Commercial Gold Nanoparticles on Untreated Aluminum Foil : Versatile , Sensitive , and Cost-Effective SERS Substrate

We introduce low-cost, tunable, hybrid SERS substrate of commercial gold nanoparticles on untreated aluminum foil (AuNPs@AlF). Two or three AuNP centrifugation/resuspension cycles are proven to be critical in the assay preparation.The limits of detection (LODs) for 4-nitrobenzenethiol (NBT) and crystal violet (CV) on this substrate are about 0.12 nM and 0.19 nM, respectively, while maximum analytical SERS enhancement factors (AEFs) are about 10. In comparative assays LODs for CV measured on AuNPs@Au film and AuNPs@glass are about 0.35 nM and 2 nM, respectively. The LOD for melamine detected on AuNPs@ Al foil is 27 ppb with 3 orders of magnitude for linear response range. Overall, AuNPs@AlF demonstrated competitive performance in comparison with AuNPs@ Au film substrate in SERS detection of CV, NBT, and melamine. To check the versatility of the AuNPs@AlF substrate we also detected KNO 3 with LODs of 0.7mM and SERS EF around 2 × 10, which is on the same order with SERS EF reported for this compound in the literature.


Introduction
Since SERS was reported about 40 years ago, this label-free, sensitive, and rapid method with significant, multiplexing potential has been in focus of expanding scientific investigation [1,2].SERS has been established as a versatile analytical technique with applications in sensing of explosives, environmental pollutants, and chemical warfare agents [3,4].Applications of SERS in biosensing include, for instance, detection of sugar in blood, enzymes, and DNA, and early medical diagnostics of major health threats for humans (e.g., cancer, tuberculosis, and viruses) and animals [5,6].As one of the most sensitive analytical techniques for identification of molecular species, SERS has demonstrated the capability for single molecule detection, including detection of a single biomolecule [7,8].
SERS is usually performed on silver, gold, and less frequently copper surfaces, which are more susceptible to corrosion [9,10].Still silver and even gold nanoparticles may change shape after just a few hours of exposure to solvents or solutions of thiols, which affects their plasmonic properties [11,12].Another problem is that the most gold and silver SERS substrates with high sensitivity and acceptable reproducibility are still costly for routine SERS detection.For instance, the price for commercially available SERS nanostructured gold substrate "Klarite" was about 100 USD for a single 2 × 2 mm SERS active address.
Aluminum, which is undoubtedly much less costly than silver or gold, has been shown as effective plasmonic material in UV-blue light range for LSPR and surface-enhanced fluorescence [13][14][15].Computationally and experimentally Al is shown as SERS active material in UV spectral range [16][17][18].There is Al 2 O 3 film formed on the metal surface, which typically has a thickness of 2.5-3 nm, and LSPR position is red-shifted in the presence of this layer [14].Aluminum is highly reflective material with more than 85% reflection efficiency in visible range, but until recently Al had not been known as effective plasmonic material in this range.That is attributed to an existence of interband transition around 800 nm (1.5 eV) resulting in a large imaginary part of the dielectric function in visible range [19].However, recently the first SERS phenomenon has been reported on nanostructured (annealed) Al film with excitation in visible range (e.g., with 785 nm laser) by Mogensen group [20].In this study as few as 3 × 10 5 -3 × 10 7 molecules of Raman active molecules (adenine, p-mercapto benzoic acid (pMBA), and 4-aminothiophenol (4-ATP)) in the probed focused laser beam area (∼1 m 2 ) were detected by SERS, but no estimates of SERS EFs or/and LOD were reported.Very recently, it was reported that the average EF for single silver nanoparticles (AgNPs) modified with 2-methoxythiophenol (MOTP) and drop-casted on gold film is just about twice as high as average EF of MOTP at AgNPs on Al film (5.0 × 10 6 versus 2.2 × 10 6 ) [21].Another group reported SERS of Rhodamine 6G, Rose Bengal, and crystal violet on the substrate of concave gold nanocubes on Al-6063 alloy [22].
Hybrid SERS substrates can achieve LODs or EFs that would match or exceed those sensing parameters obtained on gold/silver films or even micro/nanostructured SERS substrates.For instance, synthesized silver nanoparticles over polyvinylidene difluoride (PVDF) membrane detected food contaminant melamine with LOD 50 nM/6 ppb [23].Hybrid substrates consisting of gold nanoparticles on microstructured gold surface (Klarite) demonstrated about 3 orders of magnitude higher EFs for detection of 4-ATP relative to EFs on the Klarite surface [24].SERS on some hybrid substrates can achieve relatively high EFs (10 7 ) as exemplified in detection of tyrosine in aqueous solution on silver nanoparticle doped filter paper [25].High accuracy and precision are reported for detection of three drugs (0.1 mM solutions) on silver spotted two-pence coins by Goodacre group.To detect those drugs in solutions, small volumes of the solution were spotted on the individually prepared silver targets at coin surface and let to dry before the Raman measurements [26].
In this paper we present inexpensive and rapid method for the fabrication of hybrid SERS substrate from commercial gold nanoparticles drop-casted on untreated and unmodified Al foil.Al foil is universally available, low-cost substrate, which is just rinsed with DI water before preparation.The only equipment needed for preparation is minicentrifuge used for 20-30 minutes.The cost of nanoparticles, needed for preparation of 1 mm 2 of this SERS active surface, is less than 2 cents.Untreated Al foil by itself, used as a substrate without gold nanoparticles, demonstrated relatively small maximum analytical enhancement factors (80-240).However, the composite substrate of AuNP@Al foil has about ×10 5 higher (10 7 versus 10 2 ) maximum analytical enhancement factors and 4 or 5 orders of magnitude lower LOD relative to LOD for crystal violet (CV) on bare Al foil substrate (∼0.35nM pM versus ∼0.01 mM).Finally, we will demonstrate versatility of this substrate in SERS detection of such chemicals as 4nitrobenzenethiol (NBT), KNO 3 , and melamine.
Crystal violet (CV) is a typical synthetic basic cationic dye, which has been widely used in the aquaculture as a veterinary medicine, additive to poultry food to inhibit propagation of harmful bacteria [27].4-Nitrobenzenethiol (NBT) is one of the most commonly applied Raman markers, which is used to prepare extrinsic Raman label preparation (ERLs) in sandwich SERS immunoassays for cancer and tuberculosis [28,29].KNO 3 is a major component of self-made explosives, which was recently detected on nanostructured substrate Klarite with maximum SERS EF calculated [30].Melamine is a toxic compound that has been found in many samples of chicken eggs, dry milk, or baby formula resulting in hundreds of thousands of hospitalizations and hundreds fatalities, especially among children [31].

