Developing a Method for Enhanced Explosive Detection by Surface Enhanced Raman Scattering (SERS) and Metal Enhanced Fluorescence (MEF)

The first manuscript, “Fabrication of SERS Substrates for Explosive Detection,” focuses on the development of a Surface Enhanced Raman Scattering (SERS) substrate for use in the area of explosive detection. The substrate was created by immersion plating of silver onto porous silicon, resulting in a Ag roughened surface with an average roughness of 135 nm as determined by AFM. When explosive solutions, such as trinitrotoluene (TNT) and dinitrotoluene (DNT) in ethanol, were applied directly to the substrate, detection of unique Raman bands was possible down to the 10 -9 – 10 -10 mol/cm 2 range. These results prove the technique is selective and shows promise for future work when vapor phase explosives will be investigated. The second manuscript entitled “Light Trapping to Amplify Metal Enhanced Fluorescence with Application for Sensing TNT” focuses on the use of the previously mentioned substrate for Metal Enhanced Fluorescence (MEF). By using a polymer layer as a dielectric spacer in between the Ag layer and the fluorophore, enhancement of fluorescent emission was possible even though the spacer thicknesses was 10 – 20 times larger than typically reported. In experiments with rhodamine 6G, the fully assembled substrate resulted in a 1600-fold enhancement of emission. Compared to the usually reported 10 1 – 10 2 times enhancement, the large enhancement is suspected to occur due to numerous effects. As enhancements were observed with and without the roughened Ag layer, light trapping is suggested as a contributing factor in addition to MEF. While less effective in enhancing the emission of the conjugated polymer methoxy-ethylhexloxypolyphenylene-vinylene (MEH-PPV) that is known to interact with TNT vapor, the quench in fluorescent emission of MEH-PPV occurred more rapidly when the light trapping polymer was incorporated. Rapid detection and increased sensitivity are important features to the detection of trace explosives.


LIST OF TABLES
In order to capitalize on the enhancements in sensitivity and selectivity provided by SERS and MEF, a combined sensor is suggested for the detection of explosives. Figure 1 represents the proposed sensor. Starting at the right side of the sensor, the flat silicon region will provide an area for background data to be collected if needed. The center portion, made of a simple porous silicon and Ag nanoparticle (p-Si/Ag) assembly, can be utilized for SERS. The left edge of the sensor will be reserved for MEF experiments, comprised of the same p-Si/Ag substrate, a polymer spacer and a fluorophore with known interactions to explosives.
The manuscripts presented herein discuss preliminary research that has been done in this effort. Manuscript 1 presents the procedure and characterization of the p-Si/Ag substrate. Also, included SERS spectra of varying explosives show that detection is possible utilizing the substrate. The second manuscript shows that large enhancements in fluorescence are possible through the suggested MEF substrate.
Continuing work will be needed to take this sensor a step further, establishing an array sensor capable of discerning a variety of explosive vapors in the field.

Introduction
In the recent years, great advances in the area of detection have been made, specifically in Surface Enhanced Raman Scattering (SERS). SERS is a technique that utilizes the surface plasmon resonance of metal nanoparticles to amplify signals that contain vibrational information of the analyte. While single molecule detection has been reported, this sort of response is not typical. Rather, signal enhancements on the order of 10 3 -10 6 are more commonly reported. 1,2 Even at this level, trace detection is possible and far exceeds the response seen by traditional Raman, with typically diminished fluorescence interference. 1 Unfortunately, creating SERS-active substrates for varying applications can be challenging, with many steps of optimization.
Growing steadily with the improvements in detection has been the number of terrorist attacks, both at home and abroad, and the use of explosive materials in these actions. These materials have proven to be difficult to detect by a standoff detection method because of their extremely low vapor pressures, making research in the area important. This research group has worked closely with EmiTech, Inc. to optimize a novel explosive sensor, known as Mark I. 3 The sensor operates by combining the Solid samples of the explosives were heated to achieve 1445 ppm 2,4-DNT and 341 ppm TNT concentration for detection. 5 To achieve enhancements on the order of 10 4 -10 5 the incident light was changed from 532 nm to 232 nm. 5 This approach also proved to be applicable to the standoff detection, making it an attractive option for research and a possible direction for future work for this research group.

