Introduction

Currently, nanotechnologies are increasingly interested in metallic nanoparticles, particularly silver ones because of their exceptional properties and the extent of their field of application. They are exploited in microbiology [1,2,3,4], catalysis [5,6,7], microelectronics [8,9,10], and, especially, in the analysis by surface-enhanced Raman scattering (SERS) [11,12,13,14].

SERS technique discovered in the mid-1970s [15] is based on the use of metals such as silver, gold, and less frequently copper, generally in the form of roughened massive substrates or particles in colloidal suspensions. Here, the size and shape of the asperities of the metal surface or colloidal aggregates play a key role in the emergence of the SERS effect and affect the value of the enhancement factor of the Raman intensity. In this context, it has been reported that the enhancement factor can reach values in the order of 1014 [16,17,18,19,20] which would make possible the detection of traces or even single molecule in complex environments such as biological media [21,22,23,24,25,26].

It has been shown in many studies that the amplification of the local electromagnetic field, due to surface plasmon resonance [27] and responsible for the enhancement effect, arises at intergranular sites, conventionally called “hot spots”, whose dimensions are smaller than the wavelength of the excitation line [28,29,30,31,32,33,34]. Hence, the current craze for controlled development of new metal nanomatrices including many “hot spots” with increasingly reduced dimensions [35, 36].

Drawing on this, one should expect to observe a huge intensification of the Raman signal for molecules adsorbed or electrodeposited on Ag or Au fractal clusters due to the presence of a large number of “hot spots” in this kind of structures with scale invariance symmetry [31,32,33,34]. This is the context in which our present paper fits.

Since several years, our research team is involved in the improvement and design of new systems origins of the SERS effect and their application to rovibrational investigations of electrosynthesized conducting polymers [37,38,39,40,41,42,43]. We have shown that this fine and powerful technique not only allows to provide precise information about the polymer structure and to follow the in situ changes occurring at the conjugated skeleton during the doping-undoping process, but also to access the spatial arrangement of the chains and their anchoring mode to the SERS-active surface. We have used (i) Ag, Au, and Cu electrodes previously activated by mechanical and electrochemical roughening procedures specific to each metal [37, 40,41,42,43]; (ii) partially aggregated Ag and Au colloids in the case of soluble conducting polymers [40, 42, 43]; (iii) multilayer systems in which the polymer is either electrodeposited on a thin layer of a SERS-active metal coating a supporting metal plate or sandwiched between the thin SERS-active layer and the working electrode [41, 42]; and (iv) organized media based on micelles whose counterion is the Ag+ cation [40, 42, 43]. Electropolymerization by cyclic scanning of potential in this electrolytic medium leads to the production of a composite polymer film containing entrapped Ag nanoparticles…

In the present work, and to focus on the influence of the fractal geometry of the SERS-active matrix on the Raman spectrum quality of a conducting polymer, we present a comparative SERS study between two polypyrrole films prepared according to two different protocols. The first protocol is electrochemical (ECP) and consists of electrodepositing a thin PPy film on an Ag electrode previously roughened by potentiodynamic oxidation-reduction cycles. In this case, to highlight the emergence of SERS effect and to evaluate the enhancement factor, a film electrodeposited under the same conditions on a Pt electrode is also analyzed. The second protocol is photochemical (PCP), in which the deposition of PPy is done by electron transfer between the pyrrole monomer and the Ag+ cation present in the solution, which leads to a composite formed of polypyrrole with fractal Ag clusters. The reaction is photocatalyzed by UV light [44,45,46], and the formation of the silver fractal structure is accelerated by the copper plate because of the large potential gap between the Cu2+/Cu (ca. + 0.34 V) and Ag+/Ag (ca. + 0.80 V) redox pairs. This last procedure gives rise to PPy SERS spectra of high quality and much more intense than those observed in the case of the electrochemical protocol.

Experimental

Chemicals

Pyrrole C4H5N (≥ 98%, Aldrich) was redistilled and stored in the dark under argon atmosphere prior to use. Anhydrous dichloromethane CH2Cl2 (≥ 99.8%, Sigma-Aldrich) packaged under nitrogen in Sure/Seal™ bottle, silver nitrate AgNO3 (≥ 99.0%, Sigma-Aldrich), and potassium chloride KCl (≥ 99.0%, Sigma-Aldrich) were used as received. Perchloric acid HClO4 (11 M, Sigma-Aldrich) was diluted to 2 M with distilled water. Tetrabutylammonium hexafluorophosphate (≥ 99.0%, Sigma-Aldrich) was dried under vacuum before each electrochemical experiment. Water was purified by passing distilled water through a Millipore purification system.

Preparation of the SERS-active samples

SERS effect only occurs if the underlying metal has undergone specific roughening mechanical, chemical, and/or electrochemical treatments. As in the case of metal colloids [47,48,49,50,51,52,53,54], SERS intensity is closely related to the shape and dimensions of the surface asperities. In what follows, we will describe two different protocols (ECP and PCP) for the preparation of SERS-active substrates. Then, to compare their enhancement factors, the two supports will be used for the investigation of polypyrrole.

