Digital Protocol for Chemical Analysis at Ultralow Concentrations by

Dec 13, 2017 - Single molecule surface-enhanced Raman spectroscopy (SM-SERS) has the potential to revolutionize quantitative analysis at ultralow conc...
1 downloads 10 Views 1MB Size
Subscriber access provided by Gothenburg University Library

Article

A Digital Protocol for Chemical Analysis at Ultra-Low Concentrations by Surface-Enhanced Raman Scattering Carlos Diego L de Albuquerque, Regivaldo Gomes Sobral Filho, Ronei Jesus Poppi, and Alexandre G. Brolo Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.7b03968 • Publication Date (Web): 13 Dec 2017 Downloaded from http://pubs.acs.org on December 15, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 8 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

A Digital Protocol for Chemical Analysis at Ultra-Low Concentrations by Surface-Enhanced Raman Scattering Carlos Diego L. de Albuquerque†,‡, Regivaldo G. Sobral-Filho†, Ronei J. Poppi‡, Alexandre G. Brolo†,§* †Department of Chemistry, University of Victoria, Victoria, BC V8P 5C2, Canada ‡Institute of Chemistry, University of Campinas (Unicamp), CP 6154, 13084-971 Campinas, São Paulo, Brazil. §Center for Advanced Materials and Related Technologies (CAMTEC), University of Victoria, Victoria, BC, V8W 2Y2, Canada ABSTRACT: Single molecule surface-enhanced Raman Spectroscopy (SM-SERS) has the potential to revolutionize quantitative analysis at ultra-low concentrations (less than 1 nM). However, there are no established protocols to generalize the application of this technique in analytical chemistry. Here, a protocol for quantification at ultra-low concentrations using SM-SERS is proposed. The approach aims to take advantage of the stochastic nature of the single-molecule regime to achieved lower limits of detection (LOD) and quantification (LOQ). Two emerging contaminants commonly found in aquatic environments, enrofloxacin (ENRO) and ciprofloxacin (CIPRO), were chosen as non-resonant molecular probes. The methodology involves a multivariate resolution curve fitting known as non-negative matrix factorization with alternating least squares algorithm (NMF-ALS) to solve spectral overlaps. The key element of the quantification is to realize that, under SM-SERS conditions, the Raman intensity generated by a molecule adsorbed on a “hotspot” can be digitalized. Therefore, the number of SERS event counts (rather than SERS intensities) was shown to be proportional to the solution concentration. This allowed the determination of both ENRO and CIPRO with high accuracy and precision even at ultra-concentrations regime. The LOD and LOQ for both ENRO and CIPRO were achieved at 2.8 pM. The digital SERS protocol, suggested here, is a road-map for the implementation of SM-SERS as a routine tool for quantification at ultra-low concentrations.

Single-molecule surface-enhanced Raman scattering (SMSERS) was recognized 20 years ago1,2 as a potential tool to revolutionize the field of analytical spectroscopy. However, although there are several examples of cleverly engineered nanostructures and approaches that allowed reliable SM-SERS measurements3-11, there still are no reports of an analytical determination that takes full advantage of the phenomenon. Typically, an analytical chemist seeks to obtain a linear calibration curve that will correlate the measurements to the concentrations of the analyte in solution. Unfortunately, there are a few caveats that challenge the linearity of the calibration curve and the use of SM-SERS as direct probe in ultra-low concentrations. The SM-SERS phenomenon occurs for molecules adsorbed in special areas in a nanostructured metal surface, called SERS “hotspots”12. A SERS hotspot is a sub-wavelength region under the area illuminated by the excitation laser that exhibit local enhanced field due to the excitation of surface plasmons12. The electric field strength distribution around a hotspot is highly inhomogeneous and large enhancements are observed only within a few nm away from the surface 2. Moreover, in a random SERS substrate, such as a metal colloid deposited in glass, a distribution of hotspots with different efficiencies is expected13. In the SM-SERS limit, only molecules adsorbed on highly efficient hotspots contribute to the overall Raman signal 14, 15. The bottom line is that highly efficient hotspots are rare and the amount of adsorbed analytes is expected to also be small at ultra-low concentrations. Hence, SM-SERS experiments are characterized by fluctuations (either temporal or spatial) in Raman intensities1,16, due to the low probability of a single molecule to find, by chance, an efficient SERS hotspot. The threshold for fluctuations in SERS intensities (SM-SERS regime) depend then on several factors, including the density of hotspots, the distribution of hotspot strengths, the SERS

