1 Introduction

Millions of people all over the world suffer from cancer and chronic diseases. This represents a main cause for human death every year. Huge efforts were spent to innovate simple, accurate, and cost-effective methods to detect different diseases. The cooperation between engineering and medical researchers produced a great health care revolution (Pal et al. 2019).

Biosensors are considered the ideal solution for detecting many kinds of diseases. Biosensors have a lot of advantages as they introduce high sensitivity, fast response, less complexity, portability, and hence the ability for mass production (Rayappan and Lee 2021).

Optical biosensors are the main candidate in the biosensing field. Thanks to the fast, accurate, and direct detection of target molecules, optical biosensors are considered as a breakthrough in this area. There are many categories of optical biosensors, like evanescent wave fluorescence, interferometric, colorimetric, surface-enhanced Raman-scattering and surface plasmon resonance (SPR) sensors (Damborsky et al. 2016).

The SPR biosensors depend on monitoring of the refractive index (RI) variation due to the molecular interaction between pre-immobilized biomolecules on or near a metal–dielectric interface and a liquid sample in a real time form. These biosensors are currently used to measure different parameters, as sample concentration, kinetics of biomolecular interaction and binding constants (Shpacovitch and Hergenroder 2020). Hundreds of researches that describe the different biological applications of SPR sensors and recent novel trends of SPR-based platforms are published annually (Akib et al. 2021; Ly et al. 2021; Moznuzzamana et al. 2020).

In the field of cancer detection, SPR sensors can be used to study the behavior of human cells, and also to analyze the change of the protein in the cell to determine if this cell is normal or cancerous (Bellassai et al. 2019; Rajeswari and Revathi 2022). On the other hand, diabetes patients can beneficiate from SPR sensors, as these sensors can detect the glucose level but in a more convenient way with very high accuracy (Shen et al. 2007). Researchers developed many SPR biosensor configurations, each of which aimed to get better sensitivity and detection accuracy values than those of the previous ones.

In this paper, we propose a simple, highly-sensitive SPR biosensor with a wide detection range using a novel material (ZnSe). The ZnSe is a well-known semiconductor material in the opto-electronic field. It is used in the fabrication of laser diodes, LEDs, photodetectors, photocatalysis, solar cells, field emitters, vivo imaging and sensors (Zhang et al. 2016). The ZnSe semiconductor is used for two reasons. The first is to protect the Ag layer from oxidation, and the second is to enhance the sensor sensitivity. The combination between Ag and ZnSe materials was studied deeply in Chu et al. (2020). The excellent matching between these materials was proved experimentally through studying different parameters, especially the charge transfer (CT). Results obtained from Chu et al. (2020) motivated us to introduce this combination between Ag and ZnSe in the proposed SPR biosensor. Till now, no reported SPR biosensor contains ZnSe material in its structure, which gives our paper the novelty and simplicity for using such a semiconductor material instead of the common 2-D and transition metal dichalcogenide (TMDC) materials to improve the SPR sensor performance.

The rest of this paper is organized as follows. Section 2 illustrates the modeling and the design of the proposed sensor. Then, Sect. 3 displays the obtained results with their discussion. Section 4 is devoted to the main conclusions.

2 Modeling and design of the proposed SPR sensor

2.1 Theoretical modeling

The proposed sensor is based on Kretschmann configuration (Kretschmann and Reather 1968). In this configuration, a metallic layer is directly deposited on a prism surface and the sensing medium is put over the metallic layer. When the light beam strikes the metallic layer, through the prism with a certain angle, the resonance occurs and the plasmonic wave travels along the interface between the metallic layer and the sensing medium. This is simply the SPR. The resonance condition can be described by the wave forms satisfying the condition that the wave vector of the incident p-polarized light is equal to the wave vector of the surface plasmon wave produced along the interface between metal layer and the sensing medium.

This matching condition can be described mathematically by (Chen and Chen 1981).

$$K_{x} = K_{sp}$$
(1)

where Kx is the wave vector of incident light and Ksp is the surface plasmon wave (SPW) vector. They can be defined by (Chen and Chen 1981)

$${K}_{x}=\frac{2\uppi }{\uplambda } {n}_{p} \times \mathrm{sin}{\theta }_{0}$$
(2)
$${K}_{sp}=\frac{2\uppi }{\uplambda } \times \sqrt{\frac{{\varepsilon }_{1}\times {\varepsilon }_{2}}{{\varepsilon }_{1}+{\varepsilon }_{2}}}$$
(3)

where λ is the wavelength of the incident light, np is the RI of the prism, θ0 is the light incidence angle at the interface between prism and metal layer and \({\varepsilon }_{1},{\varepsilon }_{2}\) are the dielectric constants of the sensing medium and the metal layer, respectively.

Any change in the RI leads to a direct change in the SPW propagation constant. So, to get the equality condition in Eq. (1), the incidence angle must be shifted. At resonance, the energy of the incident light is transferred totally to the SPW, and at this time, there will be a minimum reflectivity. The SPR sensor is intended to detect the angle at which there will be  a minimum reflectivity (a resonance), and any change in this angle means a change in the RI.

The transfer matrix method (TMM) is one of the most effective tools used to describe the SPR sensors reflectivity. The TMM does not contain any approximation. So, it is considered a more accurate method (Katsidis and Siapkas 2002).

The calculations of the transmittance or the reflectance for the multilayer model supposes are performed under the assumption that the layers are thin homogeneous films stacked on each other in cascade. It is also supposed that all the layers are symmetric, non-magnetic and uniform (Yuan et al. 2011).

The reflectance of the incident p-polarized light can be defined as (Mostufa et al. 2021)

$$R_{p} = \left| {r_{p} } \right|^{2}$$
(4)

where \({r}_{p}\) is the amplitude of the reflected wave that can be calculated for the N-layer model from (Mostufa 2021)

$${r}_{p}= \frac{{(F}_{11}+{F}_{12}{q}_{N}){q}_{1 }- {(F}_{21}+{F}_{22}{q}_{N}) }{{(F}_{11}+{F}_{12}{q}_{N}){q}_{1 }+ {(F}_{21}+{F}_{22}{q}_{N})}$$
(5)

According to TMM, the wave propagation parameters for the N-layer model that describes the light propagation from layer (N) to layer (N + 1) can be calculated through the Fi j matrix (Mostufa 2021, Rahman 2018)

$${F}_{ij}= \left[{\prod }_{k=2}^{N-1}\left(\begin{array}{cc}\mathrm{cos}{\beta }_{k}& \frac{-i\mathrm{sin}{\beta }_{k}}{{q}_{k}}\\ -i{q}_{k}\mathrm{sin}{\beta }_{k}& \mathrm{cos}{\beta }_{k}\end{array}\right)\right]= \left[\begin{array}{cc}{F}_{11}& {F}_{12}\\ {F}_{21}& {F}_{22}\end{array}\right]$$
(6)

where \({\beta }_{k}\) and \({q}_{k}\) are the random phase constant and the admittance at the kth layer, respectively, and both can be calculated by (Fouad et al. 2017)

$${\beta }_{k}= \frac{2\pi }{\uplambda }{d}_{k} \sqrt{{n}_{k}^{2}-{(n}_{p}\mathrm{sin }{\theta }_{0}{)}^{2}}$$
(7)
$${q}_{k}=\frac{\sqrt{{n}_{k}^{2}-{(n}_{p}\mathrm{sin}{\theta }_{0}{)}^{2}}}{{n}_{k}^{2}}$$
(8)

where nk and dk are the RI and thickness of the kth layer, respectively, np is the RI of the prism and \({\theta }_{0}\) is the incidence angle at the prism-metal interface.

The sensor performance can be evaluated with several parameters like its angular sensitivity (S), detection accuracy (DA), and quality factor (QF). The angular sensitivity (S) is defined as the amount of the resonance angle shift (ΔθSPR) with respect to the change in the RI of the sensing layer (Δnsens), and it is expressed by (Hossain et al. 2020)

$$S= \frac{\Delta {\theta }_{SPR}}{{\Delta n}_{sens}} deg/RIU$$
(9)

The sensor DA calculation is based on the full width at half maximum (FWHM) of the reflectance curve. It indicates the ability of the sensor to detect the RI variation, precisely. It is calculated by (Hossain et al. 2020)

$$DA= \frac{1}{FWHM} {deg}^{-1}$$
(10)

The QF of the sensor depends on S and   FWHM. It can be expressed by (Hossain et al. 2020)

$$QF= \frac{S}{FWHM} {RIU}^{-1}$$
(11)

2.2 Structure of the proposed sensor

The proposed sensor structure is shown in Fig. 1. It contains BK7 as a coupling prism, a metallic layer of silver (Ag) and ZnSe as a sensitivity enhancement material, and at the same time, an oxidation protection film for the silver layer. An He–Ne laser source (λ = 633 nm) is used to produce TM-polarized incident light on the prism surface. It was suggested in (Hossain and Bin Abdul Akib 2019) that the utilization of this wavelength enhances the system sensitivity, as it provides a minimum Kerr effect. The sensor uses a low-refractive-index prism BK7. The RI of the prism is 1.51508 (Brahmachari and Ray 2013). Over the prism surface, the silver layer is deposited with an optimized thickness of (dAg). The RI of silver is 0.06656 + 4.0452i (Monzón-Hernández et al. 2018). The sensitivity enhancement layer ZnSe has a thickness of (dZnSe), which is optimized for each application, and an RI of 2.5780 (Marple 1964). This layer is grown on the silver layer. The top layer is the sensing medium, which contains the test sample.

Fig. 1
figure 1

Schematic diagram of the proposed sensor

Full size image

Biosensor performance is evaluated using Matlab software. Every part of the sensor structure is investigated deeply to determine the best combination of elements.

2.2.1 The prism

To clarify the effect of the prism type on the performance of the sensor, Brahmachari and Ray (2013) provided a deep study for the effect of the prism glass type on the sensor performance. Different types of prisms (BK7, BAK1, BAF10, SF5, SF10, SF11) with different RIs (1.51508, 1.5704, 1.66707, 1.66846, 1.72312, 1.77858, respectively) were presented. The results proved that for lower-RI glass (BK7), the SPR angle and the sensitivity are high while for higher-RI glass (SF11), the SPR angle and the sensitivity are low. Regarding the dynamic range, it was noted that for lower-RI glass (BK7), the dynamic range of the sensor is low while for higher-RI glass (SF11), the dynamic range is high at the operating wavelength of 633 nm.

As the detection range for the proposed sensor is not too wide and the sensitivity is a crucial parameter, a low-RI prism (BK7) will be preferred in our sensor structure.

2.2.2 The metal layer

Gold and silver are the common metals that are widely adopted in SPR configurations. Both metals are studied to get the suitable one for the proposed sensor applications. The sensor construction used for our study is composed of a BK7 prism, a metal layer (Au or Ag) with a thickness of 50 nm, a layer of ZnSe of 1 nm thickness, and a sensing medium with an RI of 1.33 to 1.40, which corresponds to the proposed sensor operating range. The RI of Au is 0.16195 + 3.20991 i (Monzón-Hernández et al. 2018). The obtained results for both configurations are presented in Fig. 2. From Fig. 2a, it is clear that the Au layer is not suitable for the proposed configuration while the Ag layer, as shown in Fig. 2b, provides excellent results with sharp resonance dip which means more DA. As silver suffers from the disadvantage of chemical interaction with the buffer solution, the existence of the ZnSe layer above it provides the required protection as well as the required sensitivity enhancement.

Fig. 2
figure 2

Performance of the proposed sensor at different metal layers within an RI range of 1.33 to 1.40. a Au, b Ag

Full size image

To achieve the best results, the sensor performance is studied at different Ag thickness, with a ZnSe layer of thickness 1 nm, to get the optimum thickness value. Results are shown in Fig. 3. From Fig. 3, 50 nm and 60 nm thickness achieve almost the same reflectance curve. Therefore, the optimum thickness value is 50 nm.

Fig. 3
figure 3

Proposed sensor performance at different Ag thickness with a ZnSe thickness of 1 nm and a sensing medium of water

Full size image

2.2.3 Sensitivity enhancement layer

To prove the feasibility of using the ZnSe layer, the sensor performance is monitored without a ZnSe layer, Fig. 4, and after adding a ZnSe layer, Fig. 5, over an Ag layer thickness of 50 nm.

Fig. 4
figure 4

Proposed sensor performance without a ZnSe layer at different sample refractive indices

Full size image
Fig. 5
figure 5

Proposed sensor performance with a ZnSe layer at different sample refractive indices

Full size image

Figure 5 shows that the sensor has an S of 225\(^\circ\)/RIU within an RI range from 1.36 to 1.39. On the other hand, S is 168\(^\circ\)/RIU without the ZnSe layer, Fig. 4, at the same RI range, which means that the sensitivity is enhanced by 34% after adding the ZnSe layer.

3 Results and discussion

The proposed biosensor is tested for the detection of different cancerous cells in the human blood, and it is also tested for the detection of the glucose concentration in the human urine.

3.1 Detection of cancerous cells in human blood

The proposed sensor configuration for cancerous cell detection is composed of a BK7 prism, an Ag layer of thickness 50 nm, and a ZnSe layer. The sensing medium comprises different blood samples with RI depending on the nature of the sample. The thickness of the ZnSe layer is optimized to detect samples within an RI range from 1.36 to 1.4. Figure 6 shows the results for the proposed sensor at different ZnSe layer thickness. From Fig. 6, it is clear that the optimum thickness that provides the best sensitivity is 3 nm. The sensor performance is evaluated by the measurement of S, DA, and QF for each case using Eqs. (9–11).

Fig. 6
figure 6

Performance of the proposed sensor at different ZnSe layer thickness to detect cancerous cells within an RI range of 1.36 to 1.40, a 1 nm, b 2 nm, c 3 nm, d 4 nm

Full size image

The breast cancer is considered as one of the most spreading types of cancer among women all over the world. This type of cancer is the second reason of cancer death between women, as lung cancer is the main cause for women mortality. Early detection of that cancer slows down the annual death rate from 2.6% to less than 1% per year (Key Statistics 2022).

MDA-MB-231 is one of the cells that are responsible for creation of breast cancer. The proposed sensor is intended to detect this kind of cancer cells. The RI of this cancer cell is 1.399 and that of the normal cells of breast is 1.385 (Hossain et al. 2020; Sani and Khosroabadi 2020). To detect such cells, a sample of a normal case (reference sample) and another malignant one are used. Figure 7 shows the resonance angle shift due to the existence of the cancerous cells. The detection sensitivity is excellent, as it is 359\(^\circ\)/RIU, where FWHM is 4.4\(^\circ\), DA is 0.22\(^\circ\)−1, and the QF is 81.59 RIU−1.

Fig. 7
figure 7

Reflectance curves of the proposed sensor for normal MDA-MB-231 cells and cancerous cells

Full size image

Basal Cancer Carcinoma (BCC) is considered as the basic type of skin cancer, and it is one of the most repeated forms of all cancers. About 3.6 million cases in U.S. alone are diagnosed with BCC every year. This kind of cancer grows slowly, and it is caused from abnormal, uncontrolled growth of basal cells (Basal Cell 2022). The RI of the basal cancer cells is 1.38, while that of the normal skin cells is 1.36 (Sani and Khosroabadi 2020). To examine the existence of such cancerous cells, two samples are tested with the proposed sensor. One sample is for a normal case (reference sample) and the other is for a malignant case. Figure 8 displays the resonance angle shift due to the presence of the cancerous cells. It is clear that the sensor could detect the presence of the basal cells with a sensitivity of 204\(^\circ\)/RIU while FWHM of 3.12\(^\circ\), DA is 0.32\(^\circ\)−1, and QF is 65.38 RIU−1.

Fig. 8
figure 8

Reflectance curves of the proposed sensor at normal Basal cells and cancerous cells

Full size image

Another kind of cancer cells that attacks women is the cervical cancer. HeLa cells are considered as the main candidate in this cancer. This type of cancer cells is associated with the presence of another harmful cells (Nambisan 2017). The RI of the normal cells is 1.368 and that of the cancer cells is 1.392 (Sani and Khosroabadi 2020). Again, two samples are tested with the proposed sensor. Figure 9 shows the resonance angle shift, when the cancerous HeLa cells are present. The sensor detects the cancerous cells perfectly with a sensitivity of 247\(^\circ\)/RIU, while FWHM is 3.84\(^\circ\), DA is 0.26\(^\circ\)−1, and QF is 64.32 RIU−1.

Fig. 9
figure 9

Reflectance curves of the proposed sensor at normal HeLa cells and cancerous cells

Full size image

From the previous results, it is obvious that for each case, the proposed sensor detects the presence of the cancerous cells with excellent sensitivity, sharp detection accuracy, and very-high quality factor.

3.2 Detection of glucose concentration in human urine

Patients who suffer from diabetes are always obligated to check the glucose level in the blood regularly. This process causes a lot of pain to the patients. Diabetes also affects the patient urine as well as the blood. So, monitoring of the variation of the glucose concentration in urine will be more convenient for the patient. When the glucose level changes in urine, the refractive index of urine also changes, which makes it appropriate to be detected by the SPR biosensors.

The proposed sensor configuration for glucose concentration detection is composed of a BK7 prism, an Ag layer of thickness 50 nm and a ZnSe layer. The sensing medium is composed of different urine samples with different RIs. The thickness of the ZnSe layer is optimized to detect samples within the RI range from 1.33 to 1.35. Figure 10 shows the results of the proposed sensor at different ZnSe layer thickness. From Fig. 10, it is clear that the optimum thickness that provides the best sensitivity is 9 nm. Again, the sensor performance is evaluated by the measurement of S, DA, and QF for each case using Eqs. (9–11).

Fig. 10
figure 10

Performance of the proposed sensor at different ZnSe layer thickness to detect glucose level within RI range of 1.33 to 1.35 a 7 nm, b 8 nm, c 9 nm, d 10 nm

Full size image

Normal people have a glucose concentration in urine of 0–15 mg/dl, which corresponds to an RI of 1.335 (Mostufa et al. 2021). Any increment in the glucose concentration leads to an increment in the urine RI.

Two urine samples are tested with the proposed biosensor, one is a normal human urine (0 ~ 15 mg/dl) with an RI of 1.335, which is a reference value, and the other with a higher glucose level (0.625 gm/dl) with an RI of 1.336. Figure 11a shows that the sensor detects a resonance angle shift due to the small increment in the urine RI. The sensor sensitivity for 0.001 increment in RI is 280\(^\circ\)/RIU, where FWHM is 4.64\(^\circ\), DA is 0.21\(^\circ\)−1, and QF is 60.34 RIU−1. Similarly, when the concentration of glucose is changed from 1.25 gm/dl, a reference value, to 2.5 gm/dl, which corresponds to a change in the urine RI from 1.337 to 1.338, the sensor detects a right shift for the resonance angle. The sensor showed a sensitivity of 280\(^\circ\)/RIU, Fig. 11b, where FWHM is 4.64\(^\circ\), DA is 0.21\(^\circ\)−1, and QF is 60.34 RIU−1. When the glucose concentration is increased significantly from 5 gm/dl, which represents a reference value, to 10 gm/dl, the urine RI increases also with an increment of 0.006 (from 1.341 to 1.347), which in consequence shifts the curve apparently. Figure 11c shows that the sensor sensitivity is increased to 366\(^\circ\)/RIU, where FWHM is 5.52\(^\circ\), DA is 0.18\(^\circ\)−1, and QF is 66.41 RIU−1. From the previous results, the sensor shows a high capability to detect a very low RI change as well as a large increment with very high sensitivity. Sensitivity achieved with the proposed sensor is superior to those obtained in Mostufa et al. (2021) and Mudgal et al. (2020). The obtained results for all cases of different glucose concentrations are summarized in Table 1.

Fig. 11
figure 11

Detection of glucose concentration in human urine in normal and patient cases at different categories. a 0–15 mg/dl ~ 0.625 gm/dl, b 1.25 gm/dl—2.5 gm/dl, c 5–10 gm/dl

Full size image
Table 1 Glucose level in human urine for normal and patient cases
Full size table

Table 2 provides a comparison between the sensitivity of the proposed sensor and several other biosensors. The superiority of the proposed sensor is clear.

Table 2 Comparison between the sensitivity of the proposed sensor and other different biosensors
Full size table

4 Conclusion

In this paper, we proposed an SPR sensor that can be used to detect different types of cancer cells and also urine glucose concentration with excellent sensitivity and quality factor values. For cancer cell detection, the sensor provides a sensitivity of 359\(^\circ\)/RIU for detection of MDA-MB-231 cells. For basal cell detection, the sensor achieves a sensitivity of 204\(^\circ\)/RIU, while for  the detection of HeLa cells, the sensitivity is 247\(^\circ\)/RIU. The detection of glucose levels in urine is achieved with a sensitivity of 366.6\(^\circ\)/RIU. The novel use of ZnSe semiconductor material shows an excellent performance in enhancing the sensor sensitivity by 34% as compared with  a bare sensor. The sensor construction is simple with very high sensitivity, which makes it a practical choice to be implemented in the near future.