To determine the S, the Linear Phase Modulation Interferometry (LPMi) is applied on the SiSPR chip using the laboratory prototype shown in Fig. 33. To precisely determine
Fig. 39 Simulation of ϕSPR retrieval.
The difference between ideal ϕSPR (black line) and the retrieved phase (red line) without noise is negligible so that two curves are actually merged together. Blue curve indicates the phase retrieved under condition of large noises and very weak contrast.
doi:10.6342/NTU201700804
86
the OPD, and thereby precisely determiner S, the SiSPR chip is placed at total internal reflection (TIR) angle for glass-air interface (41.8 degree). TIR angle if firstly obtained by the laboratory set-up without the SiSPR chip, by observing the vanishing of refracted beam. After arriving at TIR angle, SiSPR chip is placed onto the laboratory prototype, and a saw-tooth function is applied to the VCSEL laser diode. The saw-tooth current ramp has an amplitude (Δi) of 1.404 mA on the VCSEL laser (𝜆0 = 850 nm) with a constant DC bas of 0.5 mA.h At TIR angle, the OPD in SiSPR is around 2.26 mm (optical thickness). The LPMi interferogram is then picked up by our detector. The resulting interferogram can be seen in Fig. 40.
As can be seen in the figure, the interferogram has a sinusoidal function atop its saw-tooth function. It is expected since in case of LPM interferogram (with 𝐼0 = 𝑐𝑡𝑒) the
signal is:
𝐼𝑑𝑒𝑡 = 𝐼0(𝑡)[1 + 𝑚𝑐𝑜𝑠(𝛥𝜙)] = 𝐼0(𝑡)[1 + 𝑚𝑐𝑜𝑠(2𝜋𝐿
𝜆20 𝑆𝛥𝑖)] Eq. 43
Where I(t) in this case is in form of piece-wise linear function. Based on above equation, we can drive that:
Δ𝜙 =2πL
λ02 SΔi Eq. 44
The advantage of this linear phase modulation interferogram is that the amount of phase
h Δi=1.404 mA is achieved by setting laser diode at Vdc=2.0 V with a rampe of 0.6 V (The a current limiting resistor of 220 ohm is connect in serie with VCSEL)
doi:10.6342/NTU201700804
87
change (𝛥𝜙) can be simply judge from the waveform peak-and-trough, since it follows a simple sinusoidal function. For example, as marked in the figure with the black arrow, as 𝛥𝑖 increases, we can observe Δ𝜙 from the waveform. Through a collection four of 𝛥𝜙/𝛥𝑖 relation ship (from 𝛥𝜙=0 to 3π), spaced by π), we are able to construct a
linear regression between 𝛥𝜙 and 𝛥𝑖 (cf. Fig. 40). From the fitted linear regression relationship in Fig. 41 , we know empirically that:
𝛥𝜙 = 12.9𝛥𝑖 Eq. 45 And we can derive from :Eq. 44 and Eq. 45 that:
𝑆 = 12.9 × 𝜆02
2𝜋𝐿 Eq. 46
where, as stated earlier, L=2.26 mm, 𝜆0 = 850 𝑛𝑚, we obtain S= 0.62 nm/mA for the VCSEL under consideration. Based on the obtained S value, we can now estimate the
Fig. 40. Laser intensity normalized LPM interferogram.
doi:10.6342/NTU201700804
88
amount of 𝛥𝑖 needed to generate 𝛥𝜙𝑎 = 3.8317 in case of SiSPR. This estimation is of vital importance as it lies at the core of SiSPR phase extraction. Typically, Kretschmann angle of water lies at 65 degree of incidence (cf. Fig. 36). Based on the result from Fig.
29, we know that the OPD is 1.267 mm for incident angle of 65 degree. We have therefore come to a 𝛥𝜙𝑎 vs 𝛥𝑖 functions for SiSPR:
𝛥𝜙𝑎 = 6.793 (𝑟𝑎𝑑
𝑚𝐴) × ∆𝑖(𝑚𝐴)
Which is for the case where Kretschmann angle is at 67 degree. Both of the estimation is plotted in the figure and the insets. In case of 65 degree Kretschmann angle, it requires
0.56 mA to obtain meet the condition of 𝛥𝜙𝑎 = 3.8317 for phase extraction.
