A high-sensitivity inkjet-printed flexible resonator for monitoring dielectric changes in meat

Academic Editors: Pedro Pinho and Christian Vollaire Received: 13 January 2025 Revised: 17 February 2025 Accepted: 20 February 2025 Published: 22 February 2025 Citation: Abounasr, J.; Gharbi, M.E.; García, R.F.; Gil, I. A High-Sensitivity Inkjet-Printed Flexible Resonator for Monitoring Dielectric Changes in Meat. Sensors 2025,25, 1338. https:// doi.org/10.3390/s25051338 Copyright: © 2025 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/ licenses/by/4.0/). sensors Article A High-Sensitivity Inkjet-Printed Flexible Resonator for Monitoring Dielectric Changes in Meat Jamal Abounasr * , Mariam El Gharbi , Raúl Fernández García and Ignacio Gil Department of Electronic Engineering, Universitat Politècnica de Catalunya, 08222 Terrassa, Spain; [email protected] (M.E.G.); [email protected] (R.F.G.); [email protected] (I.G.) *Correspondence: [email protected] Abstract: This paper introduces a flexible loop antenna-based sensor optimized for realtime monitoring of meat quality by detecting changes in dielectric properties over a six-day storage period. Operating within the 2.4 GHz ISM band, the sensor is designed using CST Microwave Studio 2024 to deliver high sensitivity and accuracy. The sensing mechanism leverages resonance frequency shifts caused by variations in permittivity as the meat degrades. Experimental validation across five samples showed a consistent frequency shift from 2.14 GHz (Day 0) to 1.29 GHz (Day 5), with an average sensitivity of 0.173 GHz/day . A strong correlation was observed between measured and simulated results, as evidenced by linear regression ( R2= 0.984 and R2= 0.974 for measured and simulated data, respectively). The sensor demonstrated high precision and repeatability, validated by low standard deviations and minimal frequency deviations. Compact, printable, and costeffective, the proposed sensor offers a scalable solution for food quality monitoring. Its robust performance highlights its potential for integration into IoT platforms and extension to other perishable food products, advancing real-time, non-invasive, RF-based food safety technologies. Keywords: flexible loop antenna; inkjet-printed sensor; microwave resonance; meat freshness monitoring; dielectric property analysis; real-time food quality assessment; permittivity-based sensing; non-invasive monitoring; 2.4 GHz ISM band; printed electronics 1. Introduction Food safety and quality are critical aspects of global public health and food security. The increasing complexity of food supply chains, coupled with growing consumer demand for fresh, high-quality products, requires robust monitoring systems [ 1 ]. Ensuring that food, especially perishable items such as meat, remains safe and of high quality throughout its storage and distribution is essential to preventing foodborne illness, reducing economic losses, and building consumer confidence [ 2 , 3 ]. Contaminated or substandard food creates serious health risks to consumers leading to illnesses associated with ingestion of food, which include acute bowel disorders, chronic illness, and mortality [ 4 ]. It is reported by the World Health Organization (WHO) that millions of people across the globe suffer from foodborne diseases every year with, contamination by pathogenic organisms such as Salmonella, Listeria, and E. coli as key factors. It is also important to realize that the spoilage of meat leads to the generation of poisonous by-products as well as the growth of pathogens, thus warranting the need for stringent supervision [ 5 ]. On a worldwide scale, food safety and food quality are important issues of food security. Spoilage and waste appear as a challenge to food security, especially in underdeveloped countries with very Sensors 2025,25, 1338 https://doi.org/10.3390/s25051338 Sensors 2025,25, 1338 2 of 16 little cold chain facilities. A developed means of supervision may help to manage food waste by tracing the most susceptible products dressed to be lost in the chain, which will help to take proper measures as early as possible. Technological advances in radio frequency (RF) and microwave technologies transform food safety and quality assurance by providing real-time, non-invasive monitoring solutions [ 6 ]. These techniques leverage the interaction between electromagnetic waves and food materials to detect certain key quality parameters such as spoilage. RF-based systems, including antenna-based sensors and RFID sensors, allow for continuous monitoring of