cross-section of the molecule, and the number of molecules illuminated by the probing laser17. In terms of quantification, the strong intensity fluctuations observed in the SM-SERS regime should affect the linearity of the calibration curve. Some groups have attacked this issue by, for instance, selectively direct the species of interest to the hotspot18,19. Here we are proposed a general method that actually embraces the stochastic characteristics of the SM-SERS intensity fluctuations. This quantification method can be applied for any type of SERS-active analyte, even non-resonant molecules, and for a variety of planar SERS substrate. The protocol is based on the concept of digital assays, a wellestablished approach in biomedical research.20 The analytical method reported here should then provide an avenue for analytical chemists to finally take advantage of the SM-SERS phenomenon for quantification, and fulfill the potential of the technique touted since its discovery 20 years ago.

EXPERIMENTAL SECTION Chemicals. HPLC degree (≥ 98%)Enrofloxacin (ENRO) and ciprofloxacin (CIPRO-C12,N14) were supplied by Sigma Aldrich. CIPRO-C13,N15 isotope in HCl, 100 µg mL-1 in methanol, HPLC degree (≥ 98%), was purchased from Cambridge Isotope Laboratories. Gold (III) chloride trihydrate (HAuCl4.3H2O, 99.9%), sodium citrate tribasic dihydrate (C6H5Na3O7.2H2O, 99%) and (3aminopropyl)trimethoxysilane (APTMS, 97%) were purchased from the Sigma Aldrich (St. Louis, US). The chemical structure of CIPRO-C12,N14, CIPRO-C13,N15 and ENRO are shown, respectively, in the supporting information file (SI file, Figure S3). Synthesis of AuNPs. The synthesis of gold nanoparticles (AuNPs) was carried out using a procedure described elsewhere21. Briefly, 37 µL of gold (III) chloride trihydrate was diluted in a 25 mL volumetric flask and then transferred to a

