IEEE SENSORS JOURNAL, VOL. 10, NO. 3, MARCH 2010 699

SERS and Density Functional Theory Study ofo-Dinitrobenzene on Cu Nanoparticles

Omar E. Rivera-Betancourt, Oliva M. Primera-Pedrozo, Leonardo C. Pacheco-Londoño, andSamuel P. Hernández-Rivera

Abstract—Cu colloidal suspensions were prepared by wetchemistry methods: chemical reduction using trisodium citrate.Steadiness varied under conditions such as stirring rate, rate ofaddition of reducing agent, and temperature control. Sizes ofnanoparticles prepared ranged from 2 to 20 nm. Morphology andsizes were characterized using high-resolution transmission elec-tron microscopy. Enhanced Raman spectra of -dinitrobenzene onprepared nanoparticles suspensions of Cu colloids were obtainedwith visible excitation at 488 and 514.5 nm. It was found thatenhanced Raman signal intensities resulted from the interactionof the analytes with the metallic surface resonance plasmon ofCu nanoparticles excited at the laser frequency. The adsorptionbehavior of the analytes on the Cu nanoparticles was modeledwith Gaussian 03 density functional theory software package. Thesimplified models used work reasonably well in describing someenhancement results for the Raman experiments on the noblemetal colloidal suspensions.

Index Terms—Cu nanoparticles, density functional theory(DFT), dinitrobenzenes (DNB), metallic colloidal suspensions,surface-enhanced Raman scattering (SERS).

I. INTRODUCTION

R AMAN SPECTROSCOPY (RS) is a popular nondestruc-tive probing tool used to characterize molecular struc-

ture and usually imposes very little constraint on the substratesize [1]. The vibrational spectrum provided by RS is a molec-ular “fingerprint” that can be used to differentiate target sam-ples from complex media or differentiate molecules with verysimilar structure and most often provides unambiguous sampleidentification. Owing to its scattering nature, fact has sponta-neous or normal RS is an inherently weak process. This facthas motivated physicist, chemists, and engineers to be in con-stant seeking for increasing the strengths of vibrational intensi-ties observed in a normal RS experiments. There are two generalschemes of increasing vibrational RS signals: resonance Ramanscattering (RRS), which depends on the probability of closely

Manuscript received April 13, 2009; accepted April 30, 2009. Current versionpublished February 24, 2010. This work was supported in part by the Centerfor Chemical Sensors Development and the Chemical Imaging Center of theDepartment of Chemistry at the University of Puerto Rico-Mayagüez sponsoredby the Department of Defense, University Research Initiative-MultidisciplinaryUniversity Research Initiative Program under Grant DAAD19-02-1-0257, andin part by the U.S. Department of Homeland Security under Award 2008-ST-061-ED0001. The associate editor coordinating the review of this paper andapproving it for publication was Dr. Neelam Gupta.

The authors are with the Department of Chemistry, University of Puerto Rico-Mayagüez, Mayagüez, PR 00681 USA (e-mail: omar_rivera83@yahoo.com;olivaprimera@yahoo.es; pachecolc@yahoo.com; sp_hernandez_uprm@yahoo.com).

Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/JSEN.2009.2038626

matching the excitation laser energy to the energy gap betweenthe ground electronic state and an excited state; and surface-en-hanced Raman scattering (SERS) in which surface enhancementfrom a roughened noble metal substrate. The largest signal en-hancements arise from the observation of these two phenomenasimultaneously in a process termed SERRS.

There is a large variety of SERS substrates other than metalliccolloidal suspensions, including electrodes, metal island films,and others. However, wet chemistry preparations provide aninexpensive and versatile approach to metal nanoparticlesfabrication. Colloidal nanoparticles are commonly used forSERS/SERRS studies either in a suspension as sols, or theanalyte-colloid system is cast onto a sustaining surface andallowed to dry on the solid support forming particles of avariety of shapes such as nanowires and aggregates. The useof colloidal suspensions for detection of a modest ( – )SERS enhancement has several advantages; the presence of asolution tends to minimize the burning of the sample, allowingthe use of higher powers and the use of more energetic laserlines (LLs). As well, the use of sols permits the acquisitionof an average spectrum due to the Brownian motion thatgoverns the colloidal suspension [2]. Common metals usedas SERS substrates include silver, gold, and copper. Variousmethods have been developed to prepare nanoparticles, usingthese metals. The procedures include chemical reduction, laserablation, and photoreduction. The most widely used is basedon chemical reactions in solutions (wet chemistry) that yieldmetal nanoparticles. Wet chemistry methods are performed bymixing metal salts, which are reduced by a chemical agent toproduce colloidal suspensions, producing metal nanoparticles[3]. Sodium borohydride and sodium citrate are widely used asa reduction agent. Both of the reducing agents have proven toproduce affective colloids, but synthesis with sodium citrate ismore common because it has given monodispersity on colloidsize and characteristics [4], [5].

