1 Introduction

Dyes used in the clothing industry are a significant pollutant in waterways [1] whose detection usually requires sensitive analytical methods such as mass spectroscopy and chromatography [2]. Typically, electrochemical [3,4,5] and optical [6,7,8,9,10] methods are used in laboratory environments to study and monitor these contaminants. Yet, spectroscopic methods are not currently effective for monitoring them in field environments where a large variety of substances are present. Therefore, detection methodologies are needed to enhance the sensitivity of photonic spectroscopy instruments from terahertz to optical, which are inherently more portable and robust than mass spectrometers. Furthermore, terahertz and millimeter wave spectroscopy, which could potentially detect dyes, are severely limited in their applicability to water samples due to high absorption in water and signal extinction [11, 12].

Recent advances in materials research have demonstrated that hydrogels can be optimized to efficiently sorb (i.e., absorb and adsorb) certain anionic azo dyes from water [13,14,15,16]. Therefore, a variety of these materials are being developed to purify the water contaminated with such types of pollutants [17, 18]. These materials have created an opportunity to capture dyes from water samples under test and immobilize them in a solid medium, practically water-free. Thus, the utilization of contaminant-impregnated hydrogels is amenable to a wide range of spectroscopic methods.

Currently, no studies have been reported in the literature that specifically address the use of hybrid hydrogels and terahertz spectroscopy for environmental remediation, particularly in the detection of water-polluting dyes. In this work, we demonstrate the detection of the anionic dye acid blue 113 (AB113), originally present in an aqueous medium, in hydrogels using terahertz time-domain spectroscopy (THz-TDS). This anionic dye is one of the most prevalent azo dyes found in textile industry effluents [19], making it a suitable model for studying these contaminants. A selection of four distinct hydrogels, each differing in origin and chemical structure, is investigated to capture the AB113 azo dye. A comprehensive study is conducted to characterize the mass of dye in each hydrogel, along with its absorption spectra in the UV, visible, and infrared regions. The THz-TDS spectra of these hydrogels are then obtained, making it possible to quantify the effect of the dye on the spectra.

2 Materials and Methods

For conducting the dye sorption experiments and subsequent spectral characterizations, hydrogel membranes from both natural (chitosan, CS) and synthetic (poly(vinyl alcohol), PVA) polymers were prepared separately. Furthermore, chemically cross-linked hybrid hydrogels formed from the CS and PVA polymeric blend were also synthesized for the same purpose. Additionally, we prepared a nanocomposite version of these cross-linked hybrid hydrogels. To accomplish this, medium-molecular-weight (190,000–310,000 g/mol) CS powder with a 75–85% of deacetylation and PVA with a 99% of hydrolysis and 89,000–98,000 g/mol of average molecular weight (\(M_w\)) were used. As solvents, we used deionized water and HPLC-grade acetic acid. Analytical grade genipin (\(\ge \)98% purity, GEN) powder was used as a chemical cross-linker reagent. In addition, multi-walled carbon nanotubes (MWCNTs)—6–9 nm in diameter and 5 \(\mu \)m in length—were used to synthesize the nanocomposite hydrogels. All chemicals were purchased from Sigma-Aldrich and used as received.

2.1 Hydrogels’ Synthesis

In order to produce thin membranes, the hydrogels were prepared by a solvent casting method as has been reported in our previous works [20, 21]. In brief, a 2.5% w/v aqueous chitosan solution was prepared by dissolving CS powder in a 1% v/v acetic acid solution under constant magnetic stirring at room temperature for 24 hours. Then, a poly(vinyl alcohol) solution was prepared by dissolving PVA in deionized water under magnetic stirring at 80 \(^\circ \)C for 2 h. These solutions were used to prepare, separately, CS and PVA pristine hydrogels, as well as a hybrid blend of CS/PVA mixed at a 3:1 (CS to PVA) volume ratio. The cross-linker reagent (GEN solution 0.5% w/v) was added to the CS/PVA polymeric mixture, and it was stirred for 1 h at room temperature to obtain a CS/PVA/GEN cross-linked blend. Moreover, the CS/PVA/GEN/MWCNTs nanocomposite blend was prepared following this same method, but the MWCNTs (0.01% w/v) were first incorporated into the CS solution. Subsequently, the CS/MWCNTs blend was sonicated using an ultrasonic bath and maintained under magnetic stirring at room temperature for 24 hours to disperse the MWCNTs. This blend was used to obtain the CS/PVA/GEN/MWCNTs polymeric solution at the same volume ratio mentioned above. The pristine (CS and PVA), cross-linked (CS/PVA/GEN), and cross-linked nanocomposite (CS/PVA/GEN/MWCNTs) polymeric solutions were poured into plastic Petri dishes and subsequently dried for 72 hours at room temperature. The films obtained were then detached from the dishes and placed in a vacuum oven at 60 \(^\circ \)C and \(-\)17 inHg for 48 hours to remove any remaining water or solvents within the films. Finally, the thickness of the cast films was measured using a digital micrometer with a resolution of 1.27 \(\mu \)m, and they were then stored in a desiccator until testing.

