Silver nanoparticle-treated filter paper as a highly sensitive surface-enhanced Raman scattering (SERS) substrate for detection of tyrosine in aqueous solution

Analytica Chimica Acta 708 (2011) 89–96

Contents lists available at SciVerse ScienceDirect

Analytica Chimica Acta journal homepage: www.elsevier.com/locate/aca

Silver nanoparticle-treated filter paper as a highly sensitive surface-enhanced Raman scattering (SERS) substrate for detection of tyrosine in aqueous solution Min-Liang Cheng, Bo-Chan Tsai, Jyisy Yang ∗ Department of Chemistry, National Chung-Hsing University, Taichung 402, Taiwan

a r t i c l e

i n f o

Article history: Received 30 April 2011 Received in revised form 1 October 2011 Accepted 9 October 2011 Available online 14 October 2011 Keywords: Raman spectroscopy Silver nanoparticles Surface enhancement Filter paper

a b s t r a c t Highly sensitive SERS substrates based on deposition of silver nanoparticles on commercially available filter paper were prepared in this work, and used to overcome problems found in analyses of aqueous samples. To prepare silver nanoparticle- (AgNP) doped filter substrates, a silver mirror reaction was used. The procedures for substrate preparation were systematically optimized. Pretreatment of filter paper, reaction time, temperature, and concentration of reagents for silver mirror reactions were studied. The morphologies of the resulting substrates were characterized by field-emission scanning electron microscopy (FE-SEM) and correlated with the SERS signals by probing with p-nitrothiophenol (pNTP). Filter papers with different pretreatments were found to have different sizes and distributions of AgNPs. The best performance was found when filter paper was pre-treated with ammonia solution before growth of AgNPs. Based on the SEM images, the resulting AgNPs had roughly spherical shape with a high degree of uniformity. The silver-coated filter paper substrates provide much higher SERS signals compared to glass substrates and the reproducibility was improved significantly. Based on statistical analyses, the relative standard deviations for substrate-to-substrate and spot-to-spot were both were less than 8% and the enhancement factors for the substrates were, in general, higher than 107. The SERS substrates were used to selectively detect tyrosine in aqueous solution. Results indicate that filter-based SERS substrates are highly suited to detection of tyrosine. Compared to glass-based SERS substrates, 50 times more SERS signal was observed in detection of tyrosine. The linear range can be up to 100 ␮M with a detection limit of 625 nM (S N−1 = 3). © 2011 Elsevier B.V. All rights reserved.

1. Introduction Since the observation of the enhancement of Raman signals for molecules adsorbed on roughened metal surfaces [1–3], Surface-enhanced Raman scattering (SERS) has gained considerable attention. The transformation of this technique into a sensitive and practical analytical tool, especially in the characterization of molecules adsorbed on the surface of metal substrates, is very promising [4–7]. A wide range of methods have been developed to prepare nanometer-scale particles for SERS applications. For instance, SERS substrates can be prepared by electrochemically roughening electrodes [8,9], metal colloids [10–13], island films prepared by vapor deposition [14], silver-doped sol–gel films [15–18], and silver-mirror (Tollen’s) reaction [19–24]. Among these methods, the silver-mirror reaction offers several advantages such as low cost, no need for delicate equipment, and virtually no limitations on substrate materials. In our previous report [24], the

∗ Corresponding author. Tel.: +886 422840411x514; fax: +886 422862547. E-mail address: [email protected] (J. Yang). 0003-2670/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.aca.2011.10.013

silver-mirror reaction produced highly sensitive SERS substrates having silver nanoparticles (AgNPs) with spherical morphology and an enhancement factor approaching 106 . The silver-mirror reaction does not limit the shape, size, and material of the substrates, which is an important advantage in terms of sensing technology. Besides glass, a wide variety of materials have been used to prepare SERS substrates. For instance, silicon wafer [20], copper foil [25], aluminum foil [26], and cellulose substrates [27–33] have been demonstrated. In general, for a solid substrate, such as glass or metal, the analysis of aqueous samples requires extensive time to evaporate the water to concentrate the analyte at the AgNPs. Several authors [27,32,34] have attempted to physically remove the water from the sample drops to speed up the analysis. For instance, Huang et al. [34] used heat to dry analyte-doped AgNPs and found an enhancement factor up to 106 with a short analysis time. Another approach utilized AgNPs in colloidal solution mixing with samples, then deposited on filter paper to drain the water. With Raman sensitive molecules of several dyes, detection limits approached a few hundreds of nanograms as demonstrated by Tran [27] Similarly, Ota et al. [32] utilized 1064 nm laser radiation to measure SERS spectra of amino acids by the same procedure but the results for detection of L-phenylalanine only approached an enhancement factor of 30.

