Enhanced thermal stability of boron nitride-coated Au nanoparticles for surface enhanced Raman spectroscopy

Accepted Manuscript Enhanced thermal stability of boron nitride-coated Au nanoparticles for surface enhanced Raman spectroscopy Xinxin Yu, Ranran Cai, Jinlong Jiao, Yunlong Fan, Qiang Gao, Junwen Li, Nan Pan, Mingzai Wu, Xiaoping Wang PII:

S0925-8388(17)33270-X

DOI:

10.1016/j.jallcom.2017.09.220

Reference:

JALCOM 43276

To appear in:

Journal of Alloys and Compounds

Received Date: 1 June 2017 Revised Date:

18 September 2017

Accepted Date: 20 September 2017

Please cite this article as: X. Yu, R. Cai, J. Jiao, Y. Fan, Q. Gao, J. Li, N. Pan, M. Wu, X. Wang, Enhanced thermal stability of boron nitride-coated Au nanoparticles for surface enhanced Raman spectroscopy, Journal of Alloys and Compounds (2017), doi: 10.1016/j.jallcom.2017.09.220. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT Enhanced thermal stability of Boron nitride-coated Au nanoparticles for surface enhanced Raman spectroscopy Xinxin Yua, ξ,∗, Ranran Caia, ξ, Jinlong Jiaoa, Yunlong Fana, Qiang Gaoa, Junwen Lic, Nan Panb, Mingzai Wua,∗ and Xiaoping Wangb,c School of Physics and Material Science, Anhui University, Hefei 230601, PR China

b

Hefei National Laboratory for Physical Sciences at the Microscale, University of

Science and Technology of China, Hefei 230026, P R China c

RI PT

a

Department of Physics, University of Science and Technology of China, Hefei

SC

230026, P R China ∗

: Xinxin Yu. E-mail: [email protected], Fax: 86-551-63861813

M AN U

Mingzai Wu. E-mail: [email protected], Fax: 86-551-63861813 ξ

: These two authors contribute equally to this work.

Abstract

During the past decades, researchers have made great efforts towards ideal surface

TE D

enhanced Raman spectroscopy (SERS) substrate, which is supposed to be reliable, sensitive and reusable. Boron nitride (BN) is electrically insulator and can sustain up to 600°C, which makes it possible that BN acts as a barrier layer to eliminate

EP

metal-induced disturbances for SERS and provide more reliable signals of the analyte. Here, we synthesis BN via chemical vapor deposition (CVD) method, investigated its

AC C

oxidation resistance and prepare BN-coated gold nanoparticles (BN/Au). Although the electromagnetic enhancement decreases exponentially with distance, the SERS performance of BN/Au substrate is comparable to bare Au after being annealed at proper temperature. Moreover, Cycle tests results indicated that the introduction of BN as a coating layer could improve the stability of the SERS substrate compared to bare Au nanoparticles film. Our SERS substrate of BN/Au was obtained via CVD method, which could ensure scalable synthesis and applications. Keywords: Boron nitride; chemical vapor deposition; surface-enhanced Raman spectroscopy (SERS); stability 1

ACCEPTED MANUSCRIPT 1. Introduction As a non-destructive analytical technique, surface enhanced Raman spectroscopy (SERS) could provide the special chemical fingerprints of the analyte, showing a wide range of applications in physics, chemistry, biology and medicine areas[1-3]. The ultra

RI PT

high sensitivity is based on well designed Raman enhancement substrate[4]. During the past decades, much effort has been put on the development of gapped noble metal structure as SERS substrate for its high sensitivity to single molecule level[5-6]. However, the catalysis property of the noble metal and metal-molecule interactions

SC

could induce transformation of the molecule structure, which resulted in instable and distorted SERS spectra[7]. To get accurate and cleaner Raman signals, a shell-isolated

M AN U

structure was proposed for the coating of noble metal nanoparticles with an inert shell or graphene[7-9]. The SERS performance of the passivated metal substrate is proved to be comparable to bare metal substrate[10]. More importantly, as it is free from metal-molecule interactions and photo-induced damage, cleaner vibrational information could be obtained. The inert shell could be SiO2, Al2O3 or polymer.