Note that S value for each VCSEL has to be measured independently for precise Fig. 41 Fitted 𝛥𝜙/𝛥𝑖 for evaluation of S value.
doi:10.6342/NTU201700804
89
estimation of S and 𝛥𝜙𝑎. The calculation demonstrated herein serves as an example, but the value may subject to change depending on the diode used.
Section 9-3. Imaging of beam profile of SiSPR
In this part, we will provide an image based analysis on SiSPR interference fringe. To this end, a IR camera is made from model C170 webcam from LogitechTM , Taiwan. As shown in Fig. 42, the IR camera is used in place of the P and S photodetector in the laboratory prototype (cf. Fig. 33). To image the beam cross sectional profile, the IR camera is placed farther than focal point of the coupling prism. This this distance from prism, the beam is diverging slightly.
The capture can SiSPR beam cross sectional profile and interference can be seen in Fig.
43. The image is taken under idc=0.968 mA (Vdc=2.2 V) for VCSEL at different incident angle without any Δi.
Fig. 42 optical set-up for imaging of the SiSPR beam cross section profile .
doi:10.6342/NTU201700804
90
For this particular set of SiSPR chip, the reflective layer is 8 nm of gold with 2 nm of adhesion layer. In the captured image, we can clearly observe the signal beam and reference beam as discussion in wave optics analysis (Red circle: signal beam/Blue circle:
reference beam). As shown in the image, we can observe clear fringes in the overlapping zone. When SiSPR is placed in air, without any micro-fluidic system, we can see clearly how signal beam is absorbed at 42.5 degree due to the SPR effect. After 43 degree of incidence angle, we can clearly see how signal beam recover its intensity.
We have also performed same image process when SiSPR chip is place under deionized water using microfluidic channel. The result can be seen in Fig. 44. As we can see, the SPR dip is around 65.40 degree, near the simulation results. This angle will be used for SiSPR sensing experiments. We have observed a shape distortion in reference beam, which may likely originated from the relatively small size of prism as compared to beam size, which could be further optimized in near future.
Fig. 43 image of SiSPR fringe and beam cross sectional profile.
doi:10.6342/NTU201700804
91
Note that effect of reflective thickness can also be observed by imaging the cross sectional beam profile of SiSPR. As shown in Fig. 45, we compare the beam profile
between SiSPR chip with different reflective layer thickness. In one chip, the reflective layer is composed of 8 nm of gold plus 2 nm of adhesive layer (right hand side of the
Fig. 44 SiSPR signal when device is under pure water.
Fig. 45 Image of SiSPR signal with different thickenss of reflective layer.
doi:10.6342/NTU201700804
92
figure), while another SiSPR chip is composed of 2 nm of Au plus 2 nm of adhesive layer).
As can be observed from the figure, when reflective layer is overly thin, the intensity of reference beam would be too small to be observed in the image. Moreover, due to low
intensity in reference beam, the contrast is strongly decreased. As result, the fringe can hardly be observed in the image. Therefore, the thickness of SiSPR’s reflective layer need
to be carefully decided via image based analysis. For present work, the reflective layer is 8 nm of gold plus 2 nm of adhesive layer.
Section 9-4. SiSPR sensing performance:
In this section, we will examine the sensitivity (∆ϕ/Δn) of the SiSPR and compare it with the past literature. For this purpose, the SiSPR chip with 44 nm and 47 nm of plasmonic layer is tested against reference solutions.
To evaluate the sensitivity (∆ϕ/Δn) of SiSPR chips, glucose solutions are used as reference solutions as their refractive index has well defined behavior at λ=850nm with relationship of n = 0.0018xC + 1.33, where C is the glucose concentration in terms of weight percentage (w/w%). In this way, the phase response of the SiSPR chip with different surface refractive index can be estimated. The gluocse concentration for reference solutions ranged from 0 % to 11 % (w/w%), which is marked in the upper inset of the figure. The measured phase response is recorded to calculate the sensitivity later on.
doi:10.6342/NTU201700804
93
Based on our calculation in chapter 9-2, we use a VDC of 2.35 V and a VAC=0.239 V to achieve a Δ𝜙𝑎 of 3.8317 rad (Where S~0.62 nm/mA for this typical VCSEL). The 𝜇 is measured to be 0.59. Before the measurement, a rough scanning of SPR angle is performed with CCD to find the dip. Then, the P & S photo-detectors are placed around 0.35 o higher than the resonance angle. A good level of contrast, as can be seem in Fig.
