Zhixiang Tao, Wei Zhao, Shang Wang, Boyu Zhao, Rushuai Hua, Ji Qin, and Zongwei Xu,a)

ABSTRACT Raman spectroscopy is a type of inelastic scattering spectroscopy that is widely used in determining and analyzing molecular structure.It also has a number of practical applications in evaluating food safety,medicine,and forensics.The Raman spectral signal is weak,but the development of the surface-enhanced Raman scattering (SERS) technique has overcome this problem and led to further developments in Raman spectroscopy.This paper describes a fundamental study of the use of focused ion beam (FIB)direct writing for preparing gold substrates for SERS.Molecular dynamics and Monte Carlo simulation methods are used to investigate the damage induced by gallium ion implantation of a gold substrate.Based on characterization by x-ray photoelectron spectroscopy (XPS) and scanning electron microscopy,the mechanism by which ion implantation and annealing influenc the damage induced by a gallium FIB is analyzed.After annealing at 350○C,a mixture of metallic gallium,its oxide Ga2O3 conforming to the stoichiometric ratio,and its sub-stable oxide (Ga2Ox) in sub-stoichiometric ratio precipitated on the surface are detected by XPS.Annealing treatment can effectively reduce the effect of gallium ion implantation on a SERS substrate fabricated by FIB direct writing.

KEYWORDS Raman spectroscopy,Focused ion beam,Annealing,X-ray photoelectron spectroscopy

I.INTRODUCTION

Raman spectroscopy is an analytical method based on the inelastic scattering effect discovered by the Indian scientist C.V.Raman.It analyzes scattering spectra with frequencies different from that of the incident light to obtain information about molecular vibration and rotation and apply it to studies of molecular structure.The intensity of the Raman signal is low,but Fleischmannet al.1discovered in 1974 that the Raman signal of pyridine molecules adsorbed on the surface of a rough silver electrode was about 104–106times stronger than that in solution,and this surface enhancement effect was named surfaceenhanced Raman scattering (SERS).With its outstanding advantages such as high sensitivity and the ability to detect substances in single molecular layers,SERS has attracted great interest from physicists and chemists and has undergone rapid development,effectively promoting the development of Raman spectroscopy itself.2–7

Analysis of the localized surface plasmon resonance (LSPR)effect based on metals such as gold,silver,and copper is vital for the optimal design and fabrication of SERS substrate structures.Effective processing of nanogap hotspot structures is the key to obtaining SERS substrates with high enhancement performance.The focused ion beam(FIB)technique can realize the fabrication of micro-and nanoscale structures by accelerating and focusing an ion beam from an ion source[usually a gallium(Ga)metal ion source]through an ion optical system.8The FIB direct writing nanofabrication technique is an important method for preparing SERS substrates,and,for example,can achieve high-precision preparation of SERS hotspot structures with a gap of less than 10 nm.9

During FIB nanofabrication,the high-speed incident Ga ions will implant into a sample after losing kinetic energy to the sample atoms through cascade collisions,which introduces a nanoscalethick ion-implantation-induced modifie layer.This layer can be used,for example,in semiconductor modificatio studies10and as a gas-assisted etching masks for single-crystal silicon.11The FIB gallium ion irradiation technique has been applied to fabricate 3D structures based on a suspended gold (Au) thin film During the exposure of the substrate to ion irradiation,the interaction between Ga ions and Au grains causes vacancies and dislocations in the Au film thereby inducing stress in the fil subsurface to realize the 3D configuration12

The ion-implantation-induced modifie layer will degrade the enhancement performance of SERS substrates fabricated by FIB nanofabrication.Despite this,there have few reports of fundamental research into the ion modificatio of SERS-active substrates by Ga FIB implantation.In this paper,the Ga ion implantationinduced damage in the Au substrate is studied by atomic simulation,thermal annealing,and x-ray photoelectron spectroscopic (XPS)characterization.

II.EXPERIMENTAL

The experiments were conducted using Au fil samples with SERS activity on quartz substrates.The samples consisted of three layers along the depth direction:a top layer of Au fil about 80 nm thick,a middle transition layer of chromium (Cr) about 10 nm thick,and a bottom layer of quartz substrate,as shown in Fig.1.