TE D

Considering the electromagnetic exponential attenuation with distance, a thickness of 2~4 nm is demanded to ensure a relative high electromagnetic field on the surface of the shell[7], which make the synthesis of pinhole-free inert shell with such thickness

EP

very difficult. Although wafer scale graphene cloud be produced by a much easier method, such as chemical deposition method[11], the reusability of graphene-based SERS substrate is limited by its low oxidation resistance[12].

AC C

Two dimensional (2D) sheet-like boron nitride (BN) is atomically thin electrical

insulator with good mechanical strength and thermal conductivity[13]. It is reported that mechanically exfoliated BN could be thermally stable up to 800°C in air[14], which is of great significance to the removal of adsorbed analyte molecules by heating treatments. Besides, it has been reported that BN has Raman enhancement effect from the interface dipole interaction between BN and adsorbed molecules[15]. This implies that noble metal nanoparticles coated with BN might combine the chemical enhancement and electromagnetic enhancement[16-17], and is supposed to 2

ACCEPTED MANUSCRIPT show a higher Raman enhancement factor. Although BN possess many advantages over inert shell and graphene, the thermal stability study of its Raman enhancement performance is still limited and are mainly base on mechanically exfoliated BN[18]. Herein, we prepared BN-coated Au nanoparticles film (BN/Au) via CVD method

RI PT

followed by annealing treatment, and study its thermal stability for the first time when worked as SERS substrate. After being annealed at proper temperature, the thermal stability of Raman performance could be obviously improved; meanwhile, its Raman enhancement performance is still comparable to bare Au nanoparticles film. As a

eliminate metal-induced disturbances for SERS.

M AN U

2. Experimental details

SC

coating layer of Au nanoparticles, the introduction of BN layer would help to

2.1 Synthesis of hexagonal boron nitride

Few layer hexagonal boron nitride (h-BN) was grown on a 25 µm thick polycrystalline copper foil (Alfa Aesar, No.46365) by atmosphere pressure method (APCVD). Prior to growth, the copper foil was electrochemically polished at 3 V for

TE D

180 s to remove surface natural oxide layer and surface contaminants. After being cleaned with ethanol and acetone, the Cu foil was transferred to a two-heating-zone system. Borazane powder was used as the source of boron and nitrogen elements for

EP

its stability and non-toxicity. The temperature of the heating zone is labeled as T1 for the precursor zone and T2 for the growth zone. T1 was set to 95°C to ensure thermal decomposition of borazane and the growth temperature (T2) was set to 1050°C. The

AC C

distance between precursor and substrate is set to 40 cm. In a typical growth process, the T1 was heated to 95°C after T2 reached 1050°C. The product of borazane was carried to the growth zone by Ar/H2 (180 sccm Ar, 30 sccm H2). After growth for 20 min, the sample was cooled rapidly to room temperature by simply open the cap with the protection of Ar/H2 (180 sccm Ar, 5 sccm H2). Then, PMMA (950 K A4, 4 wt%, Micro Chem) was coated on the surface of BN at 3000 rpm for 40 s by spin coating method. After the underlying copper foil was etched by 1g/ml FeCl3 solution, the resulting BN/PMMA film was thoroughly washed and transferred to SiO2/Si or Au film. The thickness of SiO2 is about 90 nm. After the remove of PMMA with acetone, 3

ACCEPTED MANUSCRIPT the BN was washed with ethanol and water thoroughly. 2.2 Synthesis of BN/Au Firstly, Au film with a thickness of 15 nm and 20 nm was deposited respectively on SiO2/Si by thermal evaporation method at a rate of 0.1 nm/s. Then, BN was

