Miniaturization limits of ceramic UHF RFID tags (2025)

The ongoing 4th Industrial Revolution¹ and the associated automation of manufacturing processes have led to a significant rise in the adoption of wireless technologies, favoring them over conventional wired approaches. Within this evolving environment, radio frequency identification (RFID) technology has become a key player, providing essential solutions for automatic monitoring and control in diverse sectors including logistics, retail, security, smart spaces, the Internet of Things, and beyond^2,3,4. RFID technology is categorized by frequency range (LF: 125–134kHz, HF: 13.56MHz, UHF: 860–960MHz, and microwave: 2.45GHz and 5.8GHz) and the operational principle (passive, active, and battery-assisted). The exact numbers depend on communication protocols and licensing countries. Hereinafter, we will concentrate on passive UHF RFID, which is a certain compromise between the cost, range, and amount of data stored on a single tag. In this scenario, as with most RFID configurations, the main hardware cost and complexity are allocated to the interrogation device (the reader), whereas the tag comprises a compact integrated circuit (IC) connected to an antenna. The reader initiates the communication by transmitting electromagnetic waves, which generate currents in the tag, thereby powering its internal memory circuit (chip) through a rectifier. In response to the query, the IC within the tag toggles between two impedance states, modulating the backscattered electromagnetic wave. This modulated signal is then captured and decoded by the reader. RFID communication protocols, as well as the permitted Equivalent Isotropically Radiated Power (EIRP) for transmitted signals, are regulated by international standards⁵. Architectures of RFID tags are often customized per application, and their performance is largely determined by the antenna design.

Generally, each manufacturer has its range of ICs, while antennas are custom-designed to meet specific requirements. The set of diverse applications, where conceptually different antenna design is needed includes reading-range extension^6,7,8, footprint miniaturization⁹, on-meta tagging^10,11,12, on-high-index material tagging¹³, omnidirectional response to a reader query^14,15, additional on-tag sensing functionalities¹⁶, operation in harsh conditions, and several others. One of the key challenges is combining multiple properties within a single device. Within this framework, we will focus on investigating compact long-range RFID tags, which hold potential for innovative applications in smart labeling of items within the domain of the Internet of Small Things (IoT), where the vision is to promote low-grade, resource-limited items to become active elements in a global network.

The notion of long-range RFID might be controversial, as this parameter not only depends on an antenna device but also on the IC and the reader, which comes with its own antenna. It is worth mentioning efforts aimed at tag size reduction and long-range tags, which, however, utilize completely different.

architectures^{17,18,19,20,21,22,23}. Hereinafter, we will focus exclusively on the tag antenna by considering a reading range of over one meter as a target while a standard Impinj R2000 reader. This setup maintains the transmitted power and reader antenna gain at 30 dBm of EIRP.

Typically, the reading range extension and the antenna footprint reduction contradict, setting an engineering tradeoff. The footprint is defined as the radius of a minimal virtual sphere (\(\:R\)), encompassing the structure. Quite a few techniques for antenna footprint miniaturization have been developed and include meandering²⁴, fractal designs^25,26, and introducing dielectric substrates^13,27 to name a few. An immediate penalty of the size reduction is the gain and bandwidth drops. The first limitation comes from an elevated near-field accumulation in the vicinity of lossy antenna materials (including lumped elements). The second limitation is that the antenna’s Q-factor increases as its size is reduced. Chu-Harrington limit (among several others) links a dipolar antenna footprint with the maximally achievable bandwidth²⁸. Specifically, the antenna Q-factor is bounded by:

where \(\:k\:\)is a wavenumber and \(\:R\) is the radius of the smallest sphere enclosing the antenna. These aspects should be considered when miniaturizing the RFID antenna.

In our recent efforts to reduce the volume of RFID tags, we have proposed and demonstrated a series of designs that incorporate high-index ceramic resonators as the primary antenna element. In particular, compact long-range ceramic tags²⁹, omnidirectional ceramic long-range^30,31 and miniaturized³² on-metal tags, and RFID sensors^33,34,35 were demonstrated.

Here we will investigate the miniaturization limits of this technology, emphasizing several practical constraints, including the choice of the material platform. We will also revise in detail the impedance matching techniques which do not involve introducing any additional lumped elements apart from the IC itself.

A ceramic RFID tag consists of a dielectric resonator coupled with a non-resonant metal loop, within which a standard RFID chip is integrated. The resonator is designed to support a magnetic dipole mode, resonant at a specified RFID frequency range. Inductive coupling between the resonator and the loop supports the interaction between the resonator’s mode and the IC. To recap, the vast majority of RFID tags rely on electric resonant antennas, which support the technology but do not allow for significant footprint reduction without degrading the reading range.

