1. Introduction

Monitoring surface reactions and changes in near field is extremely important for many applications in physics, chemistry, biomedicine fields, etc. Considering the electromagnetic wave penetration, two depth scales are discussed here for the definition of near field: the penetration depths of evanescent fields in total internal reflection (TIR) and surface plasmon resonance (SPR).

Fluorescence microscopy based on TIR has demonstrated to be an effective technique for real-time study of cell-substrate in near field [1,2]. However, the fluorescence labeling may destroy the specimen. In recent years, as a well-developed imaging technique, digital holographic microscopy (DHM) can simultaneously provide quantitative amplitude and phase information of specimens with dynamic, nondestructive as well as full-field observation capabilities [3–7]. Thus, DHM based on TIR was proposed for quantitative phase imaging of cell-substrate interfaces, adhesions, and tissue structures at the near field of prism surface [8, 9]. And this technique was further developed to dynamically measure two-dimensional (2D) refractive index (RI) distributions of dielectric specimens [10, 11]. On the other hand, the technique based on SPR shows higher sensitivity in measuring near-field tiny variations compared with the characterization technique based on TIR [12–15].

Surface plasmon microscopy (SPM) [16], which uses the evanescent field in SPR as the illumination light, has been widely applied for reading biomolecular binding events of DNA or protein microarrays in the near field [17]. Traditionally, the SPM systems are based on amplitude-contrast SPR imaging [18]. While, the phase-contrast imaging technique has shown greater application potentials because of the higher detection sensitivity [19]. Furthermore, surface plasmon holographic microscopy (SPHM), which combines SPM with DHM, can simultaneously obtain the amplitude- and phase-contrast SPR images [20–22]. Consequently, this technique has been applied for quantitatively monitoring tiny variation of dielectric RI, imaging biological tissue [23] and mapping thin film thickness in near field with high sensitivity and temporal stability [24]. However, the current SPHM with single wavelength is limited to a narrow detectable range of RI variation, which significantly hinders its applications in sensing and imaging.

In this paper, an improved SPHM with the wavelength multiplexing technique is proposed. Particularly, we introduce two laser sources in the prism-coupling SPHM and design a common-path DHM configuration using a single beam splitter to simultaneously record the two off-axis holograms of different wavelengths. Then, through establishing the models for the extension of measurement range in monitoring RI variation and the compensation of detection sensitivities in 2D SPR imaging, performing the experiment with standard specimen and properly processing the experimental data, the measurement range in quantitatively monitoring the tiny variation of dielectric RI can be extended without decreasing the high sensitivity. Furthermore, imaging onion tissues at two wavelengths is performed to compensate the detection sensitivities in both amplitude- and phase-contrast SPR imaging.

2. Experimental setup

In order to simultaneously record the SPR images at two wavelengths, we design an experimental setup for the wavelength-multiplexing SPHM, as shown in Fig. 1. The light beams with wavelengths of 632.8 nm and 660 nm are coupled into a fiber coupler C, expanded and collimated by the beam expander BE and lens L, respectively. Then, the coupled beam is adjusted as vertical polarization in free space by a polarizer P so that it is p-polarized when reflected by the prism of the SPR configuration, which is made up with a prism, a gold film and one (in three-layer SPR model) or several layers (in multi-layer SPR model) of media [25]. The prism is accurately adjusted to ensure that the incident angle of the beam satisfies the SPR condition. After that, the beam is magnified by a long working distance microscope objective LWDMO (Mitutoyo M Plan Apo 5 × ) and goes through the imaging lens IL.

 

Fig. 1 Experimental setup of wavelength-multiplexing SPHM. C: fiber coupler; BE: beam expander; L: collimating lens; P: polarizer; LWDMO: long working distance microscope objective; IL: imaging lens; BS: beam splitter; F_{1, 2}: optical filters; insert on the left shows the three-layer SPR configuration; insert on the right shows the side view of BS.