environmental conditions and dielectric property changes in food [ 7 ]. By offering a sophisticated and scalable approach to food monitoring, RF and microwave techniques pave the way for safer and more efficient food quality management. In the context of food safety and quality assurance, dielectric properties, particularly permittivity, serve as reliable indicators of structural and compositional changes in perishable foods [ 8 ]. As food ripens or spoils, its chemical composition undergoes significant alterations, leading to measurable changes in its dielectric properties. These changes make permittivity a cornerstone of radio frequency (RF)- and microwave-based sensing and monitoring technologies [ 9 ]. By accurately measuring these properties, advanced systems can deliver real-time insights into food freshness, enhancing safety, reducing waste, and ensuring better quality control throughout the supply chain. A trained individual, such as a store manager, can assess the quality of meat by sight, smell, and touch. However, this approach poses potential health risks to the inspector. In addition, relying solely on subjective judgment leads to the possibility of errors that are not supported by objective data. Furthermore, when touch and smell are used in the inspection process, it becomes impossible to assess the packaged products [ 10 ]. The existing meat quality monitoring systems are costly to implement and not widely utilized. Despite their presence, cases of health issues and fatalities resulting from the consumption of spoiled meat are still occur. To address these challenges, several studies have demonstrated the effectiveness of RF and microwave techniques for monitoring perishable foods such as meat, dairy, and seafood [ 11 – 13 ]. These methods offer advantages over traditional approaches, including faster processing times, non-destructive analysis, and improved scalability. For example, a UHF RFID system was used to monitor frozen meat using received signal strength indicator (RSSI) data, as reported in [ 14 ]. This method demonstrated monotonic relationships between RSSI values, temperature, and hardness during defrosting, enabling effective cold chain monitoring and safety. Another technique was introduced in [ 15 ], where dielectric measurements were used to assess meat freshness by monitoring changes in permittivity and conductivity over storage intervals. Using capacitance and conductance data measured with an LCZ meter across frequencies from 10 kHz to 1 MHz, the study demonstrated a decline in these dielectric properties with increasing storage time. In addition, an IoT-based system combining cameras and air quality sensors, using deep learning models to analyze color changes and predict meat freshness in real time, was reported in [ 16 ]. This system requires complex calibration for different types and conditions of meat, and its applicability may be limited to specific use cases. However, all of the studies mentioned above rely primarily on rigid substrates, and these kinds of substrates may not be ideal for applications that require flexibility, such as integration into smart packaging or wearable systems for real-time monitoring. This paper presents a flexible antenna-based sensor for monitoring meat (beef) freshness over six days using microwave signals. The sensor, a circular loop antenna printed on a flexible polyamide substrate, detects shifts in resonance frequency caused by changes in the meat’s dielectric properties as it degrades. These shifts alter the antenna’s resonance Sensors 2025,25, 1338 3 of 16 characteristics, enabling non-invasive freshness detection. Measurements were conducted using a Vector Network Analyzer (VNA) to observe resonance frequency shifts over time. The analysis was limited to six days because the meat was visibly rotten beyond this period, making further investigation unnecessary. Additionally, resonance frequency results on subsequent days showed no significant changes compared to Day 6, confirming complete spoilage, which is evident through visual and sensory inspection. Extending the analysis further was, thus, unnecessary for food quality assessment. 