Electrochemical Biosensor for the Monitoring of Phages of Lactococcus lactis in Milk-Based Samples

— Lactococcus lactis (LL) bacteriophage infections in milk prevent proper lactic fermentation, leading to the production of unsaleable low-quality products and great economic losses in the dairy industry. In this work, we present an innovative biosensing approach for the cost-effective detection of LL phages (PGs) in milk-based samples through electrochemical impedance spectroscopy (EIS). The detection is based on the evident parametric shifts in the charge transfer resistance and the impedance phase at 100 Hz caused by different bacteria proliferation due to PG activity. The EIS results are compared with optical absorbance measurements at 600 nm, in order to validate the proposed method. Preliminary experimental tests with filtered milk-based samples confirm the sensor capability to detect PGs at different concentrations in milk-based solutions in less than 4 h. In order to reach a higher sensitivity, we propose a new milk pretreatment adding calcium chloride (CaCl 2 ) to the samples. The EIS results with CaCl 2 -treated milk evidence an enhanced PG activity, which leads to a much larger parametric shift. Lastly, the sensor is tested with CaCl 2 -treated milk-based samples at different PG concentrations, obtaining an enhanced performance reaching the limit of detection (LOD) of 10 3 PFU/mL.


I. INTRODUCTION
T HE sudden degradation of dairy fresh products along the production chain is one of the main causes of economic losses and food waste in the agrifood industry [1], [2].The Lactococcus lactis (LL) is the most common type of lactic acid bacteria (LAB) responsible for milk fermentation in dairy processes thanks to its metabolic activity [3].However, these bacteria are frequently affected by bacteriophage infection, which causes bacterial lysis.The LL death leads to the failure of the fermentation process and so to unsaleable low-quality dairy products and to significant lags in the production chain [4].Hence, the timely detection of the presence of bacteriophages in fermenting milk is essential to avoid heavy production losses in dairy industries.The traditional phage (PG) detection techniques usually rely on expensive laboratory molecular test, such as q-PCR [5], [6], on lengthy and outdated microbiological test, like the spot test [7], or on optical absorbance spectroscopy [8].While being quite sensitive, these methods require specialized personnel, specific equipment, and a response time over 6 h.Hence, in order to avoid production delays and analysis costs, industries often prefer to reduce the number of quality controls, resourcing to recurring sanitation protocols even when not strictly necessary.
Electrochemical biosensors are a promising alternative to these approaches, allowing a cheap and fast detection of the analytes.In particular, they offer the possibility of a quick detection directly on field performed by nontrained personnel, completely bypassing all the traditional methods drawbacks [9], [10].Numerous biosensors for dairy industry applications have been reported in [11] and [12], for example, evaluating the fermentation quality [13], or the antibiotic quantity in milk [14], [15], or to detect milk adulterations [16].Some of these works proposed biosensing methods for the detection of health-hazardous bacteria in milk samples [17], [18], also exploring the possibility of using bacteriophages as a recognition element by exploiting their lytic activity to detect the bacteria presence [19].However, only few biosensors have been developed for bacteriophage detection.A novel approach based on the impedance imaginary component shift at low frequency was proposed in [20] for the detection of the PG PhiX174 of Escherichia Coli WG5.Electrochemical biosensors were also reported for LL bacteriophage detection.In [21], interdigitated electrodes were used to detect the PG infection in standard phosphate saline buffer (PBS) solutions by evaluating double-layer capacitance and interdigitated capacitance variations.Our recent work [22] presented a proof of concept for an innovative LL PG biosensor based on charge transfer resistance and impedance phase difference shifts.The detection of 10 7 PFU/mL PGs was performed in less than 5 h and tested with laboratory-made milk-based solutions.
In this work, we propose an optimized detection method for LL bacteriophage in milk-based samples using electrochemical impedance spectroscopy (EIS).Our results provide the consolidation of a novel simple and cost-effective biosensor with improved sensing performance, better sensitivity, low limit of detection (LOD), and fast response time, taking a further step forward the final application directly on field.

A. Biological Elements and Chemicals
All the used chemicals were of analytical grade.The water was filtered at 0.22 µm and de-ionized to ultrapure Milli-Q water (MQ, with conductivity <2000 µS/cm).The calcium chloride (CaCl 2 ) was purchased from Sigma Aldrich (Germany) and diluted at 100 mM in MQ.The LL and its Bacteriophage P008 were provided by DSMZ (Germany).The M17 broth (HiMedia Laboratories) was selected as the culture medium for the bacteria growth.The PGs were stored in a saline PG buffer (pH 7.5) composed of 100-mM NaCl, 8-mM MgSO 4 , and 50-mM Tris-HCl.The milk was commercial offthe-shelf milk.