1 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

250 mL Erlenmeyer flask, where the volume was adjusted by adding water up to 100 mL. The solution was heated until boiling at constant stirring and then 3 mL of citrate sodium (1%) solution was added. A change in color to red-wine indicated the formation of the gold nanoparticles and the heating was stopped. The AuNPs colloid was characterized by UV-vis and TEM (see SI, Fig. S4). Immobilization of AuNPs on coverslip glass. Microscope coverslip glass (18 mm (W) x 18 mm (L) x 0.15 mm (T)) were purchased from Fisher Scientific®. The coverslip glasses were carefully washed in piranha solution (3:1 concentrated H2SO4:30% H2O2) for 1 h (CAUTION! This reaction releases large amounts of corrosive fumes) and then rinsed thoroughly with high amounts of water. Then, the coverslip glasses were allowed to dry in an oven (~ 1 h). The cleaned coverslip glasses were placed in a Petri dish and completely covered by an 30% APTMS solution in toluene. The immersed coverslips were left immersed overnight to ensure uniform molecular packing at the surface. The coverslip glasses modified with APTMS (glass-APTMS) were then removed from the Petri dish, thoroughly washed with ethanol (HPLC grade) and then immersed in (HPLC grade) ethanol overnight. This ethanol washing/immersion procedure was repeated once a day for 5 consecutive days. After the 5th day, the glass-APTMS was removed from the ethanol and dried in an oven at 110 oC for 3 h. The drying step facilitates cross-linking of the siloxane (SiO-Si) bonds. Additional details about the glass functionalization procedure can be found elsewhere22. The glass-APTMS substrates were immersed in the AuNPs suspension (diluted 1:1 with ultrapure water) for 2 h under constant stirring. This procedure led to a highly homogeneous coverage of AuNPs on the glass surface. Finally, the glassAPTMS-AuNPs substrates were rinsed with water, dried using a gentle N2 flow, and stored under air. The homogeneity of the resulting substrate (glass-APTMS-AuNPs) was probed by both AFM and SERS (see the SI file). SERS measurements. SERS mapping measurements, depicted in Figure 1, were collected on a dispersive spectrometer inVia Raman confocal microscope system (Renishaw) equipped with a He–Ne laser source at 632.8 nm, a 50x objective (NA = 0.75), and a motorized stage control. The measurements were carried out using the StreamLine® (Renishaw) operation mode. The mapped spectral range was from 1086 to 1667 cm-1. A SERS band observed at ~1390 cm-1 was assigned to the strongest O-C-O stretching mode23, which is common for all analytes (CIPRO-C12,N14, CIPRO-C13,N15 and ENRO). All spectral acquisition was performed using 50% of the laser power (around 10 mW at the laser head), 5 s acquisition time and 1 accumulation. Aqueous analyte solutions, ENRO or CIPRO-C12,N14 or a CIPRO-C12,N14 and CIPRO-C13,N15 mixture, were added to the SERS substrate (glass-APTMS-AuNPs) as illustrated in Figure 1 . For each experiment, 1 µL of the analyte solution was dropped onto substrate (the average diameter of the drop on the slide was ~1 mm). Then, the drop was dried using a gentle N2 gas flow. The dry sample avoids adsorption dynamics24, dilution factors25 and cross contamination since a high numerical aperture (NA) objective can be used without risking contact between the lens and the solution, which could severely contaminate the results at ultra-low concentrations. Notice that, since we are using a dried sample, the edges of the sample stain in the substrate were excluded to avoid bias due

Page 2 of 8

to, for instance, coffee ring effects. Our approach shows reasonable homogeneity only when the middle of the dried sample was probed. The laser was then positioned to the middle of the dried drop using the motorized stage control. The mapped areas were 50 x 50 µm2 with 5 µm step size, totaling 121 SERS spectra (pixel) per map. The process was repeated for each one of the concentrations in different regions around the middle of the dried sample spot.

Figure 1. Scheme illustrating the experimental procedure carried out in this work.

Data analysis. The raw spectra were stored in .txt format and directly exported to the Matlab (version 7.12.0) environment. The raw spectra were preprocessed as follow: 1) Outlier spectra were removed; 2) the remaining spectra was smoothed using a Savitzky-Golay filter26; 3) The baseline was corrected using an asymmetric least square method27. After preprocessing, the data were first analyzed by PCA to determine the number of factors, and then the non-negative matrix factorization with alternating least square algorithm (NMF-ALS) resolution method was employed 23,28,29. Finally, the SERS response in each pixel was “digitized” by attributing values of “1” or “0” when the SERS response was either “above” or “below” a predetermined threshold. A comprehensive description of the data analysis procedure, including the original spectra and the resulting NMF loads, can be found as supporting information.

RESULTS AND DISCUSSION The SERS substrate chosen for this work was simply Au colloids21 immobilized on an aminated glass slide (glassAPTMS-AuNPs). This common substrate assures that the procedure developed here can be widely tested, since it does not require high-end fabrication tools. The molecular probes used for the method development, enrofloxacin (ENRO) and ciprofloxacin (CIPRO), are nonresonant molecules (they don’t have electronic absorptions that coincide with the wavelength of the laser excitation). As one of the most widely used type of fluoroquinolone antibiotics, ENRO and CIPRO are prominent emerging contaminants that can be found as pollutants in surface waters22-24. The problem of quantification at ultra-low concentrations in SM-SERS can be viewed as a typical sampling