2.2 Dye Sorption Experiments

To evaluate the dye sorption performance of the CS, PVA, CS/PVA/GEN, and CS/PVA/GEN/MWCNTs hydrogels, batch sorption experiments were conducted using solutions of the azoic dye acid blue 113 (AB113, \(C_{32}H_{21}N_{5}Na_{2}O_{6}S_{2}\), \(M_w\)= 681.65 g/mol, 50% purity, from Sigma-Aldrich) as a model of pollutant anionic dyes present in textile effluents. Thus, samples of 1 cm\(^2\) of the distinct hydrogel films were soaked, separately, in 2.6 mL of an AB113 dye solution with an initial concentration of 6.4 mg/L (\(\approx \)10\(^{-5}\)M) and pH= 4.7±0.2 for 24 hours. Afterward, the hydrogel samples were removed from the dye solutions and were then dried on a flat surface at room temperature to prevent the formation of wrinkles or wavy patterns on their surface. Subsequently, the hydrogel samples impregnated with AB113 were placed in a vacuum oven at 60 \(^\circ \)C and \(-\)17 inHg for 24 hours to extract the residual water inside the films. Based on previous swelling experiments and hydrogel dehydration assays, these vacuum-drying conditions enable the complete removal of bound and interstitial water trapped in the samples’ polymeric network. It is worth noting that even minor amounts of water in the samples could potentially modify their THz spectra [22, 23]. Lastly, the samples were stored in a desiccator and subsequently analyzed with THz-TDS and FT-IR spectroscopic techniques.

Furthermore, to determine the hydrogels’ dye sorption capacity (\(q_t\)), aliquots of the dye solutions were collected before and 24 hours after hydrogel immersion, and their absorbance spectra were measured using a Shimadzu 1800 UV–Vis spectrophotometer. The correlation between absorbance and AB113 dye concentration was established at the dye’s characteristic maximum absorbance wavelength (i.e., \(\lambda \)= 566 nm). A reference dye solution of AB113, with the same initial concentration and pH, was used as a control. The \(q_t\) of the hydrogels were calculated using Eq. 1 [13].

$$\begin{aligned} q_t (mg/g)= \frac{(C_0-C_f)V}{m} \end{aligned}$$
(1)

Here, \(C_0\) and \(C_f\) (mg/L) represent, respectively, the initial and final (i.e., after 24 hours) concentration of AB113 dye in the solution; V (L) depicts the volume of the dye solution, and m (g) refers to the weight of the hydrogel samples.

2.3 Fourier Transform Infrared Spectroscopy (FT-IR)

To examine the captured AB113 dye and identify the interactions between the dye molecules and the functional groups in the hydrogel polymeric networks, FT-IR spectra of powdered AB113 dye and dry hydrogel samples (before and after the dye sorption assays) were then acquired. For this analysis, attenuated total reflectance (ATR) with a diamond crystal module was used (Smart-iTX ATR, Nicolet iS50, Thermo Scientific Inc.). The wave number ranged from 4000 to 500 cm\(^{-1}\) with a resolution of 0.5 cm\(^{-1}\), and 32 scans were performed per sample. The baseline of the measured FT-IR spectra was corrected, and then major vibrational bands were identified and associated with the main chemical groups present in CS, PVA, GEN, and AB113 dye.

2.4 Terahertz Spectroscopy

In this study, terahertz time-domain spectroscopy (THz-TDS) in transmission mode was used to detect the presence of AB113 dye within the different polymeric networks, through possible changes in their THz transmittance spectra and complex refractive index.

2.4.1 Terahertz Time-Domain Spectroscopy System

The THz-TDS was performed in a transmission mode configuration using a custom-designed free-space THz spectrometer (Fig. 1) [24]. This spectrometer is equipped with a femtosecond frequency-doubled Erbium fiber laser (ELMO 780, Menlo Systems) whose emission is directed by dielectric mirrors (DM1-6) and divided into two beams through a beam splitter (BS). One of the beams (pump beam) is used to excite the emission antenna and generate THz radiation, while the other (probe beam) is used as a probe to detect the THz signal at the receptor antenna. The THz emitter and receptor are integrated by two LT-GaAs photoconductive dipole antennas (TERA8-1, Menlo Systems). Two convex lenses (FL) are used to focus the laser within the gap of these antennas, and two other hyper-hemispherical lenses (HL) are used to collimate and collect the THz radiation in free space. This lens arrangement yielded a THz beam focal spot size of \(\approx \)2 mm in diameter. A motorized delay stage is used to generate a temporal delay between the THz and the probe beam. Thus, the spectroscopic system can register the electric field of the THz pulse as a function of time by measuring the signal detected synchronously with a lock-in amplifier.

Fig. 1
figure 1

Schematic representation of the custom-designed free space THz time-domain spectrometer in transmission configuration

Full size image

2.4.2 THz-TDS Experiments

For the THz measurements on hydrogels, we utilized the vacuum-dried film samples of CS, PVA, CS/PVA/GEN, and CS/PVA/GEN/MWCNTs, both before and after sorption of AB113 dye. To obtain the THz transmission spectra of the pure azo dye, we used pressed pellet samples. Uniform discs of AB113 dye, with a diameter of 13 mm and 1.39 mm of thickness, were formed by compressing 158 mg of AB113 dye powder using a manually operated hydraulic pellet press with a pressure force of 40 kN. To prevent nonuniform spatial distribution, density fluctuations of the dye within the pellets, or any unwanted contribution, the AB113 dye powder was not combined with a solid matrix (i.e., 100% w/w of AB113).