46, is of fundamental importance to obtain low noise sensing results.
Fig. 47 and Fig. 48 reveals the phase response of SiSPR chips with 44 nm and 47 nm of plasmonic layer thickness. The sensing process are comprised of three consecutive cycles for obtaining statistically significant data. Each cycle starts with D.I water (n as a baseline, followed by several glucose reference solution. For example, in Fig. 47, solution of 0.75% (n=1.33135), 0.75% (n=1.33135), 1.00% (n=1.33180), 1.50% (n=1.33270),
Fig. 46 Interferogram from p-polarization and s-polarization of SiSPR signal.
doi:10.6342/NTU201700804
94
Fig. 47 Phase and Amplitude response of SiSPR chip with 44 nm of plasmonic layer
Fig. 48 Phase and Amplitude response of SiSPR chip with 47 nm of plasmonic layer
doi:10.6342/NTU201700804
95
1.75% (n=1.33315), 2% (n=1.33360) and 2.5% (n=1.33450), 3% (n=1.33540), 4%
(n=1.33720), 6% (n=1.3408), 7% (n=1.34260), 9% (n=1.34620) and 11% (n=1.34980) glucose concentraion with a total refractive index change of 0.0198 are used to retrieve
∆ϕ/Δn of the SiSPR chip. The phase steps, for one measurement cycle (out of three), are marked with the corresponding concentration in the inset for easy comprehension. The red trace indicates the P-S phasogram, while the blue trace marks the P/S amplitude sensorgram. As can be seen from the figures, when the glucose concentration in the microfluidic channel rises, we observe a monotonic increase in the P-S phase signal. In P/S amplitude sensorgram, we observed first a reflection dip followed by increase in reflection. This trend is expected as we start with an incident angle a bit higher than the resonance angle. Being near the center of the reflective dip, it allows us to give a good estimation of maximum phase sensitivity. In case of 44 nm plasmonic layer, we have observed 3.00 radian over 0.0198 refractive index change, while it is 3.25 radian for 0.0108 refractive index change when plasmonic layer is 47 nm.
Based on the phasogram shown in Fig. 47 and Fig. 48, ∆ϕ vs Δn calibration curves are built to discuss sensitivity (∆ϕ/Δn), dynamic range and minimum resolvable signal of the SiSPR chip. The result can be seen in Fig. 49. Linear fits are applied to the curve to obtain the sharpest response of the chip. The fitted results indicates that:
∆ϕ = 226n − 304 for SiSPR with 44 nm plasmonic layer
doi:10.6342/NTU201700804
96
∆ϕ = 475n − 637 for SiSPR with 44 nm plasmonic layer
The noise of the phase noise for our measurement is typically around 0.0002 rad over a period of 150 seconds. We have therefore obtain theΔn𝑚𝑖𝑛 for SiSPR chip around 2.1 x10-6 RIU at 44 nm and 1.2x10-6 RIU at 47 nm. And the linear range for the SiSPR chip is 0.0072 RIU for 44 nm of SiSPR chip and 0.0018 RIU for chip with nominal thickness of 47 nm.
The result, which indicate sharper phase response with increasing plasmonic film thickness, is as predicted from simulation based the matlab code mentioned in Section.8-5 which is made to calculate the reflectivity of an arbitrary multilayer stack with varying
Fig. 49 P-S phase response vs refractive index
doi:10.6342/NTU201700804
97
surface refractive index (to simulate the reference solutions), from the knowledge of the thicknesses of the materials and the Fresnel coefficients (determined by the complex refractive indices). In the simulation, SiSPR chip is consit of a reflective layer of 2 nm ZnO with the plasmonic layer thickness ranging between 44 to 49 nm. The wavelength is 850 nm. The coupling prism is BK-7 (n=1.514). The refractive index for ZnO is 1.9540, and it is 0.1649+5.374i for the gold layer. The simulation reveals the effect of changing refractive index from 1.3315 to 1.330 with different plasmonic film thickness, as shown in Fig. 50. As indicated by the simulation, the slope of the phase response increase dramatically with increasing plasmonic layer thickness. Judging on the dynamic range and the sensitivity of the film, as compared to the simulation results, we consider that there are further spaces for greatly improve the sensitivity of the chip.