In the ECP protocol, a preliminary roughening procedure is applied to an Ag electrode. Initially, a first rapid polishing is carried out using 1200 grit sandpaper, aiming to achieve the planeness of the metal surface, followed by a succession of three steps of fine polishing using 6, 3, and 1 μm diamond pastes, respectively. This mechanical pretreatment gives the Ag surface a shiny mirror-like finish.

The silver electrode is then introduced into an electrochemical cell containing 0.1 M KCl aqueous solution. Then, three potential sweeps between − 0.4 and + 0.2 V vs SCE are applied to the working electrode with 10 mV s−1 scan rate. The voltammogram corresponding to the oxidation-reduction cycle (Fig. 1) is characterized by an abrupt growth of the anodic current at + 0.1 V which corresponds to the oxidation of the polished Ag surface and the formation of silver chloride. On the return, a cathodic peak appears at − 0.15 V and corresponds to the reduction of the silver chloride deposited on the surface of the electrode.

Fig. 1
figure 1

Electrochemical set-up used for roughening the silver electrode, comprising a single compartment electrochemical cell (c1) equipped with a three-electrode system connected to a computer-controlled potentiostat (c2). The potentiodynamic curve (c3) and SEM micrographs of the electrode surface (a, b) obtained after the SERS-activation procedure are also shown

Full size image

Thus, the consecutive and repetitive dissolution-redeposition of silver creates topographic defects on the electrode surface which are responsible for the SERS effect produced by excitation of the surface plasmons. The micrograph obtained by scanning electron microscopy (SEM) shown in Fig. 1 is that of the Ag electrode after the electrochemical activation treatment. The surface asperities are in the form of spheroids bonded to each other to form rods that themselves form connections in a “coral-like” structure whose distribution is uniform over the entire surface of the electrode.

The PCP protocol is based on an in situ oxidation of pyrrole by Ag+ cations on a copper plate in aqueous solution of silver nitrate and under UV irradiation: A carefully polished copper plate is introduced into 0.5 M aqueous pyrrole solution. Then, under constant stirring and UV exposure (365 nm), a silver nitrate solution AgNO3 0.2 M is slowly added dropwise. The procedure takes about an hour and can also be performed under daylight without UV lamp; in this case, a longer time is required for polymerization (more than 24 h). The used concentrations and operating conditions were retained after several optimization tests. An electronic transfer occurs between the monomer and the Ag+ cations and leads to a photo-assisted chemical polymerization of pyrrole and at the same time to a deposit of Ag0 particles on the electrode surface.

Topographic analysis of the electrode surface reveals a particular structure of the silver deposit. On the micrograph of Fig. 2, the growth of the Ag clusters is done by repetition of the same “pine leaves” forms at increasingly smaller scales, giving a fractal geometry on the surface of the Cu plate. The importance of the fractal behavior in the observation of SERS effect has been shown by several authors [47, 48, 55,56,57,58,59,60].

Fig. 2
figure 2

Schematic representation of the photochemical procedure (c) used for the preparation of polypyrrole with embedded fractal leaf-shaped silver crystals on a copper plate. The SEM micrographs (a, b) obtained at the end of the experiment show the growth geometry of the silver matrices responsible for the huge SERS effect

Full size image

During the formation of a PPy chain of length n (=l x k), the photochemical polymerization follows the Scheme 1 below:

Scheme 1
scheme 1

Photochemical polymerization of pyrrole by AgNO3

Full size image

where j, k, l, m, and n are the non-zero positive integers: n = l x k is the average chain length, 1/l represents the doping rate (≈ 33% for fully doped PPy), and j = m-k.

Generally, the obtaining of a fully doped polymer involves about 2.33 electrons for each monomer unit: 2 e− removed to allow coupling in the two α and α′ positions and 0.33 e− corresponding to the fraction of the positive charge carried by the ring and compensated by one doping anion for each three monomer units. The reaction leads to the formation of a PPy-Ag composite doped with nitrate anions NO3−. Nitric acid is also obtained as bypass product.

Electrochemical apparatus and characterization techniques

Electrochemical apparatus

The potentiodynamic and galvanostatic electrochemical experiments were carried out in a one-compartment cell using a Voltalab potentiostat/galvanostat model PGZ-301. The working electrode (WE) was either platinum or roughened silver plates (40 mm × 8 mm). The auxiliary electrode (AE) was a stainless-steel grid and potentials were measured vs. SCE reference electrode (Ref).

The in situ SERS experiments were performed using a homemade Teflon micro cell equipped with three miniaturized electrodes: WE was a SERS-activated silver disk (Ø = 4 mm) coated with a thin PPy layer. A stainless-steel cap-shaped grid with a small circular aperture at its center to allow the passage of the excitation laser beam served as AE. A thin silver wire (Ø = 0.25 mm) covered with silver chloride (Ag/AgCl) was used as reference electrode.

Raman spectroscopy

Raman spectra were recorded using λe = 514.5 nm excitation green line from of a Spectra-Physics model 165 argon ion laser. The scattered light, collected at 180° from the incident beam (retro-Raman), was focused onto the entrance slit of a Dilor XY spectrometer. This apparatus consists of a double monochromator used in subtractive mode (i.e., with no dispersion) to select a given spectral range, followed by a spectrograph to complete the dispersion and by a 1024 × 256 Jobin-Yvon CCD matrix multichannel detector, cooled by the thermoelectric effect.

