In our previous work, commercial Pd-ink-modified electrodes with a diameter of functional electroactive particles of 600–800 nm were used to screen glycerol in a cultivation medium of bacterial cells (Escherichia coli) [6]. However, the electroanalytical performance of Pd-ink-based electrodes was reduced due to a non-electroactive polymer layer utilized in their design. The use of a polymer as a binding agent was justified by the applied drop coating methodology to form functional sensing layers.

Here, as a possible alternative to commercial Pd-ink-modified electrodes, we used electroplating of Pd particles [27] from PdCl2-contaning electrolyte allowing to obtain Pd deposits in their refined form, e.g., without polymers or other non-electroactive substances (i.e., phosphates, ammonium complexes, etc.), ESI, Fig. S1 (shown for Sensor 3 as a case study).

Characterization of electroplated Pd particle–modified sensors

First, the impact of synthesis parameters, e.g., the applied cathodic current and deposition time on the architecture of electroplated Pd-functional layers of sensors, was investigated. SEM studies highlight the formation of Pd deposits with a size between 100–200 nm (Sensor 1), 600–800 nm (Sensor 2), and 1.0–1.5 µm (Sensor 3) (Fig. 1).

Fig. 1

SEM images of sensors modified by electroplated Pd deposits synthesized from the acidic PdCl2-based electrolyte at various experimental conditions: A at − 2.5 mA for 30 s; B at − 6 mA for 120 s; C at − 6 mA for 240 s

The morphology of electroplated Pd particles indicates that larger globules were formed by aggregation and overgrowth of the individual smaller structures with increase of a current and deposition time. It means that the synthesis route of Pd particles can readily be instrumentally controlled. This is especially important to support a reproducible architecture of functional sensing layers. Only a reproducible design of sensing layers can guarantee reproducible and reliable electroanalytical performance of amperometric sensors during subsequent electrochemical analysis of a target analyte. Notably, relative standard deviation (RSD) for all electrodes with electrodeposited Pd particles did not exceed 10% (Table 1). In contrast, the reproducibility of the baseline of Pd-ink-modified electrode reached the level of 31% that is most likely explained by the used preparation approach of functional layers (drop coating methodology).

Table 1 Electrochemical characterization and reproducibility of the baseline (row data were extracted from the second scans of three batches of each sensor) of sensors modified by Pd particles (polarized from − 0.4 to 0.8 V at 20 mV/s in buffer, pH 12)

The obtained SEM data were in line with the calculated mass of electroplated Pd deposits (Table 1). As expected, the increase of the deposition current and time leads to the increase of deposited palladium weight from 25 ± 3 µg (Sensor 1) to 115 ± 3 µg (Sensor 3). At the same time, ECSA decreased in a row: from 10.8 m2g−1 defined for Sensor 1 to 2.5 m2g−1 found for Sensor 3. Hence, it can be expected that the analytical response for Pd-modified electrodes with the highest ECSA should have an advanced performance. At the same time, glycerol oxidation process demonstrates a complicated mechanism with a participation of the adsorbed oxygenic species and intermediates [33, 34]. Therefore, the analytical merit of the small Pd particles is not so obvious and cannot be predicted [35].

Figures of merit of one-step electroplated Pd particle–modified sensors

To make the recorded trends more pronounced, the electroanalytical performance of the produced sensors modified by electroplated Pd particles was examined in model solutions at high concentrations of glycerol. The obtained analytical merit was compared with commercial Pd-ink-based electrodes.

The electrochemical behavior of Pd sensors differs depending on the used preparation mode (Fig. 2). Briefly, commercial Pd-ink-modified electrodes and electroplated Pd-based electrode (Sensor 1) demonstrated two anodic peaks, W1 and W2 (at approx. − 0.1 V and 0.6 V), corresponding to the glycerol oxidation process (i.e., these peaks are absent in background solution) on reduced (− 0.1 V [6]) and oxidized (0.6 V) Pd surfaces, while Sensors 2 and 3 with larger Pd particles showed only one anodic peak, W2, at 0.6 V. The cathodic peak that corresponded to the oxygen reduction on the Pd surface was visible for Pd-ink-modified electrode and Sensor 1 and absent for Sensor 2 and Sensor 3. In the case of Sensor 1, the cathodic process of reduction of the adsorbed oxygen interfered with the anodic process during backward scan (wave Wb) that leads to a compromise current in this potential range. The latter effect of anodic and cathodic current superposition is typical for alcohol oxidation processes on Pd [6].