The focused ion beam/scanning electron beam dual-beam system of an FEI Nova 200 NanoLab was used in this experiment.This is able to perform high-precision processing using the FIB,together with high-resolution characterization by field-emissio scanning electron microscopy (SEM).The FIB bombardment area was 300 × 300μm2,the processing parameters were 10 kV accelerating voltage and 15 nA beam current,and the processing times were 80,160,and 240 s,with corresponding ion implantation doses of 8.3×1015ions/cm2,1.6×1016ions/cm2,and 2.5×1016ions/cm2.A Tianjin Zhonghuan high-temperature tube furnace with 200,350,and 400○C annealing temperatures and nitrogen gas protection was used for the subsequent thermal annealing experiment.The surface shape of the machined area was examined by SEM,and XPS tests were performed to analyze the machined area under different annealing conditions.

TABLE I.Parameter settings of MD simulation.

III.RESULTS AND DISCUSSION

A.Molecular dynamics simulation of Ga ion implantation into Au

The Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS)13was used to perform the molecular dynamics(MD) simulation of Ga ion implantation into Au.The specifi parameter settings of the MD simulation process are shown in Table I.Open Visualization Tool (OVITO)14was used for visualization and analysis of the Ga ion implantation results.

OVITO and Origin were used to obtain the damage distribution and the average stress after the implantation of the last Ga ion (the 30th).Through the common neighbor analysis (CNA) of OVITO,the crystal structure type of the Au substrate can be clearly identified In the absence of any defects,the Au substrate has an FCC structure.However,the implantation of Ga ions leads to local changes to non-FCC structures.Figure 2 shows the number of atoms in non-FCC structures from the maximum of the potential energy to the fina stable state,and it can be seen that this number decreases from 2980 to 1185 as the potential energy of the system gradually falls.

FIG.1. FIB cross-sectional analysis of SERS Au film samples.(a)Schematic of a sample.(b)and(c)FIB cross-sectional analysis of Au film samples.

FIG.2.Number of atoms in non-FCC structures from the maximum of the potential energy to the final stable state.

By compiling the stress equation in OVITO,the trend of variation of the average stress of the atoms in the non-FCC structures can be calculated.Figure 3 shows the evolution of the distribution of atoms in non-FCC structures after implantation of the 30th Ga ion,with the local stress (as represented by color coding) ranging from 3.9 to −1 GPa.Here,the negative values represent compressive stress.Through the use of CNA analysis and the expression select command of OVITO,the number of vacancies is found to be stable at around 3,and there are no vacancies around the compressive stress area after implantation.As can be seen from Figs.2 and 3,the average stress of the atoms in the non-FCC structures gradually decreases with the decline in potential energy of the system and the decreasing number of these atoms.Influence by the local stress,the self-healing capability of the non-FCC structures will be reduced as the number of atoms in these structures decreases.

B.SRIM simulation of ion implantation

Figure 4(a)shows the distribution of damage to an Au substrate due to cascade collisions induced by bombardment by 100 000 Ga ions with 10 keV energy incident along the normal direction,calculated by Monte Carlo simulation with Stopping and Range of Ions in Matter(SRIM)software.15The version of SRIM used in this paper is SRIM2013.The displacement damage induced by ion implantation is mainly distributed in the depth range of 0–140 Å.Although Monte Carlo simulation cannot show the dynamical evolution of damage caused by ion implantation,it can calculate with high effi ciency the distribution of implanted ions and their depth profile Figure 4(b) shows the depth distribution of implanted ions in an Au substrate as a result of bombardment by 10 keV Ga ions.The depth distribution curve can be roughly divided into four periods:(1)the concentration of Ga ions increases rapidly at depths between 0 and 35 Å;(2)the concentration reaches its peak value at a depth of 20–50 Å;(3)the concentration then decreases gradually between 50 and 140 Å;(4)there are almost no Ga ions at depths greater than 140 Å.

FIG.3.Non-FCC atoms and average stress distributions in the system at different times in Fig.2:(a)at the maximum of the potential energy(1617.5 ps);(b)during the attenuation stage of the potential energy(1632.6 ps);(c)at the final stable state(1670.1 ps).

FIG.4.SRIM simulation of 10 keV Ga ions bombarding Au.(a)Damage curve of cascade collisions.(b)Distribution curve of Ga ion implantation.