B. Test Conditions and Devices
The proposed biosensor was based on screen-printed electrodes (DRPC223AT, DRP) produced by Metrohm DropSens, Spain.The devices were produced with a conventional three-electrode configuration based on working (Au), counter (Au), and reference (Ag) electrodes on a ceramic substrate (34 × 10 × 0.5 mm), as shown in Fig. 1(a).The working electrode was a disk with 1.6 mm diameter.All the used devices were enclosed into a hermetic customized 3-D-printed PDMS cell [Fig.1(b)] to avoid undesired evaporation phenomena.The biosensor was electrically characterized by using the EIS using PalmSens EmStat PICO and CH-404A potentiostat.EIS technique was chosen for this application since it allows to analyze the electrochemical signal different contributes, i.e., faradic and nonfaradic components, determining quantitative parameters characteristic of electrochemical processes of interest at the electrode/solution interface.The EIS measurements were performed using the two-electrodes configuration in a frequency range between 1 Hz and 100 kHz, with the ac signal peak-to-peak amplitude of 10 mV and the dc bias V DC of 91 mV, corresponding to the equilibrium redox potential E 0 retrieved by CV measurements.All the experimental tests were validated by observing the bacterial growth through optical absorbance measurements at 600 nm carried out with the spectrophotometer ONDA (UV-30 SCAN).
After the measurements, the EIS responses were fit through the equivalent electric circuit [Fig.1(c), inset].The circuit modeled the solution/electrode resistance (R el + R s ) in series to the parallel of constant phase element (CPE1), which represents the nonfaradic contribute, and the charge transfer resistance (R ct ) and another constant phase element (CPE2), which represent the faradic contribute.As reported in Fig. 1(c), the experimental data obtained for 10-mM FeCN in milk-based solution are well fit by the equivalent circuit, allowing the extrapolation of all the circuit parameters.