X-ray photoelectron spectroscopy

XPS analyses were performed with a Vacuum Generators Escalab MKI spectrometer equipped with Mg Kα (1256.6 eV) and Al Kα (1486.6 eV) sources operating at 200 mW. X-rays, focused onto an area of 4 mm × 10 mm, are weak enough not to damage the sample. Spectra were calibrated against the 1 s carbon electron peak (285 eV binding energy) of the polythiophene ring carbon atoms as internal reference.

Results and discussion

SERS analysis of PPy electrodeposited on roughened silver electrode

The electropolymerization of pyrrole was carried out on Pt and Ag electrodes by the galvanostatic method at 0.5 mA cm−2 current density, in CH2Cl2 + 0.1 M N (Bu)4PF6 + 0.1 M pyrrole electrolytic medium. Beforehand, the silver electrodes were subjected to mechanical polishing and electrochemical activation treatments according to the aforementioned ECP protocol.

In order to prevent the oxidation of the silver electrode which would hinder the pyrrole polymerization reaction, the constituents of the electrolytic medium have been carefully selected after tests carried out on several solvents and supporting electrolytes. A total of 0.5 mA cm−2 is the threshold value that leads to the formation of a PPy without exceeding the oxidation potential of the monomer (0.8 V) (Fig. 3b, c). The obtained film is electroactive and its electroactivity voltammogram is characterized by a couple of well-defined oxidation and reduction peaks at − 0.05 and − 0.40 V, respectively (Fig. 4). Values below 0.5 mA cm−2 do not allow electropolymerization (Fig. 3a), while high values (> 1 mA cm−2) give poor quality films since the electropolymerization plateaus occur at high potentials (Fig. 3d, e) for which the polymer is overoxidized and undergoes partial degradation [61,62,63,64].

Fig. 3
figure 3

Chronopotentiometric curves recorded during pyrrole electropolymerization in CH2Cl2 + 0.1 M N (Bu)4PF6 + 0.1 M pyrrole electrolytic medium on SERS-activated Ag electrode. Applied current densities: a 0.1 mA cm−2, b 0.5 mA cm−2, c 1 mA cm−2, d 2 mA cm−2, and e 5 mA cm−2

Full size image
Fig. 4
figure 4

Electrochemical behavior of polypyrrole film recorded in CH2Cl2 + 0.1 M N (Bu)4PF6 electrolytic medium with 20 mV s−1 scan rate. PPy was obtained galvanostatically on SERS-activated Ag electrode with 0.5 mA cm−2 current density

Full size image

Ex situ SERS experiments

Ex situ Raman analysis of the PPy films covering the two metal substrates with λe = 514.5 nm excitation wavelength reveals great differences. Since platinum is not a SERS-active material, the Raman spectrum obtained on this metal with 10 mW laser power is of weak intensity (Fig. 5). On the other hand, the film deposited on roughened Ag shows a very important SERS effect; the spectrum recorded in this case with reduced laser power (1 mW) is more intense with better resolved peaks (Fig. 5). By comparing the intensities of the most intense band in the two spectra, considering the laser powers used in each case, the SERS enhancement factor was evaluated at ca. 100.

Fig. 5
figure 5

SERS spectrum of polypyrrole film electrodeposited on roughened silver electrode (a) as compared to Raman spectrum of polypyrrole film electrosynthesized in the same conditions on platinum electrode (b). The inset is the same as b. Excitation wavelength: λe = 514.5 nm. Laser power: 10 mW in the case of Pt and 1 mW for roughened Ag

Full size image

Moreover, all bands observed on Pt electrode are present when the Ag electrode is used. In addition, thanks to the SERS effect, other vibrational modes, inactive on Pt, are visible on the spectrum obtained on Ag.

As can be seen from Fig. 5, the Raman spectra of PPy, recorded on Pt and activated Ag electrodes, respectively, have overall same behaviors. The SERS spectrum is dominated by intense and broad signals in the 1200–1600 cm−1 domain. The asymmetrical shapes of their envelopes suggest that they result from the overlapping of several vibrational modes. Table 1 gathers the wavenumbers of the various vibrational modes pointed on the spectra of PPy/Pt and PPy/Ag with those calculated by Faulques et al. [65] using valence force field. The bands observed in the spectra were assigned according to literature.

Table 1 Wavenumbers and assignments of the Raman bands of PPy films electrodeposited on Pt and roughened Ag electrodes, as compared to the calculated ones [65]
Full size table

The main band at ca. 1610 cm−1 is assigned to C=C ring stretching [65]. It is well known that the transition between the reduced and oxidized states of polypyrrole during the doping process is accompanied by the transformation of some rings from the aromatic to the quinoid structures. According to theoretical values reported by Faulques et al. [65], the frequency of the C=C stretching vibration in the quinoid nuclei is higher than that of the same vibrational mode belonging to the aromatic rings (Table 1). Therefore, by comparing the positions of the Raman bands of PPy/Pt and PPy/Ag spectra and particularly the νC=C main band, which is more shifted towards high frequencies in the case of PPy/Ag, it appears that the polymer deposited on Ag contains more quinoid nuclei than that elaborated on Pt. Therefore, because the number of quinoid units increases at the expense of aromatic ones when the oxidation degree of the polymer increases, PPy on Ag would be more oxidized than on Pt.