2. Sensor Design and Working Principle 2.1. Sensor Design and Manufacturing The sensor, based on a loop antenna design, was developed to operate within the 2.4 GHz ISM band and optimized using CST Microwave Studio 2024. The geometrical parameters of the proposed antenna are detailed in Table 1and illustrated in Figures 1and 2 . These parameters were tuned to achieve optimal performance and ensure efficient electromagnetic coupling with the target samples. Polyimide (PI) films, commonly known by the brand name Kapton, were selected as the substrate material due to their excellent dielectric properties ( ϵr= 3.5, loss tangent = 0.0027), lightweight structure, and durability under extreme environmental conditions. To ensure consistency for high-performance microwave applications, the electrical properties of the substrate were verified using a Q-meter, as shown in Figure 1a. Figure 1. Overview . of the experimental setup and fabrication process: (a) Q-meter used for extracting electrical properties of the substrate; (b) Voltera NOVA inkjet and extrusion printer used for antenna fabrication; (c) Memmert oven for drying and curing the printed ink to ensure proper adhesion and distribution; (d) final fabricated loop antenna; (e) 10 g of meat sample; (f) vector network analyzer (VNA) for antenna performance validation. The conductive traces of the loop antenna were fabricated using an inkjet and extrusion printer (Voltera NOVA, Figure 1b) with a 225 µ m nozzle. This system simplifies the calibration process by automating parameters such as dispensing height, ink pressure, and temperature, making it highly user-friendly and efficient. The printing process typically requires less than 15 min to complete. Inkjet printing was chosen for its rapid prototyping capabilities, minimal material waste, and suitability for flexible substrates, making it an ideal technique for disposable and scalable food quality sensors. After printing, the antennas were dried in a Memmert oven (Figure 1c) at 50 ◦ C for 15 min to ensure proper ink adhesion and uniform distribution. This step is critical for Sensors 2025,25, 1338 4 of 16 maintaining reliable conductivity and achieving stable sensor performance. The final fabricated loop antenna is shown in Figure 1d, and it demonstrated excellent electrical matching to the standard 50 Ω impedance. This performance was validated using a vector network analyzer (VNA, Figure 1f). To assess its functionality, the sensor was tested with a meat sample (Figure 1e) placed directly on the loop in the sensing area illustrated in Figure 2. Measurements were collected over multiple days to analyze the impact of permittivity variations on the reflection coefficient ( S11 ) and resonance frequency behavior. This robust and systematic design ensures the sensor’s reliability for dielectric sensing applications in meat quality evaluation. Figure 2. The geometrical parameters of the proposed antenna. Table 1. Geometrical parameters of the proposed sensor. Parameter Value (mm) Description WSL50 Width of the substrate LSL60 Length of the substrate WL3 Width of the loop RL20 Radius of the loop Wf10 Width of the feedline gL4 Gap between loop and feedline Lf13 Length of the feedline hLoop 0.19 Substrate thickness t0.008 Conductor thickness 2.2. Working Principle The proposed loop antenna sensor operates by generating a localized electromagnetic (EM) field in its near-field region. When a material, such as meat, is placed within this field, the electromagnetic waves interact with the dielectric properties of the material, such as its relative permittivity ( ϵr ). This interaction modifies the antenna’s impedance by either absorbing or reflecting the EM energy, resulting in measurable changes in the reflection coefficient (S11) and a shift in the resonance frequency (fr). At the resonance frequency, the antenna’s impedance is minimized, enabling strong energy coupling with the surrounding medium. This occurs when the inductive reactance ( XL ) of the loop antenna matches the capacitive reactance ( XC ) of the surrounding system. The two reactances cancel each other out, leaving only the resistive component in the total impedance: Ztot =R+j(XL−XC)(1) At resonance, since XL=XC, the impedance simplifies to Ztot =R(2) Sensors 2025,25, 1338 5 of 16 This purely resistive impedance results in the lowest impedance value, allowing maximum current flow in the loop. The strong current flow enhances the interaction between the antenna and the surrounding material, amplifying the sensor’s sensitivity to dielectric changes. As the dielectric properties of the material affect the EM field distribution, the resonance frequency and reflection coefficient shift accordingly. The interaction between the loop antenna and the material is illustrated in Figure 3, which highlights how the loop antenna generates a localized EM field that interacts with the material. The highest intensity regions, shown in red and orange, are