2 ACS Paragon Plus Environment

Page 3 of 8 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

issue, which is common in analytical chemistry, and recently has been framed in the context of SERS-based assays with nanoparticles27. Sampling errors can be curtailed by increasing the sample size. For instance, increasing the laser illuminated area to several hundreds of m2 would guarantee that a larger number of molecules adsorbed to efficient hotspots are probed, even at ultra-low concentrations. However, this approach might suffer from lower excitation power density, and a decreased efficiency in the collection optics relative to a high NA objective microscope lens. Raman microscopes are commercially available and they allow tight focus in small areas (~ 1 m2 – depending on the excitation wavelength). Prolonged laser illumination in such a small area in a SERS substrate can lead to photodecomposition. Therefore, the best approach to increase sample size is through spatial mapping28, 29. In the case of spatial mapping, depicted in Figure 1, a diffraction limited laser excitation spot is scanned (X-Y directions) through the surface of the SERS substrate, and a SERS response is recorded from each laser position (pixel). SERS maps were then performed in a central 50 x 50 µm2 area on the dried sample spot on the substrate. The SERS intensities from all maps from different concentrations (121 data points each) were then evaluated using the non-negative matrix factorization with alternating least squares algorithm (NMF-ALS)30. The NMF-ALS is a multivariate curve resolution (MCR) method that have already been applied in SERS (see details in the SI)23, 31. Evaluation of distribution of scores in ensemble and SMSERS conditions. Figure 2 shows normalized NMF scores (equivalent to SERS intensities) histograms obtained from the SERS mappings (details in SI) at two concentrations of ENRO. At high concentrations (278 M), shown in Figure 2A, every pixel illuminated by the laser in the map was expected to contain a large number of adsorbed species (the probability of observing a SERS signal in every illuminated area was high). Ideally, the molecules of the analyte were homogeneously distributed on the surface and the variation in scores values in Figure 2A should reflect the spatial distribution of SERS efficiency expected from a random SERS substrate. The red line is the mean score, suggesting that the central limit theorem was followed in Figure 2A (“quasi-normal” distribution). A typical spatial variation in SERS scores obtained from the maps was of the order of 15% RSD. At lower concentrations, such as in Figure 2B, the shape of the SERS distribution histograms changes from a quasi-normal behavior towards a tailed behavior (noted that a true Gaussian was not obtained in Figure 2A due to the small sample size (number of pixels)). The tailed distribution observed in Figure 2B is a consequence of a decreased probability of molecules to find highly efficient hotspots1,15,17. In that case, a large number of pixels yielded small score values (SERS intensities) than the red line (average), while a few pixels produced SERS signal much larger than the average15.

Figure 2. Histogram showing the frequency of normalized scores from “Factor 1”. “Factor 1” corresponds to the spectral signature of the analyte obtained by PCA. (A) high concentration regime (278 µM or 2.13 x 108 molecules/µm2) and (B) ultra-low concentration regime (278 pM or 213 molecules/µm2). The red line is the mean value for each distribution. More details can be found in the SI.

Evaluation of calibration curves in ensemble and SMSERS conditions. Pseudo-calibration curves were plotted using the average values of the NMF-ALS scores at 95% confidence level and the surface density of the analyte (proportional to the solution concentrations). Typical pseudocalibration curves are shown in Figure 3. A good linear relationship is observed in Figure 3A. This is in agreement with several previous reports that demonstrated that reliable calibration curves can be obtained by SERS30. The variation in scores at high concentrations, Figure 3A, should reflect the spatial variability of the SERS efficiency of the substrate (assuming that the surface coverage is uniform at high concentrations). Moreover, Figure 3A shows that the SERS signal tracks the concentrations within a very wide dynamic range. The estimated error in a concentration determination using this relative simple substrate was ±13%.

3 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 8

Figure 3. Pseudo-calibration curves for ENRO. (A) “high concentration” regime (B) “low concentration” regime. The confidence level was 95%.