90

M.-L. Cheng et al. / Analytica Chimica Acta 708 (2011) 89–96

In general, the number of AgNPs in a colloidal solution is limited by the need to prevent coagulation. Also, tuning of the particle size to match the laser frequency generally is limited due to the complexity of the chemical system needed for production of colloidal AgNPs. Meanwhile, the employing colloidal for SERS measurement usually associates with low stability and reproducibility caused by the difficulties in maintaining the colloidal aggregation state. To obtain the advantage of filter paper in dealing with aqueous solutions, a low cost but highly sensitive approach to AgNP synthesis and deposition is proposed in this work. To demonstrate the feasibility for detection in aqueous solutions, tyrosine is selected due to its high polarity and importance in biological systems. For instance, tyrosine is one of the three aromatic acids and a semiessential amino acid that is involved in several genetic metabolic disorders, including phenylketonuria [35,36], tyrosinaemia type II, and tyrosinosis [37]. Since tyrosine plays important roles in biological systems, a number of analytical methods have been developed such as chromatographic [38–42], enzymatic [43–45], and optical [46–49]. However, weaknesses are found in these methods, such as requiring pre-cleaning of the samples in the separation-based methods, degradation of the enzymatic activity in some methods, and complicated optical arrangements in others. To reduce the limitations and weaknesses of current approaches, a method based on SERS measurements is proposed and demonstrated herein. The success of this proposed method benefits SERS applications in general, and also the specific need for determination of tyrosine. To obtain the advantages of cellulosic materials in handling aqueous solutions, commercial filter papers were selected as substrates for deposition of AgNPs for use in SERS measurements. To avoid the weaknesses of using colloidal AgNPs for detection and to gain the advantages of cellulose substrates, a silver mirror reaction was used to form AgNPs directly on the filter paper. This offers better control of the particle size of the AgNPs to meet the requirements for SERS requirements. Also, improvement of the capacity to uptake analytes is expected due to large surface area of fibril substrates. To help optimize the conditions for the preparation of AgNPs on the filter papers, the prepared substrates were probed with p-nitrothiophenol (pNTP), p-aminothiophenol (pATP), and pmercaptobenzoic acid (pMBA). The SERS signals were correlated with the morphologies of the AgNPs observed by field-emission scanning microscopy (FE-SEM).

2.2. Instrumentation The morphologies of the AgNPs formed on the substrates were measured with a JSM-7600F (JEOL, Ltd., Tokyo, Japan) field emission scanning electron microscope (FE-SEM) operated at 3.0 kV. Fourier transform (FT) SERS spectra were acquired with the use of a Bruker FRA-106/S FT-Raman spectrometer equipped with a liquid nitrogen-cooled Ge detector (Bruker Optics Inc., Germany). The 1064-nm excitation line was provided by a Nd:YAG laser. Laser power was 100 mW at the sample position. SERS spectra were collected to a spectral resolution of 4.0 cm−1 and the number of scans coadded was 16 unless otherwise specified. The X-ray photoelectron spectroscopy (XPS) analyses were carried out with a ULVAC-PHI PHI 5000 VersaProbe spectrometer (Kanagawa, Japan) using a monochromatic Al K␣ source operated at 15 kV and 20.9 W. 2.3. Preparation of AgNPs substrates The details of the silver mirror reaction procedure are similar to our previous report [24], except that the glass slides were replaced with filter paper. In brief, the filter papers were cut to 20 mm × 20 mm and placed vertically in the reaction solution to prevent accumulation of precipitates from the bulk solution. 10 slides were placed parallel to each other with a space of 2 mm in between using a home-made holder, which was fabricated from polystyrene foam. Silver nanoparticles were grown by immersing the filter papers into the plating solution (Tollen’s reagent), which contained 50 mM AgNO3 , 300 mM ammonia and 500 mM glucose (reducing agent). The plating solution was prepared in an ice-water bath. After placing the substrates into the plating solution, the container was moved to a 55 ◦ C water bath. After the designated reaction time, the substrates were removed from the reaction vessel and rinsed with distilled water for 10 min, and subsequently in methanol for 1 min. After air-drying, the substrates were immersed in 1 mM solutions of either pNTP, pATP, or pMBA methanoic solution for 30 min to form an adsorbed layer on the substrate. After rinsing off the un-reacted probe molecules with methanol for 5 min, SERS spectra were measured as described above. 3. Results and discussion

2. Experimental

3.1. Basic properties of filter paper in Raman spectroscopy

2.1. Chemicals

Typical Raman spectra of un-treated (blank) FP-A, 7.4 ␮g cm−2 of pNTP deposited on non-treated FP-A, and freshly prepared AgNPs@FP-A are plotted in Fig. 1A. Due to the weak Raman signals, these spectra were collected by coadding 512 scans. The AgNPs@FPA was prepared by reaction in a plating solution containing 50-mM silver nitrate, 500-mM glucose and 300-mM ammonia for 6 min at 55 ◦ C. As can be see in Fig. 1A, filter paper shows a very weak scattering feature in the FT-Raman spectrum. After depositing 7.4 ␮g cm−2 of pNTP on the filter paper, no significant variation of the spectral features can be seen in the spectrum. To check for any increase of the spectrum by a band due to the filter paper, AgNPs were grown on the surface of FP-A. The spectrum of this substrate did not show any enhanced features due to the filter paper. However, some spectral features can be observed as one band located around 1400 cm−1 increased, which likely is caused by the adsorption of carboxylic acid byproduct from the silver mirror reaction. As can be seen in Fig. 1A, all of the spectra exhibited very weak features revealing that there are no serious spectral interferences for analytical purpose using filter paper substrates for SERS measurements.

