Differential Plasmonic Sensors

Mustafa FIRAT Department of Physics, Bilkent University, Ankara, 06800, Turkey

Abstract – In this project we are aiming to fabricate a surface plasmon resonance sensor inside a microfluidic channel. Microfluidic systems provide simple means to control and manipulate fluids at micrometer scales. Integration of microfluidic systems and plasmon resonance sensors could provide very sensitive refractive index sensors with very small detection volume. In this project we will use a grating surface coated with a 50nm Ag layer will be used to excite surface plasmons. Furthermore we would like to improve detection sensitivity by performing a phase sensitive measurement.

Introduction use gold, because of its dielectric properties. Also, gold does not react with The aim of this project is to generate a oxygen and its surface is not oxidized in sensor that can measure physical, chemical time. Since we need its surface in order to and biological properties of fluids. By generate a surface plasmon, the surface combining the special characteristics of should not be oxidized for a good sensing. surface plasmons and microfluidic These are basic advantages of using a gold channels, we can get information about the layer to generate a surface plasmon. fluids that flow inside the microchannels. First of all, we inject two fluids into the Indeed, there is a distinction between microchannels. For the first tries, we will surface plasmons and surface plasmon use water and 10% Ethylene Glycol polaritons. At the boundary of the metal, because they are both transparent and there there are a huge number of electrons due to is a sensible difference between their index the free electrons of the metal. Since these of refraction, so in their dielectric are not binded to nucleus, they can be constants. Dielectric constants of the oscillated easily with an external electric medium determine the resonance condition field. If we can apply correct frequency of the surface plasmon sensor. We will and give the correct energy with the correct take advantage of the laminar flow inside momentum to these free electrons, they the microchannels to change the medium start oscillating with an exact frequency. on the surface plasmon sensor. In order to Actually they generate a plane wave in this do so, we should know Reynolds number. electron sea. This is called surface plasmon. And if the surface plasmons are Surface Plasmon Resonance coupled with a photon, the resulting hybridized excitation is called a surface Surface plasmons, also known as surface plasmon polariton (SPP). And we will plasmon polaritons, are the surface generate SPPs by sending photons from a electromagnetic waves that propagate in a Helium-Neon laser. And the technique of direction parallel to the metal/dielectric (or excitation of surface plasmons by light is metal/vacuum) interface. [1] In our setup, denoted as a surface plasmon resonance we use gold as the metal surface and we (SPR) for planar surfaces. use water-10% Ethylene Glycol laminar flow as the dielectric surface. Since all the The electric field of a propagating metals have free electrons, actually surface electromagnetic wave, here our light beam plasmon can be generated with almost all coming from Helium-Neon laser, can be metals. The condition is; dielectric expressed as constant of the metal must be negative and most of the metals have this property. We (1) where k is the wave number and ω is the the coming photon, from eq (4). If we frequency of the wave. By solving equate k x and k SPP we get Maxwell's equations for the ⁄ electromagnetic wave at an interface (9) between two materials with relative dielectric constants ε1 and ε2 (see figure 1) Here we can see that k 0 and cancels we get each other according to (7). Since this is the horizontal component of the total 0 (2) momentum at the same time, we can add g factor of the grating to this equation and according to (5). And finally we get 1,2 (3) ⁄ (10) where c is the speed of light in the vacuum. Solving these two equations, the dispersion This angle is the plasmon angle for a relationship for a wave propagating on the plasmon that that will be take place surface is between gold - 10% Ethylene Glycol interface. The only thing that we should do ⁄ is to obtain the correct angle that gives the (4) correct momentum in x direction. From the property of the grating, this momentum is shifted by g factor of the grated surface. (5) Here, m is a positive integer that describes first, second, third... order shifting to the momentum (comes from the grating) and g is the g factor of the grating, given by (6) Figure 1 : Coordinate system for two material where, λ is the wavelength of the coming interfaces and drawing of the vectors that were used photon and is the periodicity of the in equations 1, 2, 3, 4 and 8. grating. We used a grating that has a periodicity = 600 nm and we used Helium-Neon laser that has a wavelength 632,8 nm. By knowing the wavelength of Reynolds Number and Laminar Flow the Helium-Neon laser, we can get the momentum of the photons as Laminar flow occurs when a fluid flows in parallel layers, with no disruption between (7) the layers. In fluid dynamics, laminar flow And one can get the horizontal component is a flow regime characterized by high of the momentum of the coming photon momentum diffusion and low momentum [2] from k 0. If k 0 is the total momentum of the convection. There can be small mixing coming photon, k x is the x component of k 0 at interference layer of the fluids due to that is horizontal to the propagation of the diffusion but there is no mixing due to SPP. We can find k x from k 0 easily by turbulence in laminar flow. Turbulent flow (8) is just opposite the laminar flow. To And we know the momentum of the understand laminar flow in microchannels, surface plasmon that will be generated by we should examine Reynolds number at first. Reynolds number is an important parameter in the equations that describe photolithography technique. The gratings whether the flow will be laminar or that we use were fabricated by a graduate turbulent. It is a dimensionless number. student in the lab. Fabricating a grating is Reynolds number is about flow geometry. not a part of this project. Furthermore, The fluid properties; density and viscosity, there are several companies fabricating plus flow velocity and a characteristic such gratings. One can design the shape length or characteristic dimension and sizes of the gratings on the computer determine the Reynolds number. By and get them prepared to those companies. characteristic length or characteristic It is the easy way. dimension, we mean diameter, width, height... etc. For cylindrical shapes diameter is the characteristic dimension but since we have rectangular channels, we will use the height of the microchanells that is in the order of 100 µm. Having one dimension in µm is enough to see laminar flow easily. That’s why laminar flow can occur even with water in microchannels but it is very hard to observe a laminar flow in pipes in daily life. According to the definition, Reynolds number is given by (11) Figure 2 : Silicon masters used to mold elastomeric microchannels. We have used pattered SU8 where: photoresist as a master. These masters were fabricated in the clean room by a graduate student in the lab. • V is the mean fluid velocity (m/s) • L is a characteristic length,