Results and Discussion
Coating of the p-Si samples with a roughened Ag layer by immersion plating has been studied and reported previously. [8][9][10] The spontaneous deposition of Ag by the reduction of Ag + occurs due to the mild reducing power of the freshly etched p-Si. Samples were produced following the procedure outlined above with varying Ag-deposition times to determine the appropriate length of growth for the porous silicon substrates. Rh6G was selected for this study due to its known high Raman cross section. 11 Figure 1 shows a sample of the spectra collected from 4.3×10 -10 mol/cm 2 Rh6G on p-Si/Ag substrates. Growth times included 0, 10, 12, 14, 16, 18, 20, and 22 minutes. While the 12 minute deposition appears to have the highest intensity at the characteristic Rh6G peaks, the height difference can be attributed to the increased baseline. Figure 1 was created with the "hot spot," or best result, of Rh6G on each growth time. As the sample is not completely uniform, it is possible that all acquired spectra taken at varying locations could produce only a fraction of the potential enhancement. Taking this fact into consideration, Figure 2, which represents the average intensity at the 1517 cm -1 Rh6G peak over three scans, was created for the determination of deposition time. On average, the 14 minute Ag deposition provided the most intense Rh6G spectrum. It is also important to observe that the 0 minute Ag growth, indicating only p-Si, showed no Rh6G spectrum, justifying that the Ag layer is responsible for the enhancement. All SERS substrates generated after this study and for the remainder of this paper were created with the length of Ag growth parameter held constant at 14 minutes.  XPS was utilized to better understand the characteristics of the Ag layer and for further clarification of the AFM image. Figure 4 shows both the full spectrum (A) and the Si region (B). The results seen in Figure 4A show strong Ag signals and minor Si signals. Figure 4B is a magnified portion representing only the bonding energies associated with Si. Noting the poor signal to noise and knowing the penetration depth of XPS is between 1 and 10 nm, it can be again concluded that the thickness of the Ag layer varies. Therefore, the generated surface is best described as a roughened Ag surface. The unlabeled peaks on the full spectrum correspond to oxygen, nitrogen and carbon.
A B A study to approximate the limit of detection for the Raman active substrates on the Agilitron instrument was completed, once again utilizing Rh6G. Even with an average Raman instrument, detection of Rh6G was possible in the 10 -9 mol/cm 2 range ( Figure 5). While the drop in intensity from 3.5×10 -9 mol/cm 2 to 1.2×10 -9 mol/cm 2 is abrupt as seen in Figure 6, typical Rh6G peaks can still be observed in the lowest concentration when multiplied by a factor of 50. This concentration represents the Intensity (a.u.) 1.2x10 -8 mol/cm 2 Rh6G 7.0x10 -9 mol/cm 2 Rh6G 3.5x10 -9 mol/cm 2 Rh6G 1.2x10 -9 mol/cm 2 Rh6G x50 Figure 5. Raman spectra obtained of Rh6G on p-Si/Ag substrates used to determine the limit of detection of the Agilitron instrument.

Conclusions
Through experimental work, a SERS-active substrate was created and optimized for the established etching procedure. It was found that a 14 minute Ag deposition results in the largest enhancement of Rh6G. Rh6G also was utilized to determine a limit of detection of the Agilent instrument with the p-Si/Ag substrates.
The Raman spectrometer was able to detect down to the 10 -9 mol/cm 2 level, providing a starting point for solution explosive testing. Finally, these substrates show that explosive detection is possible and with further optimization vapor phase explosive detection may be achieved. Table 1. Further height parameters for the AFM image of the p-Si/Ag sample seen in Figure 3.