Download Full Size | PDF

The dielectric specimen acting as the last medium layer of SPR configuration is put on the left side of gold film, resulting in that only top side of the beam exiting from the imaging lens carries the specimen information and acts as the object beam O, while the bottom one acts as the reference beam R. Then, the object and reference beams are separated by the beam splitter BS under the condition that its half reflecting layer is inclined relative to the optical axis with a small angle, as shown in the right insert of Fig. 1. The reflected object beam and transmitted reference beam from the BS are filtered by optical filter F₁ with the central wavelength of 660 nm and bandwidth of 10 nm (FBH660-10, Thorlabs Inc., USA). Meanwhile, the transmitted object beam and reflected reference beam from the BS are filtered by optical filter F₂ with the central wavelength of 633 nm and bandwidth of 5 nm (FLH633-5, Thorlabs Inc., USA). Finally, two sets of interference fringes h₁ and h₂ at two wavelengths are simultaneously recorded on the CCD camera with a large target (Imaging Source SVS16000MFGE, 4892_H × 3280_V pixels, pixel size 7.44 μm × 7.44 μm).

The designed experimental setup of wavelength-multiplexing SPHM presents simple structure by easily introducing a few optical components. And it shows high temporal stability against vibration because the object beam propagates along with the reference one almost in the same way to realize common-path hologram recording. In addition, the LWDMO induces a spherical phase curvature to the object wave in traditional DHM system based on Mach-Zehnder interferometer. But in the present experimental setup, the reference and object beams pass through the same LWDMO, resulting in that the spherical phase curvature can be physically compensated. Furthermore, since the compact experimental setup leads to a very small optical path difference between the object and reference beams, lower-coherence light sources can be applied to obtain the SPR images with less coherent noise.

3. Principles

3.1 Amplitude- and phase-contrast SPR imaging using SPHM

A p-polarized light beam with wavelength λ is incident on the SPR prism with angle θ which is greater than the critical angle of TIR, resulting in that an evanescent wave extends into the gold film. The wave vector of the evanescent wave will match that of the surface plasmon wave (SPW) at a specific incident angle and thus SPR occurs. This angle is called SPR angle. The sequence of the optical media in the SPR configuration from the prism to the last medium layer is denoted as 1, 2, …and N. According to the Fresnel’s formula [26], the complex reflection coefficient of the reflected wave r_{1, N} is given by

The reflected wave O(x, y) which carries the specimen information on the gold film can be recorded by use of DHM and then numerically reconstructed from the hologram with angular spectrum method [27–29]. Then, the corresponding intensity and phase distributions I_O(x, y) = |O(x, y)|², ϕ_O(x, y) = arctan{Im[O(x, y)]/Re[O(x, y)]} can be obtained to realize simultaneous amplitude- and phase-contrast SPR imaging. According to the principle of double-exposure DHM [30–32], phase difference distribution of the reflected waves at time t (with specimen) and the initial time t = 0 s (without specimen) is Δϕ_Ot(x, y) = ϕ_Ot(x, y)-ϕ_O0(x, y), where, ϕ_Ot(x, y) and ϕ_O0(x, y) are the phase distributions of the reflected waves at time t and the initial time, respectively.

3.2 Extension of measurement range in monitoring RI variation

In three-layer SPR configuration as shown in the left insert of Fig. 1, the thickness of the dielectric layer on the gold film is much larger than the penetration depth of the SPW. According to Eqs. (1)-(4), if the other parameters are determined, the reflectivity R and reflection phase shift φ are uniquely decided by the dielectric RI n₃. When the value of n₃ satisfies the SPR condition, R goes down to the minimum and φ shows a very steep slope, as shown in Fig. 2(a). Tiny variation of n₃ introduces a great change of the reflection phase shift Δφ, which eventually results in phase change of the reflected wave Δϕ_O. That is, the reflection phase shift difference Δφ(x, y) is equal to the phase difference of the reflected waves Δϕ_Ot(x, y) before and after SPR occurring, which is measured using double-exposure DHM [23]. Based on this principle, tiny variations of dielectric RI in near field can be quantitatively monitored with high sensitivity. However, the RI measurement range of single wavelength is limited, resulting in that the dynamic process with a larger variation range of RI cannot be monitored with high sensitivity all the time.