(traveled length of fluid) (m)

• µ is the dynamic viscosity of the Design of Microfluidic Channel fluid (Pa·s or N·s/m² or kg/m·s)

• ν is the kinematic viscosity ( ν = µ / For design of microchannels, again we use ρ) (m²/s) silicon masters. These masters are prepared • ρ is the density of the fluid (kg/m³) with the same procedure as grating • Q is the volumetric flow rate (m³/s) masters. Photolithography technique is the • A is the pipe cross-sectional area most commonly used method. As shown in (m²) [3] figure 2, we get a silicon master that has a raise with height 100 µm. When we pour In our microchannels, if the Reynolds liquid Polydimethylsiloxane (PDMS) onto number is less than 2300, we get a laminar our master, and keep it on the hot plate at flow. 70 oC for one hour, PDMS becomes amorphous solid. And we get our

microchannels at the bottom of this elastic PDMS structure. This is the upper part of Grating Fabrication the channels. We need a cover to the bottom of the channels so that water and We used silicon grating masters to get a ethanol will pass through. Also, this part is grating on our PDMS. These gratings are the layer where we will generate the SPs. periodic on the silicon master and the So, grating and the golden layer will be in periodicity of the grating is 600 nm. These this part. We coated this second layer of silicon grating masters are prepared by PDMS by gold with a thickness 50nm. Detail illustration is given in the figures 3 and 4.

Figure 5 : Gold coated PDMSs with grating and Figure 3 : The schematic representation of the film masks used during the shadow evaporation. proposed microchannels integrated with a plasmonic sensor. The microfluidic channel is a simple Y-junction. The grating surface is molded It should be the same for two PDMS on a PDMS and this PDMS layer is sealed on a surfaces. However, when we tried to do so, microfluidic channel. we could not establish that chemical bonding. We could not stick two PDMSs Golden coat is established only through the to each other easily. Most probably, channels, not all over the surface, because chemical property of the PDMS was we need to stick it to the first layer of affected during the shadow evaporation PDMS. And PDMS does not stick to any coating process. This is the most material, including gold. In order not to challenging part of the project because if coat the entire surface by gold, we used we cannot stick two PDMS layers, we film masks, which will be used during the cannot get our microchannels. shadow evaporation coating process. These equipments are shown in figure 5.