The bands observed around 1362, 1249, and 985 cm−1 are assigned to the ring deformation, antisymmetric C–N stretching, antisymmetric C–H in plane bending, and ring in plane deformation. These bands have been reported to be due to bipolarons which are dications extended over three or four pyrrole nuclei of quinoid structure [66].

The double peaks of C–H in plane deformation at 1086 and 1057 cm−1 were investigated by Liu and Hwang [67]. The authors stated that the intensity ratio of the two bands that form the doublet reflects the oxidation state of the polymer. When the oxidation degree is decreased, the intensity of the peak located at the higher wavenumber side of the doublet decreases and almost disappears for fully reduced PPy [67]. In our case, by qualitatively comparing the relative intensities of these two bands in the spectra of PPy/Ag and PPy/Pt (Fig. 5), it appears that the film on roughened silver is more oxidized than that on platinum, in agreement with the behavior of the νC=C main band explained above. This very important observation will be verified in the in situ SERS section.

In addition, two bands are particularly affected by the effect of intensity enhancement, the 937 and 1418 cm−1 bands attributed to the Cβ–H and Cα–N stretching modes, respectively. Taking the SERS selection rules into account, an approach to the spatial orientation of the pyrrole cycles with respect to the metal surface can be made. Indeed, these rules predict that the most enhanced modes are those involving a significant variation of the αzz component (along the oz axis normal to the surface of the SERS-active substrate) of the polarizability tensor during their vibration. This implies that the polypyrrole chains are oriented so that the Cβ–H and Cα–N bonds form almost a right angle with the metal surface. Therefore, the pyrrole rings would be nearly in “standing” position, since they belong to the same plane as these bonds (Scheme 2). In previous works devoted to the investigation of polythiophene and its derivatives using activated Ag or Au electrodes or metal colloids, we have observed the same orientation of thiophene cycles with respect to the metal surface [37, 41, 42]. It should be noted that the same “standing” orientation was reported by Ye et al. [68] in a paper dedicated to surface-enhanced infrared absorption (SEIRA) investigation of polypyrrole-coated silver nanoparticles, albeit with the nitrogen atoms of the pyrrole rings pointing towards the outside of the PPy shell. However, the strong affinity of metals (Ag, Au, ...) for heteroatoms such as nitrogen and sulfur [69,70,71,72,73,74,75,76,77] corroborates with our observation.

Scheme 2
scheme 2

Spatial representation of pyrrole units in “standing” and “lying” positions with respect to the SERS-active surface

Full size image

The marked improvement of the Raman signal obtained using a roughened silver plate as working electrode for the electrosynthesis of PPy allowed the observation of high-quality SERS spectra and facilitated the identification of various vibrational modes of the polymer. Consequently, these operating conditions seem favorable for performing in situ SERS experiments coupling Raman spectroscopy to electrochemical techniques. Since the SERS effect only affects the inner layers of the polymer in direct contact with the metal substrate, the contribution of the vibrational modes of the electrolytic medium (solvent + supporting electrolyte) crossed by the laser beam will be negligible.

In situ SERS experiments

PPy film electrodeposited on roughened Ag electrode was polarized at various potentials in HClO4 2 M aqueous medium and excited with green excitation line, λe = 514.5 nm, with only 1 mW laser power. Figure 6 illustrates the applied potential-time function (E = f(t)) and the resulting chronoamperometric response (j = f(t)). Despite a drop in intensity related to the in situ conditions, the overall aspect of the obtained spectra at different oxidation degrees (Fig. 7) is similar to that of the open-circuit ex situ spectrum.

Fig. 6
figure 6

Chronoamperometric response j = f(t) obtained during the in situ SERS experiment. The double-arrows delimit the recording duration of each SERS spectrum. The applied potential-time function E = f(t) is also shown. a − 0.6 V, b − 0.4 V, c − 0.3 V, d − 0.2 V, e − 0.1 V, f 0.0 V, g 0.1 V, and h 0.2 V vs Ag/AgCl

Full size image
Fig. 7
figure 7

Evolution of the in situ SERS spectrum of polypyrrole electrodeposited on roughened silver electrode as a function of the applied potential: a − 0.6 V, b − 0.4 V, c − 0.3 V, d − 0.2 V, e − 0.1 V, f 0.0 V, g 0.1 V, and h 0.2 V vs Ag/AgCl. Electrolytic medium: H2O + 2 M HClO4. Excitation wavelength: 514.5 nm. Laser power: 1 mW

Full size image

All the bands described in the ex situ SERS analysis are also observed in the spectra recorded during the in situ procedure. A first comparative analysis of the SERS spectra overlaid on Fig. 7 reveals several modifications in the shapes, the positions, and the relative intensities of the bands.

When the potential of the working electrode is increased, the oxidation degree of the PPy film increases and leads to the creation of polarons (cation radicals) and bipolarons (dications) along the polymer chains. These charged entities, responsible for the conduction in the polymer, each consist of three to four quinoid-structured pyrrole units, resulting from transformation of an increasing number of aromatic-like pyrrole nuclei. This process is accompanied by the insertion of counter-ions from the electrolytic medium to ensure electroneutrality. This results in distortions and conjugation ruptures in the chains and increases the disorder in the polymer.

