Millimeter-Wave Choke Ring Antenna with Broad HPBW and Low Cross-Polarization for 28 GHz Dosimetry Studies

Abstract

A choke ring horn antenna has been designed for use as an RF applicator in a compact range in vitro 28 GHz bioelectromagnetic exposure system. The 30 mm × 50 mm horn antenna was fabricated and measured to operate from 27.75 GHz to 34.5 GHz with a −20 dB measured S11 and a measured antenna gain of more than 10 dBi. A wide sectoral (flat top) and symmetric E- and H-plane pattern with a half-power beamwidth of more than 60 degrees was achieved with a cross-polarization discrimination of better than 28 dB. Electromagnetic slots were introduced in the antenna to suppress excess cavity mode radiation which inherently impacts the cross-polarization levels of choke ring antennas. The proposed antenna was successfully integrated into the compact measurement chamber in partnership with the Korea Telecommunication Research Institute (ETRI) and is currently in use for real-time 5G millimeter-wave dosimetry studies.

1. Introduction

In the wake of the call for increased data rates and system throughput, the millimeter-wave (mmWave) band has been of great focus for researchers in recent times. Although much of this frequency band from 30 GHz to 300 GHz is vastly untapped, there are many applications to which these bands are currently being utilized. These applications range from airport security scanning devices to IoT device-to-device and wearable gadget communications. Others include biomedical applications such as MRI and CT scans, mobile and broadband backhaul systems, military systems as well as the prospects for 5G mobile communication [1,2].

For the latter application, the 28 GHz band in particular, is a promising candidate for the current generation of IMT-2020 5G wireless and mobile communication standards. Most researchers consider 28 GHz a mmWave frequency band due to its close air interface resemblance characteristics to the mmWave band.

The migration to this unused frequency band for commercial purposes will, however, not be complete without the proper dosimetry of human cells. The most widely used methods for human tissue analysis are the in vivo and in vitro [3,4,5]. Unlike other commercially licensed frequency bands, there is much work to be done for both in vivo and in vitro dosimetry experiments in the 28 GHz frequency band. For the latter, the penetrative effects of the mmWave frequency on human tissues are studied with the cells placed on cultured plates or transwells inside a controlled radiation chamber.

An area of concern is the geometry of the chamber sizes at 28 GHz and how it influences the antenna and field properties. As will be seen in Section 2 of this communication, certain key performance indicators (KPIs) are used to establish the relationship between the field properties and antenna as well as the complete geometry of the chamber. It is worth noting that the short wavelength of the mmWave, coupled with strict chamber condition requirements, make the chamber geometry and the culture plate’s properties and antenna performance very sensitive and critical [2,6].

Figure 1 shows a proposed exposure chamber for the intended experiment. A horn antenna or any suitable RF applicator with a certain electrical performance illuminates mmWave onto the test samples for a given time at a given exposure distance. In most mmWave in vitro propagation systems, 35 mm diameter Petri dishes are placed in the farfield region of the RF applicator, generating a plane wave as demonstrated in [7,8,9].

According to [7], most radiating systems do not use enclosures due to the complexity of controlling the environmental conditions inside the chamber. Nonetheless, in mmWave dosimetry, a proper enclosure system is required for the analysis of both thermal and ionizing experimentations [8]. Figure 1 thus shows a radiation box closing the antenna used for the EM wave emission. The enclosure mimics a typical incubator system or an anechoic chamber to trap fields within a controlled space.

The main challenge with the conventional radiating system approach is the use of a standard directive horn antenna which has an inherent drawback of poor exposure efficiency and also suffers from insufficient power density uniformity over the surface under test (SUT) [7,8]. This then requires a large test chamber with a small surface area; an option that is least desirable for commercial use. The best approach, as described by [8] and others regarding in vitro radiating and propagating dosimetry research at mmWave, is a modified horn antenna with broad beamwidth. To design a compact in vitro dosimetry chamber measuring 100 cm × 100 cm and operating at 28 GHz will require a specialized RF applicator, which is not available commercially. The compact chamber will require space for high frequency wave suppressants, ventilators, thermal cameras, positioners, etc., which then leave a compact space for the RF applicator. Thus, a 50 × 50 mm applicator is proposed. From the specified physical dimension of the radiator, more details on the values of

Table 1. Invitro RF applicator specification.

Parameter	Specifications
Dimensions	≤50 mm × 50 mm
Bandwidth	27.75~28.25 GHz
Gain	≥8 dB
Sidelobe levels	≤−20 dB
Polarization	E-H symmetric polarization
HPBW	≥60 degrees
Exposure efficiency	≥60%
Spot size	≥80 × 80 mm
S11	≤−20 dB

are elaborated in Section 2.1. A detailed chamber specification has been outlined in [10].

Table 1 shows a summary of the proposed RF applicator requirement for an in vitro mmWave exposure system. The antenna structure suitable to meet the criteria is usually a horn antenna with an oversized aperture and/or with several chokes. These structures provide a sectoral or secant pattern with a sharply descending slope. The choke ring antenna design is a known concept in antenna design [11,12,13,14,15,16]. The purpose of the technique is to alter the electric fields at the edges of a horn antenna to provide a secant beam (flat top) pattern.

The CRA in [17], the smooth-walled horn antenna of [18] and the NASA X-band multiple CRA [12] may be suitable examples of antenna systems that can meet the stringent specifications in Table 1. These were both fabricated with 3D printing of a specific dielectric material and metalized in silver foam. The foam metallization with silver reduced the weight of the antenna, but more importantly suppressed unwanted higher cavity modes which are inherent to the all-metal design of such antennas.

To this end, an all-metal choke ring horn antenna has been proposed in this communication with excellent field properties for use as an RF applicator in a mmWave exposure system. A key feature of this design is the achievement of a highly sectoral beam pattern with a low-profile horn antenna, offering excellent cross-polarization to ensure a symmetric E-H pattern.

Section 2 outlines the bioelectromagnetic requirements of the chamber system and by extension, the antenna requirement. Section 3 introduces the choke ring horn antenna with a proposed technique to suppress the cross-polarization of the all-metal design. Subsequent sections show the measured performance of the proposed antenna for both impedance and radiation field performance, making it suitable for use as an RF applicator in the proposed chamber.

2. 5G Dosimetry Specifications

In the complete system design, certain key performance indicators (KPIs) will be considered to properly characterize the performance of the proposed mmWave exposure system. Such indicators may be categorized into three broad sections, namely the SUT, the field quantities and the RF applicator or antenna. SUT considerations will include the material geometry, be it a Petri dish or transwells, the material properties or the volume and size of the transwells.

Field properties indicative of the antenna performance within the confines of the chamber will include the E-field magnitude and phase, the power density and exposure efficiency. Others may include the illumination spot size and exposure distance. These quantities are influenced directly by the gain and radiation efficiency of the antenna, the beamwidth, cross-polarization and the E-H pattern in the farfield.

A detailed description of the compact exposure chamber has been presented in a previously published work by the same authors [10]. For brevity and focus on the antenna design, field properties of power density, exposure efficiency and illuminated spot size will be discussed.

2.1. RF Applicator/Antenna Specifications

With the given antenna specification in Table 1, the traditional horn antenna, as already mentioned, will not provide the wide HPBW to evenly distribute the required power to the surface area under test. The relationship between the directivity of a standard aperture antenna, beamwidth and height will limit the design of an antenna with 50 × 50 mm size to achieve a 60-degree HPBW and E-H symmetry. From [19], a broader HPBW and low gain horn antenna require a much shorter horn with a large flare angle. These two conditions are unrealizable at 28 GHz due to the small wavelength. An approach used by some researchers including [8,20] was to use dielectric loadings and choke insertions to obtain the required antenna properties.

Furthermore, the antenna must possess symmetric E-H plane patterns with low cross-polarization. These are essential for accurate power density and exposure uniformity calculations, as inference from neighboring polarization reduces the power density at a given exposure distance and hence the exposure efficiency.

2.2. Key Performance Indicators (KPI)

2.2.1. Power Density

The power density (P_D) is calculated by using the Poynting vector as described in Equation (1), where E and H are the electric and magnetic peak phasors, respectively. That is, the time-varying average energy transferred per unit area within the exposure enclosure of Figure 1, assuming a radiated power (Prad) of 1 W.

$$P_D={\left|0.5\left(E\times H^{*}\right)\right|}_{Prad=1\mathrm{W}}$$

(1)

References [9,21] as well as other researchers, use the specific absorption rate (SAR), which is specific to the cell type under test. The power density metric was used because the tissue sizes will be ordered by micrometer thicknesses and thus require a system performance that is independent of the samples under test.

The power density profile example as shown in Figure 2 measures the concentration of the radiated power or the field intensity at a predefined illumination spot size (S) on the SUT, which is usually at the center of the antenna. Ideally, the curve should be above the −0.5 dB or any specified uniformity level. In most in vitro mmWave dosimetry tests, −0.5 dB and −1.0 dB are specified [8], but they are subject to the system requirements. That is, a −0.5 dB uniformity translates to about 89% of the input power, incident on the SUT. It can, therefore, be deduced that larger distances can reach better uniformity levels with the standard horn antenna. This means a larger incubator is required when using a narrow beamwidth antenna. In this regard, the HPBW of the antenna will directly influence the field intensity on the SUT at a given distance.

The power density P_D in Equation (2) is calculated from the received power (Prec), where P_rec is the received power of the radiator in dBW. Ae is the antenna effective aperture, G is given as the antenna gain and λ is the wavelength in free space.

$$\begin{array}{l}{P}_D={\displaystyle \raisebox{1ex}{${\mathrm{P}}_{rec}$}\!\left/ \!\raisebox{-1ex}{$A_e$}\right.}\\ A_e={\lambda }^{2}G/4\pi \end{array}$$

(2)

The above equation holds for farfield received power [22]. In the nearfield, the value of P_rec is influenced by the nearfield path loss (P_loss) shown in Equation (3).

$$P_{loss(E)}={\displaystyle \frac{{\mathrm{P}}_{rec(E)}}{P_{tx}}}={\displaystyle \frac{G_{tx}{G}_{rec(E)}}{4}}·\left({\displaystyle \frac{1}{{\left(kr\right)}^2}}-{\displaystyle \frac{1}{{\left(kr\right)}^4}}+{\displaystyle \frac{1}{{\left(kr\right)}^6}}\right)$$

(3)

For Equation (3), P_tx is the transmitted power of an open-ended probe used in verifying the RF applicator performance, r is the exposure distance between the antenna aperture opening and the surface of the SUT and k is represented by the wavenumber (2π/λ). Gtx is the gain of the open-ended probe and G_rec is the gain of the RF applicator. This loss term at higher frequencies is thus negligible and can thus be ignored if necessary.

2.2.2. Exposure Efficiency

The exposure efficiency is defined as the energy flux incident on the SUT. This is akin to the SAR since the chamber will be designed irrespective of the tissue used.

The energy flux is calculated with respect to the area of the plane under test. That is the area of a single dish or the surface area of any number of dishes. In the initial test, however, the rectangular area of the plane is used, giving an L × L sized SUT, as seen in Figure 3.

The exposure efficiency is given by Equation (4), where P_rad is the total radiated power, S_m = time-averaged Poynting vector and A = surface area of the SUT calculated from Equation (5).

$${\eta }_e={\displaystyle \frac{1}{P_{rad}}}{\displaystyle \underset{A}{\iint }{S}_m\delta A}$$

(4)

$$A={\displaystyle \frac{\pi D^2}{4}}$$

(5)

$$S=2h \ast \tan \left({\displaystyle \frac{\phi }{2}}\right)$$

(6)

2.2.3. Illuminated Spot Size

This defines the area of uniform illumination on the SUT at a given uniformity. The optimal spot size for a given HPBW will be determined from Equation (6), where S is the spot size, h is the distance from the antenna aperture to the SUT and φ is the HPBW. At the given distance h, shown in Figure 4, a wider φ will achieve the same uniformity level but at a shorter exposure distance. Considering this, RF applicators with wide beamwidth are most desirable in building a compact exposure system. The value is obtainable from the power density profile curve. At a given uniformity level of −0.5 dB, a spot size of 26 mm is incident on the SUT for a given exposure distance and HPBW.

These parameters will become important in the subsequent sections in determining the performance of the proposed horn antenna.

3. Antenna Design

With little research available for horn antenna design for use in an in vitro dosimetry at 28 GHz, it was necessary to remodel existing exposure system antennas, mostly greater than 28 GHz frequency bands to check their performance at 28 GHz. Four antenna models were chosen from several existing antennas at different millimeter-wave frequencies.

Figure 5 shows the models reoptimized at 28 GHz. The conical antenna model of Figure 5a without any dielectric loading is used for dosimetry analysis by [23,24,25] at 35 GHz/107 GHz, 60 GHz and 46 GHz, respectively. The pyramidal horn antenna shown in Figure 5b, which is commonly used for exposure system tests, were designed at 50 GHz, 60 GHz and 60.4 GHz by [9,26,27], respectively.

Ad hoc horn antenna models in Figure 5c,d were also modeled from [11,18] at 29.5 GHz and 60 GHz, respectively. The 29.5 GHz model in Figure 5d was designed for enhancing a directive beam pattern with a small form factor, but not for dosimetry analysis. As will be seen later, however, it proved capable after a few modifications to the dielectric contours.

A summary of the four antenna models is presented extensively with their corresponding E-field profile curves and power density profiles at 28 GHz. The conical horn has a nearly linear E-field magnitude curve with more of the power concentrated at the center of the antenna. The exposure uniformity and power density profiles show narrow spot sizes at the −0.5 dB uniformity level for different exposure distances (h).

The result of the pyramidal horn was similar to the conical one with an asymmetrical E-H plane pattern, but with a wider beamwidth compared to the latter. The exposure uniformity is also narrower at the center of the SUT. The choke ring antenna (CRA) model in Figure 5c has a more conformal r-squared E-field profile curve with a wide beamwidth of 50 degrees at an 11.5 dB gain. The exposure uniformity in Figure 5d is much larger due to the wider beamwidth. The E-field power, however, drops much faster in the farfield region. This indicates a limitation to the chamber size as it cannot be increased arbitrarily. The contoured smooth-walled model has a 55.5-degree beamwidth in both planes and has large uniformity values in the farfields above 4λg (guided wavelength).

Although not sufficient in illuminating the entire SUT at −0.5 dB exposure uniformity, the model in Figure 5c possessed a good potential to obtain the desired antenna properties in this research in that it has a second ring or choke to provide a secant farfield pattern for a broad and stable HPBW.

3.1. Choke Ring Horn Antenna (CRHA)

The concept of the secant beam shape is to design a theoretically “ill-fitted horn antenna” with a poor radiation pattern. That is, instead of a directive horn antenna, the dielectric and/or choke insertions create a secant beam in the farfield [8] thus making the HPBW broad with a nearly flat farfield radiation.

A rigorous analysis of the sectoral beam synthesis was performed in [28], but will not be covered in this communication. The chokes or extra rings in the horn antenna are designed to give rise to 180-degree phase shifts (λ/2) between each of the lobes with reference to the primary lobe generated by a traditional horn antenna. The depth of every successive choke is also chosen with the 180-degree phase shift. The beam coupling effect of the main beam and the successive chokes give rise to the sectoral beam. As the chokes increase, there is less coupling with the main beam and thus a reduced gain.

In the design analysis with Ansys software [29], the single choke ring naturally achieved high coupling with the center horn ring. At a 180-degree phase shift, the beam is out of phase with that of the main beam thus distorting the total beam formation and creating the secant beam pattern.

Figure 6 shows the initial design concept of the CRHA with ideal dimensions from [19]. The inner ring measured 14.75 mm at a flared opening of 31 degrees. The distance between the two rings was approximately λ/2, which provides a 180-degree phase shift. The 12.7 mm ≈ λ depth of the antenna also enabled the 180-degree phase shift of the choke to be realizable.

The E-field magnitude at a 150 mm exposure distance is also shown in Figure 7. The distribution is more uniform compared with [17,18]. However, the HPBW was observed at 54.4 degrees, which is less than the 60-degree requirement. Figure 7 shows the 3D radiation pattern using segmented FE-BI boundary condition in HFSS in which the radiating horn antenna is segmented from a 100 × 100 × 150 mm boundary box. The latter mimics an ideal boundary for the chamber system.

The inset farfield view at 28 GHz of the conical CRHA in Figure 7 shows an elliptical field distribution and a 45-degree tilt. A gain of 11 dB was realized with good E-H plane symmetry as shown in Figure 8. The 45-degree cut is also symmetric with a sharper secant slope. By minimizing the flare angle of the horn antenna and widening the choke ring, the elliptical field distribution was removed and a symmetric (E-H/45) farfield pattern is observed, as shown in the parameter sweep of Figure 9. From the sweep, the optimal symmetry was achieved at 8.76 mm ≈ WG-28 waveguide dimension (8.3 mm). The WG-28 waveguide will later be used as the coaxial to waveguide transition component. An optimal S11 of better than −35 dB is also achieved in Figure 10.

As the flared opening is reduced and approaches the dimensions of the center horn, the over-coupling between the center ring and the choke increases, thus increasing the secant angle and widening the beamwidth.

Although the cylindrical CRHA had a slightly reduced gain to 9.9 dB, as shown in Figure 11, the increased HPBW to 65.5 degrees was more desirable. This is an improvement from the conical CRHA and changes the conical choke horn into a cylindrical choke horn purposely for obtaining a wider beamwidth and secant pattern.

Before fabrication, the cylindrical CRHA model is modified with a standard flange (50 mm flange diameter) and screws for ease of mounting. The best feeding method adopted was with the use of a circular to rectangular waveguide adapter operating from 21.7 to 33.0 GHz. In addition, a rectangular waveguide to coaxial transition is attached to the adapter to complete the feeding of the horn antenna. This latter transition also operated within 26.4 to 40.1 GHz band with a VSWR of better than 1.15.

Figure 12 shows the optimized CAD drawings of the cylindrical CRHA and flange. The stem of the antenna is reduced from 19.3 mm (Figure 6) to 10.2 mm (Figure 12); since the accompanying transition and adapter offered a wavelength longer than one at 28 GHz, which is necessary for the cavity TEM mode transition. This length reduction in the horn antennas does not reduce the performance, provided the transition distance is in multiples of λ at 28 GHz. The optimized model in Figure 13 includes an electromagnetic slot, which was introduced for the suppression of the cross-polarization.

3.2. Cross-Polarization Suppression

The cross-polarization for most choke ring horn antennas is poor, as has been recorded by [11,12,13,14,15]. This is a result of the stronger coupling of the choke’s surface currents with that of the center horn. Thus, it is a natural trade-off for the highly sectoral beam patterns. As such, a novel approach is introduced to suppress the cross-polarization of an all-metal choke ring horn antenna.

The design concept was inspired by observing the surface current distribution within the cylindrical CRHA as shown in Figure 13a,b. At a 180-degree phase, the choke had strong surface currents which created a highly sectoral radiation beam pattern. This, however, reduced the cross-polarization as higher-order modes are generated within the choke. As can be observed, there was a strong current distribution converging at the slots. The slots, therefore, provided an “escape” or alternate route to loosen the coupling within the choke, thus suppressing the cross-polarization. The loosening effect was due to the altering of the phase by the slots, which distorted the 180-degree phase shift required for a sectoral pattern. Therefore, the cross-polarization sectoral pattern was heavily distorted.

Figure 12 shows the CRHA with electromagnetic (EM) slots for cross-polarization suppression. This has four circular holes at the base of the choke. They are offset from the edge of the antenna and are 90 degrees apart. Another transverse slot was made at the base of the choke ring perpendicular to the circular slots and had a tapered aperture.

Figure 14 and Figure 15 show the radiation patterns of the CRHA models with and without the transverse slots, respectively. The 4-transverse slot model achieved a cross-polarization reduction of more than 28.9 dB in the E-H plane and better than 10.5 dB in the 45-degree cut.

Table 2. Cross-pol. suppression performance with EM Slots at 28 GHz.

Model	Direction	Cpol. (dB)	XPD (dB)	HPBW (deg.)	S11 (dB)
No slot	E-plane	9.9	11.5	65.5	<−32
	H-plane	9.9	11.6	65.6
	45 deg.	10.2	11.6	65.9
2-slots	E-plane	9.6	23.7	70.7	<−28
	H-plane	9.5	23.4	67.4
	45 deg.	9.6	25.3	69.3
4-slots	E-plane	9.8	38.7	67.6	<−27.5
	H-plane	9.8	39.4	66.3
	45 deg.	9.8	38.3	67.2

details the comparison of the CRHA without EM slots, with two slots and with four slots. As can be seen, there is minor change in the impedance values across the frequency band of interest. There is, however, a slight broadening of the HPBW with the slots compared with no slots, proving that the EM slots are an efficient and low-profile method of suppressing cross-polarization in the choke ring horn antenna. This, however, also results in marginal gain drops of about 0.5 dB. The cross-polarization discrimination (XPD) at the boresight also shows excellent results for both the 2-slot and 4-slot models.

4. Fabrication and Measurement Discussion

The CAD model shown in Figure 12 was fabricated with braze (an alloy of copper and zinc) metal machining. The assembled antenna is shown in Figure 16. Braze material was used to give structural stability to the antenna and offer excellent radiation with anti-rust properties due to its copper and zinc content. Figure 16b also shows the cylindrical to rectangular waveguide adapter connected to the waveguide to coaxial transition mount used in the verification of the proposed radiator.

As can also be observed in Figure 16a, the fabricated brass was polished with a sanding finish. Sanding is an industrial process of blasting the milled metal with a special abrasive material to smoothen the edges and remove surface contaminants. At high frequencies of 28 GHz, small imperfections in the milling can create unwanted modes, especially in the choke cavity. This process changes the polished color of the brass from bright gold to a dull but smooth finish.

The CRHA was placed in an anechoic chamber for farfield measurement as depicted in Figure 17a. The S-parameters were also measured in the setup shown in Figure 17b. Both measurements were performed at the millimeter-wave chamber facility of the Korea Radio Promotion Association (RAPA) facility in Yongsan, Seoul, Republic of Korea.

Figure 18 shows the measured S11 data from the setup in Figure 17b, including the simulation results. The measured S11 between 27.5 GHz to 28.5 GHz was better than −20 dB S11. The ripples in the measured results are a normal impedance response from the included attachments for measuring setup which were well-captured by the large frequency sampling of the network analyzer. The two attachments as shown in Figure 16 are the cylindrical to rectangular waveguide transition and the waveguide/coaxial adapter. Also, the mismatch between the simulation and measurement can be attributed to the loss factor from the aforementioned attachments, both of which were not considered in the simulation setup. The impedance results were nonetheless satisfactory.

Figure 19, Figure 20 and Figure 21 show the farfield measurement and simulation results for 27.75 GHz, 28 GHz and 28.25 GHz, respectively. The simulation results were optimized values from the final CAD model shown in Figure 12.

At 28 GHz, the E-H plane was nearly symmetrical with a difference of about 1.4 degrees in HPBW, as seen in Figure 20. In all three frequencies, a 2.0 to 2.5 dB increase in gain is observed in the measured results compared to the simulation. This is theoretically about a 40% increase. A look at the radiation patterns shows conformal simulated and measured beam patterns, except for the increase in gain in the measured results.

As can be observed in

Table 3. Cross-pol. Supression performace with EM slots at 27.75 GHz, 28 GHz, and 28.25 GHz.

Freq. (GHz)	Direction		Cpol. (dBi)	XPD (dB)	HPBW (deg.)
27.75	E-plane	Sim.	9.4	29.4	81.3
		Meas.	11.6	27.6	63.4
	H-plane	Sim.	9.9	29.5	79.9
		Meas.	11.6	30.4	50.0
28.0	E-plane	Sim.	9.5	28.5	79.3
		Meas.	10.8	28.4	59.9
	H-plane	Sim.	9.8	29.2	79.3
		Meas.	11.0	31.2	58.5
28.25	E-plane	Sim.	9.8	28.7	76.0
		Meas.	11.0	28.7	73.0
	H-plane	Sim.	9.0	29.2	77.3
		Meas.	11.0	30.8	58.8

, the increase in gain has a corresponding decrease in the HPBW for the measured results. HPBW remained at a 60-degree average across the measured frequencies. The measured cross-polarization discrimination (XPD) also remained consistent with the simulation.

5. Bioelectromagnetic Dosimetry Implementation

As already noted in the introduction, the accuracy of the exposure chamber hinges heavily on the proper radiation field synthesis during the antenna design simulation and in the measurement. Using the measurement setup shown in Figure 22, an open-ended waveguide probe is positioned at different distances (depicting the exposure distance) from the CRHA. At a specific exposure distance, z, the probe is panned in the xy direction to collect the received power from the CRHA.

From the received power, the E-field power and the power density were obtained and synthesized in Mathworks Matlab [30]. This case is true for both nearfield and farfield exposure distances. It must be noted, however, that both field calculations follow slightly different assumptions.

The measured power density profiles (PDPs) are plotted in Figure 23a–d. For each curve, two horizontal curves are shown for −0.5 dB and −1.0 dB uniformity. The exposure uniformity levels are used to determine how much power is emitted on the SUT per unit surface of a given distance. A −0.5 dB (0.89 W of 1 W input power) uniformity denotes the area on the SUT under which about 89% of the exposed output power is illuminated. The illuminated area will also influence the exposure efficiencies shown in

Table 4. Measured field data summary at 28 GHz.

Exposure Distance (mm)	Spot Size −0.5 dB (mm)		Spot Size −1.0 dB (mm)		Peak Power Density/ob. Plane (mW/cm2)	Average Power Density/ob. Plane (mW/cm2)	Average Power Density/Spot Size (mW/cm2)		Z (Ω)	Peak E-Field (V/m)	Exposure Efficiency (%)
	E	H	E	H			−0.5 dB	−1.0 dB			−0.5 dB	−1.0 dB
50	16	14	26	18	73.89	32.98	68.28	64.01	376.56	527.48	41.45	81.18
80	26	18	38	26	34.17	21.50	31.73	29.50	376.83	358.85	40.24	78.99
100	34	22	54	34	22.58	16.18	20.91	19.04	376.89	291.74	42.38	94.75
150	48	34	76	50	10.27	7.78	9.48	8.71	376.95	196.77	41.92	89.73
200	65	45	95	65	5.69	3.56	5.26	4.89	376.97	146.41	41.34	80.80
250	86	58	126	82	3.56	2.17	3.30	3.07	376.99	115.90	44.61	85.84
300	108	68	153	98	2.47	1.25	2.29	2.13	376.99	96.41	44.94	85.73

.

From the exposure efficiencies, it is observed that more of the radiated power is within the −1.0 dB uniformity area (about 80% of radiated power), which has more than 70% exposure efficiency. Close to 50% of the power is also within the −0.5 dB uniformity level.

Except for 27.75 GHz, the other frequencies (28.0 and 28.25 GHz) have average PDs greater than 2 mW/cm² for a 300 mm exposure distance. It can, therefore, be concluded that the average power density per spot size at 2.0 mW/cm² is likely to be beyond 300 mm distance. The 2 mW/cm² value is an acceptable level of power density exposure on human tissue. This value was also used as a reference by [8].

The E-field magnitude as seen in Figure 24a has a consistent r² asymptotic curve. This result offers insight into the plane wave property of the CRHA as having a linearly decaying phase from the nearfield into the farfield region.

Also, from Figure 24a, the spot size is slightly asymmetric for both E-H planes. The slight dips in the values are because of insufficient exposure area in the measurement. This limitation is time-resource based, as one exposure distance measurement requires more than 3 h to collect sufficient data.

Figure 24b shows the exposure efficiency across the observable exposure distance. The slightly asymmetric E-H plane thus affected the exposure efficiency curves, especially for the −1.0 dB uniformity. A couple of fluctuations are also observed in the curve because of the insufficient exposure area. The average value of the exposure efficiency at −0.5 dB and −1.0 dB uniformity is consistent with that of [8].

The exposure efficiency reported in [8] is a simulated value and has an input power of 0 dB (1 W). With a 2-dB cable loss, the input power calculated was −2.0 dB (0.63 W). Thus, the efficiency calculations were lower than that of [8]. However, with a zero decibel input, the values of the measured exposure efficiency are higher than [8].

The peak power density (PD) and averaged power densities are also presented in Figure 24c. The results also show an r² decay, consistent with the E-magnitude curve in Figure 24a.

The impedance (Z) values in Figure 24c were recorded from the middle of the plane (0, 0, z). The impedance curve is also proportional to the exposure distance R. This confirms the assumptions made in [31,32,33] for the relationship between the impedance (