can generally neglect droplet distortion due to gravitational effects. Extrand and Moon [7] calculated that gravitational effects can be neglected for a water droplet sitting on a hydrophilic surface (θ=5°) if its volume is less than 5 μL and less than 2.7 μL on a hydrophobic surface (θ=160°). If we assume the drop to be spherical, the contact angle can be estimated by multiplying the angle between the base and the height of the droplet by 2. This is referred to as the θ/2-method and is implemented by e.g. the Contact Angle Measurement app [4] for iOS. Sessile drop measurements are generally limited by the experimental setup and operator error, but typically has a precision of ±3°.[8] Image-processing algorithms relying on curve fitting of the droplet outline can enhance reproducibility. ImageJ [2] with DropSnake-plugin [3] uses active contours (energy minimization) to track the outline of the drop and calculate contact angles. This increases precision, but is slower and currently requires analysis on a separate computer.


The Research Council of Norway is acknowledged for the support to the Norwegian Micro- and Nanofabrication Facility, NorFab (197411/V30).


  1. S. Bhattacharya, A. Datta, J. M. Berg and S. Gangopadhyay, J. Microelectromech. S., 2005 14, 590–597
  2. ImageJ software
  3. DropSnake ImageJ-plugin for contact angle measurements
  4. Contact Angle Measurement iOS app (Japanese)
  5. W. M. Lee, A. Upadhya, P. J. Reece, and T. G. Phan, Biomed. Opt. Express, 2014, 5, 1626–1635
  6. Y. Yuan and T. R. Lee, Surface Science Techniques, Springer, Berlin/Heidelberg, 2013, 51, 3–34.
  7. C. W. Extrand and S. I. Moon, Langmuir, 2010, 26, 11815–11822.
  8. A.F. Stalder, G. Kulik, D. Sage, L. Barbieri and P. Hoffmann, Colloids and Surfaces A: Physicochem. Eng. Aspects, 2006, 286, 92–103.
Digg This
Reddit This
Stumble Now!
Share on Facebook
Bookmark this on Delicious
Share on LinkedIn
Bookmark this on Technorati
Post on Twitter
Google Buzz (aka. Google Reader)

Universal and Multi-material Bonding Method for Rapid and Low-cost Assembly of Microfluidic Devices

Ya-Yu Chiang, Nikolay Dimov, Nicolas Szita

Department of Biochemical Engineering, University College London, London, United Kingdom


Why is this useful?

The packaging of micro-systems relies strongly on the capability to bond different types of materials reliably whilst maintaining the microstructures and their dimensions. However, the bonding of different materials each with their specific physical and chemical properties frequently turns into a tedious, thus time consuming operation; often, the choice of materials and microfabrication techniques are limited by the bonding technique. Particularly challenging for bonding can be combinations of quartz, glass or silicon with polymers and metals.

Here we demonstrate a rapid, low-cost, UV-irradiation based bonding method, which is suitable for the bonding and assembly of quartz-to-silicon, quartz-to-metal, quartz-to-polymer, quartz-to-quartz devices.  We demonstrate in detail on the more challenging combinations, namely the bonding of a quartz slide to an aluminum sheet. In our example, the aluminum sheet contains the microfabricated structure. The same procedure is applicable for the other material combinations, i.e. quartz-to-silicon, quartz-to-polymer, quartz-to-quartz or quartz-to-metal for a metal other than aluminium; the main requirement for implementing our method is that at least one material is transparent to UV light.

What do I need?

  • Aluminum sheets, thickness of 1 mm (e.g. AW6082-T6, Smiths Metal Centres, UK)
  • Micro milling machine (e.g. CNC MicroMill GT, Minitech, US)
  • Flat head end-mills 0.25 mm, and 2 mm (e.g. PMT Endmill, US)
  • Plasma Cleaner (e.g. PDC-32G-2, Harrick Plasma, UK)
  • UV-curing adhesive (e.g. NOA 61, Noland Products, UK)
  • UV lamp, 100 W, 365 nm (e.g. B-100 AP, UVP, Cambridge, UK)
  • Quartz microscope slide, fused quartz, 25.4 × 76.2 x 1 mm3 (e.g. 42297, Alfa Aesar, UK)

What do I do?

1. Device design and leveling of the metal substrate

  • Draw your device design in any available computer aided design (CAD) software. As the surface roughness of the metal substrate can vary, a polishing step is recommended prior to the actual fabrication.
  • Generate the G-code for the CNC machine using any CAM/CAD software. Two separate files are required: one for the polishing of the substrate, and one for the actual design.

2. Micro milling

  • Clamp the aluminum substrate on the table of the milling device. Make sure that you do not bend the material.
  • Set the initial coordinates (X0, Y0, Z0) for this work.
  • Polish the aluminum substrate with 2 mm flat head end-mill.
  • Change to the smaller diameter tool (0.25 mm).
  • Mill the designed structure in the aluminum sheet with the 0.25 mm end-mill.