 

Fig. 2 (a) Reflectivity R and reflection phase shift φ versus dielectric RI n₃ at wavelength of 632.8 nm. (b) Reflection phase shift difference Δφ versus dielectric RI n₃ at two wavelengths. The parameters for plotting the theoretical relation curves are listed in Table 1.

Download Full Size | PDF

To solve this problem, we employ the proposed wavelength-multiplexing SPHM to extend the measurement range in quantitatively monitoring tiny variations of dielectric RI. Figure 2(b) shows the plots of reflection phase shift difference Δφ and the dielectric RI n₃ at wavelengths of 632.8 nm and 660 nm. Each measurement range of RI Δn_3-1 (from the labeled point “D*” to “C*”) and Δn_3-2 (from the labeled point “B*” to “A*”) is 1.3245-1.3434 and 1.3426-1.3535, respectively. If the reflection phase shift differences Δφ at the two wavelengths can be simultaneously measured, the RI variation will be jointed together seamlessly by extracting the measured Δφ located in the measurement range at each wavelength. Consequently, the measurement range of monitoring RI variation Δn₃ will be almost doubled relative to the technique with single wavelength.

Note here that for the dynamic experiment of monitoring RI variation mentioned above, it’s impossible to detect the Δφ at each wavelength sequentially. Consequently, simultaneously imaging at two wavelengths using the proposed wavelength-multiplexing SPHM can effectively solve this problem. In addition, other two wavelengths can also be chosen to extend the measurement range of dielectric RI as long as the incident angle is changed. In the present paper, the wavelength 632.8 nm is firstly chosen because it is commonly used. Then, according to the theoretical calculation for the extension of measurement range, the wavelength 660 nm is passively chosen to work together with the former one.

3.3 RI retrieval algorithm

The RI value n₃ of the dielectric specimen at every time point during a dynamic experimental process can be calculated from the measured reflection phase shift differences Δφ at two wavelengths. However, the relationships between n₃ and Δφ in the measurement ranges of Fig. 2(b) cannot be described by certain equations. In general, a fitting equation can be accurately built from the theoretical relation curve on the condition that the line type of the curve is a priori. Previously, experimental experience formulas were employed to calculate the RI values [34]. In addition, a polynomial fitting equation based on the theoretical curve was built for the calculation [23].

Here, we propose a relatively simple and direct algorithm to retrieve the RI value. As seen from the measurement ranges in Fig. 2(b), each n₃ corresponds to a certain theoretical reflection phase shift difference Δφ. By calculating the deviations between the experimental value and each theoretical value of Δφ and then searching for the minimum deviation, the corresponding theoretical value of n₃ can be reasonably regarded as the RI value to be measured. The detailed retrieval flowchart is described in Fig. 3. The smaller of the interval between every two adjacent theoretical RI values is divided, the more precise the measured RI will be.

 

Fig. 3 Calculation flowchart to retrieve the RI value. Δφ__Exp, Δφ__Theo: measured and theoretical Δφ; Diff: deviation between Δφ__Exp and Δφ__Theo; Diff__min: minimum deviation; N: number of the theoretical values; RI__Theo, RI__Exp: the theoretical RI and RI to be retrieved.

Download Full Size | PDF

3.4 Compensation of detection sensitivities in SPR imaging

Besides monitoring tiny variations of dielectric RI in near field, SPHM shows great potentials in 2D imaging of biological tissues thanks to the full-field observation capability of DHM. According to the relation curves in Fig. 2(a), both the reflectivity and reflection phase shift present high responses with the variation of dielectric RI, thus, the specimen with a low-contrast RI distribution can be imaged with high contrast by reconstructing both the amplitude- and phase-contrast SPR images using SPHM. However, the detection sensitivities of single wavelength are higher in only a limited RI range, resulting in that practical specimens probably cannot be imaged with high contrast.