To highlight these structural transformations, it is necessary to perform a detailed analysis of some PPy key bands, to identify those most affected by the SERS enhancement effect, and to follow their evolution depending on the oxidation degree.

Nevertheless, the overall raw spectra in Fig. 7 are difficult to interpret because they consist of bands, most of which are asymmetric or include shoulders, which implies that they consist of several overlapping lines. It is therefore useful to proceed with the deconvolution of these spectra in order to access information relating to the structural changes, the geometry, and the anchoring mode of the polymer to the metal surface. For reasons of simplification, the spectral range is divided into three zones denoted H, M, and L corresponding to high, medium, and low wavenumber domains, respectively (Fig. 7).

In zone H, the first observation concerns the most intense band νC=C. When the PPy film is progressively doped, this band shifts towards high frequencies while widening. The plot representing the evolution of the νC=C wavenumber as a function of the polarization potential (Fig. 8) shows two characteristic regions. The first potential zone, between − 0.4 and − 0.25 V vs Ag/AgCl, takes the form of a plateau in which the νC=C band keeps a relatively fixed position. The second domain starts at ca. − 0.25 V and is characterized by a linear variation of \( \overset{\sim }{\nu } \) versus E with a positive slope of about 17.5 cm−1 V−1. Here, it should be noted that this behavior is not limited to 0.2 V as the upper limit imposed by the low oxidation potential of the Ag electrode, but continues beyond this value; suffice it to note the position of this band for the oxidized sample analyzed ex situ (Fig. 5).

Fig. 8
figure 8

Plot of νC=C wavenumber as a function of the polarization potential during the in situ SERS experiment. νC=C is assigned to stretching vibration of the C=C double-bonds of pyrrole rings

Full size image

Bukowska and Jackowska [78] have assigned the frequency shift of this band to the charge quantity introduced by doping into the polymer chains. These authors have assumed that the force constants of the pyrrole units forming the polymer could be sensitive to the electron density on the rings, which depends on the oxidation state of the polymer.

The second interesting feature of this band is its widening when the potential imposed to the working electrode increases. At first sight, this enlargement could be associated with changes in the distribution of the conjugation length in agreement with the work of Kuzmany et al. on polyacetylene [79, 80]. Indeed, the higher the doping rate of the polymer increases, the larger the number of structural defects on its chains becomes, thus causing a non-uniform distribution of the conjugated segments and leading to wider Raman bands. However, a second hypothesis seems more appropriate to explain both the shift and widening of this band: further analysis of the νC=C band reveals the existence of a shoulder on the high-frequency side of the main band and whose profile changes with the applied potential (Fig. 9). This massif is, in fact, made up of the overlapping of at least two lines that we have attributed to the same stretching mode of C=C double bonds belonging to the nuclei of aromatic and quinoid structures.

Fig. 9
figure 9

In situ evolution of the Gaussian components of the 1480–1680 cm−1 spectral region with the applied potential. The in situ experiment is performed in H2O + 2 M HClO4 with 514.5 nm excitation wavelength and 1 mW laser power. Applied potentials: a − 0.6 V, b − 0.4 V, c − 0.3 V, d − 0.2 V, e − 0.1 V, f 0.0 V, g 0.1 V, and h 0.2 V vs Ag/AgCl

Full size image

When the Ag electrode is brought to increasing potentials, the oxidation degree of PPy increases and the number of nuclei which pass from the aromatic to the quinoid form becomes larger. This change in nuclei structure is reflected in the SERS spectrum by a growth of the high-frequency shoulder at the expense of the low-frequency component (Fig. 9).

It is well known that PPy spontaneously oxidizes in air, which implies that all spectra recorded ex situ are those of the oxidized form of PPy. This is verified by a simple observation of the νC=C band in the ex situ SERS spectrum (Fig. 5); on the one hand, it is very displaced towards the high frequencies and on the other hand its low-frequency side is characterized by a less steep slope than its high-frequency side, which means that in this case, the component \( {\nu}_{C=C}^{quin.} \) is more intense than the component \( {\nu}_{C=C}^{arom.} \).

The combination of these data concerning the position of νC=C signal and the intensity ratio of its two aromatic and quinoid components could be used as a criterion for evaluating the oxidation state of the polymer. However, it is necessary to take into consideration other parameters such as the electrosynthesis conditions including the electrochemical method and the various constituents of the electrolytic medium, especially the nature of the doping anion which has been shown to strongly influence the behavior of the Raman spectrum [81, 82].

In zone M, deconvolution and interpretation of the SERS signal are not straightforward since several bands, most often broad, falls within this narrow wavenumber range. The poorly resolved M massif, with multiple overlapping components, includes in addition to the normal modes indicated in Table 1, other modes probably due to vibrations of structural defects. The presence of the defect bands is supported by the evolution of the spectrum behavior in this domain (Fig. 7). In the reduced state (spectrum (a)), the bands are relatively sharp and less overlapped. But as soon as the oxidation degree is increased, the bands tend to widen and overlap in a broad and noisy envelope (spectrum (h)). This is the signature of the growth of disorder in the polymer chains.

