Ag@ZIF-67 decorated cotton fabric as flexible, stable and sensitive SERS substrate for label-free detection of phenol-soluble modulin

Although strategies of compositing noble materials and Zeolitic imidazolate frameworks (ZIFs) have been used to enhance the performance of ZIF-based surface-enhanced Raman scattering (SERS) substrates, the enhancement process still remains unclear and the sampling process with powder-form substrates is challenging for practical applications. In this study, a flexible SERS substrate with silver (Ag) nanoparticle decorated ZIF-67 as the active materials and cotton fabric as the supporting framework is developed with a facile method in two steps. The proposed flexible SERS substrate not only can reduce difficulties during the sampling process, but also expands future applications for sampling towards irregular target substances. Meanwhile, the enhancement mechanism has been investigated by using Methylene Blue to probe the molecular interactions. The constructed SERS substrate shows the highest enhancement factor of 6.25 × 106 and an excellent detection sensitivity with a limitation of detection (LoD) of 10–14 M/L. Because of its excellent performance in SERS tracing, the proposed SERS substrate shows exceptional ability in detecting and identifying phenol-soluble modulins and is considered to be a label-free approach that requires no adding markers/tags or antibodies. The proposed substrate expands the potential applications of SERS technology for rapid detection of bacterial toxins during clinical diagnoses and treatment. Ag decorated ZIF-67 particles are anchored onto cotton fabric to fabricate flexible SERS substrates. The proposed flexible SERS substrate achieves high SERS activity with high enhancement factor, low LoD, good reproducibility and stability. The flexible SERS substrate shows great potential in detection and identification of Phenol-soluble modulin.

enhancement mechanism has been investigated by using Methylene Blue to probe the molecular interactions. The constructed SERS substrate shows the highest enhancement factor of 6.25 9 10 6 and an excellent detection sensitivity with a limitation of detection (LoD) of 10 -14 M/L. Because of its excellent performance in SERS tracing, the proposed SERS substrate shows exceptional ability in detecting and identifying phenol-soluble modulins and is considered to be a label-free approach that requires no adding markers/tags or antibodies. The proposed substrate expands the potential applications of SERS technology for rapid detection of bacterial toxins during clinical diagnoses and treatment.

Introduction
Surface-enhanced Raman scattering (SERS) is an ultrasensitive vibrational spectroscopic technique that is used to detect probe molecules by localizing molecules of interest on an enhanced surface. This technique has been widely explored and used in numerous applications, such as detecting the pH of biological fluids (Bi et al. 2018), monitoring pollutants (Castro-Grijalba et al. 2020), ensuring food safety (Danaei et al. 2011), detecting cancer biomarkers (Abbosh et al. 2017), nucleic acids (Jiang et al. 2018) and DNA (Cohen et al. 2017), etc. The so-called enhanced surface area is known as a SERS-active substrate, which is the key component of this detection technique. The substrate mainly includes noble metals (Xiong et al. 2017), semiconductors (Yan et al. 2018), graphene (Ling et al. 2010) Mxenes (Soundiraraju and George 2017), or their complexes (López-Lorente et al. 2018).
The properties of SERS-active materials, including their sensitivity, stability, reproducibility and portability, play a crucial role in the level of detection and a wide range of practical applications. Meanwhile, those properties have been enhanced through the advancement of SERS-active materials over years.
In terms of sensitivity, ultrahigh sensitivity which is determined through the limitation of detection (LoD) can be mainly achieved by using noble metal nanoparticles [such as silver (Ag) or gold (Au)] (Zhao et al. 2008). Furthermore, the LoD is determined by the number, shape, size, strength and location of ''hot spots'' on the substrates (Sharma et al. 2012). Therefore, numerous experiments have been carried out in the literature, and ultrasensitive detection of certain probe molecules, even a single molecule, has been realized with the use of optimized SERS substrates (dos Santos et al. 2019;Moeinian et al. 2019). However, the relatively high cost and low commercial value have inhibited the practical applications of noble metal SERS-active substrates.