The optical properties of these nanoparticles depend on thesize, shape, and dielectric constant of the medium. Sphericalmetallic nanoparticles exhibits a single surface plasmon (SP)band attributed to the coherent oscillation of the conductionelectrons caused by oscillating electric field when they are ir-radiated by light [6], [7]. The size and shape determinationscorrelate with the visible absorption data, and the full-width athalf-maximum (FWHM) can be used to estimate nanoparticledispersion [8].

The SERS is used to dramatically increase scattering signalsof analytes adsorbed on the nanoparticles surfaces. Large en-hancements ( – ) of the Raman scattering cross section can

1530-437X/$26.00 © 2010 IEEE

700 IEEE SENSORS JOURNAL, VOL. 10, NO. 3, MARCH 2010

be obtained for adsorbed molecules. The origin of this effect hasbeen in dispute for several years, but recent advances in produc-tion of nanostructures with very precise control of their size,shape, and spacing have allowed both greater understanding ofthe origin of the effect and its control [9]. The effects occur dueto the strong electromagnetic fields and field gradients in avail-able so called “hot spots” or areas of very high enhancementsof the metal colloidal nanoparticles clusters. The molecules in-volved are those adsorbed on the aggregates favorable for SPresonances. Small metal nanoparticles or nanostructures have alocalized SP resonance (LSPR) that appears to provide most ofthe Raman enhancement, maximized when the LSPR frequencyoverlaps that of the excitation laser. There also appears to bea modest “chemical” contribution to the enhancementcaused by changes in the bond polarizability of an adsorbedmolecule [9].

Even though SERS has been used to investigate the informa-tion of certain molecules, uncertainty still remains concerningthe adsorption geometry of analyte molecules on the surfaceof metal colloidal particles. The symmetry of adsorption ofthe adsorbate molecule is also an important factor to considerwhen explaining the SERS spectra obtained: relative intensitiesof the bands could be different from the neat solid or calculatedspectra. Quantum chemical investigations seem to treat theproblem in two ways, one being periodic density functionaltheory (DFT) and the other using a cluster model of the surfacestudied [10]. The DFT-based calculations have been used inmany areas [11]–[13].

In the present study, we report on the Raman scatteringspectra of o-dinitrobenzene (o-DNB) adsorbed onto Cunanoparticles with visible excitation at 514.5 and 488 nm.DFT-B3W91 with LanL2DZ basis set was used to calculatethe Raman theoretical frequencies of an o-DNB moleculeinteracting with a single copper atom model and with a coppersurface model. The theoretical results were compared with theexperimental values and the reliability of the models and thecalculation methods were analyzed.

II. PROCEDURES

A. Instrumentation

Plasmon band characterization of Cu colloids was performedin a UV-VIS Agilent 8453 spectrophotometer. The morphology,size, and shape of copper seed nanoparticles were investigatedby transmission electron microscope (TEM). A high-resolutionZeiss model 922 was used at 200 kV accelerating voltage. Vi-brational analysis was conducted using a RM2000 RenishawRaman Microspectrometer equipped with a Leica microscopewith a 10x objective.

The SERS spectra of o-DNB was performed with single linesat 488 and 514.5 nm, using laser power at sample mW.The excitation source was a Coherent, Inc., INNOVA 305-8argon ion laser with a variable output power of up to 2 W. TheRaman spectra were collected in the range of 100–1800 Ramanshift (cm ), with an integration time of 10 s/scan. The Ramanspectra were the result of the average of three scans.

B. Reagents

Sodium borohydride (NaBH , 99.9%), sodium hydroxide(NaOH, 99%), copper (II) sulfate, pentahydrate (CuSO 5HO, 99.999% Cu) were purchased from Strem Chemicals, New-buryport, MA. The nitroaromatic explosive 1,2-dinitrobenzene,99% pure was obtained from ChemService, West Chester, PA.Methanol was purchased from Fisher Scientific in Fair Lawn,NJ. Trisodium citrate (C H O Na , 99%) was purchased fromSigma Aldrich, Nanopure, Barnstead International, Dubuque,IA (Thermo-Fisher Scientific) ultra-high-purity water (UHP,18.2 M ) was used for nanoparticles and all aqueous solutionspreparation. All glassware was cleaned thoroughly with aquaregia (HCl:HNO ), 3:1 %v/v, and rinsed with UHP (18.2 M ).