Figure 4 : Cross sectional view of the micro channel integrated with the grating structures. The surface of the grating is coated with 50nm thick gold layer. We have patterned the metal in to 1mm wide stripes using shadow evaporation. Figure 6 : The fabricated microfluidic devices with an integrated plasmonic sensor. Micro channels are As seen from figure 4, it is clearer that we ready to use after sticking two PDMS layers. The stick two separated PDMSs. However, colorful reflection from the channels is due to the grating structure. very well known property of PDMS is that it did not stick to any surface, to any We overcame this problem by putting the material, as mentioned above. Actually, gold-coated PDMS to another plasma there is a way to stick it to a glass surface. system. By trying this method, we could If we interact both PDMS and the glass stick them together. with oxygen plasma, a kind of chemical bond, Van der Waals bond, is established After manufacturing the devices with a between them. gold-coated grating and microchannels, we tried to observe surface plasmon on the gold layer. However, we could not! There same time. In our RIE machine, these was again a problem. Most probably, values were optimized for Si and OG 146 shadow evaporation technique affected not polymer in our clean room. That’s why we only the chemical properties of the PDMS, chose using Si as a grating master. Finally, but also physical properties of gold. We we cleaned the wafer with piranha solution understood that it would not be achieved again. So, our grating is ready for coating by coating the PDMS surface by shadow on a Si wafer. Gold was coated to this Si evaporation process but we cannot coat surface by a post-doc scientist in our lab. PDMS surface with any other mechanism. We got gold-coated grating on our Si So, we started a new process from the wafer. We still have to stick our second beginning. We decided to use silicon wafer layer of PDMS, which has our instead of first layer of the PDMS because microchannels inside it. And it is easy to we need a transparency at only one side of stick PDMS on Si surface. It is the same as the channels. And we can coat Si wafer by sticking PDMS to glass. We stuck them another technique. together by putting them into oxygen plasma for 4 seconds. Finally, SPP was observed on our new device.

Grating Fabrication (2 nd technique)

Firstly, we cleaned the surface of the Si wafer with piranha solution (3 units of sulfuric acid and 1 unit of hydrogen peroxide). This solution cleans a surface both organically and inorganically. And we could not do the same thing to PDMS, which is an organic polymer. We coated the surface of the cleaned wafer with UV Figure 7 : At the left, the grated Si wafer by curable OG 146 polymer. We dropped the applying second technique. It is ready to stick polymer on the wafer and we rotated it in PDMS on it. At the right, the fabricated microfluidic devices stuck on grated Si wafer. the spinner at 5000 rpm for 40 seconds. Micro channels are ready to use after sticking This makes the thickness of the polymer PDMS to Si wafer. homogeneous. On this OG 146 polymer, we put our PDMS mould that has a grating on it as a grating master. Then, we put Modulation of the Flow Inside the them to the UV oven for 7 minutes. At this Microfluidic Channels and Phase step, OG 146 polymer becomes amorphous Sensitive Detection solid, having our grating on itself. The final step is sending laser beam to As usual, PDMS does not stick to any microfluidic channels and take the data. As surface easily and it does not stick to OG mentioned before, data will be taken in a 146 polymer as well. We just lift the way that we will find the correct angle of PDMS from our wafer. After leaving the coming light. At that exact angle, PDMS, we etched our wafer in RIE in horizontal component of the coming order to get rid of OG 146 polymer on it. photon’s momentum will match the We put them to RIE under 200 W RF for 3 momentum of the surface plasmon that minutes. During this process, a plasma, will be generated after coupling. The final consists of 27 sccm SF 6 and 7 sccm O 2 configuration of the experiment setup is gases at 26 microbars pressure, reacted shown below in figure 8. By putting a with Si wafer and OG 146 polymer at the detector, we can measure the energy of the reflected light and understand if SP was as 18,8 o . This makes sense because we generated. When SP is generated, some of used a very low concentration of Ethylene the energy of the incoming light will be Glycol and the difference is about 1 o spent to generate SP, and the energy of the between this and water - Au interface reflected light will decrease in a sensible plasmon angle, which is acceptable. The order. data for 10% Ethylene Glycol – Au interface is also plotted in graph 1, with the red curve.

