Nanomesh-patterning on multilayer MoS2 eld- effect transistors for ultra-sensitive detection of cortisol

Heekyeong Park Kyung Hee University Seungho Baek Sungkyunkwan University Bongjin Jeong Electronics and Telecommunications Research Institute (ETRI) https://orcid.org/0000-0001-6536-7215 Anamika sen Sungkyunkwan University Sehwan Kim Sungkyunkwan University Yun Chang Park National Nanofab Center Sunkook Kim (  seonkuk@khu.ac.kr ) Sungkyunkwan University https://orcid.org/0000-0003-3724-6728 young jun kim Electronics and Telecommunications Research Institute (ETRI)


Introduction
Field-effect transistors (FETs), with intrinsic advantages that allow label-free, rapid, and ultra-sensitive detection of target molecules [1][2][3] , are highly desired as biosensors for a wide range of medical care 4 , agricultural 5 , and environmental 6 applications. Under suitable conditions, FET-based biosensors (bio-FETs) are capable of more sensitive conversion of the speci c interaction between target molecules and receptor elements. More importantly, the detection performance of bio-FETs largely depends on the structural and electrical characteristics of the channel materials, thereby opening opportunities to further enhance assay performance by devising speci c nanostructures. One-dimensional (1D) nanomaterials, including silicon nanowires and carbon nanotubes, have been investigated to improve sensing performance; however, their complex manufacturing and integration processes are major hindrances to their commercialization 7 . Among two-dimensional (2D) nanomaterials, transition metal dichalcogenides (TMDs) are considered promising materials for ultra-sensitive biosensors because of their intrinsic carrier transport and modulation in contrast to graphene, which has zero band gap 8-10 .
However, in spite of the intrinsic properties of their 2D structure, the use of TMD materials for applications in biosensors has not been very successful mainly due to the limited number of functional groups on the TMD surface. Conventional bio-FETs based on TMDs contain a dielectric layer such as hafnium oxide (HfO 2 ) or aluminum oxide (Al 2 O 3 ), as a grafting layer to functionalize the bioreceptors on the top of TMD channels 8, 11,12 . The surface of the oxide layer enables the chemical functionalization of bioreceptors using 3-aminoproplytriethoysilane (APTES) and glutaraldehyde. However, the dielectric effect on the screening of biomolecular charges deteriorates the sensitivity of bio-FETs 13,14 . In addition, interface defects between the channel and dielectric layer can create incidental electric elds or parasitic coupling, further reducing the sensitivity and reliability of bio-FETs 15 . In an effort to circumvent the problem related to functionalization, surface defect engineering is applied to generate dangling bonds on the TMD surface. Sim et al. 16 reported the arti cial formation of sulfur vacancies on molybdenum disul de (MoS 2 ) for the direct attachment of functional molecules on the MoS 2 surface. In addition, Lee et al. 17 reported defect generation on a tungsten diselenide (WSe 2 ) surface using oxygen (O 2 ) plasma treatment to functionalize the bioreceptors on the surface.
In this work, we applied a nanomesh structure on MoS 2 via block copolymer (BCP) lithography, where newly generated dangling groups on the edges of the perforated area would provide a rich source for functionalization. Moreover, direct covalent linkage between MoS 2 and the biorecognition element could avoid sensitivity and reliability losses. Indeed, periodically arranged nanoholes on the MoS 2 nanomesh consisting of abundant edge sites were con rmed by scanning transmission electron microscopy (STEM), Raman, and X-ray photoelectron spectroscopy (XPS) spectroscopic analysis. These edge defects allow direct functionalization of the receptors on the MoS 2 nanomesh channel for ultra-sensitive detection of biomolecules. Cortisol, a target biomolecule, is a glucocorticoid steroid hormone secreted through the hypothalamus-pituitary-adrenal (HPA) axis. Repeated activation of the HPA axis has been reported to negatively affect mental health, causing major depressive disorder 18 , anxiety disorder 19 , and bipolar disorder 20 . Aptamer-functionalized MoS 2 nanomesh FETs exhibit excellent detection properties for cortisol biomarkers with a low limit of detection (LOD) of 10 − 18 g/mL, in environments including arti cial saliva and real human serum. Furthermore, high selectivity for other molecules and steroid hormones was also con rmed, thus, verifying the high reliability of the proposed nanomesh architecture for high-performance biosensors.

Results
Nano-scale patterning of multilayer MoS 2 Various nanomaterials have been used for high-performance biosensors by designing sophisticated schemes but often with complicated procedures. Multilayered MoS 2 , an extension of 2D materials, although endowed with intrinsic morphological advantages as a sensor, is unsuitable for applications as biosensor due to the absence of functional groups upon which the capture molecules can be attached. In this study, the functionality problem was overcome by generating innumerable nano-sized holes of uniform diameter on MoS 2 . We established a procedure for the fabrication of the MoS 2 nanomesh structure using the BCP self-assembly layer as a nanomesh template (Fig. 1a). Multilayered MoS 2 nanosheets, physically exfoliated from its bulk, were passivated by depositing a 10 nm thick silicon oxide (SiO 2 ) layer, which plays an important role in preventing chemical damage on the MoS 2 surface during the patterning process. The spin-coated BCP layer underwent a cylindrical microphase separation between domains of polystyrene (PS) and polymethyl methacrylate (PMMA) at an elevated temperature of 230 °C. The minor phase of the PMMA domain was selectively removed through ultraviolet (UV) exposure, leaving the PS phase intact in mesh morphology. The nanomesh fabrication procedure was completed by washing with acetic acid. A scanning electron microscopy (SEM) image of the BCP template is shown in Supplementary Fig. 1a, indicating the characteristic ordering of nanoporous structures over a large area with a uniform hole diameter (20.7 ± 1.3 nm). The nanomesh template was treated with O 2 plasma reactive-ion etching (RIE) to control the hole size of the BCP template. The nanomesh structure of MoS 2 was constructed using the BCP template as a patterning mask, in which the SiO 2 layer was etched away by sulfur hexa uoride (SF 6 ) plasma RIE and the MoS 2 was perforated by boron trichloride (BCl 3 ) plasma RIE. The remaining SiO 2 was easily removed by dipping the substrate into buffered oxide etchant (BOE), leaving no trace of contaminants. The detailed process is described in the Methods section. The SEM images of the MoS 2 surface for each process are shown in Supplementary   Fig. 1. Figure 1b shows a low-magni cation STEM image of the multilayered MoS 2 nanomesh lm, revealing periodically organized nanoholes in hexagonally packed ordering. The structural uniformity of the nanoholes was con rmed by measuring the diameter of 2,200 nanoholes, which was calculated to be 23.36 nm with a standard deviation of 1.5 nm (Fig. 1c). Further, the vertically oriented nanohole structure across the multilayered MoS 2 was con rmed by a cross-sectional STEM image, as shown in Fig. 1d, where the nanohole areas look brighter than the unperforated areas owing to differences in the distance from the MoS 2 atoms to the objective lens of the STEM. In the brighter nanohole area, 10 MoS 2 layers can be clearly observed. To study the effects of the newly formed morphological changes in MoS 2 , a comparative analysis was performed for both the nanomesh and pristine MoS 2 lms using Raman and XPS. In the Raman spectra ( Fig. 2e), 2 characteristic Raman peaks of the in-plane, E 1 2g , and out-of-plane, A 1g , vibrations were observed at 380 and 407 cm − 1 , respectively, in both the pristine and the nanomesh MoS 2 21 . After perforation, the relative intensities of the in-plane and out-of-plane vibrations (E 1 2g /A 1g ) decreased from 0.802 for the pristine to 0.613 for the nanomesh MoS 2 . The lowering of E 1 2g /A 1g in the nanomesh MoS 2 supports the presence of the newly generated nanoholes and increase in the edge sites, because A 1g vibration is preferentially excited to the E 1 2g vibration for edge-terminated lms 22,23 .  24,25 . In the doublet of i-MoS 2 , the maximum peak at around 229.10 eV (Mo 4+ 3d 5/2 ), which corresponds to 2H stoichiometric MoS 2 (ratio of S/Mo = 2), was observed in both the pristine and nanomesh MoS 2 lms with high intensity. The maximum peak of the d-MoS 2 at ~ 228.50 eV (Mo 4+ 3d 5/2 ) is relatively small, corresponding to the nonstoichiometric MoS 2 (S/Mo ratio < 2) introduced by atomic defects on the lm, such as vacancies and exposed edges 24 . These defect sites have electronic structures different from those of the intrinsic MoS 2 owing to the unstable energy state, resulting in peak excitation at lower binding energies. The atomic fraction of d-MoS 2 among the total Mo ligands was calculated to be 6.18% for the pristine MoS 2 lm, which signi cantly increased to 15.82% for the MoS 2 nanomesh lm (

Operation of the MoS 2 nanomesh FET
To explore the electrical characteristics of the multilayer MoS 2 nanomesh, we fabricated FETs using MoS 2 nanomesh lms as channels (Fig. 3a). The SEM image of the MoS 2 nanomesh FET shows uniform nanoholes across the channel area positioned between the two titanium/gold (Ti/Au) electrodes (Fig. 3b). The sensing behavior of the MoS 2 nanomesh bio-FET was studied by gradually adding cortisol stepwise so that the target concentration could be adjusted as required. Highly sensitive variation of I DS -V GS curves was observed upon exposure to cortisol, as shown in Fig. 4b. To better understand the effect of nanohole formation in MoS 2 on the sensing behavior, the sensing characteristics of the nanomesh and pristine MoS 2 FETs were compared by repeating the experiment using multiple devices. The sensing behavior of pristine MoS 2 is shown in Supplementary Fig. 4 with an enlarged view of the vertical scale.
The sensitivity was calculated based on (I base -I cortisol )/I base х 100, where I base and I cortisol are the values before and after the addition of cortisol, respectively, at V GS of 10 V. While a typical S-shaped response pattern with a wide linear range from 10 -18 g/mL to 10 -13 g/mL was observed for the nanomesh MoS 2 FET, in the case of pristine FET, no apparent variation in the response signal was observed except over Page 7/16 performance 10 9 times that of pristine MoS 2 . Such a superb enhancement in sensing behavior is thought to have been caused by the increased edge area provided by the newly generated nanoholes in MoS 2 , which in turn enabled direct chemical bonding between the aptamers and MoS 2 with concomitant augmentation of polarity modulation.

Sensor reliability
The reliability of the nanomesh bio-FET was examined by studying the extent of nonspeci c binding and selectivity using potentially interfering biomolecules. To understand the level of nonspeci c interaction that might have been generated due to nanohole formation in MoS 2 , cortisol detection was performed using nanomesh FETs in the absence of the aptamer capture molecules, speci cally the nanomesh FET in which the surface chemistry was altered only up to the functionalization step of glutaraldehyde. No appreciable signal change was detected (Fig. 4d), suggesting that nonspeci c interaction was negligible.
In addition, the selectivity of the MoS 2 nanomesh biosensor for target cortisol was evaluated by measuring the electrical signals on exposure to potentially interfering biomolecules, including steroid hormones of progesterone and aldosterone, and protein biomarkers of alpha-fetoprotein (AFP), prostate speci c antigen (PSA), and carcinoembryonic antigen (CEA) (Fig. 4e). The differences in the signal intensity before and after exposure (|ΔI DS |) to the protein antigens were below recognizable levels, whereas in the case of the steroid hormones, only slight differences in signal intensities were observed. It is worthwhile to note that the chemical structure of cortisol is very similar to that of progesterone and aldosterone.
Saliva and serum test One major advantage of using cortisol as a biomarker is the diverse availability of its biological uids, including serum, urine, sweat, and saliva. In particular, measuring cortisol concentration in saliva has multiple merits, including being a patient-friendly noninvasive detection method and direct measurement of free unbound cortisol, which is a biologically active form; 90% of cortisol circulates in complexation with globulin 29 . To study the applicability of the nanomesh MoS 2 bio-FET in the clinical environment, the detection behavior of the bio-FET was examined in the media of arti cial saliva and human serum. Figure 5a shows I DS -V GS curves when the detection test was performed by varying the amount of cortisol from 10 − 18 g/mL to 10 − 8 g/mL in arti cial saliva including the control test. Well separated signals are evident especially at low concentration range with gradual decrease in distance between signals at higher concentration window. This tendency was somewhat more pronounced in the case of arti cial saliva than human serum (Fig. 5b). The I DS -V GS curves for real humans are shown in Supplementary Fig. 5.
When compared with the results from the buffer test, the detection performance in the arti cial saliva and serum seemed to have decreased to some extent, especially in the low concentration range. However, the LOD remained at 10 − 18 g/mL with linearity ranging up to 10 − 12 g/mL in those biological uids.

Conclusion
Superb enhancement in sensing performance was achieved by providing nanohole structures in MoS 2 .
BCP nanotemplates produced periodically organized nanoholes on the MoS 2 surface, thus introducing abundant edge sites. The generation of edge sites on the MoS 2 surface resulted in degradation of the electrical performance for FET while offering extremely high sensitivity for bio-FET by inducing direct functionalization of the edge sites using cortisol aptamer. We con rmed the ultra-high sensitivity by comparing it with the sensing properties of bio-FETs based on pristine MoS 2 . In addition, the nanomesh bio-FET functionalized with aptamer exhibited reasonable selectivity for cortisol compared to other steroids and antigens, including progesterone, aldosterone, AFP, PSA, and CES. Clinical applicability was also con rmed by performing the test on arti cial saliva and real human serum. The excellent detection performance of the MoS 2 nanomesh bio-FET was con rmed by realizing an ultra-low LOD of 10 − 18 g/mL, which veri ed the potential of our effective platform for future biosensor applications and new diagnostic processes. lms were transferred onto Cu grids coated with a lacey carbon lm. A cross-sectional view was observed by milling the sample using a single-beam focused ion-beam (FB-2100, Hitachi). Low-magni cation surface images were obtained by SEM operated at an acceleration voltage of 15 kV and working distance of 8 mm. Raman spectroscopy (ALPHA300, WITec) was used to identify the formation of edge sites on MoS 2 nanosheets. A laser beam with a spot diameter of 1 µm and excitation wavelength of 532 nm was used, and the instrument spectral resolution was approximately 1 cm -1 . The chemical states of the MoS 2 lms were explored by XPS (K-Alpha, Thermo Fisher Scienti c) using an Al Kα source. All electrical measurements were carried out using a semiconductor characterization analyzer (4200-SCS, Keithley) equipped with a probe station for sample loading and electrode contact.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.