C. Preparation of Colloidal Suspensions

Cu colloidal suspensions were prepared according to themethod described by Sanchez et al., with some modifications[14]. Metallic copper colloids in the zero valence state wereprepared by chemical reduction. The amount of the aqueoussolution of trisodium citrate (C H O Na ) was varied in thedifferent colloidal copper suspensions.

Copper Seed 1: A typical experiment consisted on adding20 mL M aqueous solution of trisodium citrate(C H O Na ) to 1670 L of M copper (II) sulfate(CuSO ) in a 50 mL centrifuge tube. Then, a 10 mL mixture ofsodium hydroxide (NaOH) M and M sodiumborohydride (NaBH ) was added. Samples were purged withdry N after mixing the three solutions. Finally the tube wassealed and placed in an ice cold bath 0 C for 24 h.

Copper Seed 2: For the synthesis of batches of “copper seed2,” the same procedure as mentioned earlier was followed. How-ever, these batches were prepared using aliquots of 40 mL

M aqueous solution of trisodium citrate instead. The otherparameters were and details were the same as for the prepara-tion of copper seed 1.

Prepared Cu colloids were found to be relatively unstable asreported by others authors [15]. At the beginning of the syn-thesis they turned brown color with different intensities, de-pending on the amount of trisodium citrate solution added. Aftera 24 h period, the colloidal suspensions turned black-olive greenwith varying intensities.

D. Sample Preparation for Analysis by SERS

For SERS analysis, 800 L of the copper colloidal suspen-sion, 100 L of the sample to be analyzed (o- or p-dinitroben-zene), and 100 L of the sodium chloride solution (NaCl, 0.1M) were combined in a small centrifuge (1.5 mL) vial.

The solution was stirred in a mini vortex agitator for a periodof 30 seconds. A small amount of this mixture (1–2 cm in length)was transferred to a capillary tube sealed on one end (meltingpoint 1.8 mm 90 mm) for SERS measurements.

E. Theoretical Calculations and Models

To determine the type of interaction between o-DNB and Cunanoparticles, two calculations were performed: first, a simplesystem of a Cu atom and second, a portion of a copper surface

RIVERA-BETANCOURT et al.: SERS AND DENSITY FUNCTIONAL THEORY STUDY OF o-DINITROBENZENE ON CU NANOPARTICLES 701

model interacting with o-DNB. The results of both models re-sults were used to explain the interaction mechanism betweenCu and o-DNB.

Two assumptions were considered to construct the simple Cuatom model. One was to assume that the primary interactionbetween a Cu atom and o-DNB should be the most important.Second, was to assume that other interactions between distantCu atoms on a surface are minimal and they were negligible.This is a vague assumption to predict physical property values,but it could be used to extrapolate so that we could determinethe interaction energy.

To construct the Cu surface, the atoms were arranged in a sim-ilar way to its crystal structure of face centered cube. All copperatoms were placed at a distance of 0.3610 nm from each other.Two models were constructed for the interaction of o-DNB andthe Cu surface. One consisted in one of the two nitro groupsof o-DNB interacting more closely with the Cu surface that theother nitro group. The other model consisted of assuming thatboth nitro groups interact closely to the Cu surface. One of thesetwo models was later chosen as the bet one based on the one thatclosely resembled the experimental Raman spectra.

The DFT-based models were used to determine the en-ergies, atomic charges, molecular orbitals (MOs) diagramsand Raman spectra for neat o-DNB, and complexes formedbetween o-DNB and a copper atom and a the correspondingcomplexes formed with a limited portion of a Cu surface.All calculations were carried out with Gaussian 03 software[16]. DFT methods employed for the optimized structures,atomic charge, MOs, and Raman frequencies were calculatedby DFT-B3PW91/Lan12DZ.

III. RESULTS AND DISCUSSION

A. UV-VIS and TEM Results

The method of preparation of the colloidal solutions was op-timized to have the maximum SERS signal on o-DNB vibra-tional signatures. The concentration of the reducing agent, theaddition rate of the mixture of sodium hydroxide, and sodiumborohydride and the stirring rate of the solution were evaluated.

Prior to UV-VIS analysis, the colloidal solution was preparedby transferring different proportions of the colloidal suspen-sion and of UHP water in a 1-cm-quartz cuvette. The UV-VISspectra of the colloidal suspensions were acquired from 200 to800 nm. Results of UV-VIS spectrophotometry measurementsare included in Fig. 1.

Copper colloids prepared at different volumes of reducingagent show little difference in their spectral absorption charac-teristics, which may be attributed to morphological similarities.The absorption maximum of the colloidal suspensions providesinformation on the average particle size, whereas its FWHMgives an estimate of particle size distribution.