We now proceed to discuss this ∆ϕ𝑚𝑖𝑛 and the corresponding ∆n𝑚𝑖𝑛. In present context, ∆n𝑚𝑖𝑛 is defined as the limit of detection of the system. Therefore, noise of the
system is firstly determined from standard deviation of a static measurement. The “static measurement” is made when SiSPR chip is placed in background solution for a lone
period of time without any perturbation in microfluidic channel. The noise in P-S phasogram is determined from a data with 300 seconds of measurement and with 3000 points of data. In case of 47 nm of plasmonic layer, where the slope of phase response is sharpest, we have measured a noise is around 2 x10-4 radian. The phase noise level is
doi:10.6342/NTU201700804
98
around 2 times smaller as compared to literature, which may be attributed to the
monolithic design in combination with the performance of generalized Lock-in Amplifier.
In context of metrology, the minimally detectable signal, i.e the limit of detection, have to be at least 3 times the size of the noise to be regarded as statistical meaningful signal.
Since ∆𝜙/∆𝑛=475 rad/ RIU, we therefore conclude, for P-S phase detection, that
∆𝑛𝑚𝑖𝑛 = 1 475
𝑅𝐼𝑈
𝑟𝑎𝑑 × 0.0006 𝑟𝑎𝑑 = 1.26 × 10−6 𝑅𝐼𝑈
Same calculation is performed on amplitude response of the SiSPR chip, based on the Fig. 51, to evaluate the claimed sensitivity advantage of phase measurement against amplitude measurement. Based on the estimated noise and the slope (∆𝑅
∆𝑛= 207 for Fig. 50 Simulation on SiSPR phase response upon different refractive index.
The color of the trace marks the phase response of SiSPR at different incident angle as marked in the legend.
doi:10.6342/NTU201700804
99
SiSPR with 47 nm thickness), we conclude that the ∆𝑛𝑚𝑖𝑛 for the amplitude measurement is
∆𝑛𝑚𝑖𝑛= 1
207 𝑅𝐼𝑈
𝑟𝑎𝑑× 0.0045 𝑟𝑎𝑑 = 2.1 × 10−5 𝑅𝐼𝑈
Therefore, we have come to a first conclusion that, using SiSPR setup, phase
measurement provides a 17 times better resolution as compared to amplitude based measurement. Since that P/S amplitude sensogram and P-S phasogram is obtained by entirely identical system and signal processing method, we consider this comparison provides a rather fair ground. We would like also to discuss the implication of the interesting data above. As shown earlier, despite the nominal 47 nm design, our chip still has less ∆𝜙/∆𝑛 slope as compared to the simulation results. According to the simulation,
Fig. 51 SiSPR P/S amplitude response vs refractive index.
The P/S amplitude response as well as the noise, are normalized by the intensity of the reflective dip for comparison of between chips.
doi:10.6342/NTU201700804
100
if we increase plasmonic layer thickness to 49 nm, the simulation result suggest a slope of 11631 rad/RIU which is 24 times greater than what we have measured herein. If such chip is feasible, through annealing or precise control of film thickness, we would have achieved a SiSPR chip with much higher sensitivity if noise remain at the same level. It would also be interesting to see what is the maximum sensitivity advantage that phase measurement offers over amplitude detection.
Effect of P-S different interferometry and temperature drift effect
In case of SiSPR, due to the use of low tunability source with long OPD chip, the temperature drift of the laser emission wavelength require special attention in order to achieve a stable measurement results. Fig. 52 , which demonstrates the difference of P-S and p-ploarized phasogram, is presented to indicate the importance of the differential phasogram. In this figure, glucose sensing between 0-2.5% is conducted with a concentration steps of 0.25%, 0.5%, 0.75%, 1%, 1.5%, 1.75%, 2% and 2.5%. As can be seen in this typical case, P-S phasogram (red trace) offers a sharp and distinct phase transition upon refractive index change, while we observe trends of baseline shifting as well as noise with abrupt changes when only p phasogram is considered.
The drift is prominent in p-polarized phasogram since the temperature induced wavelength drift is not negligible in case of SiSPR. The VCSEL, in present work, has
doi:10.6342/NTU201700804
101
temperature dependence of lasing wavelength (Δ𝜆(𝑇)) is 0.06 nm/Ki and a current dependence of lasing wavelength of (S) of 0.6 nm/mA. In SiSPR phase extraction process, we have a sinusoidal wavelength modulation amplitude of 0.34 nmj. In other words, it would lead to a 0.68 radian of phase drift if there exist only a 1 degree temperature perturbation in laboratory environment. It is therefore strongly required to has differential interferometry set-up to taken out this temperature drift effect.