Figure1 depicts the activation process of a cubic-shaped dielectric RFID tag when it encounters a horizontally polarized wave, generated by a handheld reader. In this layout, the magnetic field has a vertical polarizaiton. The reader activates a magnetic dipole mode within the dielectric resonator of the tag by inducing displacement currents in the resonator. These currents excite conduction currents within the loop through inductive coupling, ultimately powering the chip’s electronics. In active mode, the chip alternates between two distinct impedance states, resulting in the modulation of the backscattered signal. The modulated signal follows the reverse path: the loop excites the magnetic mode in the ceramic resonator, which subsequently radiates the signal to the far field. To an extent, such configuration can be related to a loop antenna on a resonate substrate. As a result, this compact and long-range dielectric tag operates compatibly with standard RFID equipment and does not require any modifications to the existing system, which is especially important from an industrial perspective.

The method for designing UHF RFID tag antennas involves attaining a complex-conjugate impedance match. In this case, the matching is done to the chip impedance in its inactive state. Commercial UHF RFID chips generally exhibit comparable capacitive impedance characteristics, and their equivalent circuit model is often represented by a resistor and a capacitor arranged in parallel. To achieve impedance matching in the system, the reactive component of the impedance needs to be compensated. Conversely, minor variations in the real parts of the antenna and chip impedances typically do not significantly impact the efficiency of the matching. Therefore, the RFID tag antenna must have a specific inductive impedance at the frequency of interest. Typically for metal dipole tag antennas, a T-shaped impedance matching element is used³⁶.

The impedance matching principle for ceramic tags is different – it relies on the geometrical arrangement between the elements and does not introduce any additional circuitry into the scheme. The tag antenna consists of an inductively coupled resonator and a metal ring with a gap (split ring). Figure2a demonstrates the equivalent circuit of the tag. The parameters of the dielectric resonator are optimized with an eigenmode solver (implemented in CST Microwave Suite) to tune the dipole magnetic mode into the RFID frequency range. In the next step, a copper split ring with an active discrete port with a complex impedance \(\:Z=12.7-140.8j\) (corresponding to the datasheet impedance at 915MHz of an Impinj Monza R6 RFID chip) is added on the top of the resonator. This construction serves as an excitation source in the frequency domain solver.

To achieve impedance matching, the following parameters play a role: (i) the ring’s radius, (ii) the wire thickness (left part of the equivalent circuit, Fig.2a), and (iii) the position of the ring relative to the resonator (through the inductive coupling coefficient M). Figure2b summarizes the strategy. As a specific example, a dielectric cube with \(\:{\epsilon\:}_{r}\) = 100 with the edge length of a = 27.9mm was optimized. The parameters of the ring that allow achieving impedance matching are the following: the wire thickness \(\:{W}_{ring}=1\) mm, external radius \(\:{R}_{ring}=10\) mm, position relative to the central axis of the resonator \(\:\left({x}_{0},\:{y}_{0}\right)=\:\left(\text{0,0}\right),\) and the height above the resonator \(\:{H}_{ring}=4.5\) mm. It is worth noting that while in numerical modeling the ring is suspended in the air above the resonator, a thin polystyrene spacer with dielectric properties very close to free space is used to support the ring in the experimental layout. Using the crossing point on the right-hand side of the resonance (Fig.2d) enables impedance matching between the antenna (resonator + ring) and the RFID chip without requiring additional matching elements. As becomes evident, the resonance is primarily determined by the dimensions of the dielectric resonator. Consequently, changing the size of the ring has only a minor effect on the overall input impedance of the antenna.

Figure2c presents a Smith chart, demonstrating the differences between the initial guess and the optimized model. Figure2d shows the dependence of the real and imaginary parts of the optimized impedance together with the complex-conjugated impedance of the Impinj Monza R6 RFID chip. A perfect impedance matching is achieved within the RFID frequency band.

Further analysis of the ring parameters, impacting the tag impedance matching, is summarized in (Fig.3). The impact of the wire thickness \(\:{W}_{ring}\) (Fig.3b), the height of the ring above the resonator \(\:{H}_{ring}\) (Fig.3c), the radius of the ring \(\:{R}_{ring}\) (Fig.3d), and the position of the ring relative to the central axis of the resonator \(\:{x}_{0},\:{y}_{0}\:\)(Fig.3e) were explored. For each panel, only one ring parameter was varied at a time, while the rest, as well as the parameters of the dielectric resonator, were constant.