Cu colloids prepared with citrate had an SP at 568 nm for Cuseed 1 and 571 nm for Cu seed 2, showing small difference in av-erage particle sizes and were comparable with previous studies[14], [15]. An FWHM of 22 nm for Cu seed 1 and 26 nm for Cuseed 2 were obtained for the plasmon band absorptions. Thesevalues are indicative of relatively high monodispersity of the

Fig. 1. UV-VIS absorption spectra of Cu colloids. (a) Seed 1. (b) Seed 2.

Fig. 2. TEM images of Cu seed 1 (100 nm scale). Magnification: (a) 20 000xand (b) 50 000x.

colloidal Cu suspensions (see Fig. 1). The formation of a secondabsorbance band in the UV (ca. 293 nm) suggests the formationof a low-stability complex species of Cu with water molecules[17].

The shapes and sizes of these particles are better character-ized using TEM images. Colloidal preparations yielded particlesthat were approximately spherical seeds, as shown in Fig. 2. Alow-magnification TEM image (20 000x) is shown in Fig. 2(a).According to the TEM image with 50 000x magnification inFig. 2(b) shows that the club-shaped Cu-seed nanoparticles varyin sizes of 5–20 nm.

Fig. 3 contains TEM images of seeds 1 and 2 with averagenanoparticle sizes of about 2 nm. All of the surrounding parti-cles show conglomerates due to the aggregation of the nearlyspherical colloidal nanoparticles. Fig. 3(a) shows a conglom-erate with a network like structure. The characteristics of col-loidal nanoparticles vary with time, which can be ascertainedfrom the color of the suspension of particles. The average sizeand size distribution of the nanoparticles of the copper colloidsalso depend strongly on colloid age and temperature. Coppernanoparticles suspensions, as demonstrated in previous results,are highly monodispersed.

Precise temperature control is required to maintain steadygrowth of nanoparticles, as the size distribution sharpens. Thisis important to achieve large electromagnetic enhancements atthe metallic colloidal surface. Colloids in aqueous media, pre-pared by chemical methods, tend to form clusters because of

702 IEEE SENSORS JOURNAL, VOL. 10, NO. 3, MARCH 2010

Fig. 3. TEM images Cu colloidal suspensions of Cu seed 2. Magnification:(a) 20 000x and (b) 50 000x.

Fig. 4. Raman spectra of o-DNB: (a) solid; (b) methanol solution; and (c) SERSon Cu surface with 514.5-nm laser.

their complicated electrical double layer structure. These clus-ters have much larger morphological heterogeneities, since theyare composed of nanoparticles with a wide spectrum of sizesand shapes. This fact correlates well with the broader plasmonresonance extinction band, demonstrating the existence of smallparticles.

Aggregation is required for obtaining large surface enhance-ments of Raman signals of analytes adsorbed on the colloidalmetallic surface. Various salts can be used for inducing aggre-gation. Although, the most common ion is Cl the effect of theCl ion on the aggregation process was also investigated.

B. Raman Scattering Spectra of o-DNB on Cu Nanoparticles

Freshly prepared colloidal suspensions of copper nanoparti-cles were used for SERS analysis of o-DNB. The vibrationalexperiments showed that there is a correlation between thedecrease of the SERS enhancements and the age of the Cunanoparticles colloidal suspensions (data not shown).

Normal or spontaneous Raman scattering spectrum of solido-DNB and SERS spectra of o-DNB adsorbed on Cu nanoparti-cles excited 488- and 514.5-nm LLs are shown in Fig. 4(a)–(d).These two laser excitation lines were used for SERS analyses,because these two wavelengths are found near the Cu SP.

Solid o-DNB posses a C point group with 42 fundamentalvibrations [18], [19]. The normal Raman spectrum of solido-DNB possesses a strong band at 1360 cm attributed tothe NO (nitro group) symmetric stretch [18]. The SERSspectra in Fig. 4(b) and 4(c) shows two bands, one at 1305 and1307 cm and a second 1436 and at 1446 cm for 514.5and 488 nm excitation, respectively. The symmetric band at1306 in the normal Raman spectra appears as a splitting inthe SERS spectra in both excitation lines. These splitting canbe explained from the possible chemical interaction betweenthe copper atom and the o-DNB, causing the rupture of thesymmetric stretch degeneration.

Some of the bands are shifted from their location in the spec-trum for the solid, with values with respect to the sym-metric stretch around 53 and 18 cm , respectively, and for theasymmetric stretch around 15 and 39 cm . SERS spectra arein good agreement with that of the solid spectrum. A weakernitro vibrational mode that appears in the SERS spectra is dueto an out-of-plane wagging that occurs at 556 and 557 cm[19]. This suggests that the molecule is attached chemicallyto the Cu nanoparticles surface through its nitro group. Theband appearing at 1160 cm corresponds to a combination ofstretching modes; the C-N stretching and to the C-H in planebend. This effect was also observed for the band at 1109 cm ,which is a sum of the contribution of C-N and C-H stretchingmodes. There is also another C-H in plane bend appearing at1020 cm .

Surface enhancement factors (SEFs) were calculated ac-cording to Kneipp et al. [20]. Methanol does not presentsignificant SERS effect. This allows its use as an internalstandard for estimation of SEF values. The baseline correctednitro symmetric stretch band at 1307 cm was used for theestimation

(1)

The SEF values were: for 488 nm and for514.5 nm. The estimated values for 488 nm excitation are con-sistent with a chemical SERS effect related to charge transfer(CT) between the copper nanoparticles and o-DNB. However,SEF values higher than most likely have an electromag-netic field contribution to the estimated enhancements.

C. Copper Atom and o-DNB Interaction

The interaction energy was calculated using DFT-B3PW91/Lan12DZ by taking the difference between the energy of com-plex (Cu-o-DNB) and o-DNB plus the single copper atom. Thevalue obtained was KJ/mol. This suggests a chemical in-teraction for the above system [21]. The atomic charges werecalculated for the oxygen and nitrogen atoms of the nitro func-tional group of the Cu-o-DNB and o-DNB. The results showedan increase of the charge in nitro group. The value increased by

, equivalently to a 90% of the charge found in the copperatom. This demonstrates that a CT complex was formed be-tween the Cu and o-DNB, in which 90% of the total charge wastransferred to the nitro group. These results are in agreementwith SEF values, which indicated a chemical effect due to CTformation. The blue shifts found in the symmetric degenerated

RIVERA-BETANCOURT et al.: SERS AND DENSITY FUNCTIONAL THEORY STUDY OF o-DINITROBENZENE ON CU NANOPARTICLES 703

Fig. 5. MO. Diagram of: (a) o-DNB; (b) Cu-o-DNB; and (c) Cu.

Fig. 6. Proposed adsorption geometries of o-DNB. (a) on a copper atom and(b) on the Cu nanoparticle surface. Colors of atoms: red, oxygen; blue, nitrogen;red-brownish, copper; gray, carbon; and white, hydrogen.

band can be explained by the charge increase in the nitro group,which resulted in a strengthening of the O-N bond, making thevibration stiffer.

A chemical enhancement process (CT) must occur betweenhighest occupied MO (HOMO) and lowest unoccupied MO(LUMO) of o-DNB for resonance Raman and for normalRaman near the LUMO. This CT is more probable when theFermi level of the metal (Cu in this case) is found approxi-mately between HOMO and LUMO of the o-DNB (see Fig. 5)[22]. In the MO diagram for the Cu-o-DNB complex, it canbe seen that new orbitals were formed in the HOMO-LUMOregion of o-DNB, increasing the probability of CT formation.

DFT-B3PW91/Lan12DZ basis set was applied to calculateRaman frequencies of both computational models. The resultswere compared to the Raman spectra obtained experimentally at514.5 and 488 nm to determine which model explains the bestadsorption mechanism of o-DNB on the copper atom and on theCu nanoparticles surface (see Fig. 6).

Two different models were developed for the interaction ofo-DNB with the Cu surface. One of the models involved the in-teraction of one nitro group directly with the surface, and theother model involved the interaction of the two nitro groups.

Fig. 7. Potential energy of interaction of nitro groups of o-DNB and the Cusurface at different distances for: (a) one NO group and (b) two NO groups.

Fig. 8. Raman spectra of o-DNB: (a) neat solid and (b) DFT calculated.

Different calculations were developed to determine the min-imum energy of interaction between the nitro groups in o-DNBat different distances to the Cu surface (see Fig. 7). From Fig. 7,for the model of one nitro group of o-DNB interacting with theCu surface, the minimum energy of interaction is kJ/molat a distance of 2.1 Å. For the other model, which involves twoNO interacting with the Cu surface, the minimum energy ofinteraction is kJ/mol at a distance of 1.8 Å. According tothe results of the DFT calculations, the interaction of both nitrogroups of o-DNB on the Cu nanoparticles surface closely re-sembles the SERS spectra obtained, both at 514.5 and 488 nm.