A. Phage Detection at Different Concentrations in Filtered Milk-Based Samples
The biosensor detection capability is initially evaluated in filtered milk-based samples using the spill-out protocol previously defined in [22].Off-the-shelf raw milk (pH 6.5) is centrifuged twice at 10.000 RPM for 5 min to remove the fat component.A solution of HCl 37% is also added to lower the pH (∼4.6) and precipitate the casein.The extracted supernatant is called filtered milk.PG-contaminated solutions are prepared by adding 120 µL of PGs in PG buffer at an initial concentration of 10 9 PFU/mL to 120 µL of filtered milk.The solutions are then diluted with 500 µL of LL (initial OD 0.5), 1000 µL of 20 mM FeCN in M17, and 260 µL of M17, to obtain a 10-mM FeCN in M17 filtered milk-based solution with a final OD of 0.15 and a final PG concentration of 10 7 PFU/mL.The LL control solutions are obtained with 120 µL of PG buffer (without PGs), while a negative control solution is prepared with only filtered milk to obtain the absorbance baseline.After 300 min, it is still possible to distinguish the semicircle for the PG sensor at medium-high frequencies and calculate its R ct (12 k ).Meanwhile, the LL signal evidences a huge impedance increase over time, not allowing the extrapolation of the charge transfer resistance.Likewise, the phase Bode diagram [Fig.2(b)] shows a large phase shift between PG and LL sensors after 300 min, resulting in a much higher phase difference φ at 100 Hz for LL (43 • ).The φ was calculated at 100 Hz since the phage shift in milk-based samples is maximized around this frequency.These large parametric variations reflect the different bacterial growth in the solutions.In the LL solution, the bacterial population grows freely, determining an evident shift of the electrical parameters.Instead, the phages present in the PG solution limit the bacteria proliferation due to their lytic effect, causing a much more limited variation of the electrochemical measurements.
In order to assess the biosensor sensitivity, we evaluate the system response at different PG concentrations.The PG solutions are prepared at 10 7 PFU/mL, 10 6 PFU/mL, and 10 5 PFU/mL.The initial 10 7 PFU/mL concentration is a typical infection value in the dairy production chain, while 10 5 PFU/mL is considered the minimum PG concentration needed to induce the bacterial growth inhibition [23], [24].
The different bacterial growth in the solutions can be easily followed from the absorbance measurements [Fig.3(a)].The milk control sample (black curve) assumes values around 0 along all testing time, confirming the absence of external contaminations.The LL OD values (blue curve) grow over time reaching the saturation plateau around 240 min.The PG presence is clearly visible since the beginning for 10 7 PFU/mL concentration (green curve), and the lytic activity is well recognizable starting from 120 min.At the end of the observation time, the PG 10 7 PFU/mL OD value is around 0.05 and almost all the bacteria died.Meanwhile, the PG 10 6 PFU/mL curve (yellow curve) shows an initial increase due to the bacterial growth, but the OD starts to dramatically decrease at 210 min due to the PG activity, reaching a final value of 0.4.On the other hand, these trends cannot be seen for the 10 5 PFU/mL concentration (red curve).The PG 10 5 PFU/mL curve is similar to LL, showing only a slight OD decrease around 240 min.However, this shift is not sufficient to allow the detection of any PG activity before the end of observation time.It is worth noting that the experimental measurements are necessarily affected by the biological behavior of the living bacteria in solution.The beginning of bacterial proliferation is highly dependent by a series of external factors, e.g., the external temperature and mechanical stirring.Hence, the exponential bacterial growth, correspondent to the LL curve slope increase, may not arise at the same time in different bacteria samples.Therefore, the experimental measurements may be affected by a higher variability during this period then the remaining observation time.
These phenomena reflect in the electrochemical measurements.The electrical parameters extrapolated at 300 min for devices tested with NC, LL, PG 10 7 , and PG 10 5 are reported in T.1.
The charge transfer resistance shift is reported in Fig. 3(b) as R ct as a function of the time normalized by charge transfer resistance at time 0 (R ct0 ).The normalized R ct clearly shows the different behavior of the PG sensors.At 180 min, the 10 5 PFU/mL concentration follows the same trend of LL and  presents an explosion of the R ct , which increases more than one order of magnitude.Meanwhile, the PG 10 6 PFU/mL and 10 7 PFU/mL sensors show a much more limited increase of R ct .At 300 min, the final R ct /R ct0 is 3.6 for PG 10 6 PFU/mL and 2 for the 10 7 PFU/mL concentration, consistent with the reported absorbance measurements and the extremely limited growth of the LL due to the PG presence.As further confirmation of these observations, the impedance phase difference at 100 Hz is reported in Fig. 3(c) as function of time.Both the PG curves at 10 6 PFU/mL and 10 7 PFU/mL present very small phase shifts and are consistent with the previous findings.Interestingly, the PG 10 5 PFU/mL shows a remarkable φ decrease (55 • ) around 180 min, almost identical to the LL curve.This suggests that the PG activity at the 10 5 PFU/mL concentration is not intense enough to affect the bacterial growth and induce a parametrical shift detectable by the biosensing system.Hence, these tests evidence the biosensor LOD as 10 6 PFU/mL in filtered milk, and the minimum time required for sensing this concentration is less than 4 h.

B. Influence of CaCl 2 -Treated Milk-Based Samples on Bacterial Growth and Phage Detection
Although the previous results may be satisfying, the obtained LOD is one order of magnitude higher than the minimum PG concentration considered dangerous for bacterial growth.In particular, the PG activity may be not strong enough to allow the detection at medium-low PG concentrations.Moreover, the sample preparation requires two centrifuging steps, which are tedious and time-consuming.These constitute a huge limitation for the final application.In order to overcome the problem and reach a better sensing performance, we modified the milk sample pretreatment protocol by reducing the preparation steps and introducing a PG activity enhancer.The off-the-shelf milk is centrifuged at 10.000 RPM for 5 min only once to remove the fat component.No acidification is performed as the casein is not eliminated from the sample.Then, a solution of 100 mM of calcium chlorine (CaCl 2 ) is used as PG activity enhancer since numerous works in literature report the positive effect of calcium on favoring the PG activity [25], [26], [27], [28].Hence, the samples under test are prepared with a solution of 12 µL of centrifuged milk and 64 µL of CaCl 2 called CaCl 2 -treated milk.
First, we test the influence of centrifuged milk and CaCl 2 on the sensors' electrochemical response.As reported in Fig. 4, the EIS signals (top plot) show little variations during the observation time and the extrapolated R ct values (bottom plot) result reproducible and stable during the four sets measurements, meaning that the working electrode is resistant to the chlorine in solution.The sensors are overall unaffected by the new protocol, with high stability and reproducibility; thus, we carry out new tests in the presence of bacteria and PGs.These parametric variations are similar to the filtered milk results and confirm the effectiveness of the new protocol in detecting the PG presence.The CaCl 2 -treated milk does not affect the bacterial growth in LL solutions and allows the PG infection in PG solutions.