C.Annealing treatment

Figure 5 shows the morphology of triangular array SERS substrates fabricated by FIB direct writing under different annealing heat treatments at 200 and 400○C.There is good shape retention of the micro/nanostructure arrays on the Au fil before and after the heat treatment,which indicates that the annealing will not degrade the shape accuracy of the SERS hotspot structures fabricated by FIB direct writing.However,there are obvious changes in the morphology and image contrast of the Au fil surface after annealing at 400○C for 30 min.Taking account of the experimental results in Fig.5,the annealing parameters for the FIB implanted Au fil substrates were chosen as 350○C for 10 min.

To heat the sample sufficiently the heating rate parameter was set to 3○C/min during the annealing.The quartz tube containing the sample was evacuated before annealing and was protected by N2during the whole annealing process.Figure 6 shows SEM images of the same area before and after annealing.There are many small white dots distributed on the surface of the FIBprocessed area after annealing,and their concentration is higher for the larger ion implant dose.Since the melting point of Ga is 29.8○C,these white small dots will be precipitates of the implanted Ga ions after annealing.It has been reported that FIB implantation also leads to precipitation of Ga ions in semiconductor materials,and even Ga agglomerates of around 10 nm diameter have been found in single-crystal silicon implanted by FIB without annealing.11

D.XPS analysis of samples

X-ray photoelectron spectroscopy (XPS) is an advanced analytical technique for the microscopic analysis of electronic materials and components.Atoms of the same element have different electron binding energies at different energy levels.Under the same excitation,the same element will have photoelectrons with different energies.This is reflecte in the existence of multiple peaks in the energy spectrum,with the main peak being the most easily identifiable The main peaks of different elements,which have different values ofEk(the kinetic energy of photoelectrons emitted in the photoelectric process) andEb(the binding energy of bound electrons in the inner shell layer) can be analyzed in terms of their measuredEkvalues,and a surface composition analysis can then be performed.The underlying principle can be expressed as follows:

whereEkis a measured value,Ebis a calculated value,hνis the incident photon energy(a known value),andφis the work function of the spectrometer(a known value).

XPS was carried out on three regions:an unprocessed Au fil region (before annealing),an FIB implant with ion dose of 8.3 × 1015ions/cm2(after annealing),and an FIB implant with an ion dose of 2.5 × 1016ions/cm2(after annealing).The areas of these XPS test regions were about 200×200μm2,which is smaller than the FIB-processed region.Peaks were measured for Au,Ga,C,O,and Cr.Figure 7 shows the XPS spectra of the three different regions.

No distribution of Ga was detected in Fig.7(a),because this region was not bombarded by Ga FIB.However,weak distributions of O and C are found,which are derived from contamination by dust and from oxidation in the air,among other factors.By contrast,metallic Ga was detected on the surface of the samples after FIB implantation with ion doses of 8.3 × 1015ions/cm2and 2.5 × 1016ions/cm2followed by annealing,with a binding energy of about 20 eV.Also,the intensity of the spectral peak of elemental O was significantl higher than that of the unprocessed sample.

FIG.5. SERS structures pattern fabricated by FIB direct writing before and after different annealing under different conditions:(a) and (b) before and after annealing at 200○C;(c) and (d) before and after annealing at 400○C;(e)Au film substrate after annealing at 400○C.

From the results of both XPS and SEM,it appears that the white spherical precipitates in the areas subjected to Ga FIB bombardment followed by annealing comprise metallic Ga together with compounds of Ga and O,including Ga2O3and Ga2Ox.A detailed explanation of this is as follows.

From the measured XPS full spectrum,it can be seen that the surface of the annealed sample in the FIB-bombarded area is mainly composed of fiv elements,namely,Cr,O,C,Au,and Ga,with atomic percentages as shown in Fig.7(b).It is clear that,under the same annealing conditions,for an ion dose of 2.5 × 1016ions/cm2compared with an ion dose of 8.3×1015ions/cm2,the relative content of C is the same,those of Cr and Au are decreased,and those of O and Ga are increased.It has been shown that Ga2O3is not easily formed in a vacuum or in an oxygen-deficien environment,and even if Ga oxides are formed,they are sub-stoichiometric,i.e.,However,Ga is easily oxidizable in air,and the metallic Ga and Ga2Oxprecipitated after annealing of the sample surface tend to form Ga2O3in accordance with the stoichiometric ratio in the air environment.The present annealing experiments were performed under a nitrogen atmosphere,which is an oxygendeficien environment.When a sample was removed after completion of the annealing experiment and left in the air,its surface was oxidized,leading to a change in the composition of the surface material.