Experimental
2.1.Chemicals and Consumables.Gold nanoparticles of various diameters (40, 60, and 80), OD = 1 in PBS buffers obtained from Sigma-Aldrich.Crystal violet (CV), 4nitrobenzenethiol (NBT), melamine, and potassium nitrate obtained from Sigma-Aldrich as well.Al foil was purchased in a supermarket.DI water was obtained from Milli-Q system.

Instrumentation.
The Raman measurements were performed with 785 nm diode and 633 nm He-Ne lasers on LABRAM Horiba microscope and ×50 objective.The instrument was calibrated with the Raman signal from the principal Raman vibration of silicon centered at 520 cm −1 .For detection of all chemicals, the laser power was set to ∼1.5 mW for 633 nm excitation and ∼3 mW for 785 nm excitation at the sample position, and the exposure times for each SERS measurement were set to be 1-3 s. Background corrected spectral intensities obtained from 4 or 8 maps were taken in different locations, but within 0.5-1.5 mm from the center of each address.They were averaged (at least 64 spectra in each map) for each sample.
The AFM measurements were done with AIST AFM in tapping mode to obtain at least 3-7 AFM maps (10 × 10 or 7 × 7 or 4 × 4 m) with total of hundreds AuNPs for each representative sample.

Preparation of Substrates.
Al substrate was prepared by attaching Al foil on the glass with a double-sided scotch.We prepared analyzed solutions of different concentrations.NBT and CV were dissolved in acetonitrile; melamine and KNO 3 are dissolved in DI water.Standard solutions with various concentrations were obtained by diluting the stock solution.We used parafilm with perforated holes to limit spreading of the drop by its hydrophobic surface.After washing the addresses with DI water, we applied 25 L of prepared AuNPs solution on the addresses of 5.5 mm diameter.The suspensions of AuNPs of two diameters (40 and 60 nm; OD = 1 in PBS buffers obtained from Sigma-Aldrich) were centrifuged (40 nm AuNPs at 5000 g for 12 min and 80 nm AuNPs at 2500 g for 10 min, resp.) and then after removal of supernatant they were resuspended in DI water.Typically those cycles were repeated 3 times and suspensions were briefly sonicated (∼3-5 min) before and after each cycle.After solution of AuNP had dried on the substrate, we drop-casted 25 L of analyzed solutions on each address.As additional control we prepared some samples for detection of CV and NBT drop-casting AuNPs onto the same area on the surface of commercial evaporated gold film (100 nm thickness) or glass microscope slides.enhancement capability of the new SERS substrates, the enhancement factor (EF) value is calculated according to the following formula (1):

Calculation of Enhancement Factors (EFs) and Limits of Detection (LODs). In order to quantify the SERS signal
where  SERS is the SERS intensity of the particular peak of NBT while  RAMAN represents the unenhanced Raman intensity measured in liquid form. SERS is the number of NBT molecules on the substrate contributing to the SERS signal and  BULK is the number of molecules contributing to the unenhanced Raman signal [32].By incorporating the number of molecules in m 3 (V BULK ), average height of focused laser beam in m ( laser ), and SERS surface concentration, number of molecules per m 2 (  ) EF can be rewritten as the second part of (1).Details of the EF calculation and results are given in Supplementary Materials available online at https://doi.org/10.1155/2017/9182025.

Performance of AuNP@ Al Foil Substrate Tested with
Detection of NBT and CV as Compared with AuNPs@Au Film and AuNPs@Glass Substrates.From the data obtained from Raman spectra of the NBT (Figure 1(a)) limit of detection and enhancement factor were determined.We calculated the LOD as a concentration of analyte at three standard deviations of the blank from the plot of blank adjusted signal versus logarithm of analyte concentration.NBT and CV ware dissolved in acetonitrile and sequentially diluted before drop-casting solutions in a broad concentration range from 0.010 mM to 0.30 nM pM or ∼33000 times.The results were analyzed by plotting the intensity of characteristic stretch vibration (NO 2 ) near 1336 cm −1 as a function of the logarithm of total NBT concentration, generating the response curves as one in Figure 2, where signal increased with increase in concentration within broad dynamic range of 8 orders of magnitude.We got blank adjusted SERS signal of similar magnitude for NBT and CV with 60 nm AuNPs as shown in Figures 1(b) and 2 and Table 1.Blank adjusted SERS signals for all chemicals are calculated by subtraction of average SERS Raman signal of the blank from average SERS signal of each standard.
As shown in Figure 2(a), both CV and NBT demonstrate a trend of switching to increasing logarithmic sensitivity (slope) in the range about 30-300 nM.However, even at lower concentration and lower slope, the linear logarithmic response spans at least two or three orders of magnitude in concentration.The plots of normalized blank adjusted Raman signal of the characteristic peak versus log of analyte concentration are used for calculation of LODs.Those plots are shown for CV and NBT in Figures 1(b) and 1(d) and plot for KNO 3 is shown in Figures 2(a) and 2(b).Raman spectra (solid and SERS) in broader range are given for analyzed chemicals in Supplementary Materials.Blank adjusted SERS signals are calculated for all analytes by subtraction of average blank (e.g., solvent: DI water or acetonitrile) signal from average signal of each analyte standard, while both are prepared on the same day and measured with the same parameters.
The maximum analytical enhancement factors (AEF) for the CV and NBT samples at 0.3 nM concentration (above LOD for both chemicals detected on Al foil) were calculated by (1) above and reported in Table 1.This table also summarizes LOD and the concentration range used for LOD calculation as well as major parameters of substrate preparation such as AuNP size and number of centrifugation cycles.
The maximum analytical SERS enhancement factors (AEF) above the limit of detection are 6 × 10 6 for CV (0.3 nM) with 633 nm excitation and 1.2 × 10 7 for NBT (0.3 nM) with 785 nm excitations.Those magnitudes of EF are on a high end of SERS analytical EFs range and they establish AuNPs@Al foil as potentially strong SERS substrate with unique cost to SERS efficiency ratio.