concentrated near the loop antenna edges, while the field weakens radially outward. The meat sample perturbs the field, as evidenced by the redistribution of field lines, confirming the antenna’s sensitivity to dielectric variations in the sample. The specific composition of the meat, along with the effective dielectric properties used in this study, is detailed in the following subsection. These properties were carefully selected to ensure accurate simulation and analysis of the interaction between the loop antenna and the meat samples. Figure 3. Simulated electric field highlighting the interactions between the loop antenna and the meat sample: (a) side view and (b) 3D view of the E field. 3. Experimental Setup and Measurement Workflow To investigate the dielectric properties of beef samples and their interaction with the loop antenna, a systematic experimental workflow was implemented. Each sample was carefully prepared to weigh exactly 10 g, with a composition of 80% muscle and 20% fat, reflecting a typical profile for evaluating meat quality. The samples were cylindrical in shape, with a radius of 1.5 cm and a height of 1.5 cm, corresponding to a volume of approximately 10.61 cm 3 . These geometric and compositional parameters were used for both experimental measurements and simulation models to ensure consistency and accuracy in the analysis. Five samples were prepared and tested daily over a six-day period, providing a robust dataset for analysis. The relative permittivity ( ϵr ) and loss tangent ( tan δ ) of the meat were calculated using a weighted average approach based on the dielectric properties of muscle and fat, ensuring accurate modeling and analysis. These properties were derived using the following equations [17,18]: ϵeff =fmuscle ·ϵmuscle +ffat ·ϵfat (3) tan δ=σ ωϵ0ϵr(4) where the variables are as follows: •ϵeff: effective relative permittivity of the mixture. •fmuscle =0.8 and ffat =0.2: weight fractions of muscle and fat. •ϵmuscle and ϵfat: relative permittivities of muscle and fat. Sensors 2025,25, 1338 6 of 16 •σ: conductivity of the mixture. •ω=2πf: angular frequency, where f=2.4 GHz. •ϵ0=8.854 ×10−12 F/m: permittivity of free space. Using these equations and the dielectric properties of muscle ( ϵmuscle ≈ 49, σmuscle ≈ 1.3 S/m ) and fat ( ϵfat ≈ 5.2, σfat ≈ 0.05 S/m ), the effective properties of the mixture were calculated as follows: ϵeff = (0.8 ·49) + (0.2 ·5.2)≈40.96 (5) tan δ=1.048 2π·2.4 ×109·8.854 ×10−12 ·40.96 ≈0.020 (6) These calculated values were used to simulate the electromagnetic interaction between the loop antenna and the meat sample. The storage conditions were carefully controlled to simulate realistic handling scenarios while minimizing environmental disturbances. Each sample was stored for 12 h at 7 ◦ C (refrigerated) to slow degradation, followed by 8 h at 25 ◦ C (room temperature) to mimic typical exposure conditions. This cycle was repeated daily to reflect practical storage variations. To mitigate the influence of ambient humidity and temperature fluctuations, all measurements were performed immediately after removing the samples from refrigeration, ensuring consistency in dielectric behavior and minimizing external interference. The experimental setup included a loop antenna designed to operate within the 2.4 GHz ISM band and a vector network analyzer (VNA) to measure the reflection coefficient ( S11 ) over a frequency range of 1–3 GHz, as presented in Figures 4and 5. Each sample was carefully positioned on the loop antenna to ensure proper coupling with the electromagnetic field. The placement of the sample on the loop antenna allowed direct interaction between the localized electromagnetic field and the dielectric properties of the meat. Figure 4. Three-dimensional model representation of the experimental protocol for assessing meat quality using microwave sensing. The model depicts fresh samples (Day 0) and aged samples (Day 5) placed in containers, connected to a microwave sensing setup and a vector network analyzer (VNA). Sensors 2025,25, 1338 7 of 16 Figure 5. Comprehensive experimental setup for analyzing dielectric property variations in meat samples from Day 0 to Day 5. For data collection, S11 measurements were recorded across the frequency band for each sample. Five samples were measured daily, starting from Day 0 (fresh meat) to Day 5 (stored meat). Measurements were repeated under identical conditions each day to track trends in resonance frequency over time. Data acquisition and analysis were performed using MATLAB 2024B, which facilitated the comparison of S11 variations across frequencies and the statistical analysis of all samples. Data processing involved analyzing the shifts in resonance frequency, identifying changes in the reflection coefficient, and performing statistical comparisons across all days, as described in Figure 4. This comprehensive approach enabled precise monitoring of changes in the dielectric properties of the meat over time, providing valuable insights into the relationship between meat quality and its electromagnetic response. 