7 %10 EtGy D I water 6

3 Figure 8 : Final configuration of the experiment setup. At the left our Helium-Neon laser sends the 2 Reflection (mW) Reflection light beam. At the middle, our detector measures 1 the energy of the reflected light. At the right, our microfluidic channels accept them and surface 0 10 12 14 16 18 20 22 24 plasmon occurs between two layers. Water and 10% Ethylene Glycol comes to the micro channels Angle (Degree) through the elastic pipes. Graph 1 : Reflection vs angle of the coming laser beam. Blue curve is the data for water - Au We found the dielectric constant of Au ( ε1) as -10,75 for a light with a wavelength interface and the red curve is for 10% Ethylene [7] Glycol - Au interface. 632,8 nm, used for generating SPP . Firstly, we tried to get SPP at water - Au After finding the each curve for water and interface. We calculated the plasmon angle 10% Ethylene Glycol, the next data was according to equation (10) for water and taken with modulating these two liquids Au. We substituted ε1= -10,75 for Au and 2 2 inside the microchannels. The core part of ε2= n = (1,33) = 1,769 for water. And we o the project is to achieve phase sensitive get θ = 17,76 . After calculating the detection from the surface plasmon sensor. theoretical value, we took the data from This goal will be achieved by modulating our device and we observed minimum o the fluid flow inside the channels. This reflection at θ = 17,6 . We are in very modulation will provide periodic signals in good agreement with the theory. The graph the reflected light. of this first data is shown in graph 1, with the blue curve. In order to do so, we put two water tanks at different heights, so that one of them has a After finding the plasmon angle for water - higher pressure, and the other one has Au interface we filled the microchannels lower. And we put a valve between these with 10% Ethylene Glycol and took the two water tanks, that permits the water data for that interface. This time we do not from these two tanks in a periodic order have a theoretical value because we are (see figure 9). We drive the valve’s using a mixture, not pure Ethylene Glycol. periodicity with lock-in amplifier. The Since we found the plasmon angle very valve has two inputs and only one output. well for water, our device is trustable for So, it sends high and low pressure water in 10% Ethylene Glycol as well. And we a changing order to the microchannels. found the plasmon angle for this solution signals are the same. And the graph 2 says exactly what we were expecting.

Graph 2: Signal of the modulated water - 10% Ethylene Glycol laminar flow between 14 o and 26 o . o We have approximately zero signal at 20,2 . Figure 9: At the top, higher pressure water tank, at the bottom lower pressure water tank, at the middle, The only concern is, we have a shift about 10% Ethylene Glycol tank; in front of it, automatic 2o. We found zero signal at 20,2 o . We did valve lies. And the function generator was used for taking data from non-modulating laminar flow (for not take this data and the data in graph 1 in the first two data). the same day. Most probably, there is an alignment problem during the setting the However, 10% Ethylene Glycol is at a experiment setup again. fixed height. When the higher pressure water passes with 10% Ethylene Glycol After this step, we fixed the angle to 21 o, through the micro channel, the gold layer where the reflection difference has highest touches to water mostly. On the other value according to graph 1. At that angle, hand, when the lower pressure water we changed the frequency of the valve passes with 10% Ethylene Glycol, the gold from 1 Hz to 21 Hz in order to find that at layer touches to 10% Ethylene Glycol which frequency of the valve we can get mostly. And this modulation is controlled the maximum signal from the modulation. by lock-in amplifier with the help of the And the data is plotted in graph 3: automatic valve. In front of the SPP device, our sensor measures the energy of the reflected light from this modulation and sends it to another lock-in amplifier. We modulate the liquids because lock-in amplifier can measure the changing rate of a signal. That’s why we call this device as “differential” plasmonic sensor.