Filter paper A (Advantec, #5) and B (Advantec #6) were obtained from Advantec (Tokyo, Japan), and filter paper C (Whatman, #1) was obtained from Whatman Ltd. (Maidstone, England). To simplify the description, filter papers are named FP-A, FP-B and FP-C hereafter. According to the venders, the pore sizes are 1, 3, and 11 ␮m and the thicknesses are 0.22, 0.20, and 0.180 mm for FPA, FP-B and FP-C, respectively. Silver nitrate was purchased from ProChem (Rockford, MI). Glucose (anhydrous) was purchased from Sigma (St. Louis, MO). Sodium hydroxide was purchased from Yakuri Co. (Kyoto, Japan). Ammonium hydroxide and pNTP were obtained from Acros Organics (Phillipsburg, NJ). pATP was obtained from Aldrich Chemical Co. (Milwaukee, WI). pMBA was purchased from TCI (Tokyo, Japan). Nitric acid was purchased from J. T. Baker (Phillipsburg, NJ). Tyrosine, Tryptophan, Phenylalanine, Histidine, Arginine, Glycine, and Methionine were obtained from Acros Organics (Phillipsburg, NJ). Methyl alcohol in HPLC grade was purchased from TEDIA. All these chemicals were reagent grade and used as received.

M.-L. Cheng et al. / Analytica Chimica Acta 708 (2011) 89–96

91

multi-layer structure of the AgNPs formed on the fibers of the filter papers. But, the large increase of the Raman signals for pATP and pMBA cannot be explained simply by the increase of the number of AgNPs, which only accounts for a signal improvement of five times. This behavior may be explainable by the formation of dynamic hot spot as observed by Aldeanuva-Potel et al. [56]. In their studies of AgNP-doped agarose gel, the dehydration and rehydration caused changes in hot spot effects due to changes in the spacing between AgNPs. For molecules of pNTP, which contain only one functional group to interact with AgNPs, the ability to attract two AgNPs closer to the molecule is weak. But, for bi-functional molecules of pATP and pMBA, the formation of dynamic hot spots is possible, and likely to cause large increases in the SERS signals observed. The reproducibility of SERS substrates based on filter papers also was studied. By comparing with the glass substrates prepared as described previously [24], the reproducibility was improved from 20% standard deviation (RSD) to better than 8% for batch-to-batch studies. The improvement is given in terms of relative RSD of the SERS signals. Because the filter papers were placed in different positions during the silver mirror reaction, the variation for different locations in the reaction vessel also was examined. The results show that the substrates gave similar SERS signals, except for the substrates located at the ends (refer to Supplementary data, Fig. S1). Therefore, substrates these two locations were discarded from each of the subsequent batches prepared. By discarding the signals from the substrates in both ends, the calculated relative standard deviation was around 4.5%. 3.3. Effect of pretreatment on the performance of SERS substrates Fig. 1. (A) Raman spectra of un-treated FP-A, 7.4-␮g cm−2 pNTP on untreated FP-A, and AgNPs treated FP-A. The spectra were coadded with 512 scans. (B) SERS spectra of pNTP, pATP, and pMBA on AgNPs@FP-A substrates. (C) SERS spectra of pNTP, pATP, and pMBA on AgNPs@glass substrates.

3.2. Basic performance of SERS substrates prepared from filter papers To observe analytes, substrates of AgNPs@FP-A were prepared with a plating solution containing 50-mM silver nitrate, 500-mM glucose and 300-mM ammonia. The reaction time was 6 min and at a temperature of 55 ◦ C. The substrates were soaked in 1 mM methanoic solutions of pNTP, pATP, and pMBA for 30 min to form a chemisorbed layer of the probe compounds. The spectra of probe compounds on the substrates were collected by coadding 16 scans as plotted in Fig. 1B. Referring to the assignments in the literature [50–55], several intense bands, such as (C C) at 1575 cm−1 , sym (NO2 ) at 1355 cm−1 , and ı(C–H) at 1112 cm−1 and 1182 cm−1 can be identified. The SERS spectrum of pATP shows several characteristic bands such as (C–S) at 1088 cm−1 , ı(C–H) at 1179 cm−1 , and (C C) at 1594 cm−1 . The observed SERS spectrum of pMBA shows the characteristic bands located at 1586 and 1075 cm−1 , which are assigned as (C C) and (C–S), respectively. The intensities of the sym (NO2 ) of pNTP at 1350 cm−1 and the (C–S) for pATP and pMBA (∼1078 cm−1 ) were used for quantitative purpose as will be discussed lately. To compare the performances of the substrates prepared in this work, glass SERS substrates were also prepared under essentially the same conditions as for filter papers, except that the reaction time was reduced to 2 min. After placing probe molecules by the same method used for the filter papers, spectra were obtained as shown in Fig. 1C. Compared with the spectra in Fig. 1B, the SERS intensity increases 5 times for pNTP on NPs@FP-A substrate. But, for pATP and pMBA, the Raman signals increase 40–200 times, respectively. The increase of SERS signals for pNTP can result from the

To enhance the performance of SERS substrates based on filter papers, paper was pre-treated with acidic and basic solutions. The pretreatments are expected to remove any residual powders in the filter papers and to enlarge spaces between cellulosic fibers to allow better growth of the AgNPs in the inner layers of the fibers. Four types of experiments were performed, with the intent of discerning the effects of chemicals in pre-treatments, pretreatment time, concentration of chemicals, and the different pore sizes. After pretreatment, the filter paper samples exposed to a mirror plating solution containing 50-mM silver nitrate, 500-mM glucose and 300-mM ammonia at 55 ◦ C for 6 min. To probe the SERS performance, the substrates were soaked in 1-mM methanoic solution of pNTP for 30 min. SERS spectra then were acquired and the results plotted as band intensities at 1350 cm−1 versus the experimental parameters, as shown in Fig. 2A–D. To examine the influences of the species in pre-treatments, three FP-A samples were treated respectively in solutions of 10% (v/v) nitric acid, 10% (w/v) sodium hydroxide, and 10% (v/v) ammonia for 3 h at room temperature. For comparison, filter papers without any treatment were soaked in water for 3 h and also examined. Fig. 2A shows that the SERS signals for most of the treatments were similar, with an exception of the treatment of ammonia. These variations, in general, should be caused by two factors; the morphologies of the filter paper and changes in the chemical properties of the cellulosic fibers, both of which can influence the growth of AgNPs. To understand the contributions of these two influences, SEM images of the filter paper FP-A, FP-B and FP-C, without any pretreatments, were examined (see Supplementary data, Fig. S2). The SEM images of FP-A after different treatments and the resulting AgNPs are shown in Fig. 3A–D and Fig. 3E–I, respectively. Based on images of un-treated filter papers, only slight differences in morphologies can be found. In particular, FP-A substrates show smaller pore size as it has a smoother surface compared to other filter papers. After treatment with acidic or basic solutions, the roughness of the surface of FP-A was increased by any of the treatments as can be clearly seen in Fig. 3A–D. After growth of AgNPs on treated