4. Results and Discussion The experimental results in this section present the capability of the loop antenna to detect variations in the dielectric properties of meat samples stored under controlled conditions, as shown in Figure 5. By monitoring changes in the reflection coefficient ( S11 ) and resonance frequency over time, the antenna’s sensitivity to changes in permittivity caused by meat aging, storage, and environmental variations is assessed. The measurements were benchmarked against simulated data to confirm consistency and establish a correlation between experimental and theoretical observations. Additionally, statistical analyses were performed to evaluate the repeatability of the measurements and quantify the observed frequency shifts resulting from the effects of aging, storage, and environmental factors on the meat samples. The results are discussed with a focus on the sensor’s performance, the influence of material properties on measurement accuracy, and potential applications in food quality assessment. 4.1. Antenna Sensor Evaluation in Free Space The reflection coefficient ( S11 ) of the loop antenna was evaluated both experimentally and through simulation to benchmark the antenna’s performance in free-space conditions. Figure 6presents a comparative analysis of the measured and the simulated S11 as a function of frequency within the 1 GHz to 3 GHz range. The simulated response exhibits a sharp resonance dip at 2.4 GHz, which corresponds to the antenna’s design frequency in the ISM Sensors 2025,25, 1338 8 of 16 band. The measured response closely follows the simulation, showing a resonance dip at approximately 2.14 GHz. This corresponds to a frequency shift of approximately 10.83%. The measured response is slightly lower in frequency than the simulated response, which can be attributed to fabrication tolerances and variations in material properties. Factors such as the actual permittivity and thickness of the substrate, which may deviate slightly from simulation inputs, could lead to this shift. Additionally, environmental conditions during measurement might contribute to further deviations. Despite this shift, the close agreement between the simulated and measured data validates the design process and highlights the effectiveness of the fabrication method. The minimum reflection at resonance confirms good impedance matching to the standard 50 Ω system, ensuring efficient energy coupling and minimal power loss at the operational frequency. This benchmark establishes a reliable reference for subsequent experiments involving varying permittivity conditions caused by meat samples. Figure 6. Comparison of measured and simulated S11 in free space. 4.2. Permittivity Variation Over Days To validate the hypothesis regarding the influence of permittivity variation on resonance frequency shifts, a parametric sweep simulation was conducted in CST Studio Suite. This simulation utilized a cylindrical meat model with a radius of 1.5 cm, a height of 1.5 cm, and a volume of approximately 10.61 cm 3 , consistent with the experimentally prepared samples. The relative permittivity ( ϵr ) of the modeled meat sample was varied, with initial and final values ranging from ϵr= 37 to ϵr= 91, as derived and justified in earlier calculations. The goal of this simulation was to observe and quantify the impact of changing dielectric properties on the resonance frequency behavior of the antenna system. Figure 7illustrates the results of the simulation, showing a clear trend in the frequency response: as the relative permittivity of the material increased, the resonance frequency consistently shifted towards lower values. For the initial permittivity value ( ϵr= 37), the simulated resonance frequency was approximately 2.1 GHz, corresponding to a state where electromagnetic coupling with the material was relatively low. However, as the permittivity increased to its maximum value of ϵr= 91, the resonance frequency decreased significantly to approximately 1.3 GHz. This marked shift reflects the increased