THz-TDS experiments were conducted under dry conditions by purging the spectrometer’s room with an environmental dehumidifier. The thickness of the hydrogel samples and the dye pellets was measured with a digital micrometer. For the THz scans, all the samples were carefully placed inside poly(lactic acid)—PLA—sample holders. Data were then acquired in the time domain, spanning a total time interval of 136.5 ps, with a time resolution of 33.33 fs corresponding to a frequency-domain resolution of 7.33 GHz. The usable bandwidth of the system allowed measurements from 0.3 to 1.5 THz. In order to minimize scan noise, we conducted multiple measurements (m\(\geqslant \)4); the presented values are the mean and standard deviation calculated from all iterations. Moreover, the samples were rotated between measurements to minimize the influence of possible bulk inhomogeneities [25]. A reference signal from an empty PLA sample holder was acquired before each scan. The measured time-domain pulse data were converted to a complex frequency-domain spectrum using a standard fast Fourier transform (FFT) algorithm [26]. The transmittance spectra \(T(\nu )\) of the hydrogels, with and without AB113, as well as that of the pristine AB113 azo dye, were calculated as a function of THz frequency (\(\nu \)) from [27, 28]:

$$\begin{aligned} T(\nu )=\frac{\left| E_{sam}(\nu ) \right| }{\left| E_{ref}(\nu ) \right| } \end{aligned}$$
(2)

where \(\left| E_{sam}(\nu ) \right| \) and \(\left| E_{ref}(\nu ) \right| \) correspond to the amplitude of the FFT spectrum of the recorded sample and reference time-based signals, respectively. Likewise, the real part of the refractive index (\(n(\nu )\)) of the samples was estimated using Eq. 3 [28,29,30].

$$\begin{aligned} n(\nu )= \Delta \phi (\nu ) \cdot \frac{c}{2\pi \nu d}+1 \end{aligned}$$
(3)

Here, \(\Delta \phi (\nu )\) is the phase difference between the reference and sample signals (i.e., \(\Delta \phi = \phi _{ref} - \phi _{sample}\)), c is the speed of light in vacuum, \(\nu \) is the THz frequency, and d represents the thickness of the measured sample. Phase retrieval was performed using the phase unwrapping method described by Jepsen [28]. From \(T(\nu )\) and \(n(\nu )\), the imaginary part of the refractive index as a function of frequency (\(\kappa (\nu )\)), also known as the extinction coefficient, can be calculated as follows [28, 29]:

$$\begin{aligned} \kappa (\nu )= \frac{c}{2\pi \nu d} \cdot \ln \left( \frac{4n(\nu )}{T(\nu ){(n(\nu )+1)}^2}\right) \end{aligned}$$
(4)

It is worth mentioning that conversion between the extinction and absorption coefficient, \(\alpha (\nu )\), implies \(\alpha (\nu )= \frac{\kappa (\nu ) 4\pi \nu }{c} = \frac{2}{d} \cdot \ln \left( \frac{4n(\nu )}{T(\nu ){(n(\nu )+1)}^2}\right) \).

3 Results and Discussion

3.1 Dye Sorption Capacity

The FT-IR spectrum and the structural formula of the anionic dye AB113 are shown in Fig. 2. Here, the transmittance bands between 3500–3200 cm\(^{-1}\) and 3000–2800 cm\(^{-1}\) correspond to the stretching vibration of N–H bonds and the C–H bonds in the aromatic compounds, respectively. The bands around 1600–1200 cm\(^{-1}\) correspond to the benzene and naphthalene rings; the band at 1453 cm\(^{-1}\) can be attributed to the vibration of the azo bonds (–N=N–), and that at 1170 cm\(^{-1}\) depicts the stretching vibration of C–N bonds. Additionally, the strong bands in the range of 1100–950 cm\(^{-1}\) are associated with the vibrations of the sodium sulfonate groups (–S(=O)\(_{2}\)–ONa) [31].

Fig. 2
figure 2

FT-IR spectrum of the anionic dye AB113. The inset in the figure corresponds to the chemical structure of the AB113 dye molecule