Zone L mainly consists of three sharp and intense peaks in addition to three other low intensity bands revealed by a spectral decomposition procedure (Fig. 10). Modifications depending on the applied potential are observed essentially at the relative intensities of the different peaks. We must especially notice the behavior of the double peaks of C–H in plane deformation (Fig. 10). As stated above, when the oxidation degree is decreased, the intensity of the peak located at the higher wavenumber side of the doublet decreases and almost disappears for fully reduced PPy [67]. This is exactly what we observe on Figs. 7 and 10: the intensity ratio of the two high- and low-frequency components IC-Hhigh/IC-Hlow increases when the applied potential is changed to more positive values.

Fig. 10
figure 10

In situ evolution of the Gaussian components of the 800–1100 cm−1 spectral region with the applied potential. The in situ experiment is performed in H2O + 2 M HClO4 with 514.5 nm excitation wavelength and 1 mW laser power. Applied potentials: a − 0.4 V, b − 0.3 V, c − 0.2 V, d − 0.1 V, e 0.0 V, f 0.1 V, and g 0.2 V vs Ag/AgCl. The lower-right curve illustrates the evolution of the intensity ratio I943.9/I924.6 of the 943.9 and 924.6 cm−1 lines with the electrode potential

Full size image

In this spectral zone, besides the C–H in plane deformation doublet, other bands show changes. By way of indication, we have illustrated on the same Fig. 10 the variation of the intensity ratio I943.9/I924.6 of the 943.9 cm−1 (ring in plane deformation) and 924.6 cm−1 (C–H stretching) cm−1 lines as a function of the electrode potential. The resulting curve can be divided into two domains, the first extending from − 0.4 to about − 0.25 V is characterized by a rapid decrease of the intensity ratio, and the second starting from − 0.25 V has a softer slope.

The bands whose intensity increases during the doping process could be related either to a splitting due to the growth of the quinoid component at the expense of the aromatic one, or to the emergence of bands related to structural defects (kinks) and which have identified and investigated in in situ SERS experiments on polythiophene and its derivatives [37,38,39,40,41,42].

Moreover, it should be noted that the middle signal centered at about 930 cm−1 has been attributed by some authors to the symmetric stretching of the ClO4− doping anion [83, 84]. Therefore, its intensity should increase with the increase of the oxidation degree associated with the insertion of a growing number of counter-ions. However, on the in situ SERS spectra (Figs. 7 and 10), an inverse behavior occurs: on the one hand, the signal is present even in the fully reduced state, and on the other hand, its relative intensity decreases slightly when the potential of the working electrode is increased. These observations call into question the assignment of this band to the perchlorate doping anion.

SERS analyses of photochemically prepared PPy with leaf-shaped Ag crystals

Elemental composition of PPy containing Ag particles with leaf-like structure

X-ray photoelectron spectroscopy (XPS) is used for the characterization of PPy-Ag composite. Through this technique, which has proved to be a powerful tool in the analysis of conducting polymers, we will attempt to answer some issues about the chemical composition of the polymer, its doping level, the nature of the doping anion, and the oxidation state of the silver particles. This will be done by deconvoluting the XPS signals of carbon C1s, nitrogen N1s, and silver Ag3d. The first two elements will allow, by comparison of relative intensities, to access the PPy structure, to prove the involvement of the NO3− oxidizing species in the doping process, and to evaluate the doping degree. The third element will allow to verify that the silver particles responsible for the SERS effect are in the metallic state Ag0.

Elemental analysis by XPS of Ag (Fig. 11c) shows a doublet Ag3d consisting of two sharp and symmetrical peaks at 368.1 eV (Ag3d5/2) and 374.2 eV (Ag3d3/2) associated with a single species of silver at zero-oxidation state (Ag0). The absence of oxidized silver components (silver oxide or hydroxide) could indicate that the silver particles are stabilized by the polymer with which they are in contact. Indeed, since the redox potential of the Ag+/Ag couple (roughly 0.8 V, depending on the activity of Ag+) is greater than that of the PPyox/PPyred (located between − 0.4 and − 0.05 V which are the potentials of the cathodic and anodic peaks of PPy, respectively, Fig. 4), the metallic Ag0 and the oxidized PPyox species are the most stable thermodynamically.

Fig. 11
figure 11

XPS analysis of photochemically prepared polypyrrole with leaf-shaped Ag crystals. a C1s, b N1s, and c Ag3d XPS signals

Full size image

The involvement of nitrate counter-ions NO3− in the PPy doping process and the determination of the polymer chemical composition and doping rate requires a thorough analysis of the C1s and N1s XPS spectra (Fig. 11). Thus, the decomposition of these signals into their different components is required. It provides access to the binding energies which reveal the chemical environment of each constituent and to the relative intensities whose ratios give the chemical composition and the doping degree of the polymer. The oxygen O1s signal has been voluntarily omitted since the narrow O1s spectral zone often consists of several broad and overlapping components originating from different sources such as doping anions, contaminants, residual water, etc. which hinder the deconvolution and the assignments of the oxygen spectrum.

C1s XPS signal has been processed into four Gaussian components of 1.5 eV full width at half maximum (FWHM). All given binding energies are referred to the C1s main peak at 285.0 eV attributed to carbon atoms of the pyrrole rings, more precisely those in α and α′ positions of the cycle [85,86,87,88,89,90]. Knowing that the pyrrole cycle contains two types of carbon, Cα,α′ and Cβ,β′ in α, α′ and β, β′ positions, respectively, with different chemical environments (Scheme 3), it is expected to observe a second band since the binding energies of the 1s electrons of the carbon nucleus in Cα,α′ and Cβ,β′ would not be the same. Thus, the peak at 284.1 eV has been assigned in several papers to Cβ,β′1s [85,86,87,88,89,90].