Here, we propose to compensate the detection sensitivities in SPR imaging using two wavelengths. For imaging biological tissues in the following experiments, the incident angle is set as the SPR angle when the dielectric is air [21]. Figure 4 shows the corresponding sensitivities of the reflectivity R and reflection phase shift φ with the dielectric RI n₃ (represented by the derivatives of R and φ over n₃) at wavelengths of 632.8 nm and 660 nm. It is obvious that the detection sensitivities of two wavelengths are higher in different but adjacent RI ranges. If the reflected waves in SPR excited by the two wavelengths can be simultaneously measured using the proposed wavelength-multiplexing SPHM, the detection sensitivities can compensate for each other in SPR imaging of the dielectric specimen with a larger RI range, which will effectively broaden its application range.

 

Fig. 4 Sensitivities of reflectivity R and reflection phase shift φ with dielectric RI n₃. Incident angles are 43.4513°@632.8 nm and 43.4504°@660 nm for the following 2D imaging experiment.

Download Full Size | PDF

4. Experiment for monitoring tiny variation of dielectric RI

4.1 Experiment results of the reflection phase shift difference

The RI variation during the volatilization process of alcohol-water mixture with initial volume ratio of 1:1 is quantitatively monitored to verify the effectivity of the wavelength-multiplexing SPHM. In the experiment, the p-polarized light beam is incident on the hypotenuse of the rectangular prism in the SPR configuration. When the incident angle is set by keeping the output and incident beams coaxial, the angle will be 72.8207°@632.8 nm according to the geometrical optics. Note here that the angle is 72.8387°@660 nm because of the dispersion effect of the prism. After the incident angle is calibrated and the experimental configuration is installed with other optical components sequentially, a reference hologram is recorded without the specimen at the initial time. Then 500 holograms are recorded successively at the frame frequency of 1 fps immediately after the mixture being adhered on left side of the gold film. At last, we numerically reconstruct the object waves and acquire their 2D phase difference distributions during the mixture volatilization process. The results of 0 s, 170 s and 500 s at wavelengths of 660 nm and 632.8 nm are listed in Figs. 5(a1)-5(c1) and Figs. 5(a2)-5(c2), respectively. They obviously display the variations of reflection phase shift difference at two wavelengths in the dynamic experiment.

 

Fig. 5 Reconstructed 2D phase difference distributions of 0 s, 170 s and 500 s at wavelengths of (a1-c1) 660 nm and (a2-c2) 632.8 nm.

Download Full Size | PDF

Since the mixture is homogeneous during the volatilization process, the reflection phase shift difference can be obtained by calculating its average in the mixture area. Figures 6(a) and 6(b) show the variations with time at the wavelengths of 660 nm and 632.8 nm, respectively. The black dots represent the measured values and the violet lines are the smooth processing results.

 

Fig. 6 Variations of measured Δφ with time at the wavelengths of (a) 660 nm and (b) 632.8 nm.

Download Full Size | PDF

It is obvious that the measured reflection phase shift difference Δφ at the wavelength of 660 nm increases rapidly in the beginning and reaches the maximum at 170 s. Meanwhile, the Δφ at the wavelength of 632.8 nm decreases slowly and reaches the minimum at 139 s. In this stage of volatilization, the Δφ variation from the labeled point “A” to “B” in Fig. 6(a) just locates in the measurement range of RI@660 nm Δn_3-2 (labeled in Fig. 2(b)). As time goes on, the measured reflection phase shift difference Δφ@660 nm keeps almost constant while the Δφ@632.8 nm increases rapidly and reaches the maximum at last. In this stage of volatilization, the Δφ variation from the labeled point “C” to “D” in Fig. 6(b) just locates in the measurement range of RI@632.8 nm Δn_3-1 (labeled in Fig. 2(b)).