Full size image

Figure 3a shows the UV–Vis absorption spectrum of the AB113 dye solution (\(10^{-5}\) M, blue line). The peak at \(\lambda \)= 566 nm corresponds to the maximum absorbance, resulting from the conjugated system formed by the azo groups and the \(\pi \)-electrons in the naphthalene rings. Additionally, we show here the UV–Vis absorption spectra of the AB113 dye solutions after 24 hours in contact with the CS, PVA, CS/PVA/GEN, and CS/PVA/GEN/MWCNTs hydrogel membranes. As can be observed, the main absorbance signal of the dye solutions diminishes at different rates depending on the polymeric material with which they were in contact, indicating that all the studied hydrogels are able to capture AB113 dye to some extent. From Fig. 3b, it is evident that CS-based polymeric materials (i.e., CS, CS/PVA/GEN, and CS/PVA/GEN/MWCNTs membranes) exhibit a higher sorption capacity per mass of material (\(q_t\)) than the pristine PVA hydrogel, highlighting the strong affinity between AB113 dye molecules and the amino groups present in the CS backbone [13, 32]. It is worth noting that pristine CS membranes, despite exhibiting the highest \(q_t\) value (1.14 mg/g), have a propensity to dissolve when exposed to aqueous media for extended periods. This behavior can be attributed to the structural instability of their network due to the absence of chemical cross-linking points among the constituent polymeric chains. The incorporation of genipin into the CS/PVA blend results in more stable sorbent materials with a \(q_t\) of 0.92 mg/g for CS/PVA/GEN and 0.79 mg/g for CS/PVA/GEN/MWCNTs nanocomposite cross-linked hydrogels.

Fig. 3
figure 3

Comparison of the AB113 dye removal efficiency of the hydrogel membranes. a UV–Vis absorbance spectra of AB113 dye solution before (blue line) and after 24 hours in contact with the studied hydrogels (color lines). b Sorption capacity (\(q_t\)) of the CS, PVA, CS/PVA/GEN, and CS/PVA/GEN/MWCNTs hydrogels after 24 hours

Full size image
Fig. 4
figure 4

FT-IR spectra of the hydrogel membranes before and after AB113 dye sorption (solid and dashed lines, respectively). a FT-IR spectra of the CS natural hydrogel membranes, b FT-IR spectra of the PVA synthetic hydrogel membranes, c FT-IR spectra of the CS/PVA/GEN cross-linked hybrid hydrogels, and d FT-IR spectra of the CS/PVA/GEN/MWCNTs nanocomposite cross-linked hybrid hydrogels. The insets in the graphs show a magnified view of the region 1700–500 cm\(^{-1}\). The blue rectangle in the insets indicates the weaker transmittance bands associated with the AB113 dye particles on the membranes

Full size image

3.2 FT-IR Spectroscopy

To detect the AB113 dye attached to the surface of the synthesized hydrogels and the changes induced by its adsorption, we recorded the FT-IR spectra of the dry membrane samples. Figure 4 presents the FT-IR spectra of each hydrogel membrane before (controls, solid lines) and after AB113 dye capture (dashed lines).

The FT-IR spectrum of the CS hydrogel membrane, before dye adsorption, is shown in Fig. 4a. Here, the characteristic transmittance bands of CS are clearly visible. The strong band in the region 3600–3000 cm\(^{-1}\) corresponds to N–H and O–H stretching vibrations, and bands at around 2920 and 2880 cm\(^{-1}\) describe the C–H symmetric and asymmetric stretching, while those around 1645 and 1316 cm\(^{-1}\) correspond to the stretching vibration mode of C=O and C–N bonds present in amides I and III, respectively. Bands at 1600–1552 cm\(^{-1}\) were associated with N–H bond bending and C–N stretching vibrations in amides II, whereas band at 1090–1030 cm\(^{-1}\) describes the C–O stretching mode. Finally, bands at 1152 and 899 cm\(^{-1}\) correspond to the saccharide rings; N–H out-of-plane vibrations were identified at 640 cm\(^{-1}\). The FT-IR spectrum of pure PVA hydrogel (Fig. 4b, solid line) also showed the characteristic bands of this polymer. Thus, transmittance bands at 3400–3000, 2940–2906, 1420, and 1325 cm\(^{-1}\) were attributed to O–H stretching vibration of hydroxyl groups, stretching C–H from alkyl groups, CH\({_2}\)–CH bending and CH–OH stretching vibration, respectively. Bands corresponding to C–O–C stretching vibrations and CH out-of-plane were located at around 1085 and 835 cm\(^{-1}\), respectively (see inset in Fig. 4b).

Figure 4c, solid line, shows the FT-IR spectrum of the CS/PVA/GEN cross-linked hybrid hydrogel before dye adsorption. As can be seen, its FT-IR spectrum shows the main transmittance bands of both blend components, CS and PVA. The major changes observed in comparison to the raw polymeric components were observed at 1377 cm\(^{-1}\) and 1090–1030 cm\(^{-1}\), which correspond, respectively, to the formation of heterocyclic amines and C–N bonds because of the cross-linking reaction between CS and GEN [20, 33]. The FT-IR spectrum of the nanocomposite cross-linked blend, CS/PVA/GEN/MWCNTs, is shown in Fig. 4d (solid line). This spectrum does not exhibit new bands or significant changes compared to the FT-IR spectrum of CS/PVA/GEN hydrogel. This result was expected because of the low concentration of MWCNTs in the polymer matrix (0.01% w/v) and their incapability to form covalent bonds with the CS or PVA polymeric chains since the carbon nanotubes were not functionalized [21].