3.  Cleaning the aluminum substrate from residues.  Dust is removed first with water, and then the surface is cleaned  first with ethanol, and then with compressed air. Finally the substrate is dried in an oven (120ᴼC, 30 min).

4. Plasma activation of the quartz microscope slide. Place the quartz microscope slide inside the Plasma Cleaner. The plasma treatment is a ‘surface process’, therefore the surface that is about to be bonded should be facing towards the center of the chamber.

  • Evacuate the chamber until a working pressure of 500 mTorr at a constant inflow of air is established.
  • Switch the plasma on at 27 W, which is the highest intensity available for the specified Plasma Cleaner.
  • ‘Turn off’ the plasma after 90 seconds.
  • Vent the chamber of the Plasma Cleaner by opening the needle valve and allowing air to enter through the flow meter.
  • Remove the activated piece of substrate from the Plasma Cleaner.

5. Bonding

  • Align the substrates (and thus enclose the micro fabricated structures) by firmly pressing the activated quartz surface to the aluminum sheet. A fine interfacial gap is forming between the quartz and aluminum surfaces.
  • In case you have a large chip or thin fragile substrates you may need to carefully clamp the substrates together.
  • Prime the gap with the adhesive while holding the two substrates of your device together. In order to do so, place a small drop of adhesive to one edge, i.e. to the gap between the two substrates. The adhesive will flow into the gap due to capillary action. Thick substrates will be held together sufficiently by the adhesive film. The flow of the adhesive will stop at the edge of the microfabricated structures as a results of surface effects (surface tension and wetting angle). Inspect whether the device is completely filled with the adhesive. Add more of the adhesive if necessary.
  • Cure the completely primed device by exposing it to UV-light, 365nm @ 100 W for 5 to 10 minutes.
  • Place the device into the oven at 50°C. According to the supplier’s specifications, the bond reaches its maximum strength after 12 hours at 50°C. Alternatively, for temperature-sensitive materials, longer incubation times at room temperature are also feasible.

The main advantages of the presented bonding method are as follows:

1. Hybrid microfluidic devices can be easily bonded.

2. The method is relatively simple and does not require clean-room conditions.

3. The method works with any UV transparent material as long as the surfaces are clean, smooth and as long as they can promote the capillary action necessary for the priming with adhesive.

4. It is an economic bonding method. An expected 30 mL of UV-curing adhesive should be enough for the bonding of over hundred microfluidic devices. Each assembly will thus cost less than £0.2 GBP (or approximately $0.3 USD).

What else should I know?

Q1. What processes do you use to create the holes in the quartz slide?

A1. The quartz slides are drilled with diamond drill bit (Eternal tools, UK), 1 mm in diameter, and a bench drill (D-54518, Proxxon , Germany) at 1080 rpm.  This is a slow operation as the process is closer to grinding rather than drilling. To avoid crack formations in the quartz slide and to cool diamond bit a droplet of water is applied on the surface of the quartz. After each cycle grinded quartz debris may be accumulating at the bottom of the hole; it can be removed by using a pipette and cooling liquid.

Q2. Have you ever tried this method with channel geometries that are disconnected? For example, a channel  layout shaped like an “O” that would prevent adhesive wetting from the edge of the slide?

A2. We had bonded successfully channels with complex, serpent geometries. For “O”-shaped channels we use additional feed, a hole, drilled in one of the substrates that allows the adhesive to spread.

Q3. Does the adhesive ever “burst” and enter the channels? If so, what methods do you use to minimize the chances of this happening?

A3. Yes, it happens occasionally that the adhesive fills the channel.

To prevent this: minimum amount of glue is applied at a time, and also the propagation of the front needs to be monitored. We wait until the glue reaches the channel edge, and then we place the assembly under the UV-light for curing.

If the channel is filled with small amount of adhesive, the glue could be washed out with a bit of ethanol or acetone.

Completely filled channel requires disassembly, cleaning with acetone or ethanol of the substrates. Afterwards, the procedure can be repeated with less glue.

Digg This
Reddit This
Stumble Now!
Share on Facebook
Bookmark this on Delicious
Share on LinkedIn
Bookmark this on Technorati
Post on Twitter
Google Buzz (aka. Google Reader)

Highly precise alignment for the rapid fabrication of Plexiglas® microfluidic devices

G. Simone

University of Naples Federico II, Piazzale Tecchio 80, 80125 Napoli, Italy.

Why is this useful?

Most of microfluidic devices use channels with rectangular cross-sections. The microfabrication of rectangular shaped channels is straightforward with standard tools such as photolithography.

Fluid dynamics, rheology, soft matter and, recently, biology-based investigations need circular cross-section microchannels; indeed, pre-fabricated capillaries are normally used to carry out the studies. However, capillaries are impractical for some investigations requiring complicated designs.