The extended linearity observed in Figure 3A is lost when the concentration of ENRO decreased to ultra-low levels (less than 70.0 nM), as shown in Figure 3B. As the concentration decreases, the number of species under the illuminated area (1 µm2) also decreases. Since highly enhancing hotspots are rare14, the probability of multiple molecules occupying hotspots under the probed area also becomes very small. In extreme conditions, the observed SERS signal from a pixel is believed to be derived from single species adsorbed at a (rare) highly efficient hotspot. The enhancement efficiency vary widely between different hotspots (and with the position of the molecule within a hotspot1); hence, a small number of SERS events should have a variety of different SERS intensity values, leading to an average with very large variations. This is well-reflected in Figure 3B, where the linearity, accuracy and precision of the calibration curve was lost. Notice that other groups observed the same sort of limitation (lack of linearity in the calibration curve) when studying ultra-low concentrations by SERS31. Digital calibration curve for SM-SERS experiment. The limitation encountered in Figure 3B can be overcome by “digitizing” the SERS response. The rationale is that, when the SM-SERS regime is achieved, the SERS signal from a particular pixel will most probably originate from a single molecule. Therefore, it is not necessary to consider the Raman intensity, but only attribute the intensity to a single molecule event. It is then possible to correlate the solution concentration with the number of pixels that present a SERS response (digital counts) within a given mapped area (instead of the Raman intensity).

Figure 4. Digital mapping for several concentrations of ENRO under SM-SERS regime. The black squares correspond to positive SMevents (assigned to 1). The white squares correspond to negative SM-events (assigned to 0). The estimate number of molecules per area (µm2) is indicated in red. The laser illuminated area was about 1 µm2.

Figure 4 shows the digitized version of the SERS maps obtained at different (low) concentrations of ENRO. Each black spot in Figure 4 correspond to a pixel that yielded a SERS response (scores above a set threshold) that can be assigned to the vibrational signature of ENRO. The white spots in Figure

4 were assigned to the null events. The estimated number of molecules/m2, calculated from the solution concentration, is also indicated in Figure 4. Notice that the digital counts (black

4 ACS Paragon Plus Environment

Page 5 of 8 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry spots) decreased with the analyte concentration, as expected (Figure 4). Figure 5 presents a revised version of Figure 3B, where the NMF scores were substituted by the “SERS digital counts” (“number of black spots”). Note that the digital counts should be equivalent to the number of single molecules identified by SERS during the mapping. A linear digital calibration curve was then recovered and plotted in Figure 5 (equivalent data for CIPRO is shown as SI). It is clear from Figure 5 that the linear range of the digital calibration curve was extended to lower concentrations using this procedure, potentially allowing quantification at ultra-low concentration levels.

Figure 5. Digital Calibration curve for ENRO at ultra-low concentration.

The limit of quantification (LOQ) for ENRO estimated from the data in Figure 5 were 2.8 pM or about 2 molecules per µm2. Proof of SM-SERS regime: Isotopologue experiment for the digital calibration approach. After the demonstration of the digital SERS calibration (Figure 5), it is appropriated to confirm the main assumption; i.e., that the SM-SERS regime was achieved at those ultra-low concentrations. There are several approaches to prove the SM regime in SERS, these included measuring Stokes/anti-Stokes ratio32; investigating changes in the shape of the SERS intensity histograms; 33 and analyzing the characteristics of the SERS intensity fluctuations17,34,35. However, the most convincing argument of SMSERS has arisen from studies with isotopologues (molecules of the same compound, so same affinity to the surface, that differ due to isotopic composition)36-38. The isotopologue SERS experiment demonstrated that the unique vibrational signature of both (isotopologues) analytes rarely appear mixed and most of the individual events show the vibrational fingerprint from either one or the other compound. This agrees with the idea that single molecule events are rare and confirms (by an unique spectroscopic signature) that the analyte producing the SERS signature is alone at a highly efficient hotspot. Figure 6 shows SERS mapping obtained from mixtures of two isotopologues of CIPRO (CIPRO-C12,N14 and CIPROC13,N15 - molecular structures are indicated in the SI) of dif-