The FT-IR spectra of the investigated hydrogel membranes did not undergo significant changes upon AB113 dye adsorption, as shown by the spectra in dashed lines in Fig. 4. For CS+AB113 dye membranes, the intensity of the transmittance bands remained virtually unchanged, showing only a slight decrease in the region corresponding to the amide bands (1650–1300 cm\(^{-1}\)). In contrast, the FT-IR spectrum of the PVA+AB113 dye hydrogels (Fig. 4b) exhibited a pronounced decrease in transmittance. The bands associated with the hydroxyl groups also exhibited a slight redshift compared to the PVA spectrum. These differences can be attributed to the dye sorption mechanism in each of these polymeric networks. In the case of CS hydrogels, the capture of AB113 dye mainly relies on electrostatic attractions between the anionic sulfonate groups (–SO\(_3^-\)) of the AB113 molecules and the cationic protonated amine species (–NH\(_3^+\)) of chitosan [13, 34]. However, it is important to note that this type of intermolecular interaction cannot be detected using conventional IR spectroscopy (i.e., 4000–400 cm\(^{-1}\) wavenumber ranges), as observed in our study. On the other hand, in PVA hydrogels, the main adsorption mechanism is based on hydrogen bonding, Yoshida H-bonding, and/or n–\(\pi \) stacking interactions between the available hydroxyl groups (–OH) of PVA and the oxygen (O) or nitrogen (N) atoms in the dye molecules, as well as with their aromatic rings [13, 35]. Unlike electrostatic attractions, intermolecular hydrogen bonding interactions can be indirectly detected by IR spectroscopy through changes in transmittance and/or shifts in middle-infrared frequencies due to the stretching of the involved covalent bonds [36].

For CS/PVA/GEN and CS/PVA/GEN/MWCNTs hydrogel membranes, both the CS and PVA adsorption mechanisms can occur. The FT-IR spectra of these materials after dye adsorption (shown as dashed lines in Figs. 4 c and d, respectively) exhibited a slight decrease in transmittance at the bands associated with the hydroxyl groups (i.e., 3500–3000, 1320, 1090, and 1030 cm\(^{-1}\)). Additionally, after further analysis of the region between 1700–500 cm\(^{-1}\), we observed weak bands at 1650–1425 cm\(^{-1}\) in all CS-based hydrogels (insets in Figs. 4 a, c, and d; blue region). These bands can be attributed to the benzene and naphthalene rings of the AB113 azo dye, indicating the presence of dye particles on the surface of these polymers after the sorption process.

These results highlight the challenges associated with the detection and identification of pollutant azo dyes within polymeric sorbent materials through conventional IR spectroscopic techniques. These challenges can be attributed to two main reasons:

  1. i.

    The azo group (–N=N–) in dye molecules leads to absorption in the same spectral range as many functional groups commonly found in hydrogel networks, making it difficult to distinguish the specific transmittance or absorbance bands of dye molecules from other functional groups present in the polymeric networks.

  2. ii.

    Molecular vibrations corresponding to weaker intermolecular interactions (e.g., hydrogen bonds, van der Waals interactions, or electrostatic attractions) associated with the hydrogels’ dye adsorption mechanisms can be observed only at lower wavenumber regimes (\(\le \)100 cm\(^{-1}\) \(\equiv \) \(\le \)3 THz) [36].

To overcome these adversities, in this work, we explore the use of terahertz time-domain spectroscopy, THz-TDS.

3.3 Detection of AB113 Azo Dye at THz Frecuencies

3.3.1 Hydrogels’ Transmittance Spectra

THz frequencies, ranging from 0.3 to 10 THz, are useful for detecting intermolecular interactions among large molecules with enhanced resolution and sensitivity. Figures 5 a–d show the transmittance spectra of the hydrogel membranes, before and after the sorption of AB113 dye, and the transmittance spectrum of the AB113 dye pressed pellet (represented by solid color circles, unfilled gray circles, and unfilled blue diamonds, respectively). These spectra cover the frequency range of 0.3–1.5 THz, wherein THz vibrational modes will mainly arise from the motion of the hydrogen bonds between the AB113 dye molecules and the polymeric chains and/or among macromolecules themselves, as well as the lattice modes originating from the crystalline domains. As can be observed, the transmittance spectrum of the AB113 dye exhibits a decreasing trend as the frequency increases, showing the lowest transmittance values (related to strong absorption) at frequencies above 0.8 THz. To precisely determine the central frequencies of the AB113 dye transmittance bands, the second derivative of its spectrum was calculated. Thus, within the studied frequency range, six distinct bands were detected at 0.36, 0.58, 0.75, 0.91, 1.28, and 1.48 THz central frequencies. It is worth mentioning that similar transmittance bands at 0.75, 1.28, and 1.48 THz have been previously observed for other anionic azo dyes [37,38,39]. While these common bands could be characteristics of this type of dyes, the remaining bands detected at 0.36, 0.58, and 0.91 THz could be attributed to specific spectral features of AB113 dye. However, it is important to note that relying solely on THz transmission spectra may not be sufficient for the accurate and confident identification of the THz spectral fingerprint of pure AB113 dye. Therefore, additional characterizations using quantum mechanical simulations are necessary [25, 40].