92

M.-L. Cheng et al. / Analytica Chimica Acta 708 (2011) 89–96

Fig. 2. SERS signals of pNTP chemisorbed on AgNPs@FP-A substrates. AgNPs were grown in plating solution containing 50 mM of silver nitrate, 500 mM glucose, 300 mM of ammonia. (A) SERS signals of pNTP on AgNPs@FP-A substrates. The FP-A substrates were used without any pretreatment (䊉) and pretreated with distilled water (), 10% (v/v) ammonia (), 10% (v/v) nitric acid (), and 10% NaOH () for 3 h. The reaction time for plating of AgNPs was varied. (B) SERS signals of pNTP on AgNPs@FP-A substrates. The FP-A substrates were pre-treated with 10% ammonia () and 10% nitric acid (䊉) for different treatment times. (C) SERS signals of pNTP on AgNPs@FP-A substrates. The FP-A substrates were pretreated with different concentrations of ammonia () and nitric acid (䊉) for 3 h. The reaction time for plating of AgNPs was varied. (D) SERS signals of pNTP on AgNPs@filter papers. FP-A (), FP-B (䊉), and FP-C () filter paper was used and pre-treated with 5% ammonia for1 h. The reaction time for plating of AgNPs was varied.

and non-treated FP-A, the resulting AgNPs were varied in size as can be seen in Fig. 3E–I. Particles formed in the inner layer of the fibers also are observable, especially after treatment with ammonia, NaOH and nitric acid. The small size of AgNPs in non-treated FP-A indicates that the FP-A paper contains small residual amounts

of the pre-treatment agents that influence formation and attraction of AgNPs. For FP-A soaked in water for 3 h, the resulting AgNPs exhibited large particle size and coagulation of the particles. No evidence of layer structure in the AgNPs on the FP-A surface is seen. For the FP-A treated in an acidic solution of nitric acid, the image in

Fig. 3. FE-SEM images of FP-A after soaking 3 h in water (A), 10% HNO3 (B), 10% NaOH (C), and NH4 OH (D). FE-SEM images of AgNPs plated on FP-A without pretreatment (E) and pretreated with water (F), 10% HNO3 (G), 10% NaOH (H), and NH4 OH (I) for 3 h.

M.-L. Cheng et al. / Analytica Chimica Acta 708 (2011) 89–96

Fig. 3G shows AgNPs in the inner layer of fibers. Also, coagulation of AgNPs also is observable in this image. When treated with basic solution of NaOH, the spaces between fibers were enlarged (refer to Fig. 3C) and the deposited AgNPs were small. After pretreatment with ammonia, particles were distributed in different layers as can be seen in Fig. 3I. The influence of pretreatment, such as the length of pretreatment time and the concentrations of pretreatment reagents, also were investigated as the data in Fig. 2B and C show. The pretreatment time did not influence the SERS signals significantly for either ammonia or nitric acid treated FP-A, as can be seen in Fig. 2B. When concentration of pretreatment reagents was varied, the resulting signals results showed no influences from ammonia, but significant decreases in SERS signals were observed when nitric acid was used. This result likely is caused by absorption of the nitric acid in the fibers altering the silver mirror reactions. The silver mirror reaction is favored in basic solution. Also, the adsorption of nitrate ions changes the surface charges of cellulosic fibers leading to strong influence by the concentration of nitric acid. The influences of filter paper pore size were examined using the filter papers designated FP-A, FP-B and FP-C. These filter papers were pre-treated with 5% ammonia for 1 h. The observed spectra of pNTP on these substrates showed similar spectral features and the calculated SERS intensities are plotted in Fig. 2D. Based on this plot, the observed SERS signals showed negligible variations between filter papers with different pore sizes. Based on this observation, it can be concluded that the types of filter papers are not important, as long as the surface of the filter papers is rough enough to allow nanoscale AgNPs to form on the inner layers. Based on these observations, ammonia is the most suitable agent for pretreatment. Also, the variation of signals between different filter papers is negligible, allowing direct comparison data collected using different filter papers. 3.4. Long-term stability of the substrates Two sets of AgNPs@FP-A substrates were prepared to examine the variation of the SERS signals caused by storage or aging. One set of substrates was air dried only and the other set of substrates was dried in an oven at 80 ◦ C for 10 min prior storage. These substrates were kept at room temperature for 0–3, 5 days without any special precautions. After the designated storage times, substrates were soaked in methanoic pNTP solution for 30 min. The relationship between the resulting SERS signals and the storage times was examined (refer to Supplementary data, Fig. S3). The results show that the SERS signals decreased rapidly after stored for a day, but the signals did not decrease further than half compared to the signals from freshly prepared substrates. For the substrates were oven-dried before storage, SERS signals were similar to those from air-dried substrates after storage. The decrease of the SERS signals after storage could be caused by the oxidation of the AgNPs in the air. To examine this effect, XPS spectra (refer to Supplementary data, Fig. S4) were collected. Characteristic bands of Ag, C, and O were observed. To quantitatively analyze the coverage of oxygen on the oven-dried and air-dried substrates during storage, band intensities of Ag3d5 (367.8 eV), C1s (286.2 eV) and O1s (532.2 eV) were calculated. Because cellulosic fiber in filter paper contributes to the signals of C1s and O1s, the ratio of the band intensities of O1s to C1s are used instead. For freshly prepared substrates, the ratio value of O1s/C1s is 0.858. For air-dried and oven-dried substrates after storage, the ratio values are very similar and the average ratio value for these substrates is 0.945 with a relative standard deviation of 1.8%. The increase of the ratio clearly indicates that the oxidation of AgNPs does happen rapidly after preparation. For an accurate comparison, the surfaces of these substrates were further cleaned with an Ar ion beam. After cleaning, the observed ratio for