The stability of SERS substrates is usually evaluated by examining their Raman signal stability and chemical stability. The former is mainly determined by the distribution of active materials and whether the probe molecules react with the substrate. The chemical stability is mainly affected by the stability of active materials. Although gold nanoparticles have shown exceptional SERS performance (dos Santos et al. 2019;Moeinian et al. 2019), the high cost of noble metal limits their practical applications. Consequently, semiconductor SERS substrates, such as titanium dioxide (TiO 2 ) (Ling et al. 2010), iron III oxide (Fe 2 O 3 ) (Xu, Li, et al. 2019a, b), copper telluride (CuTe) (W. Li et al. 2013) and copper I oxide (Cu 2 O) (Lin et al. 2018) substrates, have been developed. Interestingly, these partial semiconductors present an outstanding LOD of 10 -7 M, high enhancement factor (EF) of 10 5 (Zheng et al. 2017), and excellent chemical and signal stability. However, the fabrication technique of semiconductor SERS substrates with an excellent performance usually involves extreme processing conditions, such as using the hydrothermal method (Cong et al. 2015) and annealing at high temperatures (Ling et al. 2010).
As for signal reproducibility, it is affected by the uniformity of the particle size and distribution of the particles. A considerable number of experiments have been conducted to control the uniformity of the size of the nanoparticles over the years (Chen et al. 2019a, b;Chen et al. 2019a, b). Therefore, signal reproducibility has been improved owing to enhanced particle uniformity. Nevertheless, the real challenge in improving the signal reproducibility is the distribution of the particles because SERS-active particles easily aggregate during synthesis. Besides, it is difficult to collect SERS-active particles absorbed with probe molecules, which greatly limits applications in the real world.
For the applications including the examination of food safety (Ogundare and van Zyl 2019), the prediction of disease (Kwon et al. 2019), the detection of the presence of bacteria ) and the monitor of the environment (Xu et al. 2019a, b, c), SERS substrates are required to be easy-to-use, light in weight, small in size and have the ability to efficiently collect the target molecules. Thus, the portability or flexibility of SERS substrates matters. Also, flexible SERS substrates, including paper-based (Yu and White 2013), polyester (PET) film-based , textile-based (Cheng et al. 2019, tapebased (Liyanage et al. 2018), silicon chip-based ) and swab-based (Gong et al. 2014), have been developed by demands of applications in recent years. Although these approaches may have improved the efficiency and reduced the cost of the fabrication of SERS substrates, more works are needed for the evaluation and development of applications with a wider range of target molecules.
To address the limitations of the previous study of SERS substrates, a novel SERS substrate is proposed in this study, taking the processing conditions, the simplicity of fabrication process as well as the SERS performance into consideration. It is reported that metal-organic-frameworks (MOFs), especially zeolitic imidazolate frameworks (ZIFs), can be used as a type of SERS substrate with a high EF of 10 6 and low LOD of 10 -8 (Sun et al. 2019). In previous studies, MOFs are mainly used as an absorption layer placed onto noble metal particles because of the porosity (Hu et al. 2014). Furthermore, there are few studies focusing on the use of MOFs as SERS substrates. Therefore, regardless of their performance or applications, further related studies need to be carried out. Correspondingly, we combine noble metal nanoparticles and ZIFs with SERS activity and assemble them onto cotton textile to improve their SERS performance. Cotton is an ideal medium to load SERS active particles, for its excellent mechanical properties, large surface area and high degree of flexibility. To clarify the novel flexible SERS substrate, the SERS performance of the ZIF-67/cotton, Ag/cotton and Ag@ZIF-67/cotton samples is systematically investigated, and the enhancement mechanism is also discussed. The application of the proposed substrate for the detection of phenol-soluble modulins (PSMs) as a label-free approach without adding markers/tags or antibodies will validate and expand the practical applications of SERS technology for clinical diagnoses and treatments.

Materials
Plain woven cotton fabric (133 9 72 counts per square inch, 40 D 9 40 D) was used as the substrate. Cobalt nitrate hexahydrate (Co(NO 3 ) 2 Á6H 2 O) and 2-methymelodoloze (2-MI) (Adamas Co., Ltd.) were used to prepare the solutions. Methanol (AR grade) (Aladdin Co. Ltd.) were also used in the solutions. MB (C 16 H 18 N 3 SÁCl) (Shanghai Aladdin biochemical Polytron Technologies Inc.) were used to probe the molecular interactions. PSM a1, a3 and a4 with a concentration of 1 mM/L were purchased from China Peptides Co., Ltd.
Fabrication of ZIF-67 and Ag@ZIF-67 coated cotton fabric First, 1.5 mmol of Co(NO 3 ) 2 Á6H 2 O and 22.5 mmol of 2-MI were dissolved into 20 ml of methanol, and labelled as Solutions A and B, respectively. The cotton fabric samples (5 cm 9 5 cm) were cleaned in an ultrasonic bath with a solution of acetone and ethanol (1:1) for 30 min at room temperature. The power level and frequency of the ultrasonic treatment were 200 W and 70 kHz, respectively. Then, the fabric samples were washed with deionized water and dried for 12 h at 50°C. The cleaned fabric samples were immersed into Solution B for 5 min. Then, Solution A was added by vigorous stirring to Solution B which contained the cotton fabric samples. The mixed system was stored at 35°C for 24 h. The treated fabric samples turned purple and were washed 3 times with ethanol and deionized water, respectively. ZIF-67 coated cotton fabric was produced after the fabric was dried at 50°C for 12 h.
Then the ZIF-67 coated cotton fabric was deposited with Ag nanoparticles for 5 s, 30 s, 60 s and 90 s by using magnetron sputtering technology with a RF power of 100 W. For comparation, cotton fabric without ZIF-67 was also deposited with Ag nanoparticles at the same conditions. The final samples coated with Ag@ZIF-67 and Ag were labelled as C-ZIF-67, C-AZ5, C-AZ30, C-AZ60, C-AZ90, C-Ag5, C-Ag30, C-Ag60 and C-Ag90, where C stands for cotton fabric, Z stands for ZIF-67, A stands for Ag, and the number stands for the coating time of Ag.

Characterization
The crystal structures of the samples were examined by using an X-ray diffractometer (XRD) (Rigaku Smart Lab). The process was carried out with an X'celerator X-ray detector in the normal mode with a Cu Ka radiation source (k = 0.154 nm) at an accelerating voltage of 45 kV, and emission current of 200 mA. Details of the crystal structure were collected at a 2h angle range of 5°to 90°with a scanned step size of 0.02°/step and a scanned speed of 5°/min. Scanning electron microscopy (SEM) of the ZIF-67 and Ag decorated ZIF-67 coated cotton fabric samples was conducted by using a TESCAN VEGA3 scanning electron microscope. The Raman spectra were collected on a BAYSPEC 3 in 1 Nomadic Raman microscope equipped with 532 nm and 785 nm excitation lasers. X-ray photoelectron spectroscopy (XPS) measurements were carried out on a Kratos XSAM800 spectrometer operated at 1486.60 eV.

Raman measurement
The Raman behaviors of the ZIF-67 and Ag decorated ZIF-67 coated cotton fabric samples were measured by using MB as the target molecules. A 10 -3 M solution of the target molecule was first prepared, which was diluted to an aqueous solution with a concentration that ranged from 10 -4 to 10 -14 M. Before measuring the Raman spectra, the sample substrates with dimensions of 1 cm 9 1 cm were soaked in the aqueous solution of target molecule for 5 min.
The Raman spectra were subsequently examined under the Nomadic TM Raman 3-in-1 microscope. The laser beam was focused onto a spot with a diameter of * 2.6 lm to excite the samples through the OLYMPUS MPlanFLN 10 9 objective lens. The Raman spectra were collected with an integration time of 3 s, and a laser power of 50 mW at room temperature. The Raman spectra from different locations were collected for each sample.

Results and discussion
Fabrication of Ag decorated ZIF-67 coated cotton fabric Figure 1 shows the fabrication process of the flexible Ag@ZIF-67 decorated cotton fabric substrate. In this study, ZIF-67 is synthesized in accordance with a typical method (Tan et al. 2010). To be emphasized, the pre-cleaned cotton fabric was first soaked in a solution of 2-MI to ensure that the cotton fabric was completely saturated with 2-MI for further growth of the ZIF-67 on the surface of the cotton fibers. Then, Ag nanoparticles were deposited onto the obtained C-ZIF-67. The size of the Ag nanoparticles and spacing between them were adjusted by varying the deposition time as done in previous report (Jiang et al. 2015). For comparison purposes, Ag nanoparticles were also deposited onto a pre-cleaned cotton fabric sample without ZIF-67.
The optical images of the samples are illustrated in Figure S1. Clearly, the color of the cotton fabric changes from white to purple after decorated with ZIF-67, and silver grey after coated with Ag nanoparticles, while the surface of the ZIF-67 is gradually covered after being coated with the Ag nanoparticles, which indicates that the ZIF-67 and Ag nanoparticles were deposited onto the cotton fabric substrate successfully. The flexibility of this substrate is mainly due to the excellent flexibility of the cotton fabric, which will contribute to increasing applications of SERS in tracing chemicals, micro-organisms, and toxins on the surface of the target object.
To further validate the color variations of the cotton fabric, which is attributed to the ZIF-67 and Ag nanoparticles, SEM images of the untreated, ZIF-67 and Ag@ZIF-67 decorated cotton fabric substrates are provided in Figure S2. As presented in Figures S2a, b, the smooth morphology of the cotton fabric with a woven structure can be observed. Moreover, a rough fiber surface with polyhedral particles is found after deposited with ZIF-67. However, the smooth morphology with the polyhedral shaped ZIF-67 particles gradually increase in roughness after coated with Ag nanoparticles as shown in Figures S2c-h, and the morphology of the ZIF-67 particles coated with Ag nanoparticles at a high magnification is shown in Figs. 2a-d. The figures show that the ZIF-67 particles are covered with Ag nanoparticles that have different sizes with different spacing between the nanoparticles. This shows that ZIF-67 and Ag@ZIF-67 are well coated onto the surface of the cotton fabric.  (200) and (004) of cellulose Ib, respectively (French, 2014). Meanwhile, the peak at 7.31°is attributed to the ZIF-67 (Tan et al. 2010). There should also be other ZIF-67 peaks; however, they might have been obscured by the background peaks of cotton and Ag, thus indicating that the ZIF-67 particles have been successfully synthesized. Besides that, the diffraction peaks of Ag located at 38.1°, 44.3°, 77.4°and 81.5°are also observed (Xu, Jiang, et al. 2019a, b, c). These results indicate that ZIF-67 and Ag@ZIF-67 are well deposited onto the cotton fabric.

Evaluation and investigation of SERS performance
After the C-Ag@ZIF-67 substrate was successfully constructed, its SERS performance was evaluated. In this study, MB, which is a common probe molecular, was used as the target dye tracer (Gao et al. 2020;Xu et al. 2020a, b). The prepared samples were first immersed into the MB solution (10 -3 M/L) for 5 min to absorb the MB molecules, rinsed with deionized water and then subjected to drying at 50°C in an N 2 atmosphere.
During the process of detecting the Raman signals, incident laser lights with wavelengths of 532 nm and 785 nm were applied to focus on the surface of the prepared substrates that were absorbed with the MB molecules, and the scattered signals from the excitation area were recorded. For comparison, C-ZIF-67, C-AZ5, C-AZ30, C-AZ60 and C-AZ90, as well as C-Ag5, C-Ag30, C-Ag60, and C-Ag90 were all measured with the Raman spectrometer in the same condition after absorbing the MB molecules. Figure 3 demonstrates the SERS signals of the MB molecules absorbed on the substrates. It can be observed that all of the SERS signals are detected on all the samples under a laser excitation of 532 nm and 785 nm; however, there are different characteristic peaks at the different laser excitation conditions for the same probe molecules. For signals excited at a laser excitation of 532 nm, a highly intense peak is found at 1624 cm -1 , while highly intense peaks of MB are observed at 449 cm -1 (P1), 503 cm -1 (P2), 1399 cm -1 (P3), and 1624 cm -1 (P4) at a laser excitation of 785 nm. The results are consistent with other reports (Gao et al. 2018;Xu et al. 2019a, b, c).
To rule out if there are any effects from the signals of the substrates, the Raman spectra of the substrates were also investigated, see Figure S3. It can be observed in the figure that there are no similar peaks like the characteristic peaks of MB, which confirms that there are no effects from the background. Furthermore, Fig. 3 shows the intensity of the peak at 1624 cm -1 with 532 nm laser excitation and P1-P4 with 785 nm laser excitation which are used to evaluate the SERS enhancement performance. For ZIF-67 and Ag@ZIF-67 coated cotton substrates, it can be observed that the intensity of the MB signals is obviously enhanced, in which the intensity first increases with sputtering time of Ag, then peaks at C-AZ60 (60 s of sputtering) and slightly declines at C-A90 (90 s of sputtering). Nevertheless, it was found that the intensity of the signals continues to increase for samples without the presence of ZIF-67. The results indicated that ZIF-67 has good SERS performance and the deposition of Ag nanoparticles can contribute to increase SERS activity. Meanwhile, the results also showed that the effect of the good SERS performance mainly comes from the presence of ZIF-67.
The EF is an important parameter to assess SERS performance and is derived by Eq. (1) (He et al. 2009;Li et al. 2018): where I sers denotes the SERS intensity of the MB molecules based on the prepared substrates; I 0 refers to the intensity of the MB molecules based on the untreated cotton fabric; N sers denotes the number of MB molecules absorbed on the surface of the SERS substrates, and N 0 represents the number of MB molecules with a certain volume that are laser excited. The details of the calculation process can be found in our previous reports (Xu et al. 2019a, b, c;Xu et al. 2020a, b;Xu et al. 2019a, b, c). It is worth noting that the peak located at 1624 cm -1 is detected in both the spectra obtained under laser excitation of different wavelengths, thus, the intensity of this peak is selected to calculate the EFs, which are then plotted and the curves are shown in Fig. 4. As shown in the figure, the highest EF of 6.25 9 10 6 appears in the C-AZ60 samples with an excitation laser of 785 nm. To compare the different enhancement effects of the samples under different laser excitations, the EFs were calculated and are shown in Fig. 4. The figure shows that the EFs of the different samples with an excitation laser of 532 nm range from 2.2 9 10 6 to 4.51 9 10 6 , while those with an excitation of 785 nm laser vary from 1.06 9 10 6 to 6.25 9 10 6 . For C-ZIF-67, an excitation laser of 532 nm offers better enhancement. After the samples were sputtered with Ag nanoparticles for 5 s, the EFs are almost the same, and those under an excitation laser of 785 nm exceed those under an excitation laser of 532 nm with longer sputtering time of Ag.
To investigate the process that produces these phenomenon, the results are assumed to be the synergistic effect of electromagnetic enhancement (EM) and chemical enhancement (CM) according to our previous work (Xu et al. 2019a, b, c), in which the former is derived from the effect of localized surface plasmon resonances (LSPRs) of the noble metal nanoparticles, while the latter is attributed to the charge transfer between the SERS-active materials and probe molecules (Kaneti et al. 2014;Yang et al. 2017). In order to obtain more insight into the enhancement performance of the EM effect, Raman mapping was carried out to identify the ''hotspots'' that appeared due to the Ag nanoparticles, which is the commonly used explanation of the enhancement effect of noble metal particles . Raman mapping is an image from the generated spectrum collection after recording each and every scanned pixel of the selected region. As shown in Fig. 5a, an area of 50 lm 9 50 lm is selected. The 3D (Fig. 5b) and 2D (Fig. 5c) mapping are obtained with pixels of 100 9 100. The bar color changes from blue to red, where blue means the low intensity and red means the high intensity. The high intensity is attributed to the enhancement effect of the Ag nanoparticles. Therefore, the mapping reflects the distribution of the ''hotspots''. The ''hotspots'' should be evenly distributed according to the SEM analysis in that the Ag nanoparticles are well distributed on the surface of cotton fabric. However, the excitation laser is focused on the surface of the sample at an angle, and the surface of the fabric appears to be interlaced. Thus, not all of the areas of the fabric surface can be adequately excited and show high intensity at the same time, and the interlaced fibers can also be observed on the 2D mapping image. To determine how the SERS performance would be affected by the sputtering time of Ag, electromagnetic images were obtained by applying the finite-difference time-domain (FDTD) method to investigate the distribution of the electrical field around the Ag nanoparticles after polarized by the used laser. The details of the calculation and simulation are described in the electronic supporting information.
Figures 5d, e show the electric field distributions around the Ag nanoparticles on the surface of the ZIF-67 particles under a laser excitation of 532 nm and 785 nm. The color of the inserted bar can be observed to be changed from blue to red, which means that the electric field changes from weak to strong. It can be observed that the electric field between the Ag nanoparticles under a laser excitation of 785 nm is higher than it was excited by a 532 nm laser, which indicates that the irradiation of a 785 nm laser will result in stronger polarization between the Ag nanoparticles. Simulated electric field distributions of the Ag nanoparticles without ZIF-67 are also provided to show whether the ZIF-67 has any effect, as shown in Figs. 5f, g. It was also found that a 785 nm laser has better polarization ability; meanwhile, the intensity of the electric field was found to be lower than that of the Ag nanoparticles distributed on ZIF-67, which shows that ZIF-67 affects the polarization of the electric field.
CE is mainly derived from ZIF-67 in this study; therefore, the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) levels of ZIF-67 are calculated to be 4.31 eV and -5.91 eV, respectively, according to the valance band XPS ( Figure S4). As shown in Fig. 5h, the charge transition from the LUMO of ZIF-67 (-5.91 eV) to the HOMO of MB (-4.55 eV, which is lower than the HOMO of ZIF-67) can be easily achieved once ZIF-67 is excited by the used light. As well known, the material can only be excited at an energy level that is equal or higher than its band gap energy. Thus, the band gap energy of ZIF-67 was also calculated ( Figure S4), which is 1.6 eV. The photon energy of 532 nm is 2.33 eV, while that of 785 nm is 1.58 eV. Obviously, a 532 nm laser beam will easily excite ZIF-67, thus resulting in a charge transition, and realizes the charge transfer between the band edges of ZIF-67 and the LUMO of the probe molecules to cause an enhancement of Raman signals, which is in accordance with the SERS performance results.
Additionally, it was found that the Ag nanoparticles loaded on ZIF-67 show a better SERS performance, which might contribute to both EM and CM effects. To properly illustrate this process, the module of the charge transfer between the Ag nanoparticles and ZIF-67 was established, as shown in Fig. 5i. Furthermore, this module was established based on the firstprinciples theory, in which a charge transfer might occur between the Ag nanoparticles and ZIF-67 due to different work functions. The work function of the Ag nanoparticles and ZIF-67 was calculated to be 4.26 eV and 6.48 eV versus the normal hydrogen electrode (NHE) (Birke and Lombardi 2018). Thus, there tends to be charge transfers between the Ag nanoparticles and ZIF-67, which will equilibrate the Fermi level to reach a balance, when the Ag nanoparticles are in directly contact with ZIF-67. Consequently, polarization and charge transfer will occur with the irradiation of the used laser, thus resulting in the enhancement of the local electromagnetic field at the interface between the Ag nanoparticles and ZIF-67, and the SERS performance will be dramatically improved. A comparison with the experimental results shows that the simulation is in good agreement with the original assumption, which demonstrates that the excellent SERS performance is derived from both electromagnetic and chemical enhancements.
SERS sensitivity evaluation of C-Ag@ZIF-67 substrates MB solutions with a concentration that ranged from 10 -14 M to 10 -3 M were used to evaluate the sensitivity of the prepared flexible SERS substrates. The obtained spectra are shown in Fig. 6a, and the spectra of the MB with a concentration that ranges from 10 -14 M to 10 -12 M are also provided in Fig. 6b for clarity. On the C-Ag@ZIF-67 substrates, MB can be detected even at a low concentration of 10 -14 M, which was not easily done in other research works. Figure 6c shows a comparison of the LOD and EF results between this study and those of other studies. It can be seen that the performance of the prepared SERS substrate in this study is higher than the others, thus indicating its excellent detection sensitivity Guselnikova et al. 2019;Li et al. 2019a, b;Li et al. 2019a, b;Yang et al. 2020). Meanwhile, the highest EF of current research is also higher than previous research that modifying ZIF-67 with Ag ? , indicating the advantages of surface plasmon resonance effect generated by Ag nanoparticles (Xu et al. 2020a, b).

Evaluation of signal reproducibility and stability
The SERS performance of the C-Ag@ZIF-67 substrates was also examined through their signal reproducibility and stability, which plays an important role in practical applications. The SERS spectra of the prepared substrates absorbed with a 10 -3 M MB solution were collected by using 30 different locations on the same sample to evaluate the reproducibility, as shown in Fig. 7a. The variations in the intensity of the peak at -1624 cm -1 were further established and the relative standard deviation (RSD) was calculated to evaluate the reproducibility. It is observed in Fig. 7b that the intensity of each peak fluctuates slightly with RSD values that range from 2.7 to 3.8%, thus indicating a good reproducibility.
To evaluate the signal stability of the C-Ag@ZIF-67 substrates, Raman spectra of the MB molecules on the SERS substrate were recorded in the same region with changes in time; that is, for 10 min at intervals of 1 min, as shown in Fig. 7c. It can be observed that characteristic peaks of MB are the same during the process. The intensity of the characteristic peaks is individually plotted to determine the variation during continuous laser irradiation (Fig. 7d). The RSD values of the different peaks are also calculated to assess the stability. The RSD values of P1-P4 are 3.7%, 5.7%, 4.2% and 5.1%, respectively, which indicate a relatively high stability of the signals.