Fig. 5
figure 5

THz transmittance spectra of the AB113 dye pellet (unfilled blue diamonds) and the dry hydrogel membranes, both with and without AB113 dye (unfilled gray circles and solid color circles, respectively). a CS hydrogel membrane, b PVA hydrogel membrane, c CS/PVA/GEN cross-linked hybrid hydrogel, d CS/PVA/GEN/MWCNTs nanocomposite cross-linked hybrid hydrogel, and e percentage difference between the transmittance spectra of the membranes with and without AB113 dye. The arrows in graphs a–d indicate the central frequencies of the probable characteristic transmittance bands of the AB113 azo dye (i.e., 0.36, 0.58, 0.75, 0.91, 1.28, and 1.48 THz). The blue bands highlight the similar transmittance bands observed between the AB113 dye and the Hydrogel+AB113 samples. The shaded areas represent the transmittance standard deviation for each THz spectrum

Full size image

Compared to the highly absorbent behavior of the pristine AB113 dye, the CS, PVA, CS/PVA/GEN, and CS/PVA/GEN/MWCNTs hydrogel membranes (solid color circles in Figs. 5 a–d, respectively) exhibit moderate transparency at terahertz frequencies. Their transmittance curves show a slight decrease tendency with frequency, presenting transmittance values between 0.93 and 0.8, approximately, across the frequency range of 0.3 to 1.5 THz. Notice that there is a correlation between the transmittance spectra and the molecular structure and chemical composition of these polymeric networks. For the case of the natural polymer (i.e., CS hydrogel, pink solid circles in Fig. 5a), the presence of numerous hydrogen bonds in polysaccharides and the predominant amorphous domains make it difficult to clearly identify resonance bands [41, 42]. In contrast, the synthetic PVA hydrogel (yellow solid circles in Fig. 5b) exhibits more pronounced and sharp transmittance bands, which are possibly associated with the lattice modes of its crystalline structure and intermolecular hydrogen bonds. On the other hand, both chemically cross-linked membranes, CS/PVA/GEN and CS/PVA/GEN/MWCNTs (blue and green solid circles in Figs. 5 c and d, respectively), present similar transmittance spectra at THz frequencies because of their comparable molecular structure. Notwithstanding, a slightly lower transmittance intensity (\(\sim \)3% less) was observed in CS/PVA/GEN/MWCNTs, which can be attributed to the broadband absorbance nature of the carbon nanotubes embedded in its polymeric matrix [43].

The center frequencies of the transmittance bands for each of these hydrogel membranes were determined by calculating the second derivative of their respective transmittance spectrum. The obtained frequencies are summarized in Table 1. It is important to mention that the molecular motions responsible for these bands are different for each hydrogel. Accurately assigning which specific molecular motion generates each transmittance band would require more advanced methods, such as molecular dynamics simulations, which are beyond the scope of this paper. However, according to the literature, the bands at around 0.37 and 0.54–0.56 THz in CS-based hydrogels can be attributed to the motion of the chitosan chains along their backbone, while that at 1.09–1.12 THz in hydrogels with PVA corresponds to the longitudinal vibration mode of the straight segments of such polymeric chains [42]. Additionally, bands at around 1.44–1.50 THz are associated with the bending vibrations of the OH\(\cdots \)O intermolecular hydrogen bonds [44, 45].

Table 1 Frequency position of the transmittance bands in the THz range for the hydrogel membranes, both with and without AB113 dye
Full size table

The THz transmittance spectra of the hydrogel membranes impregnated with AB113 dye are presented as unfilled gray circles in Figs. 5 a–d, while the center frequencies of their main transmittance bands are compiled in Table 1. It is evident that the presence of AB113 dye molecules in the polymer matrices significantly modifies the features of their transmittance spectra, especially in the non-cross-linked networks. Thus, the CS+AB113 and PVA+AB113 hydrogels (Figs. 5 a and b, respectively) present weaker and broader bands compared to the membranes without AB113 dye. The attenuation and broadening of the bands could be attributed to the rearrangement of the polymer chains due to the sorption of dye molecules. This rearrangement may lead to disruptions in the crystalline domains, subsequently causing a decrease in the intensity of lattice modes [36]. Conversely, the transmittance spectra of the CS/PVA/GEN+AB113 and CS/PVA/GEN/MWCNTs+AB113 cross-linked hydrogels (Figs. 5 c and d, respectively) exhibit similar shapes to those of the hydrogels before dye adsorption, but with more pronounced transmittance bands. Additionally, a slight blueshift of the resonance bands at frequencies below 0.9 THz is observed, while those above 1.0 THz show a redshift. Since both hydrogels are chemically and mechanically stable due to the cross-linking of their polymeric networks, the observed blueshifts and increased intensity of the transmittance bands could be attributed both to the increment of intermolecular interactions and the overlap of some transmittance maxima of the AB113 dye molecules with those of the polymer matrices. On the other hand, the redshifts in the characteristic region of hydrogen bonds (1.0–1.5 THz) might be attributed to the weakening of the hydrogen bonds among the polymeric chains due to the expansion of the network to accommodate the absorbed AB113 dye molecules [13]. A similar behavior has been observed in natural and synthetic polymeric membranes filled with graphene oxide [42].