Fig. 2

CV plots recorded from Pd-based sensors at 20 mV/s in 100 mM of model glycerol solution at pH 12: A commercial Pd-ink-modified electrode, B Sensor 1, C Sensor 2, D Sensor 3

In terms of the analytical merit, the onset potential for the second anodic wave is increased in the row: Sensor 3 < Sensor 2 < Sensor 1 < Pd-ink sensor, corresponding to 0.18 V, 0.22 V, 0.39 V, and 0.43 V, respectively (Fig. 2, see also ESI, Fig. S2). It means that the oxidation of glycerol species at 0.6 V is facile on the large Pd particles formed via electrodeposition synthesis approach. Thus, as it was supposed above, the complicated mechanism of glycerol oxidation leads to leveling the role of the specific ECSA of sensing Pd layers (Table 1).

Moreover, as it is typical for the adsorption processes, the electrochemical behavior of sensors with small and large Pd particles depends on glycerol concentration. Thus, with the increase of glycerol concentration from 1 to 100 mM, the anodic peak W2 potential of Sensor 1 shifted to higher values and the anodic peak currents at − 0.1 V (W1) and 0.6 V (W2) are growing while the current of the cathodic peak, C1, decreased (ESI, Fig. S3). It means that the overpotential of W2 for glycerol oxidation increases for Sensor 1 from 0.50 to 0.79 V with the increase of glycerol concentration from 1 to 100 mM, respectively. These data were in accordance with a mechanism of alcohol oxidation on palladium in alkaline media, viz. participation of oxygenic species (OH−, Oads, OHads, etc.) [34]. When the concentration of OH-ions in solution is comparable or lower than the amount of glycerol, the deficit of oxygenic reaction species leads to an overpotential of the reaction.

In contrast, both Sensor 2 and Sensor 3 with the large Pd particles demonstrate unusual properties. For Sensor 2, the first anodic peak W1 at low potential and cathodic peak C1 disappeared with the increase of glycerol concentration; at the same time, the anodic wave W2 at 0.6 V practically did not depend on glycerol concentration (ESI, Fig. S3). For Sensor 3, the first anodic W1 and cathodic peaks C1 are absent while the second anodic peak W2 potential increased from 0.3 to 0.6 V with the increase of glycerol concentration from 1 to 100 mM (Fig. 3).

Fig. 3

CV plots recorded at 20 mV/s in model glycerol solution at pH 12 from Sensor 3: 1, 2, 3, 4, 5, 6, 7 — 1 mM, 5 mM, 10 mM, 15 mM, 20 mM, 50 mM, 100 mM, respectively

The obtained results confirmed that the electrooxidation of glycerol is induced by oxygenic species (Oads, OHads, etc.) on the surface of the functional electroactive layer [34]. Apparently, with the increase of glycerol concentration, there are not enough oxygen groups on the surface of the electrocatalyst. However, the adsorption energy of oxygen and glycerol species can be different on Pd particles of various sizes. It means that the exact mechanism of glycerol electrooxidation on Pd particles is still under discussion and needs to be clarified.

Since the second anodic peak W2 was present in CV regardless of the applied glycerol concentration, this peak could be used as an analytical read-out range for the comparison of the analytical merit of tested sensors (see next section).

Calibration and validation in model aqueous solutions

The read-out of the signal from Pd-ink-modified electrode, Sensor 2, and Sensor 3 in model glycerol solutions at pH 12 was conducted at 0.55 V in the whole concentration range of glycerol (Table 2). In contrast, glycerol quantification by Sensor 1 with the appropriate regression coefficient (≥ 0.98) could be conducted at 0.55 V only in the concentration range between 20 and 100 mM. To support glycerol quantification by Sensor 1 below 20 mM, the signal read-out had to be carried out at 0.3 V.