Detection and identification of phenol-soluble modulin peptides Staphylococcus aureus (S. aureus), especially superbugs, is a severe threat causing infections. PSM peptides, which are produced by S. aureus, are the main virulence factor that causes inflammation, lysis of human cells by altering the cell cycle and structuring biofilm during the treatment of infections (Cheung et al. 2014;Schwartz et al. 2012). Its PSM family members, including PSMa1, PSMa3, and PSMa4, are formed from four peptides with about 20-residues in length. These PSMs play an important role in contributing to the emergence of methicillin resistant S. aureus, which is responsible for infections that are resistant to treatment, otherwise known as superbugs (Wang et al. 2007). Although many studies have been done to investigate the composition and structure of PSMas, there is still an absence of a rapid detection method that can identify their presence during clinical diagnosis and treatment. Therefore, the current structured SERS substrate in this study is investigated to determine its ability to identify PSMas based on provided information (Salinas et al. 2018;Tayeb-Fligelman et al. 2017;Yao et al. 2019). Here, PSMa1, PSMa3, and PSMa4 are selected as the target objects. Solutions made of PSMas with a concentration of 1 mM/L were prepared, and 10 lL of the PSM Fig. 7 Plots of a and b signal reproducibility, and c and d signal stability solution was dripped onto the surface of C-AZ60, followed by testing through Raman spectroscopy.
The structure, peptide sequence and corresponding Raman spectrum of PSM a1, a3 and a4 are shown in Fig. 8. As shown in the figure, the different PSMs have a different permutation structure with similar amino acids. Without referring to the obtained SERS spectra, the PSMs can only be identified using protein Fig. 8 Structure, peptide sequence and Raman spectrum of a PSM a1, b a3, c a4, d local Raman spectra ranged from 760 cm -1 * 1060 cm -1 , and e local Raman spectra ranged from 1330 cm -1 * 1540 cm -1 sequencing technology to analyze their different amino acid combinations. However, such a sequencing analysis is a complex and time-consuming process which requires the use of special equipment and can only be completed by a professional technician. Therefore, it is not conducive to rapid clinical diagnoses. Here, it can be observed in the SERS spectra that different PSMs have been detected and identified by the different peptide fingerprints as marked in the Fig. 8a-c with different color for different PSMs at the same range of Raman shift. For clarify, the spectra ranged from 760 cm -1 * 1060 cm -1 and 1330 cm -1 * 1540 cm -1 were individually presented in Fig. 8d, e, respectively. As shown in Fig. 8d, the difference of fingerprints of different PSMs can be told that there is a peak located around 829 cm -1 for PSM a4 and two peaks located around 829 cm -1 and 1003 cm -1 for PSM a3, while there is no obvious peaks for PSM a1. Meanwhile, as shown in Fig. 8e, PSM a4 shows a peak located around 1458 cm -1 , PSM a3 shows two peaks located around 1396 cm -1 and 1455 cm -1 , while PSM a1 contains four peaks located around 1358 cm -1 , 1400 cm -1 , 1460 cm -1 and 1519 cm -1 respectively. However, it is difficult to pinpoint the specific characteristics of the SERS spectra of these peptide fingerprints because the obtained SERS spectra are not only affected by the composition, but also subjected to the 2D or 3D structures of these peptides (Guerrini et al. 2013). Despite this, this study has confirmed that SERS technology can be applied to identify the PSMs of S. aureus by using a label-free technique that does not require adding markers/tags or antibodies. This study therefore provides important experimental evidence to increase the potential applications of SERS technology, which has never been reported in the literature to date.

Conclusion
In summary, a flexible SERS substrate, which shows great potential for detecting and identifying PSMs, has been successfully constructed with a simple method within two steps. ZIF-67 was first assembled on the surface of cotton fabric, and Ag nanoparticles were deposited onto the surface of ZIF-67 by modifying the particle size and gap between the particles by varying the deposition time. Then the SERS performance was evaluated via Raman spectroscopy of the MB molecule. With different sputtering times to deposit the Ag nanoparticles, we found that the C-ZIF-67 sample sputtered with Ag nanoparticles for 60 s had the best response to the MB molecule (as the probe) during Raman measurements. Besides, the C-AZ60 SERS substrate also demonstrated the ability to trace the signals of PSMs, which means that this substrate has excellent potential for detecting and identifying bacterial toxins during clinical diagnoses and treatment.