D. Comparison of SERS and Calculated Raman Spectra

Normal or spontaneous Raman scattering spectrum of solido-DNB and theoretical spectra of o-DNB are shown in Fig. 8.Aside from the expected shift in frequencies throughout all fin-gerprint region, based on the fact that the selected basis setis represents a compromise between the ideal set for organiccompounds and the corresponding set of functions for metallicsubstrates, a reasonable good agreement was found betweengas-phase molecule and the solid-phase o-DNB. Differences in

704 IEEE SENSORS JOURNAL, VOL. 10, NO. 3, MARCH 2010

TABLE IEXPERIMENTAL AND CALCULATED RAMAN FREQUENCIES OF O-DNB ADSORBED ON CU NANOPARTICLES

Fig. 9. Comparison of SERS and theoretical spectra of interactions of o-DNBon Cu. (a) SERS spectrum excited at 514.5-nm LL. (b) o-DNB interacting witha single Cu atom. (c) Interaction with a modeled.

intensity patterns are not significant since polarizability compo-nents for the two molecular systems have not tractable compar-ison. Fig. 8 is actually a validation of the computational method-ology used. A good agreement between SERS and the calculatedfrequencies can be assessed by comparing SERS spectra withmodel-based calculated spectra that could lead to the interpre-tation of the interaction process between o-DNB molecules andCu colloidal nanoparticles on the surface of the later. In partic-ular, DFT calculations on the interaction processes can be a verypowerful tool to get a better insight of the correlation betweenthe observed frequency shifts, going from the normal or spon-taneous Raman spectrum to the SERS spectrum of o-DNB andshedding light on the mechanism of the interaction between theanalyte and copper surface.

Experimental SERS frequencies of o-DNB compared withthe calculated frequencies of the single atom and surface models

Fig. 10. SERS and calculated spectra: (a) SERS spectrum excited at 514.5 nm.Calculated Raman spectra for NO are attached to the Cu surface; (b) one grouptwo NO groups; and (c) two NO groups.

of the two nitro groups on the Cu surface using DFT-B3W91with basis set LanL2DZ are shown in Table I. Correspondingspectra are illustrated in Fig. 9. As can be seen from Table II,the difference (delta) in values for the Cu surface compared tothe Cu single atom model are small, providing better agree-ment with experimental frequencies. The calculated spectrumthat closely resembles that of SERS spectra is the one in whichthe two nitro groups interact closely with the Cu surface, as canbe seen in Fig. 9. The deviations of the Raman shifts from theCu single atom model are larger than those found for the Cusurface model. This can be taken as an indication that the Cusurface model provides a more adequate model to predict theadsorption behavior of o-DNB on the Cu nanoparticle surface.

Fig. 10 shows the comparison between SERS spectrum ex-cited at 514.5 nm LL and the corresponding calculated spectrausing a modeled Cu surface comprising of nine atoms, all placedat the face-centered cubic-structure atomic distances. Fig. 10(b)

RIVERA-BETANCOURT et al.: SERS AND DENSITY FUNCTIONAL THEORY STUDY OF o-DINITROBENZENE ON CU NANOPARTICLES 705

TABLE IICOMPARISON BETWEEN OBSERVED AND CALCULATED FREQUENCY SHIFTS

shows the results of the calculated SERS spectrum when oneNO group of o-DNB is anchored to the Cu surface. When twoNO groups are attached to the Cu surface [see Fig. 10(c)] agree-ment between experimental and simulated Raman spectra is sig-nificantly improved.

Information from the relative intensity of the bands can beused to judge for the explanation of the polarizability compo-nents orientation on a surface. According to Fig. 10, the nitrovibration of o-DNB on the SERS spectrum decrease in intensitycompared to the solid one, according to SERS selection rules[23], this indicates that N-O mode is not totally perpendicularto the surface of the Cu nanoparticles.

IV. CONCLUSION

In this contribution, Cu nanoparticles with optimum SERSproperties were successfully synthesized. They were used asRaman substrates for detection of o-DNB. The SEF values ob-tained demonstrate that a chemical mechanism is responsiblefor most of the SERS effects observed with this analyte on Cunanoparticle suspensions.

Experimental and theoretical work was aimed at modelingthe formation of a chemical complex between o-DNB and Cuatoms. Computational models together with the electromagneticmechanism do not completely explain the SERS phenomenon,as has been widely used in the SERS literature. In the explana-tion of the adsorption mechanism, postulation of formation of a

CT complex sets a different point of view for the obtained exper-imental SERS results: SEF values, vibrational band shifts, andstrengthening/weakening of chemical bonds, and theoretical re-sults: MOs analysis and CT analysis. Based on the comparisonof the two computational models used, it was assumed that thenitro groups in o-DNB do not interact strongly (chemically)with the copper nanoparticles. A stronger interaction could giverise to higher SEF values: stronger SERS effects.