Furthermore, the most evident change observed in all membranes after dye sorption is a decrease in their transmittance values, particularly at high frequencies. To visualize this difference more clearly, the percentage difference between the transmittance spectra with and without AB113 dye, \(\Delta \)T, was calculated, and the results are shown in Fig. 5e. As it can be seen, most of the hydrogel membranes exhibit a significant transmittance drop (\(\ge \)10%) at frequencies above 1.0 THz, showing a decrease of 18–36% at 1.5 THz. This remarkable transmittance decrement might be primarily attributed to the strong absorbance of the AB113 dye molecules in this frequency range, as well as the intermolecular interactions between the polymeric chains and the AB113 molecules through hydrogen and/or ionic bonds. These results suggest that the AB113 dye removed by the hydrogel membranes from aqueous solutions could be effectively detected through the THz transmittance spectra of the dry membranes in the frequency range of 1.0 to 1.5 THz.

By analyzing in detail the central frequencies of the hydrogels with AB113 dye (Table 1), characteristic transmittance bands associated with azo dyes were observed in all the investigated samples at 0.75 and 1.48 THz. This finding suggests the presence of an azo dye in the matrices. However, it is important to note that solely relying on these transmittance bands may not be sufficient to distinguish AB113 dye from other azo dyes. Further analysis and characterization using a wider frequency bandwidth than that of our system are required to accurately identify specific dyes.

Fig. 6
figure 6

Refractive index, n, of the AB113 dye pressed pellet (unfilled blue diamonds) and the dry hydrogel membranes, both with and without AB113 dye (unfilled gray circles and solid color circles, respectively). a CS hydrogel membrane, b PVA hydrogel membrane, c CS/PVA/GEN cross-linked hybrid hydrogel, d CS/PVA/GEN/MWCNTs nanocomposite cross-linked hybrid hydrogel, e percentage difference between the refractive indices of the hydrogels with and without AB113 dye, and f percentage difference of the refractive index, \(\Delta \)n, as a function of the sorption capacity, \(q_t\), of the hydrogel membranes. The shaded areas represent the refractive index standard deviation for each sample

Full size image

3.3.2 Refractive Index

The plots in Figs. 6 a–d depict the real part of the refractive indices for the dry hydrogel samples before and after AB113 azo dye adsorption (solid color circles and unfilled gray circles, respectively). The refractive index of the AB113 azo dye pellet is also included as a reference (shown as unfilled blue diamonds in Figs. 6 a–d). It can be observed that its refractive index values, n, range from 1.67 to 1.74 within the frequency range of 0.3–1.5 THz. On the other hand, the refractive index of the hydrogel samples without AB113 dye exhibits a dispersive behavior, with values varying between 2.05 and 1.76 as the frequency increases. This dispersive behavior and the observed refractive index values are characteristic of polar polymers [46]. According to the literature, the reported refractive indices for CS and PVA, at 1.0 THz, are approximately 2.2 and 1.74–1.79, respectively [42, 47, 48]. At this frequency, our experimental results showed a refractive index of 1.82 for CS and 1.77 for PVA hydrogels. The difference between the CS refractive index value obtained in our samples and the value reported in the literature could be attributed to variations in the CS source, deacetylation degree, polymer concentration, or hydrogel processing conditions, which can influence the structure of the polymer network, subsequently affecting the rate of light transmission through the membranes. The CS/PVA/GEN and CS/PVA/GEN/MWCNTs cross-linked hydrogels exhibited refractive indices of 1.84 and 1.70, respectively, at 1.0 THz.

With exposure to the AB113 dye, the refractive index curve of the hydrogels, represented by unfilled gray circles in Figs. 6 a–d, exhibits a flatter shape and lower refractive index values across the studied spectral range, in contrast to the samples without AB113 dye. These changes can be attributed to both the sorption of dye molecules within the hydrogels and the resulting modifications in the polymer structure because of their presence. As mentioned above, the AB113 dye molecules bonded to the polymeric chains generate a larger distance between the macromolecules in the network [13]. This increase in intermolecular spacing yields in a higher specific volume of the sample, leading to a decrease in density which in turn might contribute to increasing the velocity of light through the effective medium [44, 46]. The difference in refractive index between the hydrogels with and without AB113 dye at 1.0 THz was 0.67, 0.25, 0.35, and 0.24 for CS, PVA, CS/PVA/GEN, and CS/PVA/GEN/MWCNTs, respectively, which corresponds to a refractive index percentage decrement of 36.8, 14.1, 19.0, and 14.1%.

In order to conduct a more comprehensive comparative analysis between the materials under investigation, the percentage reduction of their refractive index, \(\Delta \)n, as a function of the frequency was determined. The obtained \(\Delta \)n values are presented in Fig. 6e. It is noteworthy that these results exhibit a similar trend as the AB113 dye sorption capacity presented by the hydrogels (\(q_t\), Fig. 3b). For instance, the CS membranes, which showed the highest \(q_t\) value, also presented the highest percentage decrement in refractive index. Conversely, the hydrogels with the lowest \(q_t\) value (i.e., the PVA membranes) exhibited the lowest percentage reduction in refractive index across the evaluated frequency range.