It can be seen that an accurate multi-variable optimization can lead to impedance-matching conditions. Considering many successful realizations with different geometries (in this and our previous studies), it might be claimed that this methodology is reliable and can be translated to other types of high-index resonators and ICs with different impedances to reduce production complexity. It is worth noting that all subsequent realizations with varying ceramic resonator parameters underwent the same impedance-matching procedure.

Tag miniaturization–bandwidth limitations

By elevating the dielectric constant of the material used, the size of dielectric resonators can be reduced while maintaining a constant resonance frequency. A practical constraint in this approach is the increase in losses, which typically rise in correlation with the real part of the refractive index. Losses adversely impact the realized gain and, as a result, lead to a reduction in the tag’s reading range. The use of high-quality ceramic materials with low-loss tangents enables the production of a compact tag capable of achieving a reading range of over 10m³¹. Following those demonstrations, an appealing objective is to explore how far miniaturization can be taken. There are two primary limiting factors: the fundamental bandwidth limitation and the practical concern related to the temperature sensitivity of high-index materials. This section concentrates on exploring bandwidth limitations.

Efforts to reduce the size of resonant antennas encounter the challenge of a corresponding decrease in their operational bandwidth. Among various criteria, the Chu-Harrington limit is one of the most frequently applied²⁸. Worth mentioning that this limit applies to a dipolar antenna resonance. In our case here, the ceramic resonator operates at a magnetic dipolar resonance, thus closely following the Chu-Harrington limit.

The minimum operational bandwidth required for UHF RFID applications is defined by the communication protocol, which divides the entire bandwidth into 50 channels, each with a bandwidth of 500kHz (for the US 902–928MHz band, in other countries RFID regulations are different). This division into multiple channels facilitates anti-collision protocols, enabling the simultaneous communication of several readers with multiple tags placed in the same area. As a result, the minimum bandwidth of a tag should be no less than 500kHz. It’s worth noting that a tag with a narrower bandwidth can still be interrogated, but the reading distance will be significantly reduced. Here we use bandwidth definition for −6 dB S₁₁ level. Lower levels of matching will prevent long-range communication.

Following this basic criterion, we numerically examined a series of tags with the resonator’s permittivity \(\:{\epsilon\:}_{r}\) ranging from 100 to 1250. The edge of the cubic resonator varied from 27.9 to 7.7mm as schematically represented in Fig.4a. Figure4b shows the resonator volume as the function of its dielectric permittivity. For each iteration with dielectric permittivity, the tag was matched to the impedance of the Impinj Monza R6 chip in the frequency range of 915–917MHz (the allocated band in Israel) (Fig.4c), following the approach from the previous section.

After designing 6 ceramic tags and matching them to the IC, their performances can be assessed. Three scenarios for material losses in the dielectric were considered: (i) lossless, (ii) loss tangent of \(\:{10}^{-4}\), and (iii) loss tangent of \(\:{10}^{-3}\). Figure4c demonstrates the dependence of the tag antenna’s bandwidth at the− 6 dB level on the permittivity of the resonator. Here, the limit of 500kHz has been set. The horizontal cut with this threshold highlights the limitations of miniaturization. The results show that tags with permittivity above 750 do not meet the criterion. The bandwidth also depends on loss tangent, i.e., losses higher than \(\:{10}^{-3}\) also critically degrade the performance. Figure4d demonstrates the realized gain of the tag antenna as the function of the dielectric permittivity of the resonator. It is quite sensitive to the losses since the field is localized within the volume of the resonator. Ceramic materials with losses lower than ~ \(\:{10}^{-4}\:\) do not critically degrade the gain performance. Figure4e visualizes the dramatic bandwidth drop with the size reduction.

To further illustrate the impact of size on antenna properties, the bandwidth and radiation efficiency were plotted as functions of the radius of a sphere encompassing the structure. Figure5a shows that the bandwidth exhibits a linear behavior, as predicted by Eq.1 when the second term is neglected, which applies to small antennas. However, as the size decreases further, losses dominate the behavior. This is evident in the radiation efficiency plots in Fig.5b, where for larger sizes, the efficiency approaches unity as the element becomes matched to radiation, and internal losses play a less significant role.