RI PT

transferred to the Au film with proper thickness (15 nm or 20 nm). The transfer method is the same as section 2.1. Before characterization of Raman enhancement performance, Au film and the BN-coated Au film was annealed at a series of temperature to transform the Au film to Au nanoparticle film, and then BN/Au

SC

(BN-coated Au nanoparticle film) or Au (bare Au nanoparticle film) was obtained. For the Raman characterization, the resulting BN/Au film was transferred to 1×10-6 M

M AN U

R6G aqueous solution and soaked for 0.5 h to adsorb R6G molecule fully. In our experiments, BN/Au15 and BN/Au20 represent the thickness of the Au films is 15 nm and 20 nm respectively. 2.3 Characterization

The morphology of the products were characterized by Field emission-scanning

TE D

electron microscopy (SEM) at 5.0 kV (Sirion200). The Raman enhancement performance was characterized using a Renishaw Raman microscope equipped with 532 nm solid state laser excitation source (inVia-Reflex, Renishaw, U.K.). For the characterization of Raman enhancement performance, the laser power density is set to

EP

1.5 mW/µm2 and a integration time of 10 s was used. To obtain the Raman spectrum of pure BN, the laser power is set to 15 mW/µm2 and a integration time of 60 s was

AC C

used. The AFM image and height profile were collected under a dynamic force microscope mode with an SPA 300HV microscope (Seiko Instruments, Inc.).

3. Results and discussion BN was synthesized on the Cu foil by APCVD method[19]. After growth for 15min, hexagonal crystal BN with a size of 5~10 µm formed on the surface of copper foil and then transferred to 90 nm SiO2 coated Si wafers. Although atomically-thin BN exhibits little optical contrast on standard oxidized Si wafers (≈300 nm SiO2), whereas it can be easily identified on the much thinner layer of SiO2 (≈80±10 nm) by 4

ACCEPTED MANUSCRIPT optical microscopy[20]. The hexagonal crystal domain with a size of several micrometer indicates a relative high quality of the BN layer. Figure 1a shows scanning electron microscopy (SEM) images of the hexagonal BN on the SiO2/Si substrate. When the growth time was increased to 20 min, the copper foil surface was

RI PT

fully covered with BN, which could be verified by simply heating the copper foil in air. After heating for 2 min, the bare copper foil was oxidized and the color changed to reddish brown; while the copper foil covered with h-BN would keep bright, as shown in the inset of Figure 1b. Raman spectrum of h-BN was used to investigate the

SC

thickness of the h-BN. The measurement result is shown in Figure 1b, showing a relatively weak G band at ~ 1370 cm-1 corresponding to the in-plane ring vibration of

M AN U

h-BN (E2g vibration mode). It has been reported that the peak position is dependent of the film thickness. The peak located at 1370 cm-1 (G band) indicates a relatively small thickness of the h-BN layer[20-22], which is further characterized using AFM. As gas and water molecules in air could unavoidably adsorbed between the BN and the Si/SiO2 substrate during the experimental process, there would be an additional

TE D

thickness of several angstroms. Considering the existence of such an interlayer spacing between the BN and substrate, a measured thickness of 1.84 nm (as shown in Figure 1c) probably corresponds to few-layer BN (3~4 layers), which is acceptable as

AC C

EP

coating layer of noble metal nanoparticles.

5

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Figure 1. (a) SEM image of h-BN after growth for 15min; (b) Raman spectrum of h-BN on the SiO2/Si substrate. The inset is the copper foils covered with BN (left) and bare copper foil (right) after heating. (c) AFM image and the corresponding height

TE D

profile obtained along the green line in the AFM morphology image. The mechanically exfoliated h-BN has been reported to show Raman enhancement effect[15]. However, its size is too small for SERS application. Comparatively, CVD

EP

method shows great advantages over mechanical exfoliation for the preparation of large area BN sheets with atomic thickness. Figure 2a shows the Raman spectrum of R6G molecules adsorbed on pure BN, and only two peaks at 620 cm-1 and 1592 cm-1

