Nanobiotechnology Project 

Nathan Cady, Samuel Chen, Andrea Kao and TJ Zieziulewicz April 9, 2002

Due to the fact that our proposed device requires pumping approximately 10ml of fluid through the device, we determined that we would need some type of pumping system to move this fluid.  We considered some of the options presented in the lecture by Scott Stelick, but decided that since the collection of blood and other samples would likely be with needles and syringes, a syringe pump system could be utilized.  Alternatively, we could inject the fluid into a reservoir into the device and move this fluid with an on-board pump.  After some deliberation about this issue, we decided that the syringe pump would be the easiest means of moving fluids in our device.

Syringe pumps are not “miniature” by any stretch of the imagination, so we have been considering the integration of a syringe pump-like system into our device.  The most important aspect of a syringe pump is the ability to accurately control the volume of fluid that is being pumped.  Therefore, whether the pump is controlled by a worm drive (simple motor and gears) or by a piezoelectric motor, there must be some controller that determines the rate of pumping.  Piezoelectric motors might be the easiest to use, especially since they are mechanically simpler than a gear driven syringe pump.  In addition, piezoelectric motors often have lower power requirements which would also make them the better choice for our device. 

The control of the syringe pump is the most complicated design issue and will likely be best integrated with the rest of the control portions of the device.  The signal analysis and output must be handled by some kind of computer or microprocessor and therefore we have decided that pump control should also be controlled by the same processor.  This would be fairly simple at first, simply integrating all of the control and analysis in a laptop computer, but later designs would likely utilize some kind of integrated microprocessor.

        We have considered most of the nanofabrication issues involved with our proposed design and have made several changes to the overall design and fabrication.  The most significant change to the original design has been the change from a glass to silicon substrate and the integration of the filter into this substrate.  Our original proposal had the filter integrated into the PDMS cover of the device.  Upon further discussion, however, we decided that the filter needed to be fabricated more accurately than PDMS could offer.  The minimal feature size of the filter is approximately 2mm.  Although this is possible to construct in silicon and possibly transfer to PDMS, the reliable transfer of the pattern into PDMS is questionable.  Therefore, we decided to simply integrate the filter into the solid substrate that will also include the electronics.  Since silicon is easier to pattern and etch than glass, we also decided that a silicon base would be used. 

        The use of a silicon base creates a problem that was initially avoided by using a glass substrate.  Because we want to evaporate gold electrodes onto the substrate, we have to somehow insulate the silicon surface to prevent shorting across the wafer.  After consideration of this problem, we decided that it would be necessary to grow a layer of insulating silicon dioxide on the fabricated wafer after microfabrication of the filter.  Once this has been done, it would be relatively easy to evaporate gold electrodes onto the wafer, patterned via lift-off lithography.  The following is a stepwise fabrication sequence that we have come up with:

        1.  Photolithographic patterning of the filter module (photoresist)

        2.  Dry etching of the filter module (reactive ion etching)

        3.  Removal of photoresist.

        4.  Deposition of silicon dioxide insulating layer (PECVD)

        5.  Photolithographic patterning of gold electrodes (photoresist).

        6.  Deposition of gold (evaporation).

        7.  Lift-off to remove excess gold and photoresist.

        8.  Possible silicon nitride deposition step to insulate electrodes

This fabrication sequence can be utilized to construct the main base of our device, but the upper capping layer of PDMS must also be fabricated.  We have decided to integrate most of the microfluidic portions of the device into the PDMS as it will be easy to make an initial master pattern and simply re-use these molds to make multiple PDMS cap structures. 

        In the consideration of these nanofabrication procedures, we tried to determine what would be the easiest method of constructing such a device, but also what would be the cheapest.  Since a device like ours would likely be single-use, it would make sense to make modules that could be mass-produced cheaply.  For this reason, PDMS was chosen as a major substrate.  PDMS is cheap to use, but also requires much less fabrication, since only a few initial molds are needed to produce multiple PDMS structures.  Along these lines, we also considered what portions of the device would be part of the “disposable” structure and what portions would be reused.  We felt that the syringe pump, controller and analysis equipment would all be reusable, but that the chip including the microfluidic channels and the active sensor area would be disposable.  This creates one difficulty, in that we have to provide a robust connection between the disposable portion of the device and the reusable portion.  This will be considered before our final design is presented/submitted. 

In order to better understand the nature of the samples that we have been using, an individual group member has undertaken an initial study to characterize possible samples and some of their inherent properties.  Included below is a short summary of some of the information gleaned from this study and with some relevant references:

Biological agents have a moderate to high potential for large-scale dissemination or a heightened general public awareness that could cause mass public fear and civil disruption. For example, anthrax has the potential for adverse public health impact with mass casualties, and requires broad-based public health preparedness efforts (e.g., improved surveillance and laboratory diagnosis and stockpiling of specific medications)[1].  A World Health Organization report estimated that 3 days after the release of 50 kg of anthrax spores along a 2-km line upwind of a city of 500,000 population, 125,000 infections would occur, producing 95,000 deaths[2].  A brief intervening period of improvement sometimes follows 1 to 3 days of these prodromal symptoms, but rapid deterioration follows; this second phase is marked by high fever, dyspnea, stridor, cyanosis, and shock. Blood smears in the later stages of illness may contain the characteristic gram-positive spore-forming bacilli. This finding is only likely to occur late in the course of disease. Confirmation is obtained by culturing B. anthracis from blood[3]. Blood sample obtain from patients can be directly injected into the biosensor which has a symmetric array filter design forced out large particle size over 2 mm. Other diseases infected by viruses can also be detected by the microchip with suitable antibody electrodes. Body fluids collection for testing run includes blood, urine, sputum, nasopharyngeal aspirate and vesicle fluid. Other samples such as stool, scrapings or biopsy require dissolved and diluted in medium before injecting to the sensor.

Inspired by several lectures given, we have envisioned a new filter design, by fabricating layers of filters at a tilted angle.  We believe this new filter can resolve the clogging problem that many traditional filters run into.  Figure 1 shows the general the layout of such a filter with pore size 2 mm wide, 2 mm high and a 10-20 mm channel depth. 

Fig. 1                                      Fig. 2

In Figure 2, we illustrated the intended design with a sample of blood coming into the filter.  The openings of the broken lines are intended to be bigger than the bacteria/viruses but small than the bigger cells in the blood sample.  In this figure, we have increased the size of viruses, so that it will be easier to see them.  Figure 3 shows the motion of the blood sample through the filter.

Figure 3

As the sample, which includes both the bacteria/viruses and the larger blood cells, is moving through the filter, the bacteria/viruses that we hope to detect will be moving through the filter due to their smaller size.  The bigger cells in the sample will be blocked by the filter.  However, the larger blood cells continue their movement along the tilted walls (broken lines) of the filter.  At the end of the tilted plane, the space is fabricated into an oval shape.  We believe vortices of the bigger cells can be created within that space, therefore trapping the bigger cells inside of that space until the rest of the sample passes through the filter – Figure 4.  This effect can be utilized to prevent the re-entrance of the bigger cells into the sample.

Figure 4

As can be seen in Figure 4, the liquid and the smaller particles, i.e. bacteria/virus, are passed through the filter and the bigger cells are being pushed aside and trapped by the vortex that is created in the oval area of the filter.  We believe that this filter can be used to successfully filter out the bigger cells while keeping the clogging problem to a minimum. 

In the previous proposal, the microchip consisted of a glass substrate with three Ti/Au electrodes insulated with a 1 mm thick layer of silicon nitride.  Electrodes D and F were reference electrodes and were initially intended to electrophoretically flow sample material past the sample electrode, electrode E.  (Please refer to the original proposal for detailed diagrams of the initial design.)  Influenced by the new filter design, we believe that the sample will be able to flow across the detection electrode without needing electrophoretic flow.  Therefore, the third electrode (F) becomes unnecessary in our improved design.  Instead of using electrophoresis, we will simply use pressure to move the sample across the detection electrode. 

The rest of the design will stay the same:  different sized areas in the silicon nitride will be etched away to expose the gold electrodes based on the size of the target to be detected.  On average, bacteria organisms are on the 1mm size scale and viruses can range from 10nm-100nm.  The detection electrode surface area was calculated to accommodate up to 100 bacterial or viral organisms. 

        Initially, we were planning to use non-specific adsorption of antibodies or Self-Assembled Monolayers (SAMs) of thioctic acid to bind antibodies to our gold detection electrode.  Upon further review, we have decided to use neither of these methods to immobilize our antibodies to the detection electrode.  We had several concerns regarding the prior two methods and therefore decided to use Protein A immobilization as a selective binding surface for antibody binding

        Concerns with regards to non-specific adsorption of antibodies to the detection electrode include: long-term stability, denaturiziation, orientation, functionality, and the ability to target these antibodies only to the detection electrode.  Many of these concerns were alleviated by the use SAMs of thioctic acid.  Antibodies can be coupled to these SAMs upon activation.  Thioctic acid has the advantage of targeting antibody binding to the detection electrode through a covalent interaction between the thiol groups on the thioctic acid and gold.  This covalent bond would also increase the stability of the antibody under a variety of conditions.  The disadvantage of using thioctic acid or other similar chemistries to bind antibodies to gold surfaces are that they require several activation steps utilizing EDC and NHS in order to covalently bind the antibodies to the SAM.  It has also been reported that covalent binding leads to a decrease in reactivity of the antibodies (Stefan et al., 2000) .

        We have therefore decided to use Protein A immobilization to the gold detection electrode as the first step of our antibody immobilization.  Protein A can be immobilized onto gold surfaces either by non-specific adsorption or covalent binding.  The non-specifically adsorbed Gold-Protein A complex is a highly stable complex and is believed to be Van der Waals in nature (Davis et al., 1989) .  However, covalent binding of Protein A to the gold detection electrode is a much more attractive alternative, and can be achieved by genetically engineering the protein to contain cysteine residues at its C-terminus.  These covalently bound Protein A molecules have been reported to have a higher IgG binding activity than physically adsorbed Protein A (Kanno et al., 2000) .  In addition, the antibodies bound to these genetically engineered proteins have also been shown to have a greater antigen binding activity (Kanno, Yanagida et al., 2000) .  We hope to use Protein A in our biosensor as a direct immobilization method due its natural affinity towards the Fc region of IgG antibodies (Babacan et al., 2000) .  This allows us the flexibility to bind any antibody without denaturing it and to orient the antibody molecule in such a manner that will not block the antigen binding sites (Deshpande, 1996) .  This method has also been shown to be functional, by its ability to bind Salmonella through the immobilization of anti-Salmonella antibodies (Babacan, Pivarnik et al., 2000) .  The fact that this genetically engineered Protein A can form covalent bonds with the gold through thiol-gold interactions increases the stability of the antibody binding to the gold detection electrode and allows us to potentially target the antibodies only to the detection electrode.  This method seemed to be the most logical solution for our biosensor in order to bind antibodies to our detection electrode that would be oriented, stable and functional.

Babacan, S., Pivarnik, P., et al. (2000). Evaluation of antibody immobilization methods for piezoelectric biosensor application. Biosensor & Bioelectronics 15, 615-621.

Davis, K. A. and Leary, T. R. (1989). Continuous liquid-phase piezoelectric biosensor for kinetic immunoassays. Anal. Chem. 61, 1227-1230.

Deshpande, S. S. (1996). Antibodies: Biochemistry, structure, and function. Enzymes Immunoassays: From Concept to Product Development. NY, Chapman and Hill: 24-51.

Kanno, S., Yanagida, Y., et al. (2000). Assembling of engineered IgG-binding protein on gold surface for highly oriented antibody immobilization. J. Biotechnol. 76, 207-214.

Stefan, R.-I., Staden, J. F. v., et al. (2000). Immunosensors in clinical analysis. Fresenius J. Anal. Chem. 366, 659-668.

Found below is our current design for the biosensor.  Please refer to the initial proposal for a review of the previous design. 

         Our group has found this to be an excellent project that is well suited to our various disciplines and backgrounds.  By splitting the project into sub-units, we have been able to tackle different areas based on our personal expertise.  Each member has worked hard to keep the other group members “up to speed” on their portion of the project and we have had reasonable success in critically reviewing and updating our design.  As can be seen from this progress report, we have made several significant changes from our initial design with the hopes of making a better and more feasible final product.  While there are still some design issues that we are working on, we feel that we have a solid design to work with from this point forward.  In the near future we hope to finalize some of our fabrication details, as well as to have a better electronics design for our final presentation.