Data analysis for optical sensors based on spectroscopy of surface plasmons

Abstract

We have examined the noise sources for a surface plasmon resonance (SPR) sensor system to facilitate optimization of SPR sensor instrumentation and data-processing methods for high-resolution SPR sensing. We found detector shot noise to be the dominant source of noise in the SPR sensor, and investigated the propagation of noise through a commonly used SPR sensor data-processing algorithm. Deriving an expression relating the detector noise level to the noise of the output data, we used measurements of the magnitude and distribution of both charge-coupled device and photodiode array detector noise to predict the noise levels observed in the sensor output. We used this method to optimize the parameters of the algorithm and introduced a modified algorithm with enhanced resistance to correlated light level noise. In addition, we evaluated the noise performance of alternative SPR sensor designs, including a sensor with no polarization optics and another with optics producing an inverted SPR absorption feature. Experimental SPR sensor data demonstrated good agreement with predicted noise levels.

Phase‐sensitive surface plasmon resonance biosensors: methodology, instrumentation and applications

“…In angular or spectral interrogation methods, the determination of SPR dip position involves a series of intensity data from different incident angles or wavelengths, and the abundance of data has helped to suppress the intensity noise and greatly improve the signal to noise ratio. Thus, the refractive index resolution in both modulation methods has been improved down to 10 −6 level [36,31]. Various averaging and mathematical treatment methods can be further used to lower the detection limit down to 10 −7 RIU level [31,30].…”

Projection method for improving signal to noise ratio of localized surface plasmon resonance biosensors

“…Now we apply all the methods to the same simulated noisy data in order to estimate the RI change, the reference set is used to obtain a calibration curve for each method, and the slope of each curve provides the sensitivity factor used in determining the RI change (known in this case to test the accuracy of each method). The dynamic-baseline centroid method [9] uses a threshold value (usually the FWHM) to estimate the resonance wavelength (λ r ) as follows…”

“…The dip-finding method can be used to process the sensor data where the dip of the resonance curve is tracked simultaneously and a sensorgram can be generated for binding events in real time [7]. Several data processing methods have targeted SNR improvement of propagating SPR sensors: the centroid and full-width-at-half-maximum tracking method [8,9], principle component and locally weighted parametric regression [10], optimal linear data analysis [11], polynomial curve-fitting of the measured curve [12], statistical hypothesis testing [13], and a double projection method [14]. The integrated response method was applied to nanohole-structures [15], which is based on the extraordinary transmission of light arising from the propagating surface plasmon resonance [16][17][18][19].…”

Optical integrated chips with micro and nanostructures for refractive index and SERS-based optical label-free sensing

“…Such a system can resolve refractive index changes down to 5 × 10 −7 RIU. For improved wavelength-interrogated SPR sensors, Homola's group designed a platform with wavelength modulation and parallel channel architecture, where the polychromatic light reflected from different channels can be collected by different output collimators [53]. More recently, a detection limit of 2.5 × 10 −8 RIU has been attained by exciting a long range SPR and combing superluminescent diode with polarization-maintaining fibers [54].…”