AC C

can be observed. It is reported that SiO2 as a spacer could improve the intensity of the Raman spectrum in some cases, for it could induce favorable interference between the direct and reflected beams[23]. However, it is supposed that the enhancement of Raman signal on the surface of BN is probably not due to the SiO2. After the deposition of R6G molecules on the surface of SiO2, there is only strong fluorescence background and no Raman signal of the molecule could be found. With the BN, the appearance of Raman signals suggests that there may be an enhancement effect. However, this enhancement is relatively small. Although pure BN is not a proper SERS substrate, it is an insulator with good stability and can serve as coating layer to 6

ACCEPTED MANUSCRIPT eliminate the interaction between molecule and metal nanoparticles, favoring the enhancement of detection accuracy. Most importantly, BN nanosheets could be prepared on scale with atomic thickness via CVD method, which is uniform and large in size.

RI PT

In addition, as the coating layer of noble metal nanoparticles for SERS application, h-BN is desired to be thermally stable, which can ensure the removal of the adsorbed molecules when heated in air. Mechanically exfoliated h-BN has been reported to show good thermal stability, however, the stability of h-BN prepared by CVD is to be

SC

examined. Thus oxidation tests were conducted for h-BN prepared by CVD. At a series of temperature, we heated the h-BN for 10 min respectively in open air. Figure

M AN U

2b shows the Raman measurement results for h-BN annealed in air at different temperatures. The characteristic peak of BN is at about 1370 cm-1. The Raman peak at about 1450 cm-1 can be attributed to third-order scattering from Si, and the Raman peak at about 1560 cm-1 is supposed to originate from the ambient molecular oxygen, which is detectable because of the higher laser powers and longer measurement

TE D

time[24-25]. After being annealed at 400 °C, h-BN shows the enhanced ratio of signal to noise, which probably comes from the removal of impurity. With the annealing temperature increases to 600 °C, the characteristic peak at about 1370 cm-1 remain

EP

unchanged, which is labeled by red star. For h-BN annealed at 800°C, the characteristic peak intensity dramatically reduces. Obviously, few-layer BN is quite

AC C

resistant to oxidation up to 600°C and show oxidative etching at 800°C. These results imply that heating h-BN in oxygen-containing gas below 600°C is not supposed to introduce defects. Compared with the oxidation temperature of graphene (250°C), few-layer BN is more resistant to oxidation. Combined its low cost and simple processing, the scalable production of BN via CVD can find promising application in SERS as the coating layer of metal nanoparticles.

7

RI PT

ACCEPTED MANUSCRIPT

Figure 2. (a) Raman spectrum of R6G molecules adsorbed on pure BN. The vibrational modes at 900~1000 cm-1 is from the SiO2/Si substrate. (b) Raman spectra

SC

of BN before and after oxidation at different temperatures for 10 min, respectively.

M AN U

As "hot spot" usually forms between metal nanoparticles, Au (bare Au nanoparticle film) and BN/Au (BN-coated Au nanoparticle film) were prepared by annealing, and the morphology evolution of Au films with 15 and 20 nm in thickness after annealing treatment were also examined. Figure 3 (a), (b), (c) and (d) are SEM images of BN/Au20, which was annealed at 400, 500, 600 and 700°C for 10 min respectively.

TE D

The annealing treatment results in labyrinth of Au film, which is not supposed to yield ideal SERS performance. Comparatively, the same annealing treatment does not obviously change the morphology (granular film) of BN/Au15, as shown in Figure 3

EP

(e), (f), (g) and (h). In addition, without the protection of BN, these two Au film suffer obvious weight loss with the increase of the annealing temperature, especially for Au

AC C

film with 20 nm in thickness, as shown in Figure 3 i-p.