Optimization of the Substrate Preparation: Impact of the Centrifugation/Resuspension
Cycles.We optimized SERS detection of those chemicals by variation, AuNP size, number of centrifugation cycles, and concentration of AuNP in the applied (drop-casted) solution.The number of centrifugation cycles makes the strongest impact on SERS detection efficiency.For example, as shown in Figure 1 for NBT, three centrifugations versus one centrifugation produce SERS with 0.3 nM * for CV/glass and 0.3 nM * NBT/Au are outside of calibration range, but it shows that 1 nM and 0.3 nM CV concentrations cannot be distinguished on this substrate.much higher sensitivity and much lower LOD (0.1 nM versus ×30 nM).Overall, two or even three centrifugation cycles produce the best SERS response for this hybrid substrate.The possible explanation is that the layer thickness of stabilizing agent decreases after each centrifugation and the "hot spots" between gold nanoparticles become "hotter" as interparticle distances decrease upon the loss of surfactant during centrifugation/resuspension cycles.Another likely contribution to this increase in SERS signal is that upon the loss of surfactant the gold nanoparticle surface, including hot spots, becomes more accessible for the adsorption of analyte molecules.Data obtained by atomic force microscopy (AFM) with relative standard error in surface concentration 4-19% are illustrated by representative unprocessed AFM maps in Supplementary Materials.Apparently, the most NPs exist as dimers, trimers, and oligomers (nonsingles count 60-90% of total number of nanoparticles).Therefore, we would expect a significant contribution to the SERS signal from hot spots generated in the junctions between agglomerated AuNPs.However, as we demonstrated in a recently published study of AuNPs agglomeration on SERS signal with combined AFM Raman imaging the average enhancement on AuNPs nanoparticles dimers is probably higher, but still likely to be on the same order of magnitude as average enhancement on the surface of single AuNPs, when both are laid on metallic film [21].

Comparison of AuNPs@Al Foil with AuNPs@Gold and
AuNPs@Glass in Detection of CV and NBT.Apparently gold nanoparticles generate most of the enhancement in the hybrid substrate because "bare" Al foil (without AuNPs) makes maximum AEF ∼240 (785 nm excitation) for NBT and ∼80 (633 nm excitation) for CV as shown in AEF calculation table in Supplementary Materials.To compare efficiency of Al foil as a bottom part of the composite substrate we also did detection of CV and NBT with the same 60 nm AuNPs on commercial evaporated gold film and AuNPs on glass microscope slide wafer.
Table 1 demonstrates SERS data output obtained with 633 nm laser excitation in this comparative SERS assay of both Raman labels on 3 different substrates, prepared on the same day with 3 cycles of 60 nm AuNPs centrifugation/resuspension.Table 1 shows that calculated analytical enhancement factors for 0.3 nM CV and 0.3 nM NBT detected with same 60 nm AuNPs are substantially higher on Al foil and on gold in comparison with AEF for assays on glass, which is especially the case for detection of NBT, where AEF on gold and Al foil is multiple of ×15 and ×8, respectively, relative to AEF on glass.The difference between AEF on Au and Al is much less significant (within ×2).In terms of LOD calculated for CV and NBT alike, Al foil outperformed both gold film and glass as substrates, demonstrating even stronger ( 2 = 0.99) logarithmic response in the concentration range 0.3-10 nM in comparison to other tested substrates.