C. Phage Detection in CaCl 2 -Treated Milk-Based Samples at Different Concentrations
The sensitivity of the biosensing system is tested with the newly defined CaCl 2 -treated milk to assess its influence on the detection capability.We prepare four different PG concentrations starting from 10 8 PFU/mL and decreasing each one of one order of magnitude until 10 5 PFU/mL.The initial 10 8 PFU/mL concentration is representative for a higher level of infection in the dairy production chain.
The bacterial growth in the solutions during an observation time of 240 min can be evaluated by the absorbance measurements [Fig.6(a)].The absence of external contaminations is confirmed by the milk control sample (OD = 0).The LL OD continuous growth evidences the correct bacteria proliferation in the solution, which begins to reach the saturation around the end of the observation time (OD = 1).Starting after 120 min, the PGs at 10 8 PFU/mL begin the bacteria lysis, leading to the complete death of the bacterial colony in the next 30 min.A similar trend is followed by the PG 10 7 PFU/mL and 10 6 PFU/mL curves.Their OD values increase at the beginning and drastically drop to 0 at 180 min.The PG 10 5 PFU/mL curve shows an analogous behavior.The OD increases until 0.7, and then the PG activity becomes evident after 180 min and leads to the total disruption of the bacteria in the next 30 min.Interestingly, the PGs presence is clearly distinguishable for all the concentrations under test.
The same behavior can be observed from the electrochemical measurements.A similar trend for all the PG sensors is reported in the normalized R ct as a function of the time [Fig.6(b)].While the LL sensors present an explosion of the R ct after 210 min, the PG sensors only show a slight increase of normalized R ct , which reaches final values lower than 2. A similar behavior can be seen from the impedance phase difference at 100 Hz as a function of time [Fig.6(c)].All the PG curves present a slight deflection of few degrees (>10 • ).Meanwhile, the φ for the LL sensors evidently decreases around 210 min, reaching a value of 44 • .
These findings clearly highlight the system detection capability in less than 4 h and shows the positive influence of the new milk preparation protocol.The presence of CaCl 2 in the samples enhances the PG activity as reported in the literature, allowing to perform the PG detection even at the low concentration of 10 5 PFU/mL.

D. Limit of Detection Assessment
The previous findings clearly highlight the system capability to correctly detect the presence of PGs down to a concentration of 10 5 PFU/mL.In order to establish the LOD of the proposed biosensor, we evaluate the sensor response at lower PG concentrations.
PG-contaminated solutions are prepared by diluting a phage solution at initial concentration of 10 5 PFU/mL with Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.The minimum detection capability of the sensor can be evaluated by the normalized R ct and φ changes during an observation time of 330 min.Since we consider solutions at low PG contamination levels, a longer period is needed to eventually observe the PG activity.As reported in Fig. 7(a), the R ct explosion for LL sensors is evident after 240 min, while PG sensors at a concentration of 10 4 PFU/mL and 10 3 PFU/mL show only a slight increase.On the other hand, the R ct for a PG concentration of 10 2 PFU/mL after 240 min follows a more rapid ever-increasing trend, similar to the LL behavior.The normalized R ct final values are still obtainable around 4.1.However, the sensor response is instable, as evidenced by the high standard deviation.These considerations are even more evident by looking at the impedance phase difference [Fig.7(b)].The final φ for 10 4 PFU/mL and 10 3 PFU/mL is 47 • lower than LL sensors.Meanwhile φ for 10 2 PFU/mL shows a decrease similar to LL, indicating a LOD of 10 3 PFU/mL.The high variability obtained with 10 2 PFU/mL at 330 min is induced by the LOD of the sensor.
Hence, these findings suggest that the sensing system LOD is 10 3 PFU/mL in CaCl 2 -treated milk.It is worth noting that lowest PG contamination actively affecting dairy production is reportedly around 10 5 -10 4 PFU/mL [23], [24].Hence, the proposed sensor LOD is at least one order of magnitude lower than the final application minimum requirement.