8

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Figure 3. (a), (b), (c) and (d) are SEM images of BN/Au20 film annealed at 400, 500,

TE D

600 and 700°C respectively; (e), (f), (g) and (h) are SEM images of BN/Au15 film annealed at 400, 500, 600 and 700°C respectively. (i), (j), (k) and (l) are SEM images of Au film with 20 nm in thickness annealed at 400, 500, 600 and 700°C respectively;

EP

(m), (n), (o) and (p) are SEM images of Au film with 15 nm in thickness annealed at 400, 500, 600 and 700°C respectively.

AC C

Then, the performances of BN/Au15 and BN/Au20 as substrate for Raman enhancement were investigated and the results are shown in Figure 4. Obviously, the intensity of Raman signals for BN/Au15 are much stronger than those for BN/Au20 annealed at the series of temperatures between 400-700°C, suggesting the formation of hot spots for sample BN/Au15 after annealing treatment. Thus, in the following experiment, sample BN/Au15 and Au with 15 nm in thickness are chosen for the comparative study of SERS performances.

9

RI PT

ACCEPTED MANUSCRIPT

SC

Figure 4. The Raman of R6G molecules adsorbed on the surface of BN/Au15 and BN/Au20, which has been annealed at a series of temperatures for 30min. The

M AN U

annealing temperature is 400°C, 500°C, 600°C and 700°C in (a), (b), (c) and (d), respectively.

To further improve the Raman enhancement performance, we annealed BN/Au film and Au film under a series of temperatures and investigate the influence of the

TE D

annealing temperature on the Raman enhancement performance. For each sample, three Raman spectra were obtained and averaged to exclude occasional or accidental factors. The results can be found in Figure 5(a-d). The Raman spectra of the adsorbed molecules obtained on BN/Au15 annealed at 400°C (400-Au/BN) is similar to that on

EP

pure BN, showing weak peaks lying at 620 cm-1 and 771 cm-1 and 1360 cm-1. Their intensities are much weaker than those on Au after annealing at 400°C. With the

AC C

annealing temperature increases, the difference between them decreases. When the annealing temperature increases to 600°C, the intensity of Raman spectra obtained on BN/Au15 is comparable to that on Au. When the annealing temperature is further increased to 700°C, the intensity of Raman spectra obtained on BN/Au15 is not compromised at all as compared with that on bare Au film. As the annealing temperature increases, the Raman signal intensity on bare Au film decreases, while the Raman signal intensity on BN/Au increases. Hence, the Raman enhancement of BN/Au15 could be obviously improved via annealing, which might benefit from a tighter contact between BN and Au nanoparticles. Besides, the Raman spectra of R6G 10

ACCEPTED MANUSCRIPT on these different BN/Au substrates maintains the same undisturbed fingerprint and the enhancement of Raman vibration modes is not selective. An explanation is proposed as follows. Before the annealing, BN sheet was placed on the top of plasmonic Au nanoparticles. Molecules adsorbed on the surface of BN might be far

RI PT

from the electromagnetic "hot spots", which are buried below BN at a distance away, as shown in Figure 5(e). Thus the Raman spectrum of the adsorbed molecules obtained on BN/Au15 annealed at 400°C is similar to that on pure BN. After being annealed at a certain temperature, the few-layer BN would follow the profile of the

SC

underlying nanoparticles better. Thus the molecules adsorbed on the surface of BN could be just very near to "hot spot", resulting in further enhancement of Raman

M AN U

signals of the adsorbed molecules by the electromagnetic "hot spots", as shown in Figure 5(f). Comparatively, Au shows much different SERS behavior from that of BN/Au15. After annealing at 400°C, the Raman signals intensity of the adsorbed molecules on Au are greatly enhanced. With the further increase of the annealing temperature, the signal intensity decrease obviously, as shown in Figure 5(b, c and

TE D

d),which indicate that Au cannot withstand annealing at high temperature. All of the above results show that BN/Au15 can work as SERS substrate and the introduction of

AC C

EP

BN can improve its SERS stability.