ACKNOWLEDGMENT

The authors would like to thank contributions from AaronLaPointe of Night Vision and Electronic Sensors Directorate,Department of Defense. High Resolution transmission electronmicroscope images were measured at the Materials Character-ization Center of the Rio Piedras Campus of the University ofPuerto Rico by O. Resto, whom we recognize for his dedicationand commitment to the scientific community in Puerto Rico.The views and conclusions contained in this document are thoseof the authors and should not be interpreted as necessarily rep-resenting the official policies, either expressed or implied, of theU.S. Department of Homeland Security.

REFERENCES

[1] P. K. Chu and L. Liuhe, “Characterization of amorphous and nanocrys-talline carbon films,” Mater. Chem. Phys., vol. 96, pp. 253–277, 2006.

706 IEEE SENSORS JOURNAL, VOL. 10, NO. 3, MARCH 2010

[2] K. Kim, S. Lee, and K. L. Kim, “Surface-enhanced Raman scatteringof 4-nitrothioanisole in Ag sol,” J. Phys. Chem. B, vol. 108, pp.16208–16212, 2004.

[3] R. F. Aroca, R. A. Alvarez, N. Pieczonka, S. Sanchez, and J. V. Garcia,“Surface enhanced Raman scattering on colloidal nanostructures,” Adv.Colloid Interface Sci., vol. 116, pp. 45–61, 2005.

[4] M. V. Canamares, J. V. Garcia-Ramos, J. D. Gomez-Varga, C.Domingo, and S. Sanches-Cortes, “Comparative study of the mor-phology, aggregation, adherence to glass, and surface-enhancedRaman scattering activity of silver nanoparticles prepared by chemicalreduction of Ag using citrate and hydroxylamine,” Langmuir, vol.21, pp. 8546–8553, 2005.

[5] R. S. Sheng, L. Zhu, and M. D. Morris, “Sedimentation classification ofsilver colloids for surface-enhanced Raman scattering,” Anal. Chem.,vol. 58, pp. 1116–1119, 1986.

[6] K. L. Kelly, E. Coronado, L. L. Zhao, and G. C. Schatz, “The opticalproperties of metal nanoparticles: The influence of size, shape, and di-electric environment,” J. Phys. Chem. B, vol. 107, pp. 668–677, 2003.

[7] J. A. Creighton and D. G. Eadon, “Ultraviolet-visible absorptionspectra of the colloidal metallic elements,” J. Chem. Soc., FaradayTrans., vol. 87, pp. 3881–3891, 1991.

[8] A. Pyatenko, M. Yamaguchi, and M. J. Suzuki, “Laser photolysis ofsilver colloid prepared by Citric acid reduction method,” Phys. Chem.B, vol. 109, pp. 21608–21611, 2005.

[9] D. S. Moore, “Instrumentation for trace detection of high explosives,”Rev. Sci. Inst., vol. 75, pp. 2499–2512, 2004.

[10] F. Tielens, M. Saeys, E. Tourwe, G. B. Marin, A. Hubin, and P. Geer-lings, “An ab initio study of the interaction of SCN with a silver elec-trode: The prediction of vibrational frequencies,” J. Phys. Chem. A, vol.106, pp. 1450–1457, 2002.

[11] M. Muniz-Miranda, B. Pergolese, and A. Bigotto, “SERS and DFTstudy of nitroarenes adsorbed on metal nanoparticles,” Vib. Spectosc.,vol. 43, pp. 97–103, 2007.

[12] M. Lazzeri and F. Mauri, “Phonon dispersion and lifetimes in MgB ,”Phys. Rev. Lett., vol. 90, pp. 95506–95510, 2003.

[13] A. Markovits, M. Garcia-Hernandez, J. M. Ricart, and F. Illas, “The-oretical study of bonding of carbon trioxide and carbonate on pt(111):Relevance to the interpretation of “in situ”,” J. Phys. Chem. B, vol. 103,pp. 509–518, 1999.

[14] S. Sanchez, J. V. Garcia, and G. Morcillo, “Morphological study ofmetal colloids employed as substrate in the SERS spectroscopy,” J.Colloid Interf. Sci., vol. 167, pp. 428–436, 1994.

[15] L. Ping and Y. Fang, “The study of resonance Raman scattering spec-trum on the surface of Cu nanoparticles with ultraviolet excitation anddensity functional theory,” Spectrochimica Acta Part A, vol. 67, pp.767–771, 2007.