Scheme 3
scheme 3

The carbons at α and β positions do not have the same chemical environment: Cα is bonded to two carbons and one nitrogen, while Cβ is bonded to two carbons and one hydrogen

Full size image

Clark et al. [91] have measured binding energies for some five membered heterocycles. They reported, in the case of pyrrole monomer, that the presence of nitrogen which is more electronegative than hydrogen leads to a shift of 0.9 eV in C1s binding energy between Cα,α′ and Cβ,β′. This is exactly the same value we obtained in the case of PPy-Ag composite.

Ideally, the conjugated skeleton of polypyrrole contains as many Cα,α′ as Cβ,β′. Therefore, the intensities of the two lines were expected to be of approximately the same order of magnitude. However, the intensity ratio I (Cα,α′)/I (Cβ,β′) gives a much higher value (ca. 2.2) than that intended to reflect the equal amounts of both carbon species. Contrary to Pfluger and Street [88], it is unlikely that this large difference in intensity can be explained only by the presence of hydrocarbon contamination. On the other hand, in a paper devoted to XPS investigation of electrochemically synthesized polypyrrole, Atanasoska et al. [85] reported that the greater intensity of the Cα,α′ arises as a consequence of some structural arrangement of doped polypyrrole and not by random contamination. In short, so far, no consistent interpretation has been stated to explain this intensity disproportion.

The coupling being mainly α-α′, one could also think of the appearance of lateral groups attached to some β carbons along the polymer backbone as has been observed in the case of other polyheterocycles [37, 40,41,42]. This would cause an intensity drop of the Cβ,β′ 1s component. The existence of such groups, mostly oxygenated as C–OH and C=O, which act as structural defects in the polypyrrole matrix is reinforced by the presence of two other bands at 286.4 and 288.0 eV, assigned to higher oxidation states carbon atoms. However, this assumption on its own would not be enough to justify the sharp decrease of the Cβ,β′ line intensity, which is about half in comparison with Cα,α′ component.

The peak at 286.4 eV is attributed to βC–OH species [87, 89] and carbons adjacent to positively charged nitrogen atoms αC–N+ [87, 92]. To confirm the former attribution and to identify the source of these covalently bonded hydroxyl groups, Lei and Martin [93] carried out infrared investigations of polypyrrole and poly(N-methylpyrrole) electrochemically synthesized in acetonitrile containing traces of water. They conclude that the emergence of the –OH defects results from nucleophilic attacks by water on the growing polymer chains during the electropolymerization process. When the organic electrolytic medium is rigorously dried, the IR bands associated with these groups are not observed.

A fourth highest binding energy peak of relatively weak intensity at 288.0 eV arises from carbonyl groups βC=O [45, 87, 94,95,96]. They are also structural defects resulting from a slight oxidation of the polymer surface due to the unavoidable oxygen of contamination [45]. Recently, we have shown in the case of polythiophene that the amount of carbonyl groups increases with the increase of the polymer oxidation degree [97]. When the polymer is subjected to higher potentials, its electrophilic character becomes more pronounced and, hence, undergoes more nucleophilic attacks leading to more substitutions.

The XPS spectrum of nitrogen N1s (Fig. 11b) is particularly interesting, it results from the overlapping of two contributions: the nitrogen atoms of the pyrrole rings in the polymer skeleton and those belonging to the nitrate doping anions. Thus, it consists of two spectral massifs, A and B, which after signal processing into Gaussian bands, reveal the presence of four different nitrogen species.

The main band at 399.7 eV is associated with the neutral N in PPy, while positively charged nitrogen atoms N+ come out with 400.8 eV binding energy [98]. The intensity ratio of N+ to total nitrogen atoms belonging to PPy chains I(N+)/I (NPPy) can be used as an indicator of the polymer oxidation state. In our case, the calculated ratio is 0.20 (20%).

The polymer doping rate can also be calculated by the intensity ratio of the band at 406.5 eV assigned to the nitrogen of the NO3− counterion (B) compared to the N1s signal of the PPy skeleton (A) I (NNO3-)/I (NPPy). A doping level of 0.29 (29%) was found, indicating that the obtained composite, stabilized by the silver particles, is in advanced oxidation state, the maximum value being 0.33 (33%). This observation will be confirmed later by surface enhanced Raman scattering analyses (see below).

It must be emphasized that the difference between the values 0.20 and 0.29 calculated using the two intensity ratios I(N+)/I (NPPy) and I (NNO3-)/I (NPPy), respectively, could be due to the existence of a quantity of non-doping nitrates trapped in the polymer matrix during the polymerization process. The coexistence of two types of doping and non-doping anions has already been observed in the case of other supporting electrolytes such as perchlorates ClO4− and other conducting polymers such as polythiophene [97, 99, 100].

A fourth peak appears at 397.9 eV as à weak shoulder in the low-binding energy side of the massif A. It has been assigned by several authors to nitrogen of C=N structural defects [45, 46, 87]. This component represents, on average, only 3.4% of the total area of the N1 s spectrum.