Table 2 Electroanalytical performance of Pd-modified sensors in model glycerol solutions at pH 12

The significant increase of a slope in the calibration formula of electroplated Pd particles versus Pd-ink (Table 2) can readily be explained by the absence of non-electroactive compounds such as polymer binding agents in the design of sensing layers (Fig. 4). Therefore, despite having almost the same dimensional factor (e.g., 500–800 nm), the electroplated Pd particles appear to be more efficient versus Pd-ink. It also explains the enhanced LDRs and sensitivity values obtained for the sensors modified by electroplated Pd particles (see Table 2).

Fig. 4

SEM images of Pd-ink functional layer (A) and electroplated Pd particles produced at − 6 mA for 120 s from the acidic electrolyte (B). Note: the organic polymer layer used as a binding agent is shown in (A) by arrows

Taken together, our data suggest that by electroplating and surface engineering, it is readily possible to achieve the controlled design of sensing layers with the advanced analytical merit, viz. LDRs and sensitivity. More significantly, the sensitivity of glycerol electrooxidation does not significantly depend on the design of electroplated Pd deposits and could be a function of glycerol adsorption features (see “Impact of the adsorption stage on the glycerol electrooxidation”).

Next, to demonstrate the reliability of the developed sensors with electroplated Pd particles for quantification of glycerol, the recovery was evaluated by standard addition approach in model solutions. For this goal, model phosphate-containing solutions with pH 12 were spiked with different glycerol concentrations (5.00, 10.00, and 100 mM). A droplet of 150 μL was then placed on the surface of sensors. After each measurement, the sensors were rinsed by DI water prior to the next run. The time taken to complete a single run was 3 min. The novel amperometric sensors with electroplated Pd deposits achieved satisfactory recovery yields ranging from 96 to 107% with an RSD of less than 5% (Table 3).

Table 3 Selected recovery data for glycerol determination in model solutionsImpact of the adsorption stage on the glycerol electrooxidation

As highlighted above, the sensitivity of glycerol electrooxidation did not depend on the design of Pd deposits (see Table 2). Therefore, it was hypothesized that the sensitivity of electroplated Pd-sensing layers can be connected with the adsorption stage of glycerol. The adsorption energy will also impact the binding efficiency and surface concentration of the electroactive form of the analyte.

Next, Gibbs adsorption energies for glycerol in comparison with EtOH were calculated at the Pd and PdO plane using DFT. As it is seen from Fig. 5, the Gibbs adsorption energy is negative for PdO plane and positive for Pd. It means that the presence of oxygenic species might influence glycerol adsorption.

Fig. 5

The Gibbs adsorption energies (Gads) and the most favorable location of glycerol (A, C) and EtOH (B, D) on Pd and PdO

It should also be mentioned that the received value of glycerol adsorption (Gads) on the PdO surface was almost 75 times more advantageous as compared to EtOH (Fig. 5). In addition, the adsorption of glycerol was accompanied by the formation of two hydrogen bonds. In contrast, the adsorption of EtOH proceeds with generation of a single hydrogen bond.

It means that for the efficient glycerol attachment (that is important for its subsequent electrooxidation), at least two active sites on the PdO surface are necessary. Hence, the absence of the oxygen reduction peak for Sensors 2 and 3 could be explained by the consumption of adsorbed oxygen in the glycerol electrooxidation process.

Electroanalytical performance of electroplated Pd sensors in real yeast fermentation mediumIs glycerol really present in yeast fermentation medium?

Depending on the medium type, adjusted biotechnological process, line of the used yeast cells, their cultivation times and cultivation conditions, optical density of cells, etc., the amount of glycerol in fermentation medium can be very different. Thus, glycerol can be formed in cultivation medium as a product or it can be consumed by the cells as a carbon source during their physiological growth. Therefore, to verify the presence of glycerol in fermentation samples used in our study, a GC–MS protocol including derivatization of samples with MSTFA was applied [6].