Performance of AuNPs@Al Foil in Detection of Self-Made
Explosive Component: KNO 3 .Table 2 summarizes SERS detection output and AFM characterization results for KNO 3 and paracetamol.The maximum EF for KNO 3 on AuNPs/Al foil is about 2 × 10 3 .This EF is about the same by order of magnitude as EF (3.6 × 10 3 ) reported in the literature for SERS on commercial nanostructured gold Klarite substrate [30].
The obtained LODs and/or EFs on AuNP@Al are at least comparable (or even better) for SERS LODs and/or SERS EFs reported in the literature for those chemicals on other, typically more expensive substrates.For instance, LOD reported for CV by SERS on MnO 2 /Au hybrid nanowall film was about 10 nM [33].LOD reported in the assay on concave cubes on Al-6063 alloy (processed for oxide removal) for CV was about 0.1 nM [22].

Detection of Melamine at
AuNPs@Al Foil and AuNPs@ Au Film Substrates.We detected melamine in aqueous solution with 60 nm AuNPs drop-casted on both Al foil and evaporated gold film.The SERS spectra of melamine at various concentrations and calibration plot of the Raman peak area as a function of log  melamine are shown in Figure 3.The limit of detection of melamine is lower on AuNPs@ Au film (4 ppb) than LOD on AuNPs@ Al foil (28 ppb).
However the range of linear response on Al foil as a substrate is much wider (4 5 = 1280) than the linear range for the assay on gold film (4 3 = 64) and this linear responds on Al foil which has even better linearity with  2 = 0.97 for 6 data points in comparison to the linear response for 4 data points for the detection of melamine on gold film ( 2 = 0.90).
Therefore, in the assays for melamine Al foil as a part of composite substrate demonstrated its competiveness against more expensive gold film when the same 60 nm AuNPs are used on both substrates.
Overall the SERS limit of detection on both substrates is by about 1.5-2.5 orders of magnitude lower than the tolerance level recommended for this toxic compound by World Health Organization of 1 ppm [34].Therefore we demonstrates that SERS on AuNPs@ Al foil substrate can confidently detect melamine at concentration well below its legal or health threatening limit.

Conclusion
In this work, we demonstrated that gold nanoparticles at Al foil as low-cost hybrid substrate is capable of highly sensitive SERS.Commercial gold nanoparticles on untreated and unmodified Al foil can be used for subnanomolar detection of Raman active molecules.Various molecules from high end (CV; NBT) to low end (KNO 3 ; melamine) of SERS enhancement factor range can be detected on AuNPs@Al foil.They are detected on other SERS substrates, including expensive commercial substrates, such as Klarite, with similar magnitude of SERS enhancement and LOD.However, this substrate has advantages over most SERS substrates in simplicity of preparation with no need for sophisticated equipment, low-cost (1-3 cents worth of commercial AuNPs per 1 mm 2 of the substrate area), and high availability of the substrate bulk in the form of Al foil.Further work will aim at application of this hybrid low-cost substrate in biosensing.
Figure 1: ((a) and (c)) Raman spectra from NBT and CV solutions of different concentrations drop-casted on 60 nm AuNP@Al foil substrate.Spectra are taken with 633 nm laser excitation.((b) and (d)) Plots of blank adjusted Raman signal versus logarithm of NBT and CV concentration (M), respectively.Signals are obtained in SERS assay on 60 nm AuNPs at various substrates.

Figure 2 :
Figure 2: Plots of blank adjusted Raman signal versus concentrations of analyzed chemicals for SERS assay on unmodified Al foil: (a) for CV and NBT; (b) for KNO 3 .(c) SERS spectra (vertically adjusted to prevent overlap) for KNO 3 at 4 detectable concentrations and blank, taken with 785 nm laser excitation.

Figure 3 :
Figure 3: Detection of melamine.(a) Raman spectra of melamine on 60 nm AuNPs@Al foil in 0.078 to 80 ppm concentration range, taken with 785 nm laser excitation.(b) Calibration graph, where 820 cm −1 characteristic vibration peak area is plotted versus log  in ppb of melamine.

Table 1 :
Raman signals, LODs, and AEFs calculated from detection of CV and detection of NBT on 60 nm AuNPs set on various substrates (Al foil, Au film, and glass).

Table 2 :
SERS assay results for KNO3 and paracetamol detection at AuNPs@Al foil substrate.