4.2 In situ determination for the thickness of gold film

To quantitatively retrieve the RI values from the measured reflection phase shift differences, the thickness of gold film d₂ in the experiment must be determined accurately. Here, we propose two approaches to determine this parameter. The first one is to measure the reflection phase shift difference Δφ when the dielectric specimens are water (n₃ = 1.3317@632.8 nm) and air (n₃ = 1.0003@632.8 nm) on the gold film, respectively. In this situation, Δφ is different when the thickness of gold film d₂ changes, as shown in Fig. 7(a). This approach provides an effective solution to determine the thickness of gold film and also test its flatness thanks to the 2D imaging ability. Because this parameter in the experiment needs to be determined in situ, we can assume that the final state of the mixture is pure water and take the final value of the measured Δφ to calculate d₂. However, this assumption may lack of verification and lead to measurement errors.

 

Fig. 7 (a) Reflection phase shift difference Δφ@632.8 nm versus thickness of gold film d₂ when the dielectric is water and air, respectively. (b) Minimum of the experimental Δφ@632.8 nm versus thickness of gold film d₂. (c) Sensitivities of Δφ and Δφ_{_min} with d₂ in (a) and (b).

Download Full Size | PDF

In the above experiment for monitoring the volatilization process of alcohol-water mixture, the reflection phase shift difference Δφ@632.8 nm reaches the minimum and then increases, as shown in Fig. 6(b). Actually, this minimum changes only with the thickness of gold film d₂ on the condition that the other parameters are determined, as shown in Fig. 7(b). By interrogating the minimum of the measured reflection phase shift difference Δφ_{_min}@632.8 nm, the thickness of gold film can be determined accurately. There is not any assumption in this approach and it is a complete in situ solution for determining this parameter. The sensitivities (represented by the slopes of the curves in Figs. 7(a) and 7(b)) of these two approaches are in the same order, as shown in Fig. 7(c). Thus, the second approach is employed in this work and the calculated d₂ = 58.9 nm in the present experiment.

4.3 Experiment results for measurement range extension

After calibrating and determining the experimental parameters accurately, the RI values of the mixture can be retrieved precisely. According to the theoretical analysis in section 3.2, the RI measurement range is expected to be extended by jointing the retrieved refractive indices (RIs) at the critical points of the two relation curves, labeled by “B*” and “C*” in Fig. 2(b). Thus, an obvious critical point should appear in each experimental curve of the Δφ variation with time in theory. However, the experimentally measured maximum of the reflection phase shift difference Δφ@660 nm in Fig. 6(a) is smaller than the theoretical value at the critical point “B*” and the measured Δφ afterwards decreases slightly and slowly. The reason lies in that the sensitivities of Δφ with n₃ near the critical points are greatly lower than those in the central range of this relation curve.

The violet curves in Fig. 8 describe the measurement sensitivities in the theoretical ranges and they are obviously getting lower and lower near the critical points. Thus, in order to monitor the tiny variation of dielectric RI with high sensitivity, we should select the central measurement range at each wavelength, which are denoted by the rectangular boxes in Fig. 8. For the above experiment, we select the measured reflection phase shift differences from 0 s to 128 s@660 nm and 188 s to 500 s@632.8 nm for calculating the RI values.

 

Fig. 8 Sensitivities of Δφ with n₃ at the wavelengths of (a) 660 nm and (b) 632.8 nm.

Download Full Size | PDF

Note here that the RI values of the dielectric specimen are different at each wavelength because of the dispersion effect. Here, we “transfer” the RI values at the wavelength of 632.8 nm to those of 660 nm. As displayed in Table 1, the RI values of pure water and alcohol at the wavelength of 632.8 nm are 0.0007 and 0.0006 larger than those at the wavelength of 660 nm, respectively. Besides, the initial volume ratio of the measured alcohol-water mixture is 1:1 and the content of alcohol is getting lower and lower, we can reasonably assume that the dispersion characteristic of the mixture matches that of pure water in the second stage of volatilization. Thus, the equivalent RI values@660 nm are obtained by subtracting 0.0007 from those@632.8 nm. Consequently, the mixture RI@660 nm has decreased 0.0195 RIU during the volatilization process, as shown in Fig. 9. The RI measurement range in the present experiment has achieved more than twice than that employing the SPHM with single wavelength [23].

Table 1. Parameters for the theoretical calculations [33]

View Table

 

Fig. 9 Retrieved tiny variation of mixture RI@660 nm during the volatilization process.

Download Full Size | PDF

The above experiment is performed under the condition of theoretically extending the measurement range by jointing the critical points of the relation curves. While, the experimental ranges are jointed with the central measurement ranges at each wavelength for RI retrieval in practice. Consequently, the RI variation near the critical points of the relation curves, i.e., the RI variation from 128 s to 188 s, has not been effectively monitored during the volatilization process. As long as the laser sources with two closer wavelengths are employed, monitoring the tiny variation of dielectric RI for an entire process in a large measurement range can be realized using the proposed experimental system and data processing approach.

5. Experiment for imaging biological tissues

We perform the 2D SPR imaging for onion tissues using the wavelength-multiplexing SPHM. Figures 10(a1) and 10(a2) are the recorded holograms containing two sub-holograms at two wavelengths, and the selected regions are magnified in the big green boxes. In Fig. 10(a1), the onion epidermis cell structures are clearly imaged at the wavelengths of 660 nm (left) and 632.8 nm (right) because of the SPR effect. Central parts of the phase-contrast SPR images at the two wavelengths are numerically reconstructed, as presented in Figs. 10(b1) and 10(c1), respectively. It is obvious that the phase contrast at the wavelength of 632.8 nm is greatly higher than that of 660 nm. Actually, the detection sensitivity of phase-contrast imaging@660 nm in this situation is relatively lower, and it can be compensated by that of the phase-contrast imaging@632.8 nm.

 

Fig. 10 Experiment results for imaging onion tissues. (a) Recorded holograms; reconstructed phase-contrast SPR images at the wavelengths of (b) 660 nm and (c) 632.8nm, respectively.

Download Full Size | PDF

Then, we change the incident angle of the experimental system, in which the SPR condition is exactly satisfied at the wavelength of 660 nm. The phase-contrast SPR images at two wavelengths can be effectively reconstructed in Figs. 10(b2) and 10(c2), though the contrast is certainly a little higher at the wavelength of 660 nm. However, as shown in the recorded hologram of Fig. 10(a2), the amplitude-contrast SPR image@660 nm is clearly shown while the cell structures cannot be imaged completely at the wavelength of 632.8 nm. As a matter of fact, the detection sensitivity of amplitude-contrast imaging@632.8 nm in this experiment is relatively lower, and it can be compensated by that of the amplitude-contrast imaging@660 nm.

6. Discussions

We have employed the proposed wavelength-multiplexing SPHM for monitoring tiny variation of dielectric RI with an extended measurement range and imaging biological tissues with compensated detection sensitivities based on the three-layer SPR model. In addition, the wavelength multiplexing technique can also be applied in the multi-layer SPR model to acquire the multiplexing information and then broaden the application range. Moreover, the proposed multiplexing technique can be introduced in the objective-coupling SPHM to obtain highly magnified and distortion-free SPR images at two wavelengths.

7. Conclusions

In conclusion, an improved SPHM with the wavelength multiplexing technique has been proposed in this paper. To simultaneously obtain the SPR images at two wavelengths, we have designed an upgraded SPHM setup with a common-path hologram recording configuration and two laser sources. The designed experimental system presents simple structure, high temporal stability and inherent capability of phase curvature compensation. After establishing the models for the extension of measurement range in monitoring RI variation and the compensation of detection sensitivities in 2D SPR imaging, and improving the approach for accurately determining the experimental parameters, we have performed the experiment of quantitatively monitoring the tiny variation of dielectric RI in near field with an extended measurement range while maintaining the high sensitivity. Also, imaging onion tissues has been performed to demonstrate that the detection sensitivities of two wavelengths can compensate for each other in both amplitude- and phase-contrast SPR imaging. The improved SPHM shows great potentials for further applications in monitoring diverse dynamic processes related with RI variations as well as imaging biological tissues with low-contrast RI distributions in the near field.