*Article* **Electrochemical Immunosensor Using Electroactive Carbon Nanohorns for Signal Amplification for the Rapid Detection of Carcinoembryonic Antigen**

**Angélica Domínguez-Aragón 1,2, Erasto Armando Zaragoza-Contreras 2,\*, Gabriela Figueroa-Miranda 1, Andreas Offenhäusser <sup>1</sup> and Dirk Mayer 1,\***


**Abstract:** In this work, a novel sandwich-type electrochemical immunosensor was developed for the quantitative detection of the carcinoembryonic antigen, an important tumor marker in clinical tests. The capture antibodies were immobilized on the surface of a gold disk electrode, while detection antibodies were attached to redox-tagged single-walled carbon nanohorns/thionine/AuNPs. Both types of antibody immobilization were carried out through Au-S bonds using the novel photochemical immobilization technique that ensures control over the orientation of the antibodies. The electroactive SWCNH/Thi/AuNPs nanocomposite worked as a signal tag to carry out both the detection of carcinoembryonic antigen and the amplification of the detection signal. The current response was monitored by differential pulse voltammetry. A clear dependence of the thionine redox peak was observed as a function of the carcinoembryonic antigen concentration. A linear detection range from 0.001–200 ng/mL and a low detection limit of 0.1385 pg/mL were obtained for this immunoassay. The results showed that carbon nanohorns represent a promising matrix for signal amplification in sandwich-type electrochemical immune assays working as a conductive and binding matrix with easy and versatile modification routes to antibody and redox tag immobilization, which possesses great potential for clinical diagnostics of CEA and other biomarkers.

**Keywords:** electrochemical immunosensor; carcinoembryonic antigen; carbon nanohorns; redox-tag

#### **1. Introduction**

Cancer is a life-threatening disease with worldwide significance for the healthcare systems and a huge economic impact. Tumor biomarkers are important tools for the detection of cancer diseases, which either originate from tumor cells or emerge from the organism as a response to it. Alterations of their concentration in the body fluids may correlate qualitatively or quantitatively with the presence of cancer cells and therefore possess important clinical value for the early detection and diagnosis of the cancer diseases and thus the prognosis of the patient [1]. In fact, some biomarkers have been routinely used in clinical diagnosis including carcinoembryonic antigen (CEA), alpha-fetoprotein, prostatespecific antigen, carbohydrate antigen 125, carbohydrate antigen 153, carbohydrate antigen 199, and so on [2]. Among them, CEA, which is a set of glycoproteins of great relevance for cell adhesion during fetal development, has been considered a common cancer biomarker in clinical diagnosis since its expression declines after birth. CEA overexpression in blood serum in adult humans is usually related to the presence or progression of different types of cancer such as colorectal, liver, breast, ovarian or lung. In addition, CEA levels can also be monitored during chemotherapy to assess the progress and result of the treatment [3]. In healthy individuals, the concentration of CEA in blood serum should be less than

**Citation:** Domínguez-Aragón, A.; Zaragoza-Contreras, E.A.; Figueroa-Miranda, G.; Offenhäusser, A.; Mayer, D. Electrochemical Immunosensor Using Electroactive Carbon Nanohorns for Signal Amplification for the Rapid Detection of Carcinoembryonic Antigen. *Biosensors* **2023**, *13*, 63. https:// doi.org/10.3390/bios13010063

Received: 16 November 2022 Revised: 15 December 2022 Accepted: 22 December 2022 Published: 30 December 2022

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

3 ng/mL [4]. Therefore, the development of simple and accurate methods for ultrasensitive monitoring of CEA is of great importance to help detect the presence of cancerous tumors without the need to use invasive or costly methods.

Immunoassays are important analytical techniques based on specific antigen-antibody interactions, which are widely used in clinical diagnosis. Numerous conventional immunoassays for CEA determination, including enzyme-linked immunosorbent assay (ELISA) [5], fluorescence immunoassay [1], and electrochemical immunoassay [3], have been reported. For instance, electrochemical immunosensors are considered being promising tools due to their simple instrumentation, portability, high sensitivity, low cost, and fast response. In conventional sandwich-type immunosensors, the detection antibodies are usually labeled with a peroxidase enzyme to generate the amperometric detection signal [6,7]. Although this method is sensitive and very useful, it has been shown that the biological tag can be replaced with nanomaterials for signal conversion and amplification [3].

Single-walled carbon nanohorns (SWCNH) are an emerging class of semiconducting nanocarbons, similar to carbon nanotubes, composed of single-layer of graphene wrapped to nanosized sheaths. SWCNH form spherical aggregates with a diameter of around 80–100 nm with different morphologies, such as dahlia, bud, and seed structures, rather than dispersing separately [8]. These unique morphologies provide special properties, such as a large surface area, small size, numerous internal nanospaces, high conductivity, and mechanical strength, making them ideal nanomaterials for application in electrochemical sensors. In addition, SWCNH can be functionalized by chemical oxidation to obtain a highly hydrophilic material and to multiply the number of binding sites on the surface for the coupling with biomolecules, metal particles, etc [9]. SWCNH can be directly used in electrochemical sensing for electrode preparation due to their superior conductivity or they can be adopted as a signal tag after their decoration with redox groups [10]. Due to the high surface area, a large number of signal molecules can be attached to the SWCNHs to facilitate strong signal amplification and consequently low detection limits.

In this work, a sandwich-type electrochemical immunosensor was developed. The capture antibodies (AntiCEA1) were immobilized on the surface of a gold disk electrode, while detection antibodies (AntiCEA2) were tethered to a nanocomposite based on SWCNHs functionalized with thionine (Thi) and gold nanoparticles (AuNPs) (SWCNH/Thi/AuNPs). Both types of antibody immobilization were carried out using the photochemical immobilization technique (PIT). In this technique, antibodies are exposed to UV irradiation, which leads to selective photoreduction of the disulfide bonds in specific cysteinecysteine/tryptophan triads (Cys-Cys/Trp) [11]. The breaking of these Cys-Cys bonds produced free thiol groups, which can interact with the proximal gold surface, resulting in a covalent bonding of the antibody. Besides, PIT ensures control over the orientation of immobilized Abs, with their binding sites exposed to the solution phase and accessible for antigen coupling [12]. The main advantage of this immunosensing system is the reduction of the fabrication time that the PIT method provided; the immobilization of the antibodies required only 15 min. Meanwhile, the PIT technique used in the present work does not require any additional surface modification steps at the electrode, which decreased the total fabrication time to only 2.25 h, notably less than that of other reported assays.

Thionine was used as a redox tag for the amperometric detection scheme and a multitude of these redox molecules was attached to the large surface area of the SWCNH. The detection was carried out through differential pulse voltammetry (DPV), where the change in the current intensity of the redox peak of thionine was related to the concentration of the biomarker CEA. The SWCNH were thus employed as conductive, high surface area but still small binding matrix for the attachment of AuNP-antibody entities and redox tags.

#### **2. Materials and Methods**

#### *2.1. Characterization*

Cyclic voltammetry (CV) and differential pulse voltammetry (DPV) were measured on a multichannel potentiostat (CHI1030B, CH Instruments, Inc. Austin, USA.) with a threeelectrode configuration. While electrochemical impedance spectroscopy was measured on a BioLogic potentiostat (SP-300, BioLogic Systems, Grenoble, France. The gold electrode (Au-disk, 2 mm diameter) was used as the working electrode, and a Pt wire and saturated Ag/AgCl electrode were used as the counter electrode and reference electrode, respectively. All potentials in this work are quoted with respect to the potential of the Ag/AgCl reference electrode.

The morphology and energy-dispersive X-ray spectroscopy (EDS) mapping was analyzed with a high-resolution Hitachi 7700 transmission electron microscope (TEM, Hitachi High-Technologies Corporation, Ibaraki, Jpan) and with a scanning electron microscope (SEM, Magellan 400, FEI, Hillsboro, OR, USA, and 1550VP, Carl. Zeiss SMT AG, Oberkochen, Germany).

#### *2.2. Materials and Reagents*

Potassium ferrocyanide K4[Fe(CN)6], potassium ferricyanide K3[Fe(CN)6], thionine acetate salt, gold nanoparticles (5 nm diameter), and oxidized carbon nanohorns were obtained from Sigma-Aldrich (Merck KGaA, Darmstadt, Germany). Phosphate buffer solution (PBS) (0.01 M) was prepared from sodium chloride (NaCl), potassium chloride (KCl), disodium phosphate Na2HPO4 and dipotassium phosphate (KH2PO4), AntiCEA1, AntiCEA2 and CEA were purchased from mybiosource.com.

#### *2.3. Fabrication of SWCNH/Thionine/AuNPs Nanocomposite (SWCNH/Thi/AuNPs)*

In this method, 2 mL of a SWCNH (I) dispersion (1 mg/mL) was mixed with 2 mL of thionine (4 mg/mL), stirring vigorously for 24 h at room temperature. The product was purified with Milli-Q water by centrifugation (12,000 rpm) to remove unbound thionine molecules. The SWCNH/Thi (II) was dispersed in 2 mL of Milli-Q water and then, 8 mL of AuNPs dispersion was added to the dispersion. The mixture was allowed to react for 48 h under magnetic stirring. Subsequently, the mixture was washed several times by centrifugation (12,000 rpm); the recovered solid was redispersed in 2 mL of 0.01 M PBS and stored at 4 ◦C.

#### *2.4. Preparation of Detection Antibody Labeled SWCNH/Thi/AuNPs/AntiCEA2*

AntiCEA2 was immobilized on SWCNH/Thi/AuNPs by covalent interaction between AntiCEA2 and AuNPs. Briefly, 350 μL of AntiCEA2 (121.42 μg/mL) was irradiated with a UV lamp (Trylight®, Promete Srl. Naples, Italy) for 30 s; afterward, it was mixed with 500 μL of SWCNH/Thi/AuNPs (III) by gently stirring for 15 min. The UV source consisted of two U-shaped low-pressure mercury lamps (6 W at 254 nm) in which a standard quartz cell could be easily housed. Considering the envelope geometry of the lamps and the cell proximity, the irradiation intensity used to produce the thiol group was approximately 0.3 W/cm2 [12].

The obtained SWCNH/Thi/AuNPs/AntiCEA2 (IV) nanocomposite was washed by centrifugation (12,000 rpm) with PBS to remove unbound material. Then, the product was redispersed in 500 μL of PBS 0.01 M. To avoid non-specific adsorption on the surface of the AuNPs, 350 μL of aqueous bovine serum albumin (BSA, Sigma Aldrich) solution (50 μg/mL) was added to the SWCNH/Thi/AuNPs/AntiCEA2, shaking gently for 1 h. Finally, the SWCNH/Thi/AuNPs/AntiCEA2/BSA (V) system was washed again, and the recovered material was redispersed in 500 μL of 0.01 M PBS and stored at 4 ◦C. For practical purposes, the term SWCNH/Thi/AuNPs/AntiCEA2/BSA will be referred to as the nanoprobe (NaPro).

#### *2.5. Assembly Process of the Immunosensor*

First, the Au-disk was polished on a micro cloth using 0.3 μm and later 0.05 μm alumina. Then, it was electrochemically annealed by 100 cyclic voltammetry scans using H2SO4 0.5 M at a potential sweep of 0.35 to 1.5 V at 1 V s−<sup>1</sup> (VI). The CV with H2SO4 did not only clean the Au surface, but also worked as a pretreatment to improve the electroactive area of the Au electrode, helping with the sensitivity of the immunosensor.

The immobilization of the AntiCEA1 antibody, on the surface of the gold electrode, was carried out using the photochemical immobilization technique (PIT). Briefly, 350 μL of AntiCEA1 (15 μg/mL) was irradiated with a UV lamp (Trylight®, Promete Srl) using a quartz cell for 30 s. The irradiated AntiCEA1 was transferred to an Eppendorf tube, and the gold electrode was immediately dipped in the solution for 15 min. Subsequently, the electrode was rinsed with PBS, obtaining Au-disk/AntiCEA1 (VII). Afterward, 25 μL of BSA (50 μg/mL) was deposited on the Au-disk/AntiCEA1 and incubated for 30 min at room temperature to avoid non-specific absorption. Subsequently, the electrode was rinsed with PBS, obtaining Au-disk/AntiCEA1/BSA (VIII). Then, 25 μL of CEA antigen was deposited at different concentrations and incubated for 45 min at room temperature. Afterward, the electrode was rinsed with PBS, obtaining Au-disk/AntiCEA1/BSA/CEA. Finally, 30 μL of NaPro was deposited and incubated for 45 min at room temperature, and the electrode was rinsed with PBS, obtaining Au-disk/AntiCEA1/BSA/CEA/NaPro. Figure 1 shows the assembly steps of the immunosensor.

**Figure 1.** (I–V) Preparation of the nanoprobe consisting of SWCNH/Thi/AuNPs/AntiCEA2. (VI–X) Immobilization of the AntiCEA2 on the Au-disk electrode by PIT and assembly of the electrochemical immunosensor. The electrochemical detection is achieved by a dependence on the thionine redox peak as a function of the CEA concentration.

#### *2.6. CEA Biomarker Detection*

The modification step-wise process of the working electrode was characterized by cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) in [Fe(CN)6] 3-/4- 10 mM in PBS 0.01 M. CEA detection was carried out by differential pulse voltammetry (DPV) in PBS 0.01 M at pH 7.4, monitoring the redox peak of thionine around −0.25 V.

#### **3. Results**

#### *3.1. SWCNH/Thi/AuNPs Characterization*

The morphology of SWCNH and SWNH/Thi/AuNPs were characterized by HRTEM and STEM. Figure 2A,B shows that a single carbon nanohorn is around 2–5 nm in diameter and 40–50 nm in length. The individual nanohorns tend to aggregate forming the typical dahlia-like nanostructure with an approximate diameter of 80–100 nm [8,13]. The STEM images of SWCNH also confirmed the dahlia-like assemblies.

Figure 2C–F show the SWCNH/Thi/AuNPs, which demonstrated the AuNPs were homogeneously distributed and anchored on the pristine SWCNH surface, providing uniform binding sites for the attachment of antibodies to the nanohorns. The average size of the AuNPs was 5–10 nm. Notably, the structure of the SWCNH was not altered during the incorporation of the AuNPs.

**Figure 2.** (**A**,**B**) HRTEM images of pristine SWCNH, (**C**,**D**) HRTEM images of SWCNH/Thi/AuNPs, (**E**,**F**) STEM images of SWCNH/Thi/AuNPs.

To verify the presence of thionine and AuNPs in SWCNH an EDS mapping was performed, and the elemental distribution within the SWNHs was verified using STEM-EDS (Figure 3). The SWCNH/Thi/AuNPs is mainly composed of carbon (Figure 3B), the presence of oxygen was also observed as a consequence of the oxidation treatment of SWCNH (Figure 3C), which likely contains a variety oxygen associated functional units such as hydroxyl and carboxyl groups [14]. Besides, the element sulfur was also observed (Figure 3D), which is attributed to the presence of thionine in the surface of the SWCNH since thionine contains a thiazinium group [15]. The presence of gold nanoparticles can be clearly seen on the SWCNH surface, which indicates that the nanoparticles observed by TEM and STEM are certainly gold nanoparticles, Figure 3E. These results prove that thionine and AuNPs were firmly tethered to the SWCNH with uniform distribution and cannot be washed off be rinsing.

Possible explanations for strong interaction between these components are on the one hand that thionine has a planar aromatic structure that facilitates strong π-π stacking interactions to the likewise aromatic SWCNH surface [16]. In addition to π- stacking interactions, also coupling via (electro)activated functional groups (-C=O- and COOH) of SWCNH can be involved in thionine linking via its amino-groups [17]. Additionally, the AuNPs were attached mainly to the SWCNH surface predominantly via unspecific adsorption, which could involve physisorption, π-π stacking, hydrophobic and electrostatistic interactions [18].

A glassy carbon electrode was modified with SWCNH/Thi/AuNPs and characterized by cyclic voltammetry (CV). Figure 3F shows an anodic peak around 0.21 V and a cathodic peak at 0 V, which are characteristic potentials of the reversible two-electron transfer process of thionine at acidic pH [19]. The difference between the anodic and cathodic peak

is 210 mV, and the current ratio *I*pa/*I*pc is 2.41 mA. Therefore, the thionine redox reaction can be considered as quasi-reversible process [20].

**Figure 3.** (**A**) Energy-dispersive X−ray spectroscopy (EDS) mapping of SWCNH/Thi/AuNPs, including SEM image (**B**) Carbon (C) element, (**C**) Oxigen (O) element, (**D**) Sulfur (S) element, (**E**) Gold (Au) element and (**F**) Cyclic voltammetry of glassy carbon electrode (GCE) modified SWCNH/Thi/AuNPs in H2SO4 0.5 M at 50 mVs<sup>−</sup>1.

Noteworthy, the redox properties of thionine were not affected by the incorporation into the SWCNH. Moreover, a prominent cathodic peak was observed at approximately 0.9 V. This peak is characteristic for the reduction of gold in the reverse potential sweep after a considerable electro-oxidation in the forward sweep to potentials higher than 1.2 V, confirming the presence of the AuNPs [21].

The SWCNH/Thi/AuNPs nanocomposite showed distinct redox activity, supporting the incorporation of Thi on the SWCNH surface. Furthermore, it should be pointed out that the high conductivity of the SWCNH facilitated the electron transfer across this carbon material and thus the redox reaction of Thi molecules at the distal side of SWCNH, which enhanced the electrochemical current [22,23]. Likewise, the multitude of SWCNH associated AuNPs offers multiple sites for antibody immobilization, promising all in all high potential as signal tag for the fabrication of the sandwich-type immunosensor.

#### *3.2. Optimization Test*

To verify that the immobilization of AntiCEA1 antibodies on the surface of the Audisk by PIT was successful, the Au electrode was characterized by CV before and after immobilization of the antibodies, using a redox probe Fe(CN)6 3- /Fe(CN)6 4- (Figure 4).

The PIT method includes an exposure of the antibodies to UV irradiation, which leads to selective photoreduction of the typical disulfide bond of the antibodies in specific cysteine-cysteine/tryptophan triads (Cys-Cys/Trp). The breaking of these Cys-Cys bonds produces free thiol groups, which can interact with the proximal gold surface, resulting in a covalent Au-S bond between the antibody and the Au surface [12].

**Figure 4.** (**A**) CV of Au−disk with (a) 0, (b) 5μg/mL, (c) 15μg/mL, (d) 30μg/mL, (e) 50μg/mL of AntiCEA1 in Fe(CN)6 3-/Fe(CN)6 4- in PBS 0.01 M at pH 7.4. (**B**) DPV of Au−disk/AntiCEA1/CEA/NaPro with (a) 5 μg/mL, (b) 15 μg/mL, (c) 30 μg/mL, (d) 50 μg/mL of AntiCEA1 in PBS 0.01 M at pH 7.4.

The Au-disk electrode showed well-defined anodic and cathodic peaks, due to the reversible oxidation and reduction of the solution phase Fe(CN)6 3-/Fe(CN)6 4- redox molecule, with a peak-to-peak difference (Δ*E*P) of 112 mV (±2.16) and an anodic current intensity (*I*P) of 54.1 μA ± (1.87). After antibody immobilization and as the AntiCEA1 concentration increased, Δ*E*<sup>P</sup> increased and *I*<sup>P</sup> decreased, confirming that the antibodies were immobilized on the Au-disk since their covalent immobilization on the surface is acting as an insulating layer, causing slower electron transfer [24]. Besides, at the concentration of 30 μg/mL of AntiCEA1, the surface of the Au-disk was practically saturated, since the change of *I*<sup>P</sup> from 30 to 50 μg/mL was minimal. It should be noted that the higher the concentration of immobilized capture antibodies, the higher the impact on the available electroactive surface area of the Au-disk. In other words, there could be a tradeoff between receptor density for binding the target molecule and the efficiency of the charge transfer between the electroactive surface and thionine redox probes. Hence, it is important to find an optimal immobilization concentration that leads to a receptor surface coverage at which the biosensor generates the highest analytical signal. Consequently, the electrodes were modified with different concentrations of AntiCEA1 and exposed to the complete detection system (CEA antigen, and the nanoprobe NaPro) at a constant concentration.

In Figure 4B, a DPV voltammogram is shown. A redox peak around −0.22 V can be observed, which is characteristic for the Thiox/Thred redox couples [20]. This demonstrates that Thi was present on the NaPro and it underwent electron transfer reactions [25]. Since the amount of attached NaPro is related with the amount of CEA present in the electrode surface, Thi works as a redox tag for the electrochemical detection of CEA. The intensity of the redox peak of Thi is related to the concentration of the biomarker CEA. The highest current intensity was obtained with an AntiCEA1 concentration of 15 μg/mL, therefore, this concentration was chosen for the further implementation of the immunosensor.

Moreover, the immobilization of the AntiCEA2 on the SWCNH/Thi/AuNPs was also carried out by the PIT. To verify the immobilization effectivity, a glassy carbon electrode (GCE) was modified with the SWCNH/Thi/AuNPs/AntiCEA2, testing AntiCEA2 concentrations of 50 and 100 μg/mL. Figure 5B shows the DPV plots of the SWCNH/Thi/AuNPs. At around −0.22 V a peak was observed that can be attributed to the redox reactions of the present thionine. The redox peak possessed a current intensity of 14.79 μA for the antibody-free SWCNH/Thi/AuNPs. This current intensity decreased after immobilization of AntiCEA2 to 5.44 μA (50 μg/mL) and 0.37 μA (100 μg/mL), confirming that AntiCEA2 was successfully immobilized. Since the concentration of 100 μg/mL significantly decreased the redox peak of thionine by blocking the charge transfer, 50 μg/mL of AntiCEA2 was chosen as the optimal concentration to maintain high analytical sensitivity.

**Figure 5.** DPV of (a) GCE/SWCNH/Thi/AuNPs/0 μg/mL of AntiCEA2, (b) GCE/SWCNH/ Thi/AuNPs/50 μg/mL of AntiCEA2, (c) GCE/SWCNH/Thi/AuNPs/100 μg/mL of AntiCEA2 in PBS 0.01 M at pH 7.4.

#### *3.3. Electrochemical Characterization by Fabrication Steps*

CV and EIS were used to corroborate the immunosensor assembly process at each modification stage and to verify the binding of the biomarker CEA and the NaPro. Both characterizations provide information on the electron transfer process and specifically, the changes in charge transfer resistance caused by anchoring the insulating biomolecules on the gold electrode.

CV tests were carried out in a Fe(CN)6 3-/Fe(CN)6 4- 10 mM solution. The Δ*E*<sup>P</sup> and *I*Pa values were determined for Au-disk (Δ*E*<sup>P</sup> = 112.4 mV ± 2.16), *I*Pa = 54.11 μA ± 1.87), Au-disk/AntiCEA1 (Δ*E*<sup>P</sup> = 335.9 mV ± 1.22, IPa = 32.7 μA ± 0.81), Au-disk/AntiCEA1/BSA (Δ*E*<sup>P</sup> = 403 mV ± 2.04, *I*Pa = 21.97 μA ± 0.49), Au-disk/AntiCEA1/BSA/AgCEA (Δ*E*<sup>P</sup> = 498.11 mV ± 0.4, *I*Pa = 16.57 μA ± 0.81), Au-disk/AntiCEA1/AgCEA/NaPro (Δ*E*<sup>P</sup> = 376.77 mV ± 6.94, *I*Pa = 29.28 μA ± 0.61).

The height of the redox peaks consecutively decreases after the addition of AntiCEA1, BSA, and CEA antigen, Figure 6A. This behavior is attributed to the fact that these biomolecules do not possess conductive properties, which on the one hand do not contribute to the transport of electrons and on the other hand block the diffusion of solutionphase redox probes to the surface of the electrode [26]. In the last step, where the NaPro is incorporated, the *I*Pa increased and Δ*E*<sup>P</sup> decreased, indicating that the addition of the NaPro improves the electroactivity, due to the good conductive properties of the SWCNH and AuNPs, similar effect was found in previous reports [27,28]. The corresponding changes observed at each stage confirm that each component was successfully implemented in the system.

Electrochemical impedance spectroscopy (EIS) is an effective tool to characterize the electrode-electrolyte interface properties. The charge transfer resistance (*R*ct) can be calculated from the semicircular section of the Nyquist plot with the axis for the real part of the impedance in EIS at low frequencies [29].

Fe(CN)6 <sup>3</sup><sup>−</sup> /Fe(CN)6 <sup>4</sup><sup>−</sup> was used as a redox couple for the EIS experiments, Figure 6B. According to the Nyquist plot, the *R*ct values were Au-disk (204.65 Ω ± 5.8), Au-disk/AntiCEA1 (1679.19 Ω ± 15), Au-disk/AntiCEA1/BSA (8361.51 Ω ± 84), Audisk/AntiCEA1/BSA/CEA (22861.61 Ω ± 116), and Au-disk/AntiCEA1/BSA/CEA/NaPro (2360.66 Ω ± 44). The addition of AntiCEA1, BSA, and CEA antigen increased the diameter of the semicircle consecutively, indicating that these biomolecules enhanced the blocking of the charge transfer at the electrode interface. Interestingly, *R*ct decreased with the incorporation of the NaPro due to the highly conductive nature of the carbon nanohorns. The result of

EIS coincided with the characteristics observed for CV; which demonstrates the successful implementation of a sandwich electrochemical immunosensor for the carcinoembryonic antigen detection.

**Figure 6.** (**A**) CV and (**B**) EIS of (a) Au−disk, (b) Au−disk/AntiCEA1, (c) Au−disk/AntiCEA1/BSA, (d) Au−disk/AntiCEA1/BSA/CEA, (e) Au−disk/AntiCEA1/BSA/CEA/NaPro in 10 mM of Fe(CN)6 3-/Fe(CN)6 4- in PBS 0.01 M at pH 7.4.

#### *3.4. Analytical Performance of the Immunosensor*

The performance of the immunosensor for the CEA biomarker detection was investigated using DPV. The CEA antigen detection was carried out in PBS 0.01 M at pH 7.4. Figure 7A shows the immunosensor response at different concentrations of CEA. The DPV signals increased as the CEA concentration rose.

**Figure 7.** (**A**) DPV of Au−disk/AntiCEA1/BSA/CEA/NaPro with different concentrations of CEA in PBS 0.01 M at pH 7.4. (**B**) The linear relationship between the current peak and the log concentration of CEA.

The sensing mechanism is attributed to the Thi used as a redox tag, since a multitude of these redox molecules was attached to the large surface area of the SWCNH. The amount of attached NaPro is related with the amount of CEA present in the electrode surface due to the formation of immunocomplex between CEA and AntiCEA2. Therefore, the change in the current intensity of the redox peak of Thi is related to the concentration of the biomarker CEA.

The calibration curve (Figure 7B) showed a linear relationship between the current intensity of the thionine redox peak and the logarithm of the CEA concentration. The linear detection range extended from 0.001 to 200 ng/mL for CEA. The calibration curve equation was *I*<sup>P</sup> (nA) = 24.726 log *C*CEA (ng/mL) + 363.24 (R<sup>2</sup> = 0.964) and the limit of detection was calculated to be 0.1385 pg/mL defined as the mean of the blank signal and 3 times the relative standard deviation. It should be noted that the concentration of CEA in blood serum is typically 3 ng/mL [4] in healthy individuals; therefore, the proposed immunosensor covers the medical relevant concentration range of the CEA biomarker and potentially facilitates practical application to monitor this biomarker. The promising performance of this sensor could be attributed to the high signal amplification capabilities of the highly conductive SWCNH/Thi/AuNPs and their decoration with a high number of redox active thionine.

Compared with other previously reported methods in the literature (Table 1), our immunosensor advanced current detection technology in the combination of exhibiting a wider detection range and lower detection limits. It should also be noted that the preparation time of the previously reported systems is typically quite long because the incubation times for the immobilization of the antibodies can take several hours while here it required only 15 min thanks to the PIT activation and enhanced the immobilization via the thiol groups of the cysteines proteins. In addition, before immobilization, other methods require a modification of the electrode surface. Meanwhile, the PIT technique used in the present work does not require any additional surface modification steps, which decreases the total fabrication time to only 2.25 h, notably less than that of other reported techniques.

Besides, the immobilization of the antibodies by the PIT is very effective since it ensured control over the orientation of the immobilized Ab, with their binding sites exposed for the formation of the antigen-antibody immune complex [12,24,30]. Indeed, Funari et al. [11] investigated the immobilization and orientation of antibodies (Abs) photoactivated by PIT. In their experiments, the photoactivated antibodies were immobilized on ultrasmooth template stripped gold films and investigated by atomic force microscopy (AFM) at the level of individual molecules. They found smaller contact area and larger heights measured in the surfaces with the antibodies immobilized by PIT than the ones immobilized by physisorption. Therefore, the activated antibodies tend to be more upright compared with nonirradiated ones, thereby providing better exposure to the binding sites. The immobilization and orientation of antibodies photoactivated by PIT enhance the binding capabilities of antibody receptors, which is a critical aspect of immunosensor development because both the number and the orientation of the immobilized biomolecules are closely related to the detection efficiency of the device [31].


**Table 1.** Comparison of the proposed immunosensor with previous similar works.

#### *3.5. Selectivity*

The high and evolutionary evolved specificity of antibodies is one advantage of immunoassays over competing biosensor concepts. To evaluate the specificity of the electrochemical immunosensor, a selectivity analysis was performed, spiking possible interfering agents such as bovine serum albumin (BSA), human serum albumin (HSA), or CA15-3 antigen to the blank sample solution (containing the nanoprobe without the presence of CEA). The tests were performed separately by incubating the Au-disk/AntiCEA1/BSA

electrode surface in 50 ng/mL CEA, 50 ng/mL BSA, 50 ng/mL HSA, 50 U/mL CA15-3, and blank solution (0 ng/mL CEA). Although the interfering substances were applied under the same conditions as the real analyte, the response currents were much lower compared to the response toward CEA (Figure 8). This result indicates that these substances do not interfere with the target detection and the high selectivity of the antibodies was conserved during the implementation of the immunosensor, resulting in an immunosensor with excellent selectivity for CEA.

**Figure 8.** Current responses of the immunosensor to CEA (50 ng/mL) and interfering substances BSA (50 ng/mL), HSA (50 ng/mL), CA15-3 (50 U/mL) and blank in PBS 0.01 M at pH 7.4 (*n* = 3).

#### *3.6. Real Sample Testing*

To investigate the performance of the immunosensor for detection in real clinical samples, human serum samples with known CEA concentrations were analyzed, Table 2. The standard addition method was used to corroborate electrochemical detection. CEA concentrations were calculated from the calibration curve and the tests were repeated three times for each sample. Table 2 shows the recovery (%) of the serum samples found in the range of 95 to 113%. The successful results demonstrate high accuracy and the feasibility of using the immunosensor for the electrochemical detection of CEA in real clinical samples. Therefore, the results confirm the potential of the proposed method to be implemented in the clinical field for the detection and monitoring of the carcinogenic biomarker CEA in patients.


**Table 2.** Results of the recovery of the immunosensor in serum samples in 0.01 M PBS.

#### **4. Conclusions**

In this work, a sandwich-type electrochemical immunosensor was developed for the quantitative determination of the CEA biomarker using a signal amplification strategy based on carbon nanohorns. The fast photochemical immobilization technique (PIT) was employed for both capture and detection antibodies to tether them onto the gold electrode and the SWCNH/Thi/AuNPs, respectively, which facilitated short assay assembly times of less than three hours. The immunosensor showed a low detection limit of 0.1385 pg/mL, a linear detection range from 0.001–200 ng/mL, and high selectivity. The remarkable performance was attributed on the one hand to the antibodies being covalently bound to the gold surfaces by PIT, controlling the orientation of their active sites. On the other hand, the large surface area, high conductivity, and manifold thionine redox activity of the SWCNH/Thi/AuNP nanocomposite enhanced the amperometric sensor signal, which resulted in a high sensitivity of the device. Therefore, the proposed strategy of PIT antibody immobilization and SWCNH/Thi/AuNP nanocomposite-based signal amplification can be used as a versatile strategy for the clinical detection of the CEA biomarker and could potentially be extended for the clinical detection of other relevant biomarkers.

**Author Contributions:** A.D.-A.: investigation, methodology, formal analysis, writing—original draft. E.A.Z.-C.: conceptualization, validation, writing—review and editing. G.F.-M.: validation, methodology. A.O.: validation, review, and editing. D.M.: conceptualization, writing—review and editing, supervision, funding acquisition. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the German Academic Exchange Service (DAAD) by an scholarship granted to Angélica Domínguez Aragón (grant number 91791585) to carry out the research stay at the Forschungszentrum Jülich (Research Center of Jülich). Furthermore, she received a Ph.D. scholarship from the National Council for Science and Technology of Mexico (CONACYT) (grant number 701397).

**Institutional Review Board Statement:** The study was conducted in accordance with the Declaration of Helsinki, and approved by the Kommission für Ethik in der Forschung (KEF) of the Research Center Jülich (protocol code Humanserum and date of approval: 06.Jan.2022).

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Acknowledgments:** We thank Mateo Martinez, Elke Brauweiler, Raul Ochoa and Paola Anchondo for their valuable collaboration during this research.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

**Chuansheng Xia 1, Jianli Sun 1, Qiong Wang 1, Jinping Chen 1, Tianjie Wang 2, Wenxiong Xu 1, He Zhang 1, Yuanyuan Li 2, Jianhua Chang 2, Zengliang Shi 1, Chunxiang Xu 1,\* and Qiannan Cui 1,\***


**Abstract:** Observing interfacial molecular adsorption and desorption dynamics in a label-free manner is fundamentally important for understanding spatiotemporal transports of matter and energy across interfaces. Here, we report a label-free real-time sensing technique utilizing strong optical second harmonic generation of monolayer 2D semiconductors. BSA molecule adsorption and desorption dynamics on the surface of monolayer MoS2 in liquid environments have been all-optically observed through time-resolved second harmonic generation (SHG) measurements. The proposed SHG detection scheme is not only interface specific but also expected to be widely applicable, which, in principle, undertakes a nanometer-scale spatial resolution across interfaces.

**Keywords:** second harmonic generation; bovine serum albumin; heterointerface; adsorption and desorption; biomolecule

#### **1. Introduction**

Biomolecular activities at interfaces are fundamental phenomena of lives. Interpreting the interfacial dynamics of biomolecules is important for constructing accurate disease models [1–3], performing effective drug screening [4,5] and understanding spatiotemporal transport of matter/energy for life systems. As a spatial region with a thickness usually smaller than 10 nm, label-free probing of the interfacial dynamics of biomolecules is challenging. Thus far, only limited label-free probing techniques have been developed, such as surface plasmon resonance [6,7], optical fiber sensors [8], time resolved sumfrequency generation [9,10], surface-enhanced Raman spectroscopy and so on [11–15]. These methods have significantly improved the performance of interfacial biosensing in terms of high sensitivity, high resolution and real-time observation, and have greatly deepened the understanding of the interfacial dynamics at the molecular level.

Surface plasmon resonance microscopy, elegantly utilizing the localized interactions between interfacial electromagnetic fields and biomolecules, possesses molecular-level sensitivity and interfacial spatial resolutions beyond the optical diffraction limit. Inspired by the physical scheme of surface plasmon resonance microscopy, we intend to develop a complementary optical spectrum sensing technique which can realize interfacial biosensing specificity for microfluidic chips in a label-free manner. In our opinion, interfacial second harmonic generation (SHG), a second-order nonlinear optical effect induced by broken inversion symmetry of an interface, is promising to fulfill this goal. Unfortunately, secondorder susceptibility of biomolecule interfaces is usually rather small, resulting in a weak SHG signal. In practice, a PMT of a single pixel is usually required to magnify the weak SHG signals. The consequence is that it is difficult to monitor the biomolecular dynamics of an interface in real-time. An alternative way to overcome such difficulty is to significantly

**Citation:** Xia, C.; Sun, J.; Wang, Q.; Chen, J.; Wang, T.; Xu, W.; Zhang, H.; Li, Y.; Chang, J.; Shi, Z.; et al. Label-Free Sensing of Biomolecular Adsorption and Desorption Dynamics by Interfacial Second Harmonic Generation. *Biosensors* **2022**, *12*, 1048. https://doi.org/ 10.3390/bios12111048

Received: 25 October 2022 Accepted: 18 November 2022 Published: 20 November 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

increase the power of a fundamental femtosecond laser, which imposes a high risk of damage to the biomaterials and biostructures.

Two dimensional (2D) semiconducting monolayers, such as monolayer MoS2, with broken inversion symmetry, possessing large second-order nonlinear optical susceptibility, can produce strong SHG signals under femtosecond laser pulse excitations, which have been comprehensively investigated in the recent decade [16–18]. In our opinion, the strong SHG of these 2D semiconducting monolayers can serve as excellent biosensors if biomolecules can interact with them and form heterointerfaces in a liquid environment. Intuitively, the formation of a heterointerface on the surface of 2D semiconducting monolayers can change the inversion symmetry and then lead to a change in SHG signals. Since strong SHG signals of 2D semiconducting monolayers can be readily detected by regular spectrometer or CMOS sensors, it will be possible to develop a real-time sensing technique to probe biomolecule dynamics at 2D interfaces.

To establish progress, we report a label-free interfacial biomolecular sensing technique by monitoring biomolecular adsorption and desorption processes on the surface of monolayer MoS2 through time-resolved SHG. Chitosan nanoclusters and bovine serum albumin (BSA) molecules have been used to form effective Coulomb attractions with the negatively charged surface of monolayer MoS2. By measuring the SHG intensity changes as a function of time, we realize the label-free real-time sensing of biomolecular adsorption and desorption processes all-optically in liquid environments. Our results open new avenues of label-free interfacial biosensing, taking advantage of the strong optical SHG of monolayer 2D semiconductors.

#### **2. Results and Discussion**

Monolayer MoS2 is an ideal platform to construct biosensors for interfacial molecule adsorption and desorption processes, considering that the monolayer MoS2 lattice undertakes a sub-nanometer thickness with broken inversion symmetry. Inspired by the pioneering work of strong SHG observations in monolayer MoS2 in 2013 [16], monolayer MoS2 can be considered as a sub-nanometer thick nonlinear optical source emitting SHG with extremely space-confined dipole moments, which can facilitate the interfacial sensing and imaging of ultrahigh spatial resolution across interfaces. Moreover, large specific surface areas and the abundant binding sites of monolayer MoS2 can further enable effective interaction with biomolecules. To be specific, negatively charged surfaces of monolayer MoS2 samples grown by chemical vapor deposition (CVD) [19–23] are expected to induce strong Coulomb attractions in liquid environments for positively charged biomolecules, which, in our opinion, are very promising for realizing label-free, real-time sensing of molecule adsorption and desorption processes.

Figure 1a shows the experimental setup detecting the SHG spectra of monolayer MoS2 embedded in a microfluidic chip. The wavelength of a fundamental laser is centered at 780 nm, with a pulse width of about 60 fs and a repetition frequency of about 80 MHz. The fundamental laser is reflected by the dichroic mirror and then focused on the MoS2 monolayer by the objective lens (NA = 0.55). SHG of the monolayer MoS2 is collected by the same objective lens. A short-pass filter is deployed behind a dichroic mirror to eliminate the reflected fundamental residual. The SHG signals are simultaneously sent to a spectrometer and a CCD by a beam splitter. In our measurements, the power of the fundamental laser was low enough to avoid optical damage of monolayer MoS2. The fine structure of the microfluidic chip is illustrated in Figure 1b. The main body of the microfluidic chip is fabricated by 3D printing with an optical glass window on top. A sapphire substrate containing monolayer MoS2 is attached to the optical glass window. The microfluidic channel enables unidirectional flow of liquid solutions of biomolecules to form a laminar flow. As a result, a homogeneous 3D fluid–2D solid interaction is constructed to facilitate adsorption and desorption of biomolecules on the surface of monolayer MoS2. Monolayer MoS2 grown by CVD on double-sided polished sapphire substrates (Six Carbon Technology, Shenzen, China) were used as received. To confirm the monolayer nature of

these samples, optical characterizations were carried out before microfluidic experiments. Figure 1c presents optical absorption and photoluminescence (PL) spectra of monolayer MoS2. Lorentzian fitting of the PL spectrum points to a resonant peak centered at about 669 nm, which agrees well with the A-exciton resonance peak of the optical absorption spectrum [24,25]. Employing the experimental setup of Figure 1a, we measured SHG (centered at 390 nm) and fundamental (centered at 780 nm) spectra of monolayer MoS2, as indicated by the purple squares and red dots in Figure 1d, respectively. Solid lines are Gaussian fittings. Full width at half maxima (FWHM) of fundamental and SHG were fitted to be about 12.8 nm and 5.5 nm. Meanwhile, no SHG signals were observed when the fundamental was focused on the substrate, as indicated by the spectrum of sapphire (black triangles) in Figure 1d. Fundamental power dependence of SHG spectra was obtained by varying the power of the 780-nm femtosecond laser, and the result is presented in Figure 1e. Fitting with a square function (solid purple line) matches well with experimental SHG results (purple dots), suggesting a quadratic dependence of SHG power with respect to fundamental power [16]. These observations validate that our MoS2 samples are monolayers, and strong SHG can be readily recorded by our experimental setup equipped with a regular spectrometer.

**Figure 1.** (**a**) Experimental setup of SHG detection with a microfluidic chip. SM: Spectrometer, BS: Beam splitter, F: Filter, DM: Dichroic mirror, OB: Objective lens. (**b**) Schematic diagram of the microfluidic chip. (**c**) Optical absorbance and photoluminescence (PL) spectra of monolayer MoS2. (**d**) The fundamental laser (red) and SHG spectra of monolayer MoS2 (purple) and sapphire substrate (black). (**e**) Fundamental power dependence of SHG in monolayer MoS2.

To justify the electrostatic adsorption effect of monolayer MoS2, we employed positively charged chitosan to interact with monolayer MoS2. Chitosan was dissolved into a water solution of acetic acid, configuring a solution with a mass fraction of 0.8 mg/mL. Before adding the chitosan solution, a pixel-to-pixel SHG mapping of a monolayer MoS2 sample in the air was taken by scanning a 2D translation stage (Physik Instrumente, P-51, Karlsruhe, Germany). The SHG image of this sample was presented in Figure 2a. The spatial scanning step was set at 300 nm, and the grey value of each pixel was obtained by integrating the SHG spectrum counts within a wavelength ranging from 370 nm to 410 nm. For monolayer MoS2 region, strong SHG leads to a distribution of triangle reflecting the spatial profile of monolayer MoS2 lattice. For the substrate region, the absence of SHG

points to a black background. Subsequently, 1μL of chitosan solution (one drop) was added onto the same monolayer MoS2 sample. When the solution completely evaporated at room temperature, pixel-to-pixel SHG mapping of the monolayer MoS2 sample was repeated. The obtained SHG image was presented in Figure 2b. Similarly, by repeating the procedures of dropping, drying and SHG mapping, Figure 2c,d are SHG images of the same monolayer MoS2 sample covered by two and three drops of chitosan solutions, respectively. We anticipate that the electrostatic adsorption process can be initiated by the Coulomb forces between the negatively charged monolayer and chitosan. Accumulated amounts of chitosan are expected to form chitosan nanoclusters on the surface of monolayer MoS2, as illustrated by the schematic diagram in Figure 2e. By carefully comparing the SHG images before (Figure 2a) and after (Figure 2b–d) adding chitosan solutions, it is clear that adsorbed chitosan nanoclusters on monolayer MoS2 can enhance SHG's intensity. To directly reveal these differences, we subtracted the SHG image in Figure 2a from that in Figure 2b–d. The corresponding differential SHG images are plotted in Figure 2f–h. These differential SHG images present randomly distributed SHG enhancements, suggesting adsorbed chitosan nanoclusters were randomly distributed, as well. We expect that this phenomenon originated from the process wherein a solution dropping action causes a rearrangement of molecules on the surface of monolayer MoS2. Rather, the differential SHG intensity shows an increasing trend from Figure 2f to Figure 2h, as added amounts of chitosan were increased. Especially, for certain edge regions of monolayer MoS2, SHG enhancement effects turn out to be stronger, suggesting that the edge region with more charged active sites tends to facilitate chitosan adsorptions [26].

**Figure 2.** SHG mapping of monolayer MoS2 adsorbing chitosan nanoclusters. (**a**) SHG mapping of monolayer MoS2 (**a**) without chitosan nanoclusters and (**b**–**d**) with adsorbed chitosan nanoclusters of increasing concentrations. (**e**) Schematic diagram of chitosan nanocluster adsorption on surface of monolayer MoS2. (**f**–**h**) Differential SHG images of (**b**–**d**) with respect to (**a**).

To visualize the adsorbed chitosan nanoclusters on the surface of monolayer MoS2, we measured the AFM image after dropping and drying a chitosan solution for monolayer MoS2 samples on the sapphire substrate. As shown in Figure 3a, it is clear that chitosan nanoclusters adsorbed on monolayer MoS2 (triangle region) form many white dots. Furthermore, on the sapphire substrate, some chitosan nanoclusters can still be absorbed, but their density of distribution as well as size are smaller than the case of monolayer MoS2. The physical reason is that positively charged chitosan molecules tend to be more efficiently adsorbed by the negatively charged monolayer MoS2 through attractive Coulomb forces [19–23]. The height profile along the red arrow in Figure 3a is plotted in Figure 3b. Observed height values of CNs on the sapphire substrate (distance range: from 0 to 1.4 μm) suggest an averaged thickness of about 7 nm. It is obvious that height values of CNs on the monolayer MoS2 (distance range: from 1.4 to 3.5 μm) could be as large as about 13 nm. The measured thickness of monolayer MoS2 is less than 1 nm (about 0.9 nm) according to Figure 3b, which agrees well with previous measurement [27]. In addition, size distributions of adsorbed chitosan nanoclusters on monolayer MoS2 and on the sapphire substrate were analyzed, as shown in Figure 3c, where two representative regions, marked in Figure 3a, were selected. Histograms of adsorbed chitosan nanoclusters in red and green of Figure 3c are statistics of region A and B of Figure 3a, respectively. After fitting these histograms with a Gaussian function, it turns out that the averaged diameter of adsorbed chitosan nanoclusters on substrate is about 41 nm. In comparison, on monolayer MoS2, the averaged diameter of adsorbed chitosan nanoclusters is about 78 nm, validating that Coulomb attraction forces between monolayer MoS2 and chitosan nanoclusters enhance the adsorption processes. The size FWHM of adsorbed chitosan nanoclusters on monolayer MoS2 is about 42 nm, which is about twice that (about 22 nm) on the substrate. At this point, we can conclude that chitosan nanoclusters can be effectively adsorbed on monolayer MoS2 by electrostatic attractions and mediate SHG intensity after drying. However, whether such an interfacial adsorption effect can induce enough SHG intensity change for biomolecules flowing in a liquid environment is still unknown.

**Figure 3.** (**a**) Atomic force microscopy (AFM) image of monolayer MoS2 with adsorbed chitosan nanoclusters (CNs). The scale bar is 2 μm. (**b**) Height profile along the red arrow in (**a**). (**c**) Size histogram of chitosan nanoclusters distributed on the surface of monolayer MoS2 and sapphire substrate.

To demonstrate the feasibility of our SHG technique towards real-time sensing for flowing biomolecules in liquid environments, we selected bovine serum albumin (BSA) molecules and performed time-resolved SHG spectra measurements employing the experimental setup in Figure 1a,b. The proposed experimental schemes of BSA adsorption and desorption are presented in Figure 4a,b, respectively. Simply speaking, protonated BSA molecules are positively charged, so that the adsorption process is expected when a negatively charged monolayer MoS2 tends to apply attractive Coulomb forces. Then, by controlling the pH of the liquid environment, protonated BSA molecules can gain electrons, and positive charges of adsorbed BSA molecules will be neutralized. Laminar flow in the microfluidic channel will take away these interfacial BSA molecules and trigger a desorption process. Before placing the monolayer MoS2, the microfluidic chip and tubes were carefully cleaned. The BSA solution (1 μg/mL) was configured in PBS buffer solution (pH = 3.6) using BSA (5%, Yuanye Bio-Technology, Shanghai, China). The power of the fundamental laser was fixed at 8 mW. By focusing the fundamental laser tightly on the center of a monolayer MoS2 sample by a 50× objective lens (NA = 0.55), we ensured that the size of the focal spot (1.7 μm) was much smaller than the lateral size of the monolayer

MoS2 sample (about 15 μm). By finely tuning the axial position of monolayer MoS2 sample, we maximized the intensity of SHG spectrum recorded by the spectrometer. A computer program was coded to record the SHG spectra every 500 ms, which integrated all non-zero SHG counts between 370 nm and 410 nm.

**Figure 4.** (**a**,**b**) Schematic diagram of BSA adsorption and desorption processes on surface of monolayer MoS2. (**c**) Time-resolved SHG signals (purple) of BSA adsorption and desorption processes on monolayer MoS2. The lower panel is time-resolved SHG signals (black) when there is no BSA. (**d**) Time-resolved fundamental (red) and SHG (purple) signals of substrate.

In our measurements, a fluid pump sent solutions into the microfluidic channel at a constant flowing rate of 22 μL/s. In the beginning, we flushed the microfluidic chip with a PBS solution (pH = 7.4) until the SHG signal of monolayer MoS2 became stable in flowing conditions, as indicated by the time-resolved SHG signals (purple spectra) before 90 s in Figure 4c. At the 90 s mark, the BSA solution (pH = 3.6) was sent into the microfluidic channel. The intensity of SHG signals started to increase and, approximately, maintained a constant after the 200 s point. The increasing evolution of SHG signals between 90 s and 200 s is caused by the BSA molecule's adsorption on the surface of monolayer MoS2. Then, at 480 s, the PBS solution (pH = 7.4) was sent to trigger a BSA molecule desorption process. Interestingly, the intensity of SHG signals started to decrease and, eventually, recovered to a constant magnitude at about 850 s, which is equal to the scenario when no BSA was added (before the 90 s mark). Furthermore, a control experiment was performed by shifting the fundamental laser focus onto a nearby monolayer MoS2 sample and replacing the BSA solution with a PBS solution without BSA molecules, while other experimental conditions were kept the same. As indicated by the lower panel of Figure 4c, at 90s, the intensity of SHG signals (black spectra) when the fundamental laser was focusing on monolayer MoS2 remained a constant. This result strongly validates that ions or other molecule components

in the PBS solutions would not induce a detectable intensity change of monolayer MoS2 SHG signals for our experimental setup. The baseline decrease is attributed to the lattice orientation difference in CVD MoS2 from sample to sample. To evaluate the contributions of adsorbed BSA molecules to the refractive index change as well as intensity change of SHG signals, we focused the fundamental laser on the sapphire substrate. By replacing the short-pass optical filters in front of the spectrometer, a small portion of the fundamental laser (780 nm) was allowed to pass. Hence, we can measure fundamental and SHG signals at the same time. The BSA adsorption experiments were repeated by adding BSA solutions at 200 s. As shown in Figure 4d, the spectra intensity of fundamental (780 nm) remained a constant after adding BSA molecules, ruling out the possible effect of interfacial refractive index change. More importantly, the spectra intensity of SHG remained zero, indicating that SHG contributions of the sapphire substrate as well as adsorbed BSA generated can be neglected compared with the monolayer MoS2. The low SHG conversion efficiency of sub-nanometer thick monolayer MoS2 and high noise-level of our spectrometer lead to a relatively low signal to noise ratio. This issue can be further improved by increasing the integration time of each SHG spectrum and optimizing the design of microfluidic chips.

To address the physical mechanism of observed SHG signal changes accompanying the BSA adsorption process, we can consider the second-order polarization of the interface with the follow model [28,29]: *<sup>E</sup>*2*<sup>ω</sup>* <sup>∝</sup> *<sup>P</sup>*(2) = *<sup>χ</sup>*(2) *<sup>E</sup>ωE<sup>ω</sup>* + *<sup>χ</sup>*(3) *<sup>E</sup>ωEω*∅0, where ∅<sup>0</sup> is the interfacial electric field and *χ*(2) and *χ*(3) are the second-order optical susceptibility of monolayer MoS2 and third-order optical susceptibility of the interface, respectively. In our case, we believe that the spatial distribution of adsorbed BSA molecules on the surface monolayer MoS2 is random. Specifically, since the initial charge distribution of monolayer MoS2 is inhomogeneous in a 2D plane defined by the flat substrate, the orientations and 3D stacking orders of adsorbed BSA molecules are expected to be highly random. As a result, directions of interfacial electric fields between the monolayer MoS2 and adsorbed BSA molecules in disorder will no longer be strictly perpendicular to the 2D plane. Therefore, the angle between the wave vector of the fundamental laser and the direction of interfacial electric fields will not be zero, so that the ∅<sup>0</sup> term can induce a non-zero second-order polarization field. When BSA molecules are dynamically adsorbed, the total magnitude of second-order polarization fields formed by superposition of polarization fields from monolayer MoS2 and interfacial electric fields will depend on their initial phase difference.

#### **3. Conclusions**

We have comprehensively demonstrated that the interfacial SHG of monolayer MoS2 can be utilized for label-free biomolecule sensing. Through static SHG mapping experiments, we show that the intensity of SHG in monolayer MoS2/adsorbed chitosan nanocluster heterostructures can be mediated due to electrostatic attractions. With a time-resolved SHG measuring system equipped with a microfluidic chip, we further realize label-free sensing of BSA adsorption and desorption dynamics in real time through the SHG intensity change of monolayer MoS2 in liquid environments, which has been tailored by Coulomb interactions between BSA molecules and monolayer MoS2. Our work provides a complementary mean of label-free interfacial biomolecule sensing, which, in principle, undertakes molecular-level spatial resolution across the interfaces for applications including, but not limited to, in vitro medicine evaluation.

**Author Contributions:** Q.C. and C.X. (Chunxiang Xu) conceived the idea, designed the experiments and supervised the project. C.X. (Chuansheng Xia), Q.C., J.S., Q.W., J.C. (Jinping Chen), W.X. and H.Z. performed the experiments. C.X. (Chuansheng Xia), Q.C., T.W., Y.L., J.C. (Jianhua Chang), Z.S. and C.X. (Chunxiang Xu) analyzed the data. C.X. (Chuansheng Xia) and Q.C. drafted the paper with input from all authors. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by National Key Research and Development Plan of China (Nos. 2017YFA0700500 and 2018YFA0209101), National Natural Science Foundation of China (Nos. 11734005, 62075041, 61904082, 61821002, 61875089, and 62175114) and Natural Science Foundation of Jiangsu Province (BK20190765).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