Figure 8 shows spectra obtained from the samples.The firs peak is associated with the O 2score level contribution.It is at 22.9 eV,which is very close to the typical value for this emission peak.17The second peak was centered at 20.32 eV,corresponding toThe third peak is centered at 19.03 or 18.36 eV,corresponding to Ga+and gallium in its metallic state(Ga0),respectively.In Fig.9,the position of the O 1speak is at 530.5 eV,corresponding to the stoichiometric

FIG.6.SEM images of FIB implantation area at 10 kV accelerating voltage after 350○C annealing for ion doses of(a)and(d)8.3×1015 ions/cm2,(b)and(e)1.6× 1016 ions/cm2,and (c) and (f) 2.5×1016 ions/cm2.

FIG.7. (a) XPS spectrum of the unprocessed region (before annealing).(b)XPS spectra of samples subjected to FIB implantation with ion doses of 8.3×1015 ions/cm2 and 2.5×1016 ions/cm2 (after annealing).

FIG.8. O 2s and Ga 3d peaks in XPS spectra of samples subjected to FIB implantation with ion doses of 8.3×1015 ions/cm2 and 2.5×1016 ions/cm2(after annealing).

Compared with the Ga 3dpeak of the annealed sample treated with an ion dose of 8.3×1015ions/cm2,that of the annealed sample with an ion dose of 2.5 × 1016ions/cm2is shifted toward a lower binding energy,with the Ga+peak shifted negatively by 0.67 eV(a large value,reflectin a change in chemical valence) to become a Ga0peak,as shown in Fig.8.In addition,Fig.8 also shows that the area under the peak corresponding to Ga3+increases,and a new Ga0peak appears on the surface of the annealed sample treated with an ion dose of 2.5×1016ions/cm2.This indicates that,under the same annealing conditions,GaO (which can be seen as Ga2O3⋅Ga2O,a mixed oxide of Ga2O3and Ga2O,or as Ga+)in the surface products is converted to Ga2O3when the FIB processing time becomes longer.Further,each O 1speak in Fig.9 can be fitte with two peaks,O(I) and O(II),corresponding respectively,to lattice oxygen of Ga2O3and surface-adsorbed oxygen or lattice oxygen of GaO.16

FIG.9.O 1s peaks in XPS spectra of samples subjected to FIB implantation with ion doses of 8.3×1015 ions/cm2 and 2.5×1016 ions/cm2(after annealing).

It can be seen from Fig.9 that the binding energy shifts of the O 1speaks for the annealed FIB-processed samples are not significant but the areas under the peaks corresponding to O in the samples processed by FIB for 80 and 240 s are different.This is mainly because the Ga0and Ga+corresponding to the peaks from the two different samples have different abilities to bind oxygen,as can be seen from the XPS spectra in Fig.8.

Therefore,to effectively weaken or even eliminate the modifying effects of Ga FIB implantation on the Au film of SERS substrates,a thermal annealing treatment can be performed on the FIBprocessed SERS-active substrate.This is expected to be an effective way to boost the performance of SERS substrates prepared by FIB nanofabrication.

IV.CONCLUSIONS

In this paper,we have used a Ga FIB to process SERS Au fil substrates,and,based on XPS and SEM characterizations,we have analyzed the mechanisms by which ion implantation and thermal annealing affect the ion-induced damage to the substrate.The XPS results show that the precipitates appearing after FIB processing and 350○C annealing consist of a mixture of Ga2O3and Ga2Ox.The thermal annealing process can effectively reduce or even eliminate the modification of the SERS Au fil substrate caused by Ga FIB implantation.

ACKNOWLEDGMENTS

The study is supported by the National Natural Science Foundation of China(Grant No.51575389)and the State Key Laboratory of Precision Measurement Technology and Instruments(Pilt1705).The authors thank the reviewers for their careful work and thoughtful suggestions.

AUTHOR DECLARATIONS

Conflict of Interest

The authors have no conflict to disclose.

DATA AVAILABILITY

The data that supports the finding of this study are available within the article.