[16] M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb,J. R. Cheeseman, J. A. Montgomery Jr., T. Vreven, K. N. Kudin, J. C.Burant, J. M. Millam, S. S. Iyengar, J. Tomasi, V. Barone, B. Men-nucci, M. Cossi, G. Scalmani, N. Rega, G. A. Petersson, H. Nakatsuji,M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida,T. Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J. E.Knox, H. P. Hratchian, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo,R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi,C. Pomelli, J. W. Ochterski, P. Y. Ayala, K. Morokuma, G. A. Voth,P. Salvador, J. J. Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D.Daniels, M. C. Strain, O. Farkas, D. K. Malick, A. D. Rabuck, K.Raghavachari, J. B. Foresman, J. V. Ortiz, Q. Cui, A. G. Baboul, S. Clif-ford, J. Cioslowski, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz,I. Komaromi, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y.Peng, A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson,W. Chen, M. W. Wong, C. Gonzalez, and J. A. Pople, Gaussian 03,Revision C.02. Wallingford, CT: Gaussian Inc., 2004.

[17] B. Fox, P. Balaj, I. Balteanu, M. Beyer, and V. Bondybey, “Single mol-ecule precipitation of transition metal (I) chlorides in water chlorides,”J. Amer. Chem. Soc., vol. 124, pp. 172–173, 2001.

[18] V. Suryanarayana and G. Ramana, “Transferable valence force fieldsfor substituted benzenes part iv. Dinitrobenzenes,” Vib. Spectrosc., vol.8, pp. 343–356, 1995.

[19] J. H. S. Green and H. A. Lauwers, “Vibrational spectra of benzenederivatives-xiii the dinitrobenzenes,” Spectrochimica Acta Part A, vol.27 A, pp. 817–824, 1971.

[20] K. Kneipp, Y. Wangc, R. R. Dasari, M. S. Feld, B. D. Gilbert, J. Janni,and J. I. Steinfeld, “Near-infrared surface-enhanced Raman scattering(NIR SERS) on colloidal silver and gold,” Spectrochimica Acta Part A,vol. 51, pp. 2171–2175, 1995.

[21] G. Attard and C. Barnes, Surfaces. Oxford, NY: Oxford, 2006, ch. 1,pp. 9–10.

[22] M. J. Banholzer, J. E. Millstone, L. Qin, and Ch. A. Mirkin, “Rationallydesigned nanostructures for surface—Enhanced Raman spectroscopy,”Chem. Soc. Rev, vol. 37, pp. 885–897, 2008.

[23] M. Moskovits, “Surface selection rules,” J. Chem. Phys, vol. 77, pp.4408–4416, 1982.

Omar E. Rivera-Betancourt was born in San Juan,Puerto Rico. He received the B.S. degree in chem-istry at the Department of Chemistry, University ofPuerto Rico-Mayaguez, Mayaguez, where he is cur-rently working towards the M.S. degree.

His current research interest include the synthesisof copper nanoparticles using wet chemical methodsfor surface-enhanced Raman scattering.

Oliva M. Primera-Pedrozo was born in Colombia.She received the BS degree in chemistry from theUniversity of Cartagena, Colombia, and the MSdegree in chemistry from the University of PuertoRico-Mayagüez (UPRM), Mayagüez. She is cur-rently working towards the Ph.D. degree in materialschemistry at the Department of Chemistry, UPRM.

Her research interests include synthesis of goldnanorods and nanocubes for surface-enhancedRaman spectroscopy (SERS) applications. Duringthe last three years, she has participated in projects

related to vibrational spectroscopy (IR, SERS, and Raman) detection ofexplosives, chemical warfare agents, and pharmaceutical ingredients.

Leonardo C. Pacheco Londoño was born inColombia. He received the B.S. degree in chem-istry from the University of Cartagena, Cartagena,Colombia, and the M.S. degree in chemistry fromthe University of Puerto Rico-Mayagüez, Mayagüez(UPRM), where he is currently working towardsthe Ph.D. degree in materials chemistry at theDepartment of Chemistry.

Samuel P. Hernández-Rivera received the B.S.and M.S. degrees from the University of PuertoRico-Mayagüez (UPRM), and the Ph.D. degree fromThe Johns Hopkins University, Washington, DC.

He is currently a Professor of Physical Chemistryat the Department of Chemistry, UPRM. He is alsothe Director of the Department of Defense-MultipleUniversity Research Initiative Sponsored Center forSensors Development and the UPRM-based Chem-ical Imaging Center. His research interests includeIR/Raman microspectroscopies in applied and fun-

damental studies and standoff and point detection of explosives, chemical andbiological agents using vibrational spectroscopy. Other interests also includesurface-enhanced Raman spectroscopy and in chemometrics enhancement ofspectroscopic data and discriminant and cluster analysis of populations.

Prof. Hernández-Rivera is a member of the American Chemical Society,American Physical Society, Sigma Xi, Phi Kappa Phi, Society for Photo-OpticalInstrumentation Engineers (SPIE), and the Puerto Rico Chemists Association.He is also a member of the editorial board of Springer’s Sensing and Imaging:An International Journal.