Figure 5. Raman spectra of R6G molecules adsorbed on BN/Au and Au which has been annealed respectively at 400°C, 500°C, 600°C and 700°C in (a) , (b), (c) and (d). The thickness of Au is 15 nm for Au/BN and bare Au film. (e) and (f) are the schemes of the Raman enhancement before and after annealing respectively. 11

ACCEPTED MANUSCRIPT

As shown in Figure 2, CVD BN also has good oxidation resistance, which is helpful to protect metal particles and can ensure the effective removal of adsorbed molecules by heating in air. The introduction of atomically thin BN is supposed to

RI PT

make SERS substrates highly reusable. For further investigation, cycle test was conducted on both bare Au and BN/Au films. Before cycle test, both have already been annealed at 600 °C to transform the Au films into Au nanoparticle films. In each cycle process, the substrates were heated at 300 °C in air for 10 min to remove any

SC

organic molecule, and then soaked in 1×10-6 M R6G aqueous solution for 0.5 h to adsorb R6G molecules again for the next cycle of SERS characterization. Raman

M AN U

measurements were performed under the same condition for both substrates, and the intensity of the peak at 620 cm-1 in each cycle is recorded and plotted in Figure 6. In the first cycle, BN/Au15 shows weaker Raman signal intensity than bare Au. However, in the second cycle, the Raman signal intensity obtained on BN/Au15 is significantly enhanced and increases to twice of the original value, which is even

TE D

stronger than that originally obtained on bare Au. At the same time, the intensity of the Raman signal obtained on bare Au gradually reduces during the cycles, which is obviously weaker than that on BN/Au15 beginning from the second cycle. After the

EP

fourth cycle, the Raman intensity obtained on BN/Au15 is still 1.3 times stronger than that obtained on bare Au. These results indicate that the BN/Au could provide stronger and more durable Raman enhancement performance than bare Au, which is

AC C

extremely desirable for real applications (where repeat detection is certainly required). Hence, this BN/Au film can serve as a reusable, more durable, and even undisturbed (precise) SERS substrate which is better than bare metal films.

12

RI PT

ACCEPTED MANUSCRIPT

SC

Figure 6. Cycle test of BN/Au15 and bare Au film. Raman signal intensities of R6G molecules adsorbed on the surface of BN/Au15 is compared with that on bare Au film

M AN U

for four cycle. The number represent the time of cycle. 4. Conclusions

In this study, we prepared atomically thin BN sheets by chemical vapor deposition and investigate its oxidation resistance. Based on Raman characterization, the as-prepared BN sheets could sustain annealing treatment up to 600°C without obvious

TE D

change. After being transferred to Au film, the annealed BN/Au showed comparable Raman performance and improved stability compared to bare Au nanoparticles film. Our method paved the path for the scalable preparation of BN/Au as a reusable,

AC C

stability.

EP

durable, and even undisturbed substrate for SERS application with high thermal

Acknowledgements

This work was supported by the National Natural Science Foundation of China (11404001, 11374013, 51672001), the PhD Start-up Fund of Anhui University (No.33190209).

References [1] K. Kneipp, Y. Wang, H. Kneipp, L.T. Perelman, I. Itzkan, R. Dasari, et al., Physical Review Letters 78 (1997) 1667-1670. [2] L.L. Qu, Y.Y. Liu, M.K. Liu, G.H. Yang, D.W. Li, H.T. Li, ACS Appl Mater Interfaces

(2016). 13

ACCEPTED MANUSCRIPT [3] X.M. Qian, X.H. Peng, D.O. Ansari, Q. Yin-Goen, G.Z. Chen, D.M. Shin, et al., Nature Biotechnology 26 (2008) 83-90. [4] Z.Q. Tian, B. Ren, D.Y. Wu, J. Phys. Chem. B 106 (2002) 9463-9483. [5] G.A. Baker, D.S. Moore, Analytical and Bioanalytical Chemistry 382 (2005) 1751-1770.