To illustrate the correlation between the hydrogels’ AB113 dye sorption capacity and their percentage change in refractive index, Fig. 6f was constructed. This plot provides a clear visualization of the relationship between these two variables at frequencies of 0.5, 1.0, and 1.5 THz. As can be observed, the correlation between \(q_t\) and the absolute value of \(\Delta \)n at the selected frequencies can be accurately described by a non-linear fitting characterized by a rational function, as indicated by the mathematical models in Fig. 6f. These models present a feasible and promising approach to quantify the concentration of AB113 azo dye in the hydrogels, only knowing the membrane weight and the percentage change in their refractive index at a specific frequency. Considering this, the studied hydrogel membranes captured between 7.8 and 15.0 \(\mu \)g of the AB113 dye from the solutions.

Fig. 7
figure 7

Extinction coefficient, \(\kappa \), of the AB113 dye pellet (unfilled blue diamonds) and the dry hydrogel membranes, both with and without AB113 dye (unfilled gray circles and solid color circles, respectively). a CS hydrogel membrane, b PVA hydrogel membrane, c CS/PVA/GEN cross-linked hybrid hydrogel membrane, and d CS/PVA/GEN/MWCNTs nanocomposite cross-linked hybrid hydrogel. The shaded areas in the spectra represent the standard deviation of the average \(\kappa \) for each sample

Full size image

3.3.3 Extinction Coefficient

The extinction coefficient, \(\kappa \), for the AB113 dye pellet and the investigated hydrogels, before and after AB113 dye sorption, was estimated using Eq. 4. The resulting extinction curves for each material are presented in Fig. 7. The higher \(\kappa \) values (\(\times 10^{-2}\)) observed for both the AB113 azo dye and the hydrogel membranes are indicative of their polar nature [46]. Note that these materials can be easily distinguished based on their extinction coefficient spectra (solid color circles in Figs. 7 a–d).

The exposure of the hydrogel membranes to the AB113 dye solution significantly modifies the \(\kappa \) spectra of those non-cross-linked polymers (i.e., CS and PVA hydrogels). Conversely, no significant changes in the pattern of the CS/PVA/GEN and CS/PVA/GEN/MWCNTs cross-linked hybrid hydrogels are observed. Thereby, the CS+AB113 and PVA+AB113 dye hydrogel membranes (unfilled gray circles in Figs. 7 a and b, respectively) present broader and weaker extinction bands in comparison to their dye-free counterparts. These modifications could be attributed to disruptions in the crystalline domains and the rearrangement of the polymeric chains caused by the retention of AB113 dye molecules. Note that due to the random reconfiguration of polymeric chains, the formation of new crystalline regions might also occur, leading to more defined extinction bands. On the other hand, the extinction bands of the CS/PVA/GEN+AB113 and CS/PVA/GEN/MWCNTs+AB113 dye membranes (unfilled gray circles in Figs. 7 c and d, respectively) become more pronounced, indicating a probable increase in the strength and number of dipole moments as a result of intermolecular interactions between the dye molecules and the available functional groups in the polymeric networks [13]. This assumption is further supported by the observed slight shift toward higher frequencies (i.e., blueshift) of the extinction bands. All these findings are in good agreement with the previously discussed transmittance behavior at THz frequencies. Nevertheless, assigning these experimental spectral features to specific molecular vibrations requires the use of advanced computational models due to the long-range and collective nature of the molecular movements, especially after the sorption of AB113 dye. In future studies, the incorporation of solid-state density functional theory (ss-DFT) simulations could be valuable for modeling the molecular structures and vibrational movements that contribute to the THz spectra [25].

4 Conclusion

We have successfully demonstrated the potential of THz-TDS to detect water pollutant dyes, particularly the azo dye AB113, using hydrogel membranes based on chitosan and poly(vinyl alcohol). The presence of AB113 dye molecules induces significant changes in the spectral features of the polymer matrices due to the rearrangement of the polymeric network, and the formation of hydrogen bonds and/or electrostatic attractions between the dye molecules and the functional groups in the polymer chains. Compared to conventional spectroscopic techniques like FT-IR, our results highlight the superior sensitivity and effectiveness of THz-TDS in detecting these weaker intermolecular interactions. In this regard, the significant decrease in transmittance above 1.0 THz in the hydrogels’ spectra, along with changes in their extinction bands due to alterations in the atomic-level structure of the hydrogels, allows us to infer the presence of dye molecules. Additionally, the observed changes in the refractive index of the membranes further emphasize the sensitivity of the THz-TDS spectroscopy to detect the different sorption capacities of the studied hydrogels for the AB113 azo dye. The derived mathematical models allow us to estimate the presence of approximately 7.8–15.0 \(\mu \)g of AB113 dye within the polymeric networks, making it feasible to conduct both qualitative and quantitative analyses for the detection of this water pollutant.

These findings provide the research basis for the analysis of different concentrations of AB113 dye solutions, as well as other pollutant dyes or water contaminants, setting a precedent for utilizing THz-TDS technology in the future development of efficient environmental sensor devices.