IV. DISCUSSION
The new milk protocol reduces the number of preparation steps and introduces the presence of calcium chlorine in the final samples.The previously reported results evidenced the stability of the measurements and biosensor detection sensitivity using CaCl 2 -treated milk.To discuss the benefits of the proposed new protocol, we compared the filtered milk results with the CaCl 2 -treated milk responses in the presence of a PG contamination of 10 8 PFU/mL.
The great influence of CaCl 2 on phage activity can be assessed by observing the absorbance measurements in Fig. 8(a).In the presence of phages, CaCl 2 enhances their lytic action, and the bacterial growth is immediately slowed down, reaching a complete stop after 150 min.On the other hand, the PGs in filtered milk need more time to start their activity, thus allowing the proliferation of a higher number of bacteria.Similar conclusions can be retrieved from the normalized R ct in Fig. 8(b).Both the PG sensors curves present a linear trend, but the CaCl 2 -treated milk has a lower pendency due to the higher PG activity.Thanks to the enhanced PG lytic action the sensor LOD can be lowered of two orders of magnitude, reaching 10 3 PFU/mL.
The comparison between the concepts of filtered milk and CaCl 2 -treated milk is illustrated in Fig. 9.The milk sample is first centrifuged and filtered to eliminate the fat component (a).
An acidification step and a second centrifuge and filtering are required to prepare filtered milk (b), while the addition of CaCl 2 is the only step necessary for CaCl 2 -treated milk (c).LL and PG-contaminated samples are prepared from these solutions.In (d), a drop of each LL sample is placed on the electrodes.The hampering of the electric current flow is most likely due to the precipitation of living bacteria on the surface, inducing an increase in the charge transfer resistance.The milk preparation has no influence on bacterial growth, and the observed behavior is the same for both samples.In (e), the filtered milk-based sample is contaminated by PGs, leading to the bacteria death and to the synthesis of new PGs.A slightly higher current flow is allowed probably due to the reduction of bacterial proliferation on the electrode.On the other hand, in (f), the CaCl 2 -treated milk is contaminated by bacteriophages.Since CaCl 2 has enhancing proprieties on the PG activity, the PGs multiply and disrupt the bacterial cells faster.A much higher current is allowed to flow from the electrode, thus inducing a higher parametric shift than filtered milk between PG-contaminated and LL sensors.
A possible point-of-care biosensing system for the final application in dairy production plants can be based on a differential measurement by exploiting these findings.The system should be able to assess the electrochemical response difference between a sensor tested with the taken milk sample, after the appropriate pretreatment, and a control sensor tested with a supplied bacteria-only solution.In the case of PG contamination of the milk sample, the detection system should observe an evident shift of the electrical parameters with respect to the control solution.

V. SUMMARY AND CONCLUSION
In this work, we proposed an innovative electrochemical biosensor for the detection of LL bacteriophages in milk-based samples.The detection method was based on different LL growths in milk-based solutions due to the level of PG contamination.The PG presence on the electrode was confirmed by detecting a different bacterial growth in solution.The bacteria proliferation was quantified through the impedance phase Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.
shift and the charge transfer resistance variation.The method was first tested in filtered milk-based samples, obtaining an initial LOD of 10 6 PFU/mL.A new simplified milk sample preparation protocol was introduced to avoid the previous tedious pretreatments and reach a better sensing performance.The milk was diluted with a solution of CaCl 2 to stimulate the PG activity.The EIS measurements were repeated at different PG concentrations, obtaining a LOD of 10 3 PFU/mL.The good sensitivity of the system allows the PG detection even at the lowest concentration found in the dairy production chain in less than 4 h.
These results constitute a promising step to develop a pointof-care device, suitable for milk analysis directly on the field in dairy production plants.The point-of-care biosensing system should be based on differential measurements comparing the sample under test with a control bacteria-only device.A calibration curve should also be provided to assess the level of PG contamination in the solution under test.Future work will aim to test the sensor performance with real milk samples, considering the perturbative external factors.Furthermore, the possibility of PG detection in milk samples will be explored with different sensing devices such as ChemFET and ISFET.

Fig. 1 .
Fig. 1.(a) Picture of the electrodes Metrohm DropSens C223AT (DRP).(b) Picture of the electrochemical cell used in this work.(c) Equivalent circuit applied for fitting the Nyquist diagram of EIS responses.

Fig. 2 .
Fig. 2. Tests in filtered milk-based samples for sensors with LL, and with LL PG contamination: (a) Nyquist diagram of EIS responses of LL and PG sensors in filtered milk-based solutions and (b) phase of EIS responses as a function of the frequency of LL and PG sensors in filtered milk-based solutions.

Fig. 2
reports the Nyquist curves and phase of EIS responses of two different biosensors tested with filtered milkbased samples.The PG sensor (red line) was tested with the PG-contaminated solution, while the LL sensor (blue line) was tested with the LL control solution to assess the bacterial growth.The different behavior over time between the two sensors is evident from the Nyquist diagram [Fig.2(a)].At time 0 min, both the curves show a similar shape, and the R ct values can be retrieved from the equivalent circuit (∼4 k ).

Fig. 3 .
Fig. 3. Test in filtered milk-based samples for sensors in milk with LL, and with LL PG contamination at different concentrations: (a) optical absorbance as a function of the time of the only filtered milk-based solution (M), LL, and PG solutions at different concentrations, (b) Normalized charge transfer resistance R ct /R ct0 extrapolated by the EIS measurements, and (c) differential phase measured at 100 Hz extrapolated by the EIS measurements.

Fig. 4 .
Fig. 4. Electrodes stability in only centrifuged milk-based solution after adding CaCl 2 is evaluated by EIS.The bottom image highlights the R ct stability and the measurement reproducibility over time.

Fig. 5 .
Fig. 5. Tests in CaCl 2 -treated milk-based samples for sensors with LL, and with LL PG contamination: (a) Nyquist diagram of EIS responses of LL and PG sensors in milk-based solutions treated with CaCl 2 , and (b) phase of EIS responses as a function of the frequency of LL and PG sensors in milk-based solutions treated with CaCl 2 .

Fig. 6 .
Fig. 6.Test in CaCl 2 -treated milk-based samples for sensors in milk with LL, and with LL PG at different concentrations: (a) optical absorbance as a function of the time of the only CaCl 2 -treated milk-based solution (M), LL and PG solutions at different concentrations, (b) normalized charge transfer resistance R ct /R ct0 extrapolated by the EIS measurements, and (c) differential phase measured at 100 Hz extrapolated by the EIS measurements.

Fig. 7 .
Fig. 7. Test for CaCl 2 -treated milk-based samples with LL, and with low concentrations of LL PGs: (a) normalized charge transfer resistance R ct /R ct0 extrapolated by the EIS measurements and (b) differential phase measured at 100 Hz extrapolated by the EIS measurements.

Fig. 8 .
Fig. 8.Comparison between the results for filtered milk-based samples and CaCl 2 -treated milk-based samples for sensors in milk with LL, and with LL PG: (a) optical absorbance as a function of the time of the only milk-based solution (black curve), LL and PG solutions at different concentrations and (b) normalized charge transfer resistance R ct /R ct0 extrapolated by the EIS measurements.

Fig. 9 .
Fig. 9. Schematic representation of the sample treatments and their effects on the phenomena on the electrode.The image highlights the difference between the plain filtered milk and the CaCl 2 -treated milk.(a) Milk sample is centrifugated and filtered.(b) Sample undergoes to acidification, centrifugation, and filtering.(c) CaCl 2 is moreover added for preparing the CaCl 2 -treated milk samples.(d) LL in M17 is added in milk-based samples and deposited on the sensor.The proliferation of LL inhibits the charge transfer at the electrode interface.(e) In case of PG contamination, the charge transfer is partially restored.(f) In case of PG contamination in CaCl 2 -treated samples, the phage lytic action is enhanced, and the charge transfer is improved compared to (e).
Manuscript received 31 August 2023; revised 25 October 2023; accepted 13 November 2023.Date of publication 21 November 2023; date of current version 2 January 2024.The associate editor coordinating the review of this article and approving it for publication was Dr. Levent Yobas.(Corresponding author: Stefano Bonaldo.)Stefano Bonaldo, Lara Franchin, and Alessandro Paccagnella are with the Department of Information Engineering, University of Padova, 35131 Padua, Italy (e-mail: stefano.bonaldo@dei.unipd.it).

TABLE I ELECTRICAL
PARAMETERS FOR NC, LL, PG 10 7 SENSORS AT 300 min