When we modulate the liquids in a laminar flow, we expect a zero data at the intersection point of the graph 1, at 18,2 o. At that point, reflections from both liquids are the same. Since the lock-in amplifier read a differential signal from the detector, Graph 3 : At fixed angle, we found most efficient it cannot read anything when the two frequency (3 Hz) to make a less noisy experiment. As seen from the graph 3, we found the most efficient frequency as 3 Hz. So, the main aim of this experiment, phase sensitive detection, will take place at 3Hz modulation of the laminar flow.

Graph 4 : Signal vs concentration. This is the sensitivity of the “Differential Plasmonic Sensor.” As seen from the graph, we have almost linear reduction up to 0.01%. We have some unexpected Figure 10: Upper lock-in amplifier drives the signals after 0.01% saturation, where we lost the valve at 3 Hz and the laminar flow modulates at sensitivity. that frequency. Lower lock-in amplifier takes the data from our photo detector. At the right, the computer drives the lower lock-in amplifier by lab- view program. It varies the frequency between 2.5 and 3.5 Hz and it waits at each frequency for 15 Conclusion seconds. Lab view saves the data automatically. (At that moment, equipments are working for another We fabricated a sensor that measures the experiment setup, this is just an illustration.) concentration of the liquid passing through it and we achieved a sensitivity at 0,01% Final step of the experiment is phase concentrations. This sensitivity can be sensitive detection. At this step, we fixed o improved by using a grating with longer the angle to 21 which has the maximum periodicity. For 550 nm periodicity, we got reflection difference between the liquids, the SP at 20 degrees. And it is very in other words, the most efficient angle. difficult to work at small angles. If we used And we fixed the modulation frequency to gratings with 600 nm or larger gratings, we 3 Hz, which is the most efficient had worked with larger angles, which are modulation frequency to take less noisy more stable. Furthermore, since we had to data. work with small angles, we could not put the photo detector close enough to the After that all, we started to dilute the microchannels. As seen from the figure 8, concentration of the Ethylene Glycol. We when we put the detector closer, it prevents started with 10% at the beginning, and the laser from the behind, so that we put it each time we diluted it to its half about 20 cm away from the reflection. That saturation. In order to do so, a lab-view is an also inefficient way for taking the program ran the second lock-in amplifier. data because the detector loses its It varied the frequency between 2,5 Hz and sensitivity as it goes away from the source. 3,5 Hz. After each diluting time, we Besides, we observed some fringes at the expected a half height peak at 3 Hz. We reflected light, due to the interface of the diluted the Ethylene Glycol from 10% to laminar flow. Light interferes because it 0.001%. According to graph 4, we saw reflects from two different liquids with two that our sensor is sensitive up to the different index of refractions. That makes concentrations about 0.01%. the data very noisy. In the future work, we will try to get rid of these problems for the development of the sensitivity of our sensor. Using a longer periodic grating will be helpful for working with small angles and also, we will be able to put the detector closer to the reflection it that way. We cannot do anything for the fringes due to the interface of the laminar flow, but it will be less effective in the data when we will put the photo detector closer to the plasmon surface. By doing these so, we expect a sensitivity about 0.0001% saturations, which will be, most probably, the best sensitivity for such sensors.

References

[1] [WWW document], URL: http://en.wikipedia.org/wiki/Surface_plasmon_re sonance [2] [WWW document], URL: http://en.wikipedia.org/wiki/Laminar_flow [3] [WWW document], URL: http://en.wikipedia.org/wiki/Reynolds_number [4] H. M. Hiep, "A localized surface plasmon resonance based immunosensor for the detection of casein in milk". “ Sci. Technol. Adv. Mater”, (2007), 8. [5] S. Pillai, K. R. Catchpole, T. Trupke and M. A. Green, “ J. Appl. Phys” (2007), 101 . [6]Barnes L. William, Dereux Alain, Ebbesen Thomas, “Surface Plasmon Subwavelength Optics”, “ Insight Review Articles ”, (2003), 825 . [7]This value calculated by software, named Lumerical FTDT Solutions . [8] S. Yee Sinclair, Gauglitz Günter, “Surface plasmon resonance sensors: review”, “Sensors and Actuators B”, (1999), 3.