Introduction
Exploitation of the plasmon properties of metal nanoparticles is currently of interest. Coupling the electric field created by a plasmon to a molecule on the surface can lead to significant intensity increases in Raman spectra (surface enhanced Raman spectra, SERS) and in fluorescent spectra (metal enhanced fluorescence spectra, MEF). [1][2][3][4][5][6][7][8] For SERS, typical enhancements can be 10 6 or greater while for MEF the enhancements are more modest, typically 10 1 -10 2 . We have been interested in using fluorescence methods for the detection of explosives. 9 A porous silicon (p-Si) substrate is used as the substrate for the fluorophore, which increases the surface area available to the analyte, thereby increasing the sensitivity. We presumed that adding a metal layer to the porous layer the sensitivity could be further improved by taking advantage of MEF. Optimized MEF structures require a dielectric layer between the metal and the fluorophore and when we used a polymer as the dielectric the enhancements increased to 10 2 -10 3 .
We report here that using a polymer layer between a fluorophore and substrate provides an enhanced emission for a variety of polymers and substrates, including those that do not include a metal layer. The enhancement appears to arise from a combination of effects since a single mechanism does not account for the observed intensities. When the substrate contains a layer of silver nanoparticles the MEF effect is amplified cooperatively by the polymer effect. We exploit this effect to demonstrate improved sensitivity for the detection of trinitrotoluene (TNT).

Experimental Section
Silver coated porous silicon substrates, p-Si/Ag, were created by electrochemical etching 9  There are also shifts of the wavelength maximum that are not typically seen in MEF.  To test the effect of different substrates, Rh6G was deposited onto polymers spin-cast onto a glass slide and a flat Si wafer. Neither of these substrates contained Ag, yet substantial enhancements were observed as shown in Figure 2. Even with a glass slide substrate there is a modest enhancement, by as much as a factor of ~35 for PVDF. When the substrate is flat Si, again with no Ag, the polymers lead to enormous enhancements, as high as 1600 for 0.3 µm thick co-polymer film. On the flat Si substrate there also is noticeable shift of the emission maximum, with maxima ranging from 560 nm for PMMA, 550 nm for PVDF, and 540 nm for the co-polymer. When Fabry-Perot modes are coupled to a plasmon resonance the absorption maximum can shift 11 and this should affect the emission maximum similarly. However, the flat Si is a semiconductor and is not expected to support a plasmon resonance. There is a considerable difference in the polarity of each of these polymers and that may contribute to the observed wavelength shift.   Reflection spectra for the various structures were measured, as shown in Figure 4. The effect of simply depositing Rh6G onto the substrate is shown in Fig. 4A using flat Si as the reference. In the absence of Rh6G, p-Si shows a fringing pattern resulting from the pore structure (~ 2 µm thick) but the fringing largely disappears when the Ag layer is added, indicating that the light does not pass through to the p-Si layer. The p-Si shows a resonance at 450 nm while the p-Si/Ag substrate shows minimal reflection throughout the visible region. Upon addition of Rh6G, reflection drops considerably in all cases and does not show any feature that could be assigned to the absorption maximum expected for Rh6G at 521 nm. The featureless response indicates that the Rh6G layer is trapping light. Fig. 4B shows the reflection spectra of different polymers, with no Rh6G, coated onto p-Si/Ag. The polymers, which are all transparent, all reduce the reflectivity compared to the bare p-Si/Ag, showing that the polymers also trap the light within the structure.
Samples were prepared using MEH-PPV on flat Si and p-Si/Ag/co-polymer substrates to be used for TNT sensing. MEH-PPV has been previously shown to be effectively quenched by nitroaromatics. 9 As expected, the p-Si/Ag/co-polymer/MEH-PPV structure showed a fluorescence enhancement compared to flat-Si/MEH-PPV, but the enhancement at the emission maximum (598 nm) was only about 2.5 times. It is not clear why the enhancement is so much less than the Rh6G, but may be related to the overlap of the excitation wavelength (495 nm) and the resonance observed in the p-Si. Some of the excitation light is undoubtedly lost into the tail of the p-Si absorption.

Conclusions
In conclusion, we have shown that by adding a light trapping layer to a MEF structure can increase the fluorescence signal intensity by a factor of several hundred to over 1000. Further, when the light trapping layer is used in a TNT sensor, the signal intensity increases and the exposure time to reach maximum quenching is reduced, both qualities that are required for improved sensors. Work to better understand the influence of all the parameters in the structures is underway.

Notes
The authors declare no competing financial interests.