For Plexiglas® (or other plastic) devices, microfabrication by micromilling is a low cost procedure, which, in the last decades, has gained popularity in the field of microfluidic applications.1-2 The fabrication and the sealing of Plexiglas® microchannels with circular cross-section can be challenging.

Here, a method of fabrication of plastic microfluidic devices with circular cross-section is presented. The method is low cost and it can be performed by minimally trained users. The alignment step builds on a procedure first introduced by Lu et al. that used circular magnets to align layers of polydimethylsiloxane (PDMS).3 The protocol is validated for circular cross-section channels, but it can be used for fabricating rectangular channels or special inlets as well.

What do I need?

  • Plexiglas® sheets (thickness 1mm) Rohm Italy
  • CNC MicroMill (Minitech, US)
  • Ball nose end-mills (0.001 inch, PMT Endmill)
  • Magnets 4 mm × 4 mm× 4 mm, DX
  • Microscope Bresser 58-02520
  • Clamps RS Italy
  • Ethanol
  • Microscope (or stereomicroscope)
  • Glass slides for microscopy (50 mm × 75 mm)

What do I do?

1. Computer aided design of the device (CAD).

  • Design the microfluidic channels with CAD software. Fig. 1a displays the top and bottom layer of the main channel and, in particular, aligned square through-holes in each layer.
  • Translate the CAD file to Computer Numerical Control (CNC) code for micromilling.

2. Micromilling

  • The endmill should be aligned to set the point z=0 by using a microscope.
  • Then, the microchannels (top and bottom) must be milled. Channel with depends on the endmill diameter.4 It is worth emphasizing that setting the correct work parameters (such as tool speed and depth) is crucial. Indeed, if the latter are not appropriate, the plastic workpiece develops internal stresses that during the bonding result in cracks and chip breakdown.
  • Finally, the through holes and the frame have to be milled. The number of through holes for locating the magnets depends on the dimension of the microfluidic device. It is good practice for the rectangular through-holes to have a pitch of roughly 10 mm (this dimension depends on the size of the microchannels and of the magnets).

3. Alignment

  • Align the top and bottom layers with a stereomicroscope (Fig. 1b). The bottom layer should be set on a solid surface and the magnets inserted into the through-holes. Then, the second layer should be placed on the first and a second set of magnets inserted in the square through-holes of the second layer.
  • The magnets located in the first and second layer naturally provide a good “first alignment” of the microchannels, while leaving freedom to slightly adjust the layers (i.e., this is a reversible sealing). Once the magnets are fixed, the quality of the alignment should be checked with a microscope.

    Fig. 1. The Fabrication. a. CAD design of a microfluidic channel with circular cross section. The length of the microfluidic device is 40mm, the width is 15mm. The depth of grooves is 50μm, width 25μm. b. Magnets employed for the bonding. C. Clamps used for sealing the microchannel.

4. Bonding

  • Clamp the Plexiglas® layers together and submerge assembly in ethanol for 15 minutes (Fig. 1c). For optimal bonding, the clamping should be done with clamps positioned at the edges of the Plexiglas®.5
  • After 15 minutes, the sealing of the microfluidic channels should be checked. If the Plexiglas® is sufficiently bonded, the magnets can be removed and glass slides placed on either side of the Plexiglas® layers. Clamp the glass slides and allow the assembly to rest for another 5-15 minutes.
  • The device is ready to be used for different applications as shown in Fig. 2a. Capillary tubing can be easily connected to the channel for modular design (Fig. 2b).

    Fig. 2. Examples of microfluidic devices. a. Perfusion of the samples through hole at the inlet. b. Perfusion through capillaries.

In conclusion, analyzing the protocol, the following advantages can be emphasized:

  1. The process is designed for different materials, but it fits perfectly with Plexiglas®
  2. The equipment necessary for the fabrication and assembly includes simply a micromilling machine and a (stereo)microscope
  3. The use of square magnets (instead of circular ones) allows for more precise alignment due to further restriction to the sliding of the top and bottom layers.


  1. G. Simone, G. Perozziello, J. Nanosc. Nanotech., 2010, 11, 2057.
  2. G. Simone, RSC Advances, 2015, 5, 56848.
  3. J-C Lu, W-H Liao, Y-C Tung, J. Micromech. Microeng. 2012, 22, 075006-075014.
  4. G. Perozziello, G. Simone, P. Candeloro, F. Gentile, et al. Micro and Nanosystems, 2010, 2, 227-238.
  5. G. Medoro, G. Perozziello, A. Calanca, G. Simone, N. Manaresi, 2010, US Patent App. 13/257,545.

Digg This
Reddit This
Stumble Now!
Share on Facebook
Bookmark this on Delicious
Share on LinkedIn
Bookmark this on Technorati
Post on Twitter
Google Buzz (aka. Google Reader)