The received chromatogram (Fig. 6) and mass spectra of a peak recorded at retention time of 5.12 min (ESI, Fig. S4) clearly indicate the presence of glycerol (after derivatization with MSTFA visualized as glycerol-tri-TMS ether) in the tested biological samples at high amount (Fig. 6). The similarity degree of the defined compounds with a library ranged from 86 to 97%.

Fig. 6

GC–MS chromatogram obtained in the absence of an internal standard for the pristine derivatized supernatant collected after contact with cells for 24 h (OD = 6.1): 1 — 2,3-bis-TMS butane; 2 — TMS ester 2-TMS oxy-propanoic acid; 3 — TMS ester 2-TMS oxy-pentanoic acid; 4 — tris-o-TMS 1,2,3 butane triol; 5 — bis-TMS ether 1-o-heptadecylglycerol; 6 — glycerol-tri-TMS ether; 7 — methyl ester 3,4-bis TMS oxy-benzeneacetic acid; 8 — D-1,2,3,4-tetrakis-O-TMS-ribopyranose; 9 — 1,2,3,4,6-pentakis-O-TMS-alpha-d-glucopyranose. Note: EtOH and BuOH possibly present in fermentation samples cannot be visualized in the chromatogram at the used derivatization procedure

Impact of interfering species on response of electroplated Pd-modified sensors

Further, a selectivity test for the proposed sensors with electroplated Pd layers towards glycerol determination at the defined electrochemical conditions in AM mode was conducted. Notably, sensors with electroplated Pd layers showed the exclusive sensing properties towards glycerol, whereas no response was recorded to interfering species (e.g., EtOH, BuOH) possibly formed during yeast fermentation (Fig. 7). The obtained experimental results can be explained by the advanced adsorption of glycerol at Pd-sensing layers as compared to interfering species, viz. EtOH (see “Impact of the adsorption stage on the glycerol electrooxidation”).

Fig. 7

Selectivity test performed with interfering analytes in AM mode from Sensor 2 (as a case study) at the applied potential of 0.55 V vs. Ag/AgCl. Note: pH of samples was 12

Application to glycerol determination in real samples

Finally, the electroanalytical performance of novel sensors with electroplated Pd particles was quantitatively evaluated in a droplet of HC fermentation medium collected after cultivation of yeasts. For this goal, the standard addition approach of glycerol solutions was used. The most sufficient advantage to be mentioned for the used approach is that it does not need a blank matrix for quantification. In addition, this approach overcomes the matrix effects and recovery rates [36, 37].

Concisely, by skipping of fermentation medium taken after 24 h of yeast cultivation, it was revealed that the total glycerol content was in the range of 71–75 mM depending on functional Pd layer used (Table 4). Remarkably, the anodic potential of glycerol electrooxidation was almost identical (i.e., ~ 0.55 to 0.6 V) in a real fermentation sample for all tested sensors (ESI, Fig. S5). Therefore, to simplify data processing, the signal read-out was conducted for all sensors at 0.55 V.

Table 4 Electroanalytical results of glycerol quantification in fermentation samples at pH 12 on Pd-modified electrodes polarized from − 0.4 to 0.8 V*

Interestingly, the glycerol content in a target fermentation sample was quantified almost at the same level with the same sensitivity regardless of the design of functional Pd layers (Table 4). This observation confirmed the earlier made assumption on the impact of the adsorption stage of glycerol on its electrooxidation at Pd surfaces.

The recovery data obtained in HC medium after yeast cell cultivation for 24 h summarized in Table 5 highlight a strong potential of the proposed assay utilizing sensors with electroplated Pd deposits for glycerol determination even in the presence of interfering species. The concentration of glycerol found by the novel sensors with Pd particles was in line with data received by commercial Pd-ink-modified electrode.

Table 5 Selected recovery data for glycerol determination in HC yeast fermentation medium after contact with cells (OD = 5.8)

Selected quantification data of glycerol present in yeast fermentation samples are summarized in Table S1. Obviously, the glycerol can readily be quantified in yeast supernatants. Next, the optimized assay utilizing electroplated Pd deposits is planned to be used in a tandem work with biochemists and biologists to establish the existing correlations between glycerol content and cultivation conditions of yeast cells.