ferent compositions. The results in Figure 6 were obtained by adding a 1 L of mixtures (at different concentration ratios) of both isotopologues (CIPRO-C12,N14 and CIPRO-C13,N15) on glass-APTMS-AuNPs, following the same procedure as in Figures 3 and 4. In contrast to previous isotopologues SERS experiments, the SERS fluctuations were not observed against time (this would not be reliable because the samples were dried and illumination on a fixed spot could lead to severe photodecomposition)32. The spectral signature from each pixel was analyzed using the NMF-ALS resolution method, which recovered the unique vibrational signature of each isotopologue. Any pixel that showed exclusively the presence of CIPRO-C12,N14 is colored “blue” in Figures 6A and 6C. On the other hand, pixels with a vibrational signature of the isotopically substituted CIPRO (CIPRO-C13,N15) is colored “red” in Figures 6B and 6D. The SERS digital counts are higher for both isotopologues at 140 pM (about 30 to 31 pixels) than at 14 pM (about 6 to 10 pixels). This is in agreement with the assumption that, at very low concentrations, the probability for a molecule to find a hotspot is small, and it becomes rarer when the concentration of specie(s) drastically decreases (140 pM to 14 pM). Secondly, it is important to emphasize that most of the pixels present either spectral noise or a signature of one of the analytes (CIPRO-C12,N14 or CIPRO-C13,N15). Nine out of 61 positive events (30 and 31 events of CIPROC12,N14 and CIPRO-C12,N14, respectively), ~13%, are represented in “purple” colour in Figures 6A and 6B. These “purple” events revealed mixtures between the two isotopologues at the same spot at 140 pM. On the other hand, only one pixel show the presence of the two compounds simultaneously when the concentration was decreased to 14 pM (Figure 6C and 6D), representing ~6% of the positive events. In Summary, the probability of observing single molecule events in a particular pixel is higher when the concentration is reduced from 140 pM (87%) to 14 pM (94%). These observations are strong indications that the single molecule regime was attained and it corroborate previous studies 38. The results reported here indicate that SM-SERS combined with chemometrics can be used to directly quantify analytes at ultra-low concentrations; i.e., without requirements for preconcentration. Separation steps might also be excluded, since SERS provided vibrational fingerprint identification 39. It is important to emphasize that the procedure described here assumes SM-SERS conditions and an uniform distribution of the analyte at the surface. Spatial variation of SERS efficiency of the substrate is also an important parameter. Substrates with spatial variation in SERS intensities smaller than 20% RSD are required for this procedure40,41. SERS substrates with low spatial variation in intensities are now commercially available and that should not limit the applicability of the method. In fact, the limiting aspect of the procedure should be related the size of the sample set or, in the proof of concept presented here, the number of SERS digital counts (N) per mapped area. Considering basic counting statistics42, the best estimate of the count should be 𝑁 ± √𝑁 and the fractional uncertainty (error, %) of the count should decrease as N increases according:

5 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 8

Figure 6. Digital SERS mapping from isotologue experiments (A) and (B) 140 pM (~100 molecules / m2) of CIPRO-C12,N14 ( ) and (B) CIPRO-C13,N15 ( ); (C) and (D) 14 pM (~10 molecules / m2) of CIPRO-C12,N14 ( ) and CIPRO-C13,N15 ( ). ( ) correspond to simultaneous detection of CIPRO-C12,N14 and CIPRO-C13,N15. error %  

1  100 N

(2)

ASSOCIATED CONTENT

Therefore, increasing the mapped area would provide more opportunities (pixels) for single molecule detection, leading to larger N. Although large area Raman mapping has been notorious time consuming, almost all of the new generation commercial Raman microscopes present options for fast mapping, which should be helpful on the verification and implementation of the digital SERS procedure proposed here.

Supporting Information Details on the quantification method; details on the experimental protocol; substrate characterization; chemometric methods; and additional data for CIPRO.

AUTHOR INFORMATION Corresponding Author *Email: [email protected] Phone: 1 (250) 721-7167, Fax: 1 (250) 721-7147

CONCLUSIONS A quantification procedure based on SM-SERS statistics was introduced. This method allows the direct determination of ultra-low concentrations by exploring the stochastic nature of SM-SERS. The concept consists of depositing a low volume of an aqueous sample at low concentration in a planar SERS active substrate. The surface of the substrate is then spatially mapped with a laser beam focused on a small area (1 µm2, for instance). The goal is to ensure that the SERS response from each illuminated spot will most probably be generated by a SM-SERS event. The SERS signal was then “digitized” and the number of pixels that provide a SM-SERS response (SERS digital count) was shown to be proportional to the solution concentration. The protocol suggested here provides an unique example of application of the concepts of SMSERS for high sensitive quantification. The method was developed using a relative common SERS substrate, and emergent contaminants were used as molecular probes. This work should provide some important guidelines for the development of analytical protocols that take full advantage of the main characteristics of SERS, including its unique selectivity and sensitivity, for ultra-low concentration quantification.

Author Contributions All authors have given approval to the final version of the manuscript.

Notes The authors declare no competing financial interests.

ACKNOWLEDGMENT This work was supported by NSERC, CFI, BCKDF and University of Victoria. CDLA thanks the Brazilian FAPESP Science Without Borders Program for an international visiting student fellowship.

REFERENCES (1) Le Ru, E. C.; Etchegoin, P. G. In Annual Review of Physical Chemistry, Vol 63, Johnson, M. A.; Martinez, T. J., Eds., 2012, pp 6587. (2) Kneipp, K.; Kneipp, H.; Itzkan, I.; Dasari, R. R.; Feld, M. S. Chem. Rev. 1999, 99, 2957-+. (3) Chen, H. Y.; Lin, M. H.; Wang, C. Y.; Chang, Y. M.; Gwo, S. J. Am. Chem. Soc. 2015, 137, 13698-13705.

6 ACS Paragon Plus Environment

Page 7 of 8 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry (4) Im, H.; Bantz, K. C.; Lindquist, N. C.; Haynes, C. L.; Oh, S. H. Nano Lett. 2010, 10, 2231-2236. (5) Yang, S. K.; Dai, X. M.; Stogin, B. B.; Wong, T. S. Proceedings of the National Academy of Sciences of the United States of America 2016, 113, 268-273. (6) Tabatabaei, M.; Najiminaini, M.; Davieau, K.; Kaminska, B.; Singh, M. R.; Carson, J. J. L.; Lagugne-Labarthet, F. Acs Photonics 2015, 2, 752-759. (7) Cabrera, F. C.; Aoki, P. H. B.; Aroca, R. F.; Constantino, C. J. L.; dos Santos, D. S.; Job, A. E. Journal of Raman Spectroscopy 2012, 43, 474-477. (8) Liu, H. Q.; Li, Q.; Li, M. M.; Ma, S. S.; Liu, D. B. Anal. Chem. 2017, 89, 4776-4780. (9) Patra, P. P.; Chikkaraddy, R.; Tripathi, R. P. N.; Dasgupta, A.; Kumar, G. V. P. Nature Communications 2014, 5. (10) Simoncelli, S.; Roller, E. M.; Urban, P.; Schreiber, R.; Turberfield, A. J.; Liedl, T.; Lohmuller, T. Acs Nano 2016, 10, 9809-9815. (11) Zhang, Y. Q.; Shen, J. F.; Xie, Z. W.; Dou, X. J.; Min, C. J.; Lei, T.; Liu, J.; Zhu, S. W.; Yuan, X. C. Nanoscale 2017, 9, 1069410700. (12) LeRu, E. C.; Etchegoin, P. G. Principles of Surface-Enhanced Raman Spectroscopy: And Related Plasmonic Effects, 2009, p 1-663. (13) Le Ru, E. C.; Etchegoin, P. G. J. Chem. Phys. 2009, 130. (14) Fang, Y.; Seong, N. H.; Dlott, D. D. Science 2008, 321, 388392. (15) Etchegoin, P. G.; Meyer, M.; Le Ru, E. C. PCCP 2007, 9, 3006-3010. (16) Pieczonka, N. P. W.; Aroca, R. F. Chem. Soc. Rev. 2008, 37, 946-954. (17) dos Santos, D. P.; Andrade, G. F. S.; Temperini, M. L. A.; Brolo, A. G. Journal of Physical Chemistry C 2009, 113, 1773717744. (18) Le Ru, E. C.; Grand, J.; Sow, I.; Somerville, W. R. C.; Etchegoin, P. G.; Treguer-Delapierre, M.; Charron, G.; Felidj, N.; Levi, G.; Aubard, J. Nano Lett. 2011, 11, 5013-5019. (19) Fan, M. K.; Cheng, F. S.; Wang, C.; Gong, Z. J.; Tang, C. Y.; Man, C. Z.; Brolo, A. G. Chem. Commun. 2015, 51, 1965-1968. (20) Zhang, Y.; Noji, H. Anal. Chem. 2017, 89, 92-101. (21) Lee, P. C.; Meisel, D. J. Phys. Chem. 1982, 86, 3391-3395. (22) Brito-Silva, A. M.; Sobral-Filho, R. G.; Barbosa-Silva, R.; de Araújo, C. B.; Galembeck, A.; Brolo, A. G. Langmuir 2013, 29, 43664372.

(23) Hong, K. Y.; de Albuquerque, C. D. L.; Poppi, R. J.; Brolo, A. G. Analytica Chimica Acta 2017, 982, 148-155. (24) Haran, G. Israel Journal of Chemistry 2004, 44, 385-390. (25) Darby, B. L.; Le Ru, E. C. Journal of the American Chemical Society 2014, 136, 10965-10973. (26) Savitzky, A.; Golay, M. J. E. Analytical Chemistry 1964, 36, 1627-1639. (27) Eilers, P. H. C. Anal. Chem. 2004, 76, 404-411. (28) Albuquerque, C. D. L.; Poppi, R. J. Analytica Chimica Acta 2015, 879, 24-33. (29) Lee, D. D.; Seung, H. S. Nature 1999, 401, 788-791. (30) Schlucker, S. Angewandte Chemie-International Edition 2014, 53, 4756-4795. (31) Costa, J. C. S.; Cordeiro, D. S.; Sant'Ana, A. C.; Rossi, L. M.; Santos, P. S.; Corio, P. Vib. Spectrosc 2010, 54, 133-136. (32) dos Santos, D. P.; Temperini, M. L. A.; Brolo, A. G. J. Am. Chem. Soc. 2012, 134, 13492-13500. (33) Bohn, J. E.; Le Ru, E. C.; Etchegoin, P. G. Journal of Physical Chemistry C 2010, 114, 7330-7335. (34) Aoki, P. H. B.; Carreon, E. G. E.; Volpati, D.; Shimabukuro, M. H.; Constantino, C. J. L.; Aroca, R. F.; Oliveira, O. N.; Paulovich, F. V. Appl. Spectrosc. 2013, 67, 563-569. (35) Aroca, R. F.; Constantino, C. J. L. Langmuir 2000, 16, 54255429. (36) Dieringer, J. A.; Lettan, R. B.; Scheidt, K. A.; Van Duyne, R. P. J. Am. Chem. Soc. 2007, 129, 16249-16256. (37) Kleinman, S. L.; Ringe, E.; Valley, N.; Wustholz, K. L.; Phillips, E.; Scheidt, K. A.; Schatz, G. C.; Van Duyne, R. P. J. Am. Chem. Soc. 2011, 133, 4115-4122. (38) Zrimsek, A. B.; Wong, N. L.; Van Duyne, R. P. Journal of Physical Chemistry C 2016, 120, 5133-5142. (39) Lussier, F.; Brule, T.; Vishwakarma, M.; Das, T.; Spatz, J. P.; Masson, J. F. Nano Lett. 2016, 16, 3866-3871. (40) Fan, M.; Brolo, A. G. PCCP 2009, 11, 7381-7389. (41) Fan, M. K.; Andrade, G. F. S.; Brolo, A. G. Anal. Chim. Acta 2011, 693, 7-25. (42) Cherry, S. R.; Sorenson, J. A.; Phelps, M. E. Physics in nuclear medicine, 4th ed.; Elsevier/Saunders: Philadelphia, 2012, p xvii, 523 p.

7 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 8

For TOC only

Digital SERS

ACS Paragon Plus Environment

8