energy storage capacity of higher-permittivity materials, which modifies the electromagnetic field distribution and results in a reduced resonance frequency. Sensors 2025,25, 1338 9 of 16 Figure 7. Simulated S11 response over the frequency range for various permittivity values corresponding to different days. The observed trend presented in Figure 8aligns closely with theoretical predictions, where materials with higher permittivity are expected to exhibit stronger electromagnetic coupling due to increased polarization. This effect reduces the antenna’s operating frequency as the system’s inductive and capacitive reactance components adjust to the new material properties. Such behavior is indicative of the sensor’s sensitivity to the dielectric properties of the surrounding medium and highlights its potential for detecting subtle variations in material composition. By performing a comprehensive parametric sweep simulation, the permittivity values were strategically selected to align with the frequencies observed in the measurement results (discussed in later sections). This approach ensures that the simulated frequency shifts correspond closely to the experimentally obtained data, thereby validating the reliability of the proposed sensing mechanism. The consistency between the simulation and measurement outcomes highlights the antenna’s capability to effectively monitor dielectric changes over time. Figure 8. Dynamic relationship between simulated permittivity and frequency across days. Sensors 2025,25, 1338 16 of 16 7. Bibi, F.; Guillaume, C.; Gontard, N.; Sorli, B. A review: RFID technology having sensing aptitudes for food industry and their contribution to tracking and monitoring of food products. Trends Food Sci. Technol. 2017,62, 91–103. [CrossRef] 8. El Khaled, D.; Novas, N.; Gazquez, J.A.; Garcia, R.M.; Manzano-Agugliaro, F. Fruit and vegetable quality assessment via dielectric sensing. Sensors 2015,15, 15363–15397. [CrossRef] [PubMed] 9. Kim, S. Inkjet-printed electronics on paper for RF identification (RFID) and sensing. Electronics 2020,9, 1636. [CrossRef] 10. VanOverbeke, D. Handbook of Beef Safety and Quality; CRC Press: Boca Raton, FL, USA, 2024. 11. Wang, M.; Yang, Y.; Mu, B.; Nikitina, M.A.; Xiao, X. Millimeter wave-based non-destructive biosensor system for live fish monitoring. Biosensors 2022,12, 541. [CrossRef] [PubMed] 12. Wang, D.; Zhang, M.; Jiang, Q.; Mujumdar, A.S. Intelligent system/equipment for quality deterioration detection of fresh food: Recent advances and application. Foods 2024,13, 1662. [CrossRef] [PubMed] 13. Beshai, H.; Sarabha, G.K.; Rathi, P.; Alam, A.U.; Deen, M.J. Freshness monitoring of packaged vegetables. Appl. Sci. 2020,10, 7937. [CrossRef] 14. Marindra, A.M.J.; Pratama, B.M.; Suroso, D.J. Non-invasive frozen meat monitoring system using UHF RFID tag antenna-based sensing and RSSI. Int. J. Adv. Sci. Eng. Inf. Technol. 2023,13, 1–7. [CrossRef] 15. Ghatass, Z.F.; Soliman, M.M.; Mohamed, M.M. Dielectric technique for quality control of beef meat in the range 10 kHz–1 MHz. Am.-Eurasian J. Sci. Res. 2008,3, 62–69. 16. Kim, D.-E.; Mai, N.-D.; Chung, W.-Y. AIoT-based meat quality monitoring using camera and gas sensor with wireless charging. IEEE Sens. J. 2023,24, 7317–7324. [CrossRef] 17. Sihvola, A.H. Electromagnetic Mixing Formulas and Applications; IET: Stevenage, UK, 1999; Volume 47. 18. Pozar, D.M. Microwave Engineering, 4th ed.; University of Massachusetts at Amherst, John Wiley & Sons, Inc.: Hoboken, NJ, USA, 2012; pp. 26–30. 19. Kim, D.-E.; Nando, Y.A.; Chung, W.-Y. Battery-free pork freshness estimation based on colorimetric sensors and machine learning. Appl. Sci. 2023,13, 4896. [CrossRef] 20. Calvet-Chautard, M.; Jaque Gonzalez, P.; Véronèse, T.; Dubuc, D.; Grenier, K. Microwave-based sensor dedicated to the characterization of meat freshness. In Proceedings of the 2020 IEEE MTT-S International Microwave Biomedical Conference (IMBioC), Toulouse, France, 14–17 December 2020 ; pp. 1–4. 21. Li, Z.; Suslick, K.S. Portable optoelectronic nose for monitoring meat freshness. ACS Sens. 2016,1, 1330–1335. [CrossRef] 22. Muradov, M.; Cullen, J.; Mason, A. Real-time monitoring of meat drying process using electromagnetic wave sensors. In Next Generation Sensors and Systems; Springer: Berlin/Heidelberg, Germany, 2016; pp. 221–233. 23. Nguyen, N.H.; Lam, M.B.; Chung, W.-Y. Battery-less pork freshness monitoring based on high-efficiency RF energy harvesting. J. Sens. Sci. Technol. 2020,29, 293–302. [CrossRef] Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.