From the results discussed above, we conclude that the optimal dielectric permittivity for ceramic tag miniaturization is around 500 with a dielectric tangent loss value of \(\:{10}^{-4}\:\)or lower. The relevant candidates are composites of BaTiO3/SrTiO3 in the ratios 50/50, 55/45, 60/40 with Mg-contained additions, such as Mg₂TiO₄³⁷. Those material platforms offer an optimal trade-off between size reduction, bandwidth, and realized gain and thus may be advantageous over other miniaturization techniques that are struggling to maintain this balance.

Beyond the intrinsic bandwidth constraints, several practical aspects further limit the applicability of ceramic RFID tags in everyday scenarios. We will focus here on the significant sensitivity of these tags to variations in ambient temperature. To explore these aspects, a set of resonators was investigated. The samples were provided by Ceramics Ltd³⁸. Cylindrical resonators with dielectric permittivities of 80, 100, 270, and 500 were designed to support a resonant magnetic dipole mode within the RFID communication range of 902–928MHz. Figure6a presents photos of the manufactured resonators. Note that the differences in geometries (cubes versus cylinders) do not play any critical role. Cylinders here are in use owing to their easier fabrication.

A Rohde & Schwarz ZVB20 vector network analyzer with a small loop antenna connected to one of the ports was used to measure the resonant frequencies of the resonators. The measurements were carried out at different temperatures ranging from 40 to 90 degrees, as shown in (Fig.6b). The resonators were heated on a plate, while temperature control was performed using a thermocouple connected to a multimeter.

Figure 6c demonstrates the S₁₁ spectra for different temperatures, \(\:{\epsilon\:}_{r}=500\) sample. This material (BaTiO₃ mixed with Mg₂TiO₄) has a strong temperature dispersion leading to a resonance shift of 3MHz per 1°C.

Ceramics with permittivity \(\:{\epsilon\:}_{r}=270\) (composite of BaTiO₃/SrTiO₃) is also temperature-sensitive leading to the resonance shift of 1.3MHz per 1°C. While these two materials are less suitable for conventional RFID applications, they found use in temperature sensing, which can be further improved by introducing a quasi-bound state in the continuum (quasi-BIC) modes³⁴. For ceramic RFID, operating at a diverse range of ambient temperatures, \(\:{\epsilon\:}_{r}=80\sim100\:\)are the best candidates, as can be seen from (Fig.6d).

The reading range of the tag and its temperature are non-trivially related. The behavior relies on several factors, one of which is the used RFID frequency band. For example, a communication channel with a 500kHz bandwidth, positioned within the 915–917MHz band allocated for communication (in Israel), will drift out of range with just a 1°C change in ambient temperature. At the same time, the 902–928MHz band (US standard) allows the tag to be read with temperature variations of ± 3°C if the tag is tuned to the central frequency.

Worth mentioning that ceramics are under intensive investigation nowadays and new materials in the future might possess high permittivity, low losses, and temperature stability, opening new possibilities for the miniaturization of ceramic RFID tags.

An extensive numerical investigation into the design strategies of passive ceramic RFID tags was undertaken on pathways to find miniaturization limits. An in-depth analysis of impedance matching without relying on additional lumped components and solely using geometrical parameters was performed. The factors, limiting the tag footprint reduction, were considered. The main limitation is the bandwidth, which prevents reducing the size of the ceramic tag to less than one cubic centimeter without losing long-range communication. Additional limiting factors include the losses and temperature stability of existing ceramic materials. All those factors, however, are not strictly fundamental.

Further size reduction of the antenna could be possibly achieved through resonant.

cascading^39,40,41 allowing the multi-resonant antenna to bypass the Chu-Harrington bandwidth limitation. Additionally, identifying temperature-stable, high-index ceramic materials will provide these prospective tags with stability against environmental fluctuations.

Author notes

1. Alyona Maksimenko and Dmitry Dobrykh contributed equally to this work.

Authors and Affiliations

1. School of Physics and Engineering, ITMO University, Saint Petersburg, 197101, Russia

   Alyona Maksimenko,Ildar Yusupov&Irina Melchakova

2. School of Electrical Engineering, Tel Aviv University, 69978, Tel Aviv, Israel

   Dmitry Dobrykh&Pavel Ginzburg

3. Qingdao Innovation and Development Center, Harbin Engineering University, Harbin, Qingdao, China

   Mingzhao Song

Contributions

A.M., D.D. and I.Y. made the numerical simulations and experimental study. P.G. wrote the main manuscript text. M.S. and I.M. prepared (Figs.1, 2, 3, 4 and 5). All authors reviewed the manuscript.

Corresponding author

Correspondence to Mingzhao Song.

Competing interests

The authors declare no competing interests.

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions