Section I Problem Definition Supporting Documents

Transcription

1 Section I Problem Definition Supporting Documents 1.1 Annotated Bibliography Hot Embossing [B1] Sung Won, Y., Toshihiko, N., Masaharu, T., and Ryutaro, M., 2008, Fabrication of Micro Mold for Hot Embossing of Polyimide Microfluidic Platform by Using Electron Beam Lithography Combined with Inductively Coupled Plasma, Microelectronic Engineering., 85(5 6), pp This is a journal article describing how a hot embosser can make mircofluidic channels in Polyimides, a heat resistant polymer. It stated that these polyimides worked well with biosensing material through microfluidic channels. This article was beneficial to the product to gain knowledge on how a hot embosser worked. This article along with examples shown by Mark Tondra showed what the microfluidic channels would look like in a polymer and how they would work in the biosensor. [B2] Wang, J., Xue, and W., Cui, T., 2004, A Combinative Technique to Fabricate Hot Embossing Master for PMMA Tunneling Sensors, Microsystems Technologies, 10, pp This article describes what a hot embosser is, the procedure used to make molds, and what parts can be made with it. The article gave the group a further understanding of the capabilities of the hot embosser. This allowed for certain concepts to be disregarded because the hot embosser was not capable of producing them. [B3] Lin, C.R., Chen, R.H., and Hung, C., 2003, Preventing Non Uniform Shrinkage in Open Die Hot Embossing of PMMA Microstructures, Journal of Materials Processing Technology, 140, pp This article shows how the hot embosser works and what the operating conditions of the hot embosser for it to function correctly. This article showed that the error of the hot embosser is at a minimum and could almost be negligible with affected the tolerance analysis. [B4] Heckele, M., Bacher, W. and Müller, K.D., 1998, Hot Embossing The Molding Technique for Plastic Microstructures, Microsystem Technologies, 4, pp This article states how the hot embosser works. 40

2 It showed that using a hot embosser instead of injection molding would be beneficial because it would reduce the stress in the walls. It also stated that the hot embosser is better than injection molding because it allows for high quality and high precision parts that our sensor required. [B5] Becker, H., Heim, U., 2000, Hot embossing as a Method for the Fabrication of Polymer High Aspect Ratio Structures, Sensors and Actuators, 83, pp Shows that the hot embosser can make parts with a high aspect ratio. This article was not beneficial because the mill has a smaller aspect ratio that the design would have to be limited to. [B6] Wang, W., and Soper, S.A., 2006, Bio MEMS: Technologies and Applications, CRC Press, Boca Raton, US, pp This chapter shows the difference between hot embossing and injection molding. This article showed that the process was better in regards to accuracy and precision, but could not produce parts as fast injection molding. [B7] Scott, J. and Broussard, L.A., 2004, CAMD 2004 Annual Report, Louisiana State University, Baton Rouge. This article showed that the hot embosser could manufacture parts with a very high aspect ratio of 60:1. This was beneficial to the project because a large aspect ratio could be used. It was then turned away because the maximum aspect ratio of the hot embosser we are using was only 5:1 and therefore was the limiting factor for the design. Microfluidic Design [B8] Xu, F., Datta, P., Wang, H., Gurung, S., Hashimoto, M., Wei, S., Goettert, J., McCarley, R.L., and Soper, S.A., 2007, Polymer Microfluidic Dies with Integrated Waveguides for Reading Microarrays, Anal. Chem., 79(23), pp This article describes how a biosensing material gets placed onto mircofluidic die. Once on the microfluidic die it would act as a biosensor. This article was used to understand what the biosensing material on the silicon die would be as well as ways to put the biosensing material on the die. This process affected how the die could be placed into the well. If it had biosensing material already on it this material couldn t be interfered with. So it was decided that the die should be placed into the well and then the biosensing material would then be added afterwards to ensure that sensing material was not damaged. 41

3 [B9] Gurung, S., 2009, Passive Alignemnet of Micro Fluidic Dies Using the Principle of Elastic Averaging, M.M.E. thesis, Louisiana State University, Baton Rouge. This article described a way for multiple sheets of individual components of microfluidic parts that were made on a hot embosser could be aligned in a reliable and accurate way. This thesis benefitted the project because it provides a method for aligning the microfluidic cover and well. The purpose of the thesis coincided with the scope of this project; Microfluidic parts were made on a hot embosser and precise alignment of fluidic ports were required. Biosensing [B10] Fredrickson, C.K., and Fan, Z.H., 2004, Macro to Micro interfaces for microfluidic devices, Department of Mechanical, Aerospace, and Biomedical Engineering, University of Florida, This article describes the interfaces between the biosesing die and its surroundings. This article did not affect the project but gave and understanding of how the system would work once all the subcomponents were assembled. [B11] Mulvaney, S.P., Cole, C.L., Kniller, M.D., and Malito, M., 2007, Rapid, Femtomolar Bioassays in Complex Matrices Combining Microfluidics and Magnetoelectronics, Biosensors and Bioelectronics, 23, pp This article examines a biosensor working to look for and analyze more than one type of anelyte. This article did not benefit the design of the project but rather it gave an understanding of the customer and their needs to create this type of biosensor. [B12] Singh, V., Desta, Y., Datta, P., Guy, J., Clarke, M., Feedback, D.L., Weimert, J., and Goettert, J., 2007, A Hybrid Approach for Fabrication of Polymeric BIOMEMS Devices, Microsystyt. Technol. 13, pp This technical review addressed how a biosensor was made to analyze human sweat. This article did not affect the end design but addressed alternative means in creating a mircofluidic biosensor. [B13] Bartling, B., Li, L., and Liu, C., 2009, Determination of Total Bile Acid Levels Using a Thick Film Screen Printed Ir/C Sensor for the Detection of Liver Disease, Analyst 134, pp

4 This article describes how a biosensor can be used to measure bile and liver disease. The author also wished to create an affordable method of biosensors that would be useful in detecting liver disease. This article did not affect the design or assembly process for the fully encapsulated die. [B14] Blanes, L., Mora, M.F., Lago, C.L., Ayon, A., and Garcia, C.D., 2007, Lab on a Die Biosensor for Glucose Based on a Packed Immobilized Enzyme Reactor, Electroanalysis, 19(23), pp This article showed how different biosensor can work. They can look for different particles and different analytes could be used in the same biosensor. This did not affect the project but was relavent because it related to Mark Tondra s end design of having a biosensor be able to measure different types of materials. [B15] Tuantranont, A., Lomas, T., Maturos, T.,Wisitsora at, A,. Thavarungkul, T., Kanatharana, P., Limbut, W., and Loyprasert, S., 2006, Development of Low Cost Microfluidic Systems for Lab on a Die Biosensor Applications, NanoBioTechnology, 2(3 4), pp This article defined oxygen plasma bonding. It was important to the design because it stated that oxygen plasma bonding was used for PDMS and not PMMA. It also stated that it was used to adhere PDMS to glass and not to another sheet of plastic. It enabled the bonding technique of oxygen plasma to be removed from the design concepts. 1.2 Patent Search Objectives: The object of the patent search was to find out if a manufacturing process for creating a biosensor would be patentable. Search Criteria: Google patent search was used to find patents that were applicable to the manufacturing of a biosensor. Key words that were used in the searching for a patent included: microfluidic, biosensor, PMMA, hot embosser, and manufacturing process. The class and subclass that these terms were mainly found in were: 435/ ; 73/233; 430/320-23; 430/ Findings: 43

5 There are a few patents that were researched using Google patents. The process of passing fluid over a biosensor is already known knowledge and cannot be patented. The part that can be patented is the geometry of the fluidic cover and well and how they are manufactured together. The finding in these sections used a different method then what was defined by Mark Tondra in the customer needs. These patents showed that smaller parts could be made with a hot embosser but they did not show that a hot embosser creating sheets of PMMA and aligning them together around a silicon die. Overall this manufacturing process for creating a microfluidic biosensor is a relatively new concept that could be patented through this project. 1.3 User Need Research While there was a variety of subcomponents that were already designed, many design needs arose during the initial phases of the project. The fully encapsulated die was to be integrated with these components as they would regulate many of the design needs and metric. The information for these parts and how the biosensor must function were given by Mark Tondra. Mark was the one with the governing information and he knew how the entire biosensor was to behave. With the data that Mark processed a list of questions were asked to define the customer needs listed which are located in Table 5. Table 5: Design Needs Number Need Why Important Importance 1 PMMA assembly must produce a fully encapsulated silicon die with System needs to be robust 5 all parts adhering to one another 2 PMMA alignment process is reliable (few defects/not align reduce wasted products 3 pieces) 3 PMMA assembly can be manufactured on a large scale Profit will increase 4 4 Finished Encapsulated die can be pressurized from the inside without becoming disassembled It will fail if it breaks apart 2 5 PMMA assembly can be scaled up Lets the company decide on how big or or down small to make a Si die 3 6 PMMA assembly can hold Can be used with different sizes of different Si dies from alternative silicon dies manufacturers 3 44

6 7 Finished Encapsulated die must have minimal fluid leakage Fluid is what is being analyzed if some of the fluid is not accounted for it will lead to a wrong analysis 8 PMMA assembly must be rigid If it breaks it will fail 4 PMMA assembly must have well Silicon die needs to be protected as well 9 5 be able to hold a Silicon die as aligned properly 10 PMMA assembly process is time efficient If assembly process takes to long it will result in a loss of profit PMMA assembly process must align Si dies in proper orientation Fluid must pass precisely over specific locations on the Silicon die 2 12 PMMA assembly process must apply adhesive to a certain thickness If glue is too thin it will affect the strength of the part PMMA assembly process must put adhesive only in desired locations Excess Glue must not make contact with the interface of the Silicon die 5 15 PMMA sheets are easy to assemble Easier to assemble allows for parts to be made faster 4 16 PMMA sheets are to be stamped using LSU hot embosser with the top and bottom molds created at C- Axis Consistency, previous designs were created with both companies and if used again total manufacturing cost for will be reduced Design must make efficient use of all surface area on the brass insert PMMA top and bottom sheets (when fastened together) must hold stay together during sawing process Top Mold must be compatible with previous design project Bonding process must not harm biocoating 22 Placement of die in wells efficient Allows for more parts to be made for the same price/ efficiency Will fail if they break apart 4 For the whole system to work each subcomponent must work succinctly with one another It will affect analysis 2 Faster placement allows for more to be produced in a shorter time Encapsulated die should be clear Need to see what is happening inside the biosensor/ aesthetically pleasing Finished die should survive temperature extremes normally seen during manufacturing and shipping BIOSENSING MATERIAL experiences < 50C process temperature Product will fail if it can not meet the temperature extremes it is to analyze If the assembly process requires temps that are no greater than 50C 26 Should be chemically resistant to Part should not be able to dissolve

7 water, salt, and ethanol 27 Fluid Port Locations Fluid must pass precisely over specific locations on the Silicon die 5 If smaller the alignment becomes more 28 Fluid Port Sizes difficult No blockage of Fluid flow Blockage of Flow will reduce will the 29 throughout encapsulated die accuracy of the analysis Scale 1-5 with 5 being the highest 2 5 After the design needs were found a way to quantify those needs was then created by the metric found below in Table 6. This metric gives a value on how each design need was to be quantified and the specifications that the design must meet. Table 6: Metric for the Design Needs Need #'s 1 Metric Importance Units Compression strength of encapsulated die Marginal Value Ideal Value 2 Mpa ,18 Adhesive (UV glue top sheet to bottom sheet) shear strength 5 N/mm^ ,27 Alignment test (microscope) [top sheet to bottom sheet alignment in X and Y, measured at 5 spots TopCenterBottomLeftRight] 4 Microns ±50 ±10 3 Cost 2 US $ <4000 < Pressure Gage 2 Mpa Scalable CAD drawing 3 Binary YES YES 6 Si Dimension: tolerance 4 Microns Method for sealing fluid cavity 3 Binary YES YES 9,17 Well Dimension: wall thickness 4 Microns < ,17,22 Well Dimension: width [this needs to be with respect to the die size, i.e. well width is die width + 10 microns (+ 50 micron / - zero)] 3 Microns <1,475 1,425 46

8 9,17,22 Well Dimension: length [same comment as for well length] 3 Microns <5,975 5,925 10,15 Time to assemble 2 Minutes Si Placement 4 Binary Pass Pass 11 Orientation Test 3 Binary Pass Pass 12 Glue thickness tolerance 4 Microns ±50 ±10 13 Glue placement tolerance 5 Microns <50 <10 17,22 Well Dimension: well depth 5 Microns > Usable with hot embosser 5 Binary YES YES 17,19 Fluidic Cover Dimension: width 4 Microns <1,475 1,425 17,19 Fluidic cover Dimension: length 3 Microns 3,200<L<4,000 4,000 17,19 Fluidic cover Dimension: wall thickness 4 Microns > ,27,28 Fluid Port diameter 3 Microns ,27,28 Fluid port location tolerance 5 Microns ±50 ±10 20,24,25 Process temperature 1 C 50 <50 23 Clarity (Able to see through) 2 Binary YES YES 26 Exposure to water safe? 5 Binary YES YES 26 Exposure to salt safe? 2 Binary YES YES 26 Exposure to ethanol safe? 1 Binary YES YES 4 Volumetric flow rate of fluid 5 ml/sec 100 >100 Scale 1-5 with 5 being the highest 1.4 Concept Alternatives This project has allowed for a wide range of concepts to be reviewed. The overall process needed to follow three steps. First, it needed to place the silicon die into the well, then align the fluidic cover and well, and finally adhere the fluidic cover to the silicon die and well. Each of these three steps produced a concept selection chart along with some sub concept selection designs. These concepts were broken down into five groups: glue bonding, placement of the silicon die into the well, alignment the fluidic cover to the well, fluidic cover interaction with the silicon die, and the fluidic cover s interaction with the well. Bonding Selection 47

9 There were a few concepts on how to secure each part to one another. The first concept was using the glue from the previous design class. They experimented with bonding of many different kinds of glue. Their design group came up with two UV glues that would work for this design. They were both strong, could be cured by UV light, could be left out for a few minutes without curing, were easy to obtain, inexpensive, and had a viscosity that would allow the glue to have a desired thickness. The other way to adhere two pieces of PMMA together was to use oxygen plasma bonding. The oxygen plasma would change the chemical bonds along the edges of the PMMA, so when two PMMA parts would be forced together they would create a seal. This process does not change the dimensions of the two pieces of PMMA but instead forms a seal between the two sheets of plastic by mating them together. Another concept to fuse the parts together would be to melt the plastic and then cool it. Once the PMMA was melted it would make liquid bonds between the two PMMA sheets that would harden to form a single PMMA part. Placement of the Die into the Well The first step is placing the silicon die into the well. Since this is not in the scope of the project on how to place the die; the location of the surface of the silicon die is important. The die then has three possible locations to sit: on the bottom of the well (Figure 13 a) somewhere in the middle (Figure 13 b), or flush with the top (Figure 13 c). Each of these locations would affect the overall design of the product. 48

10 Silicon Die Silicon Die Silicon Die Well Well Well (a) (b) (c) Figure 13: Possible Placements of a Silicon Die in a Well The first of these situations would be with the silicon die lying flat on the bottom of the well (Figure 13 a). In this situation the surface of the silicon die would be within a desired location in regard to the well floor. This would be achieved by placing the silicon die directly into the well. In this situation there were four main design concepts: a straight box, two and four slanted walls, and ledges. The ideal situation would be to have the well be the exact size as the silicon die itself. Here the well would have flat sides so that when the silicon die gets placed into the well it is always at its desired orientation and position (Figure 14). Silicon Die Well Figure 14: Well the Exact Size of the Silicon die 49

11 Another concept was to design the well to have all four walls that slanted down to the size of the silicon die. As the silicon die was placed into the well it would slide down the edge of the well s wall and into place at the bottom surface of the well (Figure 15). Silicon die Die Well Figure 15: Well with 4 Slanted Walls A third concept for this design was similar to the first but instead of four slanted walls there would be only two slanted walls: one for the end of the well and one for the side (Figure 16). This design allowed would allow for a greater control of the silicon die. As the silicon die got placed into the well it again would slide down the slanted walls but once the side of the silicon die hit each vertical wall it would be set in place. This allows for a great directional control over the silicon die without the silicon die sitting at an angle. 50

12 Well Figure 16: Well with 2 Slanted Walls The final way of have the silicon die sit on the bottom of the well would be to have four ledges coming out from the bottom of the well (Figure 17). These ledges would allow for glue to be placed between so when the silicon die was placed into the well the glue would have a place to disperse inside the well. There would allow for a greater directional control of the glue in the bottom of the well. Figure 17: Well with Ledges 51

13 The second of these situations would to have the silicon die suspended in between the top and bottom of the well. This would be done by the using glue as a suspender. Once the silicon die is placed in the well the glue would act as a means of positioning it to a certain depth, specified by the thickness of the glue (Figure 18). Well Figure 18: Silicon Die Placed in the Center of the Well The final placement of the silicon die inside the well would be to have the silicon die be flush with the top of the well (Figure 19 b). This would allow for no variation of the die s surface in relation to the top of the well. This process would be achieved by placing a silicon die onto a larger piece of glass with a type of reversed UV curable glue. The piece of glass would then transfer the silicon die into the well. As the die goes into the well the bottom of the glass will eventually make contact with the top of the fluidic cover. Once this occurs the reversible UV glue can be released allowing the top of the silicon die to be flush with the top of the well (Figure 19 a). 52

14 Glass Sheet With UV Reversable Glue Silicon die Glass Sheet With UV Reversible Glue Silicon die Glue Well Glue Well (a) Figure 19: Process for Aligning the Silicon Die to be Flush with the top of the Well (b) A concept for dispersing the excess glue after the silicon die was placed in the well would be to have walls that had channels cut out of them (Figure 20). This would allow the silicon die to be suspended into the well by always controlling the glue depth. For the case of having the silicon die be flush with the top of the well, it would allow for excess glue to flow out of the well as the die was pressed into the well. This is very beneficial because it does not allow for excess glue to flow onto the silicon die or the top of the well. 53

15 Figure 20: Well with Channels Aligning the Fluidic Cover with the Well After the silicon die is placed into the well and it is stamped with a biosensing material, the fluidic covers must then be placed and adhered onto the silicon die and well. To do this there needs to be precise alignment techniques, for each part to function correctly. There were two main groups of alignment techniques: optical and mechanical. Concepts for these two groups were also broken up into individual and overall alignment features. Optical alignment features work by taking two separate PMMA sheets that have the same feature on each one, such as a cross, and are located on two identical locations on each mold. With the use of a microscope these two features can be aligned in a precise manner. Once these crosses were aligned with the microscope could create a contact between the two PMMA sheets by lowering the two molds onto each other (Figure 21). 54

16 Figure 21: Overall Optical Alignment The other concept would be to use overall mechanical alignment features. Mechanical features work by aligning and then mating two structures located on each sheet. The first concept using an overall mechanical feature would be to implement a structure onto the mold itself. The structure on one of the molds would have its negative on the other mold, so that once those two overall mechanical features are aligned and mated all of the fluidic covers and wells on the sheet will also be aligned (Figure 22). Using this concept allows for a quick method of aligning two molds together, without expensive machinery. 55

17 Figure 22: Ball and Groove Joint (Source: Gurung Thesis, edited by author) The other concept for aligning the features is to use cut out a number triangles, in the same location on each PMMA sheet (Figure 23 a). Those two molds can then be placed on a jig where they will slide together, which will allow each individual part to be precisely aligned to each other (Figure 23 b). This concept is good because it is accurate and also does not require the use of expensive machinery. (a) Figure 23: Overall Mechanical Alignment Triangle Grooves (b) (Source: Gurung Thesis, edited by author) The alternative to having the overall alignment features would be to have individual alignment features. These would be located on each well and fluidic cover instead of the whole mold 56

18 itself. This would work well to align every piece one by one and ensure that all pieces are aligned very precisely. Microfluidic Cover Interaction with Silicon Die After the pieces are aligned, the fluidic cover needs to have contact with the surface of the silicon die as well as the walls of the well. The first of the contact points must be with the silicon die to provide an adequate seal for the fluid to flow over the silicon die (Figure 24). Fluidic Cover Silicon Die Interaction Points Between the Fluidic Cover and Silicon Die Well Figure 24: Interaction Points for the Silicon Die The first concept for this design was a simple flat surface to interact with the silicon die (Figure 25 a). This would provide a large area for the glue to adhere to making the seal strong. This concept is also easy to manufacture since it does not require more than one tool to make. The second concept was to have a vertical wall and then a shallow angle rising slowing away from the biosensing material on the silicon die (Figure 25 b). This would maintain a great amount of surface area for the glue to make contact with the silicon die. This also allows for a directional control of the glue so when the glue gets compressed it will travel away from the sensing material on the silicon die rather than towards it. 57

19 A third concept would be to have a vertical wall with a half circle rising away from the sensing material (Figure 25 c). This would allow for directional control of the glue away from the silicon die, as well as making the contact point quite strong. Manufacturing this would be simple because the mold would only require a circular bit to be indented slightly into the mold insert. The final concept for this contact point would be to have a full circle or a v groove contact point (Figure 25 d). These concepts are similar to the two previous concepts but have a mirrored image of the geometry on the other side of the contact point. This would provide contact area and directional control of the glue as well as the manufacturability would be increased due to the symmetry of the design. Silicon Die Silicon Die Silicon Die Silicon Die (a) (b) (c) (d) Figure 25: Fluidic Cover and Silicon Die Interaction Concepts Fluidic Cover interaction with Well In order to meet the design requirements the fluidic cover also needs to interact with the well (Figure 26). This interaction will provide rigidity to the product. It is important to have these two parts interact because when a force is applied on the fluidic cover it can then be dispersed between the well and not just the silicon die. If all the force was on the silicon die it could break causing failure for the part. Having the interaction between the fluidic cover and well will reduce this risk and allow for a more robust product. The feature will also allow for minor corrections of the alignment features. If the alignment feature is not aligned properly the interaction between the well and fluidic cover could compensate for the error. 58

20 Silicon Die Interaction Points Between the Fluidic Cover and Well Well Figure 26: Interaction Points for the Well The first concept for a structural support would be to use a tongue and groove on each piece Figure 27 a). This is one of the best bonding techniques used in many manufacturing facilities. It allows the glue to form around the edge of the tongue and provides a strong and tight seal. Manufacturing the tongue and groove can be easily performed by a machinist. The second concept for a structural support system would be to have a chamfered edge Figure 27 b). This chamfered edge allows the fluidic cover to be position directly above the well without allowing the fluidic cover to move in a lateral direction in relation to the silicon die. This edge will be able to be machined easily as well and it also allows for directional control of the excess glue. The final concept for a structural support system would be to have two arms come from the well and mate with their reverse image on the fluidic cover Figure 27 c). This would allow for a strong wall support system as the glue would not only adhere to the bottom of the fluidic cover and top of the well but also it would permit glue to bond to the outside walls creating a strong bond between to two pieces of plastic. 59

21 Fluidic Cover Fluidic Cover Fluidic Cover Well Well Side of Well (a) (b) (c) Figure 27: Fluidic Cover and Well Interaction Concepts 1.5 Concept Selection The concept selection process was done by the use of concept selection charts and knowledge of the manufacturing processes to be used. The concepts that were placed in the charts were the concepts that were the most plausible. Although there were more concepts to consider the best concepts according to the knowledge of the team were the ones that made it into the concept selection charts. Each of the five main concept categories could be addressed independently since they were independent of the whole design. All of the criteria were weighted on a scale from one to ten with ten being the highest score. Bonding selection There were three main types of bonding selections to choose from: oxygen plasma, melting, or UV glue. As the project advanced further it could be seen that oxygen plasma bonding would not work with our project. Oxygen plasma would require a different type of plastic, PDMS, instead of PMMA. PDMS would form to the silicon die and another piece of PDMS, but in order for the fully encapsulated silicon die to be functional with other parts of the biosensor (mainly those made out of PMMA) oxygen plasma bonding would not work for this design. 60

22 Melting the two halves of the PMMA together also would not be a viable option. This would violate two of the design criteria. The first would be that the assembly process would reach beyond 50 C when melting the PMMA and therefore would damage the biosensing material on the silicon die. The second reason is that the seal between the fluidic cover and silicon die would not be strong enough because the PMMA would not bond very well to the silicon die. The UV glue was then decided to be used for the design. This glue could be placed to a desired thickness, could be cured quickly, assembly process could take a few minutes without the UV glue curing on its own, and also had an adequate bonding strength to both PMMA and silicon. Placement of the Silicon Die into the Well The first concept to be chosen was the geometry of the well. The criteria were measured in seven categories: contact area, directional control of glue, empty space for glue expansion, alignment of die in Z direction, ease of die placement, amount of glue used, and ease of machining (Table 7). The first metric was defined by how much contact the glue made with the silicon die. The more contact area the better since it would allow for a stronger bond. The second metric was defined on how much control there was in determining the direction that the glue would flow as the silicon die compressed the glue inside the well. If there was greater directional control the concept would receive larger score. The third metric was weighted by how well the geometry could disperse excess glue, as the silicon die was placed into the well. If the design could disperse an infinite amount of excess glue, without affecting the silicon die placement, it would get score of ten. The fourth metric was the most important metric and it determined where the silicon die would be placed relative to the z direction. The greater accuracy the silicon die could be placed in a desired location on the well the better. 61

23 The fifth metric was the least important metric but it depended on how simple it would be to place the silicon die into the well with the correct orientation along with the correct alignment. The easier this was the greater the score would be. The sixth metric was the amount of glue to be used. The less glue that would be wasted would be the best scenario, so each criterion was measure on how much glue would be used. The less glue the better and therefore would require a higher score. The final metric would be how easy it could be machined. A part that would require many cuts, tools, and accuracy would be more expensive and therefore would receive a lower number. Table 7: Metric for Silicon Die Placement 4 Slanted Walls 2 Ledges Simple Well Size of Die Channels in End of Wall 2 Slanted Walls Metric Criteria Weight Silico 1 Contact Area Directional Control of Glue Empty Space for Glue Expansion Alignment of Die in Z Direction Ease of Die Placement Amount of Glue Used Ease of Machining

24 Totals Using this weighted criteria the best concept was to have two channels cut into the ends of the walls, and have the silicon die be placed between the top and bottom of the well. It allowed for good alignment in the z direction, a large contact area between the silicon die and the glue, and machining the part would only require one pass of the mill. Aligning the Fluidic Cover with the Well In order to find the best way to align the fluidic cover to the well a selection matrix was designed. This was done by choosing six criteria to calculate which alignment features would be the best to include for the design (Table 8). The first metric to be measured is the possible accuracy. This is the most important criteria because the fluid ports on the fluidic cover needs to be aligned very precisely to those on the silicon die. The best alignment feature with the smallest amount of error will be given the highest score. The second metric would be considering if there was a manufacturing error in the alignment feature, and how many parts surrounding that feature would be affected. The concept that affects the least amount of parts would be given the highest score. For the case of the overall mechanical features, if they those features do not fit together, it would affect all the other pieces on the mold and therefore would cause most of those pieces to be misaligned. The third metric relates to how it can be machined. The easier and quicker it is to machine or if the alignment feature requires less features the higher the score will be. The fourth metric is in regards on how easy it is to align the parts. If the two PMMA sheets could be aligned without additional equipment or if the time process to align the feature is small the concept would receive a higher score. 63

25 The fifth metric relates to how many wells could be on the mold. If the alignment feature takes up space on the mold it would require fewer parts to be produced and therefore a lower score. The final metric is in regards to the availability of resources. If there needs to be expensive equipment purchased, rented, or hired to use, the matrix score would be lower than if there was no equipment to be used. Table 8: Metric for Aligning the Fluidic Cover with the Well Metric Criteria Weight Overall Optical Overall Mechanical Individual Optical Individual Mechanical Overall Optical Individual Mechanical 1 Possible Accuracy Rick of Affecting Other Parts Ease of Machining Ease of Assembly Piece Per Mold Insert Optimization Resource Availability Totals Using this weighted concept selection matrix it can be seen that the best alignment would be to use an overall optical alignment feature. This feature allows for a great accuracy as well as it is easy to manufacture and assemble. Fluidic Cover Interaction with Silicon Die 64

26 The fluidic cover interaction is important because it defines whether the silicon die can be sealed without obstructions. The first metric in choosing the interaction for the silicon die seal was the amount of contact that the seal would make with the silicon die. More contact would allow for a stronger bond between the fluidic cover and silicon die which also means a greater chance that the seal would work properly. The second metric is regarding how the direction of the glue can be controlled once the fluidic cover is place onto the silicon die. A concept that forces the glue away from the components on the silicon die would receive a higher score than a concept that has no control in the direction of the glue. The third metric is the amount of space the glue has to disperse once the fluidic cover is compressed onto the silicon die. A concept that has more room for the glue to flow will receive a higher score because if the glue can flow into a recess freely it will less likely flow into the components on the silicon die. The fourth metric would be how well the concept can be machined. A concept that can be machined quickly and easily will be more cost efficient and will receive a higher score. The final metric is how easily the parts can be assembled. Parts that require specific locations and contact points will be scored lower than those. Table 9: Metric for the Fluidic Cover Interaction with the Silicon Die Wide Flat High Angle Wide Low Angle Narrow Low Angle V Groove Half Round Full Round Metric Criteria Weight 1 Contact Area

27 Direction al Control of Glue Empty Space for Glue Expansion Ease of Machinin g Ease of Assembly Total Using the selection matrix the wide low angle was the best design for the interaction between the fluidic cover and silicon die. This concept had a large amount of contact area making the bond between the die and the fluidic cover strong without using a lot of material. This concept allowed for a place for excess glue to go to as well as be dispersed away from the silicon die once the two parts were compressed together. Although this concept scored the highest it was found that this part could not be manufactured by the milling machine. The part that could be manufactured was the full round. This part had much of the same qualities as wide low angle and had the second largest score making it an easy choice. Fluidic cover interaction with Well The fluidic cover interaction with the well is important because it adds structural support for the fully encapsulated die. The selection matrix for these concepts is shown below in Table 10. The first metric was how well the features would align the well and fluidic cover. The better they aligned them the higher the score would be. The second metric was how easy the two features could be assembled. The features that would require precise precision to interact with one another would receive a lower score. 66

28 The third metric revolved around the ease of machining. If the part could not be machined it would receive a zero for a score. The less amount of machine time spent on making the features would receive a higher score. The fourth metric was in regards to the contact area. The more contact area would require more glue and therefore produce a stronger bond. This would allow the part to be more structurally sound. The stronger the bond the higher the score the concept would receive. The final metric dealt with controlling the direction that the excess glue would flow. If the flow would be towards the silicon die or if could affect another part of the design it would receive a lower score. Table 10: Metric for the Fluidic Cover Interaction with the Well Matrix Criteria Weigh t Chamfer Inverse Chamfer Wing Groove Alignment of Cover and Well Ease of Assembly Ease of Machining Contact Area for Gluing Directional Control of Glue Totals

29 This matrix showed that the concept of the chamfer was the best design followed closely by the inverse chamfer. This design was not only easy to machine but also easy to assemble. Final Concept The final concept consisted of using glue to bond the individual components together. An overall alignment feature would be used to ensure that the fluidic cover sheet and well sheet are lined up precisely. The well would also have channels located at both ends. The fluidic cover interaction with the silicon die would be implemented by using a half circle, while the interaction with the well would use a chamfer. Once assembled the final design of the well and bottom of the fluidic cover would like Figure 28a and Figure 28b respectively. (a) Figure 28: Geometry of the Well (a) and Fluidic Cover (b) (b) This is the best concept for the design. This concept allows for excess glue to leave the well as interaction points on the fluidic cover will force the silicon die into the well to its desired depth. This process also allows the silicon die to be at a desired location despite of errors in the manufacturing process due to tolerances. By having two interaction points on the fluidic cover it will allow for precise alignment for the fluidic cover on the well and silicon die. The interaction point of the silicon die and fluidic cover 68

30 will determine the depth that each silicon die is in the well. At the same time the interaction points of the fluidic cover and well will make the fully encapsulated die robust. Glue could be applied to a certain thickness at the bottom of each fluidic cover and top of each well. So when the well and fluidic cover are pressed together it would form a bond between all of the parts. By having these two interaction points instead of one it can be seen that excess glue will have a tendency to go towards the gap between the two points of contact. The geometry of the interaction point of the silicon die and fluidic cover will force the fluid to be expelled toward the vacant space due to the physical properties of fluid following a path of least resistance. The interaction point between the fluidic cover and well would also force the glue to this recess forming a strong bond between each part. The concepts found here work well in a manufacturing process to enable the part to be manufactured as well as assembled, while meeting each design requirement. 69

31 Section II Design Description Supporting Documents 2.1 Manufacturing Plan Manufacturing Overview **Note: see Volume I Section II for a complete overview of the manufacturing process** The design is made by adhering two PMMA sheets of plastic that will be affixed together with a silicon die being trapped between them. The well and the fluidic cover will then need to be made into a mold. This process will be done by using a CAD program and taking the negative image of each part and then pattering them on a disk approximately 4.5 inches in diameter. Doing this will then produce two sheets; one for the fluidic cover and well, which will have hundreds of each part on each sheet. After the molds are designed they will be sent to a machine shop that will use a computer program to CNC machine the design into two blanks of brass. The brass insert will then be sent to the hot embosser who will take the negative image for the fluidic cover and well and produce a PMMA sheet of the actual parts. After the sheets are produced glue, followed by a silicon die will be placed into each well. A sheet of covers will then be used to force down the silicon die to the desired height and force all excess glue out of the well. After the glue cures and the silicon die is secure, biosensing material will then be printed onto the die turning the silicon die into a biosensor. Glue will then be placed on the lip of the cover and on the top edge of the well. The fluidic cover and well sheets will then be placed in an alignment microscope. The microscope will allow the two sheets to be aligned as well as bring the pieces into contact with one another so they will adhere. After the two sheets are secured the saw alignment marks previously incorporated into the mold design will be utilized to produce fully encapsulated dice. 70

32 Part Drawings Figure 29: Well Drawing 71

33 Figure 30: Fluidic Cover Drawings Figure 31: Bottom of Fluidic Cover Drawing 72

34 Figure 32: Well Mold Drawing Figure 33: Fluidic Cover Mold Drawing 73

35 Figure 34: Well Assembly Drawing 74

36 Bill of Materials Table 11: Bill of Materials Number Item Description Supplier 1 PMMA sheet (12") Fluidic cover and well Price ($ per item) Quantity Total Cost ($) McMaster Silicon Wafer Silicon die Roguevalleymicro Biomaterial Coating die Unknown Unknown Unknown Unknown 4 5 Brass alloy 260 (5") Loctite 3921 (250ml) Brass insert for hotembossing UV glue to Bond PMMAs and Die Megapalm Loctite Total

37 Section III Evaluation Supporting Documents 3.1 Evaluation Reports Tolerance Analysis Introduction The design criterion being evaluated is that each part should be aligned accurately to within 50 microns of its desired position in space. The other design criterion is to confirm that the depth of the well floor and channels are deep enough to ensure that all three parts; microfluidic cover, well, and silicon die will be in contact with one another. Method These design criterions are to be evaluated by using tolerance calculations. For each part being manufactured there is a manufacturing error associated with it. These parts could be bigger or smaller depending on this manufacturing error is. Show below, in Figure 35, is a cross section view of an encapsulated die with these manufacturing errors. 76

38 Figure 35: Dimensions of the Parts with their Corresponding Uncertainties In order to find the total uncertainty each uncertainty in each coordinate direction will need to be summed together by using the governing equation: Each uncertainty has a confidence level associated with it, which depends upon the manufacturer. A confidence level of 95% states that 95% of all the parts manufactured are within the given uncertainty range. The confidence level for U total will then be dependent upon U 1 and U 2. If those two values of U 1 and U 2 are in the confident range of 95%, then U total will also have a 95% confidence range. These uncertainties will then be used to find the maximum well dimensions. This will be done by taking the largest uncertainty and adding it to the nominal values for each dimension as shown here: 77

39 Chip Dimension ( length) length U total ( length) l1 l2... ln ln 1 U total ( length) In order to find if the alignment of the fluid port locations are within the design requirements of 50microns the silicon die would have to be placed at the farthest location away from its intended position of the center or the well. This would be located at the corner of the well shown below in Figure 36. Figure 36: Maximum Displacement of the Silicon Die from the Fluidic Ports on the Fluidic Cover This is the location furthest away from the silicon dies intended location. This distance away is the alignment error and can be found by adding the space between the silicon die with the largest tolerance values shown in the equations presented here: Width Gap Desired Spacing U total(width) Length Gap Desired Spacing U total(length) Optical Alignment Error error x error 2 direction 2 y direction 78

40 Since the width and length gap are going to be the same value the Pythagorean Theorem can be to used find the furthest displacement of the silicon die from the fluidic ports on the fluidic cover. Max Displacement ( Length Gap 2 Width Gap 2 Optical Alignment Error Results Using the equations shown above the uncertainties for each dimension can be calculated. To find the actual length and width of the well, to allow a 95% chance of all the dies fitting inside the well, the uncertainty values will need to be added to the nominal values. The spacing gap is just the space between the silicon die and die wall (50microns) plus the uncertainty. Table 12 shows that maximum well dimensions as well as the largest gap between the silicon die and a well wall. Table 12: Maximum Well Dimensions: Nominal Uncertainty (±) Maximum Die Dimension Gap Width (microns) length (microns) The optical alignment has an error in the x and y direction of 5microns which makes the optical alignment error to be 8.2microns. With the optical alignment and spacing gap error the max displacement of the silicon die inside the well is 122 microns. The depth of the glue is calculated quite in the same way but the depth of the glue needs to be decided. The depth of the glue is the height of the bottom of the well to the bottom of the channels. The glue thickness is determined by using twice the uncertainty of the depth. By taking this into account all silicon dies that are on the large end of the uncertainty analysis can fit into the well without touching the bottom; and all silicon dies that are created smaller will sit 79

41 on top of the glue and be flush with the channel ledge. So the final depth of the entire well is the nominal values plus the glue depth, and can be found in Table 13. Table 13: Depth Dimensions for the Well Depth Nominal Value Uncertainty (±) Glue thickness Required Die Depth microns Discussion This evaluation determined the minimum inside dimensions for the well. Using these dimensions there is a 95% confidence rate that all the parts will fit and adhere to each other. By finding the minimum dimensions they can be compared to the actual dimensions of the well. By comparing the minimum well dimensions to the actual dimensions in Table 14 it can be seen that well dimensions are adequate to hold more than 95% of the silicon dies because they are larger than that of the minimum die dimensions. Table 14: Actual vs. Minimum Well Dimensions Dimensions Actual (micron) Minimum (microns) Width Length Depth Channel Depth The alignment test did have an error of 122microns when the silicon die was forced into the corner, which is much larger than the desired requirement of 50microns. This is also the worst case scenario because it is calculated with the largest uncertainty. Also since the placement of the silicon die is not part of this project there could be a chance that the die could be placed directly into the center of the well. If this occurs the maximum alignment error would be due to 80

42 the optical alignment features and would only be 8.2microns. This value is even smaller than the ideal value of the error being less than 10microns. Using this knowledge it can be assumed that the fluidic ports would have a large chance of being aligned with each other and the design requirement would be met. Structural Analysis Introduction Needing to ensure the structural integrity of the encapsulated die during the assembly process, it is desired that the dimensions of this encapsulated die assembly (well, silicon die, fluidic cover) not allow significant deflection or excessive stresses to build up and exceed the yield stress of the material. Methods To accurately simulate this situation, a CAD model of the assemblies to be investigated must be created. However, not all dimensions are known, most notably, the wall thickness of the well. To obtain a rough estimate of the required wall thickness, the longest wall was modeled as a column. With a zero buckling condition, the following equations can be solved to obtain an expression for the wall thickness. Uniform pressure, P, on area b h Fixed Free Boundary Condition L Figure 37 Simplified model for buckling prediction 81

43 4 Equation 1 12 Equation 2 Equation 3 Where: h = wall thickness b = length of wall L = height of wall E = modulus of elasticity of material A = Cross sectional area of wall as from above (b*h) viewed I = Moment of inertia of wall P = Pressure acting on a single wall (1/2 total pressure applied) 82

44 Solving these equations for the wall thickness needed it is found that 48 Equation 4 It has been observed that the finger pressure applied to a surface by a normal individual ranges from 0.03 to 0.10 Mpa (Source: 3M Tape specification). This part is designed to withstand ten times this finger pressure and therefore a total pressure of up to 1 Mpa (0.5 Mpa on each wall). It is known that the modulus of elasticity for PMMA is approximately 3.2 Gpa (Source: LSU thesis) and the yield strength is approximately 72 Mpa (Source: matweb.com). After solving for the approximate thickness required and finalizing CAD models for structural simulation, the models should be imported into ANSYS for full simulation analysis. Results Following the rough calculations for a total pressure of 1 Mpa, a wall height of 0.30 mm (0.20 for die, 0.06 for sealing lip and 0.04 for glue thickness), and the known modulus of elasticity, the required wall thickness is merely mm or 8.3 μm. To make machining easier and cheaper, a minimum thickness will be defined as mm or 100 μm. This was the value used in the ANSYS simulation for the thickness of the walls. After importing the geometry of the model into ANSYS, the model is meshed, a pressure on the entire upper surface of the fluidic cover was applied and a 0 displacement condition was applied on the bottom surface of the well. Lastly, simulation results of the displacement in the vertical direction as well as the stresses present in the part are collected. The following figures summarize these results. 83

45 Figure 38 Mesh view of the chamfer model Figure 39 Von Mises stress in chamfer model 84

46 Figure 40 Y direction displacement of chamfer model Figure 41 Mesh view of inverse chamfer model 85

47 Figure 42 Von Mises stress in inverse chamfer model Figure 43 Y direction displacement of inverse chamfer model 86

48 Figure 44 Mesh view of wing model Figure 45 Von Mises stress in wing model 87

49 Figure 46 Y direction displacement of wing model From these results it can be observed that the silicon die was removed from the model geometry. It is assumed that the silicon die will significantly reduce the amount of deformation observed in the fluidic cover and simulation of this case supports this assumption leading to a maximum deflection with the silicon die of 0.3 μm. The case where the silicon die is not placed into the well before assembly represents the worst case scenario for the structural rigidity of the encapsulated dye and is of the most concern. From the results shown, it was observed that only one model (inverse chamfer) exceeded yield stresses (72 Mpa) by showing stress levels in excess of 100 Mpa. This high level of stress is observed at the top of the well wall where the fluidic cover and the well interface. Lastly, excessive deformation of the fluidic cover is very evident in the Y direction displacement images for the three simulated models. While ANSYS plots show this deformation as exaggerated, the actual deformation of the cover is very small and never exceeds 25 μm. While this is important for these images, the actual parts that will be created will have more PMMA on this surface of the fluidic cover. This is due to the need to drastically increase these dimensions because of inconsistencies in the flywheel grinding done on the finished parts after hot embossing. 88

50 It should be noted that one model was not able to be simulated at this time. Errors were encountered during the meshing process of the groove model and were unable to be solved. More investigation into the model and meshing will take place to rectify the errors. Discussion After reviewing the results of this analysis, it is clear that the structural integrity of the walls is very high. While a thickness of only 8.3 μm is required with a safety factor of 10, the walls are designed to be at least 100 μm. Along with this, it effectively eliminates the inverse chamfer design since local yielding of the walls takes place at a critical place in the design. The Von Mises stresses seen in the chamfer design, in a truly worst case scenario, are less than half of the yield stress. This gives credence to the structural rigidity of this design as well as its superiority in displacing the stresses applied to it. Seal Integrity Analysis Introduction In order to ensure the structural integrity of the design while under pressure from fluid flow, the strength of the seal between the silicon die and the microfluidic cover and the pressure that will be seen within the fluid chamber must be assessed. If the force generated by the pressure in the chamber is less than the amount of force the bond between the lip and the die can withstand, the structural integrity will be validated. The pressure within the chamber will be assessed based upon a known flow rate. The strength of the seal is dependent on three factors: the width of the lip which creates the seal, the total length of the lip, and the bond strength of the adhesive used to hold the lip to the silicon die. Figures 47 and 48 show the dimensions used for analysis of the seal. 89

51 Figure 47 Initial fluidic cover dimensions used in seal strength analysis, top view (dimensions in mm) Figure 48 Initial fluidic cover dimensions used in seal strength analysis, view from end (dimensions in mm) 90

52 Methods The requirements were tested using a two step process. First, an estimation of the pressure distribution within the chamber was found using CFX and validated with simplified hand calculations. Second, the strength of the seal between the lip of the microfluidic cover and the silicon die was estimated. The maximum pressure found inside of the chamber was then multiplied by the surface area of the top of the chamber to obtain a force. This force represents the worst case scenario, as the maximum pressure seen in the chamber was assumed to act on all parts of it. The force obtained in this step was then compared to the strength of the bond between the lip and the die. For the CFX simulation, the blood or other bodily fluid that would normally flow through the device was approximated as water. This simplification was suggested by our project sponsor. A central fluid velocity in the chamber of 100 mm/s was used, as this is an optimal flow rate for sensing to occur. Again, this information was provided by our project sponsor. The following simplified hand calculations were used to verify the results of the CFX analysis. Central area of bath tube Fluid flow Figure 49 Simplified model for central velocity at bath tube Equation 5, Equation 6 91

53 , ; 998 Before using simulation, we had estimated the bottom pressure of the fluid, where the fluid generate the pushing force the Si die, by using Bernoulli s equation, equation 3. Inlet, point 1 Bottom surface, point 2 Figure 50 Simplified model for inlet and bottom surface Equation 7,, 998,,,, The calculated value was The pressure acts on the negative y direction, so was a positive value which meant the absolute value of was greater than absolute value of. From this estimation, we estimate that the bottom surface would have higher pressure than the inlet pressure, atmospheric pressure, of 1 atm. Next, analysis was done to ensure that the seal between the lip of the fluidic cover and the silicon die did not fail in the z direction due to the pressure found in the first step. This can be seen infigure

54 Fluidic Cover Z direction Fluid Port Pressure From Fluid Fluidic Chamber Glue Silicon Die Figure 51 Pressure from fluid on fluidic cover Results Using the central velocity of 100mm/s, we determined a mass flow rate of / at the inlet. This value was used in the CFX simulation for the fluid pressure at both bottom and wall surface. The CFX can show various types of results such as velocity contour, pressure at different surfaces, and the actual pressure values. The calculation in CFX is performed by using Navier Stokes equation, Equation 8. The results of the simulation are presented in Figure 52 to Figure 58. Equation 8,,,, 93

55 Figure 52 Bath tub central velocity contour Figure 53 Vector Contour 94

56 Figure 54 Inlet and outlet velocity contour Figure 55 Inlet and outlet velocity 95

57 Figure 56 Bottom pressure contour Figure 57 Bottom surface pressure 96

58 Figure 58 Wall pressure contour From these results the fluid pressure on the bottom, and wall surfaces, were determined. The maximum value of fluid pressure was found to be Using this maximum value of fluid pressure, the total force acting upward on the microfluidic cover can be estimated. The total surface area of the top of the chamber (excluding the fluid ports) multiplied by the maximum pressure inside of the chamber will give an estimate of the worst case scenario force that the microfluidic cover will see. This force is: / ^2.. This represents the force that the seal between the lip of the microfluidic cover and the silicon die needs to withstand. 97