Effect of the fractal character of the silver crystals on the SERS enhancement factor

The SERS spectrum shown in Fig. 12 is that of PPy obtained photochemically simultaneously with the deposition of Ag on a Cu plate (PCP protocol). In comparison with all the Raman spectra of PPy presented so far, that of Fig. 12 is of exceptional quality. Indeed, the background noise is virtually nonexistent and the enhancement factor reaches, in this case, 200 instead of 100 obtained for PPy electrodeposited on SERS-activated Ag plate. This dramatic SERS effect is attributed to the fractal nature of the analyzed surface, in agreement with other published works [47, 48]. This point will be discussed later using FDTD simulation.

Fig. 12
figure 12

SERS spectrum of polypyrrole obtained photochemically simultaneously with the deposition of fractal Ag on a Cu plate. The inset is a decomposition in four bands of the 1100–1400 cm−1 spectral domain. The spectrum is recorded with 514.5 nm excitation wavelength and 1 mW laser power

Full size image

In addition, all the peaks belonging to the spectrum are highly resolved and easily discernible. In this context, the massif observed around 1300 cm−1 in the case of PPy/Pt and PPy/Ag appears to consist of four highly visible components in the spectrum of Fig. 12; the inset included in the same figure is a decomposition in four bands of this spectral region, which until now was impossible to perform because of the overlapping of these four Raman bands in a relatively low-resolution signal.

It should be noted that the polymer obtained by this technique has a high oxidation degree. Indeed, the νC=C band located towards the high frequencies is split into its two components at 1526 cm−1 and 1620 cm−1 in good agreement with the values calculated by Faulques et al. [65] (Table 1) and confirms what we have assumed in this context. Here, the \( {\nu}_{C=C}^{quin.} \) line is clearly shifted towards the high frequencies and it is more intense than the \( {\nu}_{C=C}^{arom.} \)line: this is exactly the characteristics of a PPy in an advanced oxidation state.

Unlike individual or sufficiently spaced particles and according to several papers devoted to the study of aggregated colloidal particles or branched clusters [23, 101,102,103,104], in entangled and complex structures, interstitial spaces act as localized “hot spots” where strong electromagnetic coupling occurs and leads locally to the intensification of the Raman signal. In our case, to highlight the presence of these so-called “hot spots” and qualitatively analyze the distribution and behavior of the local electromagnetic field and, therefore, the exaltation factor (EF) at these intergranular sites, we have used a finite difference time domain method (FDTD). The simulation was performed using the Lumerical FDTD Solutions software, typically on a ramified silver branch of about 300 nm length (Fig. 13) and excited at 514.5 nm.

Fig. 13
figure 13

FDTD simulation for a leaf-shaped silver crystal pattern. The enhancement factor in the “hot spots” decreases in the order A > B > C > D

Full size image

The aim here is not to quantify the SERS enhancement factor, whose value is generally expressed as the ratio of the local field Eloc by the incident field Einc raised to the power of four (Eloc/Einc)4 [105], this could be the subject of a more detailed investigation taking into account the connections and the overlap of crystals of various sizes and shapes. The purpose is, rather, to map out, through a simple example, a spatial distribution of EF with a scale ranging from 0 (deep blue) corresponding to SERS-inactive zones to 100 (dark brown) representing the “hot spots” seats of the highest intensification of the incident field.

Figure 13 shows the existence of several cavities more or less “hot” spread across the leaf-shaped fractal object. It follows that the PPy chains incorporated in the interstices A of a few nanometers are the most concerned by the enhancement effect of the Raman signal. It should be noted, however, that the number of these sites would be much larger if one considers the propagation of the fractal geometry and the tip effects due to the interconnections between the various entangled silver clusters.

From this set of results, it appears that the gain in intensity of the PPy SERS spectrum observed in this case in comparison with that of the film electrodeposited on a roughened massive electrode can be attributed to the fractal nature of the Ag crystal growth, which promotes the formation of “hot spots” and tips with high abundance.

Conclusion

The electropolymerization of pyrrole on roughened Ag electrode has led to an important amplification of the Raman intensity whose enhancement factor reached 100 in comparison with the resonance Raman spectrum of PPy obtained in the same conditions on Pt. This significantly improved the resolution of the PPy spectra and facilitated the identification of the various vibrational modes. According to the SERS selection rules, the intensification of the Cβ–H and C–N modes suggests that the nuclei of the polymer are perpendicular to the surface of the electrode. This behavior, already observed in the case of other conducting polymers such as polythiophene and its derivatives, is reinforced by the strong affinity of silver for heteroatoms.

The in situ SERS experiments carried out on this polymer have enabled to follow several structural changes which accompany the transition from the reduced to the oxidized form, in particular the transformation of certain nuclei from aromatic to quinoid structure.

Moreover, the importance of the fractal behavior of the surface in the observation of SERS effect is dramatically confirmed by the analysis of PPy with a fractal distribution of Ag crystals photochemically obtained on a Cu plate. In this later case, the fractal leaf-shaped silver matrices exhibited significantly higher enhancement of the Raman intensity compared to the roughened silver plate because of the high amount of “hot spots” and tips as evidenced by a brief finite difference time domain simulation.