93

freshly prepared substrates remains 0.858, but the ratio decreased to 0.874 with a relative standard deviation of 2.4% for aged substrates. These results indicated that AgNPs oxidize rapidly, but the oxide coating stops changing and become stable for longer storage times. The sensitivity compared to freshly prepared substrates is roughly a two-fold decrease in sensitivity. Thus, freshly prepared substrates were used for their better sensitivity in the later studies. 3.5. Evaluation of enhancement factors The enhancement factors (EF) were calculated by use of the following equation: EF =

 I  N  SERS bulk IRaman

×

Nads

where ISERS is the intensity of a specific band in the SERS spectrum of the probe molecule and IRaman is the intensity of the same band in the Raman spectrum of the probe molecule in bulk solution. Nbulk is the number of probe molecule in the bulk solution and Nads is the number of probe molecule adsorbed on the prepared substrate. Based on results described above, filter paper with ammonia pretreatment is suggested for preparation of substrates. FP-A pieces were treated with 5% ammonia for 3 h followed by growth of AgNPs for 6 min at 55 ◦ C in a plating solution containing 50 mM of silver nitrate, 500 mM glucose, and 300 mM of ammonia. The intensity of the sym (NO2 ) of pNTP at ca. 1350 cm−1 was used to calculate the enhancement factors. After deposition of certain amount of probe molecules to cover the surface of AgNPs@FP-A substrate to form a coverage of 500 ng cm−2 , the EF values for pNTP, pATP, and pMBA were found to be 1.9 × 107 , 1.4 × 107 , and 3.2 × 107 , respectively. These values indicate that SERS substrates based on filter paper provide one order of magnitude increase of SERS signals relative to glass slides. 3.6. Detection of tyrosine by SERS substrates based on filter papers To demonstrate the advantages, including sensitivity, of filter paper in analysis of aqueous samples, tyrosine was used. Spectra of solid tyrosine, 0.1 M tyrosine, and 100-␮M tyrosine on AgNPs@FP-A substrates and tyrosine on AgNPs@glass substrate were measured as shown in Fig. 5A. AgNPs on glass and filter paper substrates were grown under standard conditions for 2 and 6 min, respectively. Tyrosine at a concentration of 100 ␮M was detected by soaking the substrates in the solution for 10 min. Typical detected spectra are plotted in Fig. 4A. The spectrum of solid phase tyrosine shows several characteristic Raman bands including ı(C–H) at 1170 cm−1 , (C–O) at 1244 cm−1 , ı(C–H) at 1492 cm−1 , and (C C) at 1595 cm−1 [57–59]. However, the sensitivity for detection of 0.1 M tyrosine in water (pH adjusted to 11.3 to increase the solubility of tyrosine) is low. The quality of the spectrum is not good enough to identify all of the characteristic bands of tyrosine. Once tyrosine was adsorbed on SERS substrates, the bands were enhanced and the features were changed, such that only a few bands can be recognized. The Raman bands are substantially broader. The SERS band located at 1594 cm−1 due to the (C C) was selected for quantitation. For comparison, tyrosine also was measured using AgNP-treated glass substrates. Typical spectra obtained by soaking and deposition methods are plotted in Fig. 4A. The SERS spectrum of tyrosine measured on FP-A, is at least 40 times greater. The increase is similar to that seen for SERS signals of pMBA and pATP on filter papers. Formation of dynamic hot spots is a possible explanation for this enlargement. Three functional groups are present in the molecular

94

M.-L. Cheng et al. / Analytica Chimica Acta 708 (2011) 89–96

Fig. 4. (A) SERS spectra of solid tyrosine, 0.1 M tyrosine (adjusted by NaOH to pH 11.3), and 100-␮M tyrosine on AgNPs@FP-A substrates and tyrosine on AgNPs@glass substrate. The laser power was changed to 140 mW for solid tyrosine and 0.1 M tyrosine. The numbers of scans were 128 and 1024 for solid and liquid tyrosine, respectively. For tyrosine on AgNPs@FP-A and AgNPs@glass substrates, the number of scan was 16 with a laser power of 100 mW. (B) SERS spectra of glycine, arginine, methionine, phenylalanine, and tryptophan on AgNPs@FP-A substrates. Substrates were soaked for 10 min 100 ␮M solutions. 16 scans were coadded with a laser power of 100 mW.

structure of tyrosine (–COOH, –NH2 , and –OH), which could serve to pull AgNPs closer together.

Fig. 5. (A) Triplicate runs in detection of 100 ␮M tyrosine. The substrates were dried in an oven at 80 ◦ C for different times (䊉). Substrates also were dried in air and the SERS signals were plotted for two substrates (first run,  and second run, ). (B) Soaking time against the detected SERS signals in detection of 100 ␮M of tyrosine. The AgNPs@FP-A substrates were dried in an oven at 80 ◦ C for 20 min.

are no strong spectral interferences from other amino acids, which allows direct measurement of tyrosine using SERS substrates based on filter paper. 3.8. Effect of drying method

3.7. Selectivity in detection of tyrosine SERS substrates were further applied to detect other amino acids with the aim of ruling out possible interferences to detection of tyrosine. Previous SERS studies have shown that aromatic amino acids give rise to more intense SERS signals than non-aromatic amino acids [60–62]. Therefore, amino acids with aromatic rings, including phenylalanine and tryptophan, were selected. Also, methionine, glycine, and arginine were selected to represent different classes of amino acids. Solutions of 1-mM amino acids were prepared and used to soak the AgNP-modified FP-A substrates for 10 min. The substrates were dried and their SERS spectra were acquired. Typical SERS spectra for these amino acids are plotted in Fig. 4B. All the examined amino acids exhibited much weaker SERS bands compared to tyrosine in Fig. 4A. Only tryptophan gave a SERS spectrum with observable spectral features. Even for phenylalanine, which contains a phenyl ring, the detected bands were extremely weak. This suggests that formation of dynamic hot spots is likely to account for the much larger signals observed for tyrosine. As phenylalanine has no other polar functional group to attract nearby AgNPs, the ability to form dynamic hot spots should be weak, leading to weak SERS signals. These results indicate that there

Unlike methanoic solutions of thio compounds, aqueous samples require substantial time to evaporate the water molecules out of cellulosic matrices. This is a significant impediment to measuring aqueous samples by SERS. Although cellulosic substrates can disperse and adsorb water molecules easily, the residual water affects the degree of swelling of cellulosic fibers and the properties of AgNPs on the fibers. For instance, the distance between AgNPs may be changed by evaporation of water. Also, the hydroxyl groups in cellulosic fiber form strong hydrogen bonds with water molecules, which could affect the ability to interact with tyrosine. To observe these influences, substrates of FP-A were prepared under standard conditions for 6 min. The resulting substrates were soaked in 100-␮M tyrosine solution for 10 min. After removing the substrates from the sample solution, one end of the filter paper was touched to tissue paper to adsorb some of the remaining water. Substrates were further dried in air and in an 80 ◦ C oven for different times followed by acquisition of SERS spectra. The observed intensities of the SERS band of tyrosine located at 1595 m−1 were quite consistent when the substrates were dried in an oven at 80 ◦ C, as can be seen in Fig. 5A. Also, 10 min of drying time was enough to remove essentially all of the water and further extension of the

M.-L. Cheng et al. / Analytica Chimica Acta 708 (2011) 89–96

95

drying time did not affect the SERS signals. On the contrary, the SERS signals for substrates dried in air showed large variations. SERS signals from first substrate differed to that of second substrate significantly in the first 30 min as can be seen in Fig. 5A. This large variation may be caused simply by variation of the residual amount of water in the substrates after soaking in analyte solution. Although after long drying times, the signals approached a certain value, the large variations seen for short drying times can cause problems for quantitative analysis. Moreover, the signals for substrates dried in air showed much lower SERS signals compared to the substrates treated in 80 ◦ C oven. This may indicate that the AgNPs coagulate slightly at 80 ◦ C to produce greater enhancement and/or the thermal energy allows AgNPs to move to form dynamic hot spots. Based on both the larger SERS signals and the long drying time required in air to approach to consistency, heating in 80 ◦ C oven for 10–20 min is suggested for filter paper substrates. 3.9. Effect of sampling processes To detect tyrosine in aqueous solutions, both methods of soaking and deposition were used to place the analytes on substrates for SERS measurements. The experiments were first conducted by soaking the FP-A substrates in 100-␮M tyrosine solution for different times. After drying in an oven at 80 ◦ C for 20 min, the SERS spectra were acquired and the results are plotted in Fig. 5B. The time to reach the maximal signal was very fast; only 5 min of soaking time was enough to reach the maximal signal. On the other hand, the fast adsorption reveals that sample can access the AgNPs easily via the large pore size in the matrices and also the cellulosic fibers allow the fast draining of water to concentrate the tyrosine on the surface. Further experiments were conducted by deposition of different amount of sample solution directly on FP-A substrates. The substrates were placed on a stack of tissues to drain the water during deposition of analyte solution. Samples were deposited two ways; deposition of an amount of tyrosine solution at once and the same amount by repeated deposition of 100 ␮L. The results are plotted in Fig. 6A. As can be seen, directly depositing a certain amount of sample solution to the substrates at once, the signals increased with volume, but reached a maximum around 500 ␮L of solution. When sample was repeatedly deposited on the FP-A substrates in small amounts, the observed signal decreased with the number of repeated depositions. This may indicate that the tyrosine molecules were not firmly interacting with either fibers or the AgNPs, and were washed out by the next deposition. Also, the residual water molecules from the previous deposition protect the AgNPs and saturate the cellulosic fibers to decrease the adsorption of the tyrosine on the surface of AgNPs. Meanwhile, repeated deposition can also increase the chance to wash off the AgNPs on the filter substrates to decrease the SERS signals. Based on above results, the soaking method is suggested for detection because it is much simpler and does not sacrifice any of the analytical signals. 3.10. Quantitative aspects To further examine detection behavior over concentration of tyrosine and to verify the applicability of this approach to the quantitative determination of tyrosine, the linearity of the standard curve was examined. SERS substrates of FT-A were soaked in aqueous solutions of tyrosine for 10 min and 30 min. After drying in an oven at 80 ◦ C for 20 min, SERS spectra were measured. Fig. 6B plots the SERS band at 1595 cm−1 against the concentration of tyrosine. The linearity of the standard curve was restricted to a region lower than 100 ␮M. Also, the change of the soaking time from 10 to 30 min gave essentially the same relationship between concentration and the resulting signals, but with a slight increase of the

Fig. 6. (A) SERS signal of tyrosine by addition of 100 ␮M solution at once (䊉) and addition of 100 ␮L of 100 ␮M of tyrosine to accumulate a designated volume (). (B) Relationship between concentration of tyrosine and SERS signal. Substrates were soaked in different concentrations of tyrosine for 10 () and 30 min (). The insert is the enlargement of the lower concentration region.

SERS signal. Based on three times the variation of blank signals, the detection limit was estimated to be ca. 625 nM for a soaking time of 10 min. 4. Conclusion In this work, highly sensitive SERS substrates based on filter paper were prepared and their capabilities demonstrated for detection of analytes in aqueous solutions. Factors including the concentration and reagents in pretreatment filter paper were systematically studied. Based on the detected signals for probe molecules of thio compounds and the observed FE-SEM images, the pretreatment method influences the sizes and distributions of AgNPs as well as the SERS signals they produce. The pretreatment reagents affected the surface properties of filter paper and potentially affected the silver plating reaction. Ammonia was found to be the most suitable pre-treatment reagent among those examined. Under the optimal conditions, the RSD obtained from triplicate runs can be less than 8% and the EF value of pNTP was calculated to be ca.107 , which is an order of magnitude higher than substrates based on glass. To demonstrate the advantages of filter paper-based SERS substrates, they were used to detect tyrosine in aqueous solution. To remove the influences from residual water, substrates were dried in an oven at 80 ◦ C for 20 min. The observed SERS signals were much more reproducible compared to air dried substrates. Also, the observed SERS signals were higher than for air-dried substrates. The drying process in an oven at 80 ◦ C may induce formation of dynamic hot spots to produce high SERS signals. In terms of

96

M.-L. Cheng et al. / Analytica Chimica Acta 708 (2011) 89–96

quantitative aspects, the soaking method is simple and more reproducible compared to direct addition method. Based on the standard curve for detection of tyrosine, a linear range up to 100 ␮M was observed and the detection limit was found to be ca. 625 nM with a soaking time of 10 min. Also, spectral interferences from other amino acids or the substrates were insignificant.

[22] [23] [24] [25] [26] [27] [28] [29]

Acknowledgments The authors thank the National Science Council of the Republic of China for supporting this work financially. Authors also thank the Center of Nanoscience and Nanotechnology of National ChungHsing University for providing critical equipments in this study. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.aca.2011.10.013.

[30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40]

References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21]

M. Fleischman, P.J. Hendra, A. McQuillan, Chem. Phys. Lett. 26 (1974) 163–166. D.L. Jeanmarie, R.P. Van Duyne, J. Electroanal. Chem. 84 (1977) 1–20. M.G. Albrecht, J.A. Creighton, J. Am. Chem. Soc. 99 (1977) 5215–5217. M. Moskovits, Rev. Mod. Phys. 57 (1986) 783–786. K. Kneipp, H. Kneipp, I. Itzkan, R.R. Dasari, M.S. Feld, Chem. Rev. 99 (1999) 2957–2976. S.P. Mulvaney, C.D. Keating, Anal. Chem. 72 (2000) 145–158. C.L. Haynes, A.D. McFarland, R.P. Van Duynee, Anal. Chem. 77 (2005) 338A–346A. A.G. Brolo, P. Germain, G.J. Hager, J. Phys. Chem. B 106 (2002) 5982–5987. X.M. Yang, D.A. Tryk, K. Hashimoto, A. Fujishima, J. Phys. Chem. B 102 (1998) 4933–4943. P.C. Lee, D. Meisel, J. Phys. Chem. 86 (1982) 3391–3395. J.A. Creighton, C.G. Blatchford, M.G. Albrecht, J. Chem. Soc., Faraday Trans. 75 (1979) 790–798. N. Leopold, B. Lendl, J. Phys. Chem. B 107 (2003) 5723–5727. Y.S. Li, J. Cheng, L.B. Coons, Spectrochim. Acta Part A 55 (1999) 1197–1207. S.B. Chaney, S. Shanmukh, R.A. Dluhy, Y.P. Zhao, Appl. Phys. Lett. 87 (2005) 031908. L. Bao, S.M. Mahurin, R.G. Haire, S. Dai, Anal. Chem. 75 (2003) 6614–6620. S. Farquharson, C. Shende, F.E. Inscore, P. Maksymiuk, A. Gift, J. Raman Spectrosc. 36 (2005) 208–212. Y.H. Lee, S.S. Dai, J.P. Young, J. Raman Spectrosc. 28 (1997) 635–639. T. Murphy, H. Schmidt, H.D. Kronfeldt, Appl. Phys. B 69 (1999) 147–150. F. Ni, T.M. Cotton, Anal. Chem. 58 (1986) 3159–3163. Y. Saito, J.J. Wang, D.N. Batchelder, D.A. Smith, Langmuir 18 (2002) 2959–2961. Z. Wang, L.J. Rothberg, Appl. Phys. B 84 (2006) 289–293.

[41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54] [55] [56] [57] [58] [59] [60] [61] [62]

H.K. Park, J.K. Yoon, K. Kim, Langmuir 22 (2006) 1626–1629. Y. Yang, A.M. Bittner, K. Kern, J. Solid. State. Electrochem. 11 (2007) 150–154. M.L. Cheng, J. Yang, Appl. Spectrosc. 62 (2008) 1384–1394. L. Qu, L. Dai, J. Phys. Chem. B 109 (2005) 13985–13990. Y. He, X. Wu, G. Lu, G. Shi, Nanotechnology 16 (2005) 791–796. C.D. Tran, Anal. Chem. 56 (1984) 824–826. T. Vo-Dinh, M.Y.K. Hiromoto, G.M. Begun, R.L. Moody, Anal. Chem. 56 (1984) 1667–1670. J.J. Laserna, A.D. Campiglia, J.D. Winefordner, Anal. Chim. Acta 208 (1988) 21–30. J.J. Laserna, W.S. Sutherland, J.D. Winefordner, Anal. Chim. Acta 237 (1990) 439–450. L.M. Cabalín, J.J. Laserna, Anal. Chim. Acta 310 (1995) 337–345. F. Ota, S. Higuchi, Y. Gohshi, K. Furuya, M. Ban, M. Kyoto, J. Raman Spectrosc. 28 (1997) 849–854. Z. Niu, Y. Fang, J. Colloid. Interface Sci. 303 (2006) 224–228. G.G. Huang, X.X. han, M.K. Hossain, Y. Osaki, Anal. Chem. 81 (2009) 5881–5888. A. Schulze, D. Kohlmueller, E. Mayatepek, Clin. Chim. Acta 283 (1999) 15–20. T. Huang, A. Warsinke, T. Kuwana, F.W. Scheller, Anal. Chem. 70 (1998) 991–997. J.C. Deutsch, J. Chromatogr. B 690 (1997) 1–6. Y. Dale, V. Mackey, R. Mushi, A. Nyanda, M. Maleque, J. Ike, J. Chromatogr. B 788 (2003) 1–8. P. Kumarathasan, R. Vincent, J. Chromatogr. A 987 (2003) 349–358. C. Bayle, N. Siri, V. Poinsot, M. Treilhou, E. Causse, F. Couderc, J. Chromatog. A. 1013 (2003) 123–130. C. Deng, Y. Deng, B. Wang, X. Yang, J. Chromatogr. B 780 (2002) 407–413. H. Orhan, N. Vermeulen, C. Tump, H. Zappey, J. Meerman, J. Chromatogr. B 799 (2004) 245–254. S. Hassan, Anal. Chem. 47 (1975) 1429–1432. Y. Mazuma, M. Maekawa, Y. Kuwabara, T. Nakajima, K. Taniguchi, T. Kanno, Clin. Chem. 35 (1989) 1399–1403. G.A. Rivas, V.M. Solis, Anal. Chem. 63 (1991) 2762–2765. F. Wang, Y.Z. Wu, Y. Qing, Y.X. Ci, Anal. Lett. 25 (1992) 1469–1478. S. Alonso, L. Zamora, M. Calatayud, Talanta 60 (2003) 369–376. C.-J. Lee, J. Yang, Anal. Biochem. 359 (2006) 124–131. G.G. Huang, J. Yang, Biosens. Bioelectron. 21 (2005) 408–418. B.O. Skadtchenko, R. Aroca, Spectrochim. Acta Part A 57 (2001) 1009–1016. Z. Zhang, T. Imae, J. Colloid Interface Sci. 233 (2001) 99–106. W. Hill, B. Wehling, J. Phys. Chem. 97 (1993) 9451–9455. M. Osawa, N. Matsuda, K. Yoshii, I. Uchida, J. Phys. Chem. 98 (1994) 12702–12707. Y. Imai, Y. Tamai, Y. Kurokawa, J. Sol–Gel Sci. Technol. 11 (1998) 273–278. C.J. Orendorff, A. Gole, T.K. Sau, C.J. Murphy, Anal. Chem. 77 (2005) 3261–3266. A. Aldeanuva-Potel, E. Faoucher, R.A. Alvarez-Puebla, L.M. Liz-Marzán, M. Brust, Anal. Chem. 81 (2009) 9233–9238. M. Wolpert, P. Hellwig, Spectrochim. Acta part A 64 (2006) 987–1001. R. Ramaekers, J. Pajak, M. Rospenk, G. Maes, Spectrochim. Acta part A 61 (2005) 1347–1356. H.I. Lee, M.S. Kim, S.W. Suh, Bull. Korean Chem. Soc. 9 (1988) 218–223. I.R. Nabiev, G.D. Chumanov, Biophysics 31 (1986) 183–190. G.D. Chumanov, R.G. Efremov, I.R. Nabiev, J. Raman Spectrosc. 21 (1990) 43–48. S. Stewart, P.M. Fredericks, Spectrochim. Acta Part A 55 (1999) 1641–1660.