Aside from temperature, differential phasogram also help to ease strain induced chip deformation when micro-fluidic channel is performing withdraw/forward action. This strain lead to phase drift (most frequently in forms of spikes), presumably due to change
i As can be found in the spec. of the VCSEL.
j Corresponding to Δ𝜙𝑎 of 3.8317, induced by a Δ𝑖 of 0.56 mA.
Fig. 52 P-S phasogram and P polarized phasogram in detection of different glucose solution.
doi:10.6342/NTU201700804
102
in optical path length when chip is deformed under fluidic pressure, which can be clearly observed with an IR camera by looking at the fringes when withdrawing solution from the microfluidic channel. Again, this kind of strain induced spikes, are presented equally in S polarization and P polarization, and can therefore be simply canceled out by extracting the P-S phase. The same idea applies to P/S amplitude information as compared to the only P amplitude information, where changes in the optical source amplitude are compensated by performing the ratio of the two amplitudes.
In brief, the P-S differential phasogram is strongly suggested in SiSPR measurement due to unique design of the phase extraction process. However, the readers should note that a temperature PID controller might still offer further stabilization in the near future, since P-S phasogram cannot exclude the temperature effect on the refractive index of aqueous solution despite this effect is undesirable.
Section 9-5. Preliminary bio-sensing Data
In this section, we demonstrate the preliminary result of biosensing using SiSPR sensor chip. To begin with, Tro4 aptamer42 is modified SiSPR chips and the reactions are monitored via the P-S phasogram. We will use diffusion limited Langmuir model to have a closer look at the surface modification condition. Finally, we will introduce our preliminary results on the cardiac troponin I detection.
“Aptamer” refers to a short strand DNA that has a functionality of antibody to a specific
doi:10.6342/NTU201700804
103
protein. Due to the room temperature stability, ease of synthesis and the potential for further chemical structure amendment, aptamer has been widely applied as probe molecule in detection of Platelet-Derived Growth Factor (PDGF)25, interferon gamma for Tuberculosis screening16 and human chorionic gonadotropin detection15.
In 2015, Hunho Jo et. al.42 has reported a series of Aptamer obtained by “Systematic Evolution of Ligands by Exponential enrichment” (SELEX)43 for detection of Cardia Troponin I (cTnI). cTnI is considered as a gold standard biomarker for screening of Acute
Myocardial Infraction (AMI)44. Based on report from Hunho Jo et. al., they have reported a “Tro4” 40 mer aptamer sequence that has a dissociation constant (Kd) of 270 pM for
cTnI monomer and has a Kd of 3.10 nM even for Troponin Complex. In the reported electrochemical sensing experiment, the aptamer can reach a limit of detection around 1 pM (S/N=3), which is lower than clinical cut-off value of 70-400 pg/mL. This suggest a high clinical value of such screening in case of AMI. Since AMI is an indication that suits the targeted application of the portable diagnostics, surface modification of Tro4 aptamer for cTnI detection is selected as a demonstration of bio-sensing efficacy in present work.
The sequence of the Tro4 is 5’-TTT TTT CGT GCA GTA CGC CAA CCT TTC TCA TGC GCT GCC CCT CTT A-3’ (46 mer). The synthesis of the sequence is carried out by
PurigoTM in Taiwan. The 5’ end of the Tro4 aptamer is modified with a thiol functional
group for reaction with SiSPR gold surface. The running buffer for the sensing process is
doi:10.6342/NTU201700804
104
1X TE buffer (10 mM Tris + 1 mM EDTA from Protech Technology Enterprise) with 1
M of NaCl. Thiolated Tro4 is firstly mixed with 20 mM of TCEP
(Tris(2-Carboxyethyl)phosphine hydrochloride) for reduction of disulfide bond. The purpose of
the reduction is to confer consistent surface modification efficiency and it typical stand
for more than one hour.
As shown in Fig. 53, before surface modification, a calibration step curve is firstly established by four reference solutions. The calibration step starts with running buffer as baseline, followed by 0.5 % glucose solution, 1 % glucose solution and 1.5 % glucose
Fig. 53 Tro4 Aptamer surface modification
A calibration is first carried out with four reference solutions, followed in influx of
A calibration is first carried out with four reference solutions, followed in influx of