3448-3457.

RI PT

[6] L.W. Nien, M.H. Chien, B.K. Chao, M.J. Chen, J.H. Li, C.H. Hsueh, J. Phys. Chem. C 120 (2016)

[7] J.F. Li, Y.F. Huang, Y. Ding, Z.L. Yang, S.B. Li, X.S. Zhou, et al., Nature 464 (2010) 392-395. [8] Y.M. Liu, Y. Hu, J. Zhang, J. Phys. Chem. C 118 (2014) 8993-8998.

[10] W.G. Xu, N.N. Mao, J. Zhang, Small 9 (2013) 1206-1224.

SC

[9] W.G. Xu, J.Q. Xiao, Y.F. Chen, Y.B. Chen, X. Ling, J. Zhang, Adv. Mater. 25 (2013) 928-933.

M AN U

[11] Z. Yan, J. Lin, Z.W. Peng, Z.Z. Sun, Y. Zhu, L. Li, et al., ACS Nano 6 (2012) 9110-9117. [12] Y.C. Lin, C.C. Lu, C.H. Yeh, C.H. Jin, K. Suenaga, P.W. Chiu, Nano Lett. 12 (2012) 414-419. [13] L.H. Li, Y. Chen, Adv Funct Mater 26 (2016) 2594-2608.

[14] L.H. Li, J. Cervenka, K. Watanabe, T. Taniguchi, Y. Chen, ACS Nano 8 (2014) 1457-1462. [15] X. Ling, W.J. Fang, Y.H. Lee, P.T. Araujo, X. Zhang, J.F. Rodriguez-Nieva, et al., Nano Lett. 14

TE D

(2014) 3033-3040.

[16] X. Ling, L.M. Xie, Y. Fang, H. Xu, H.L. Zhang, J. Kong, et al., Nano Lett. 10 (2010) 553-561. [17] X.X. Yu, H.B. Cai, W.H. Zhang, X.J. Li, N. Pan, Y. Luo, et al., ACS Nano 5 (2011) 952-958.

EP

[18] Q.R. Cai, L.H. Li, Y.L. Yu, Y. Liu, S.M. Huang, Y. Chen, et al., Phys. Chem. Chem. Phys. 17 (2015) 7761-7766.

AC C

[19] K.K. Kim, A. Hsu, X. Jia, S.M. Kim, Y. Shi, M. Hofmann, et al., Nano Lett 12 (2012) 161-6. [20] R.V. Gorbachev, I. Riaz, R.R. Nair, R. Jalil, L. Britnell, B.D. Belle, et al., Small 7 (2011) 465-8. [21] R.Y. Tay, M.H. Griep, G. Mallick, S.H. Tsang, R.S. Singh, T. Tumlin, et al., Nano Lett 14 (2014) 839-46.

[22] M.H. Khan, Z.G. Huang, F. Xiao, G. Casillas, Z.X. Chen, P.J. Molino, et al., Sci Rep 5 (2015) 8. [23] A. Baptiste, A. Bulou, J-F. Bardeau, J. Nouet, and A. Gibaud, Langmuir 20 (2004) 6232-6237. [24] S. Caneva, R.S. Weatherup, B.C. Bayer, B. Brennan, S.J. Spencer, K. Mingard, et al., Nano Lett. 15 (2015) 1867-1875. [25] P. G. Spizzirri, S. Rubanov, E. Gauja, S. Prawer, Mater. Forum. 34(2008) 161−166.

14

ACCEPTED MANUSCRIPT Highlights 1. The CVD BN is resistant to oxidation and is promising as a coating layer. 2. BN/Au for undisturbed SERS characterization was constructed based on CVD BN. 3. The SERS performance of BN/Au can be controlled by annealing

RI PT

4. BN/Au is more durable to high temperature compared to Au. 5. BN/Au can be used as a reusable, durable, and even undisturbed substrate for

AC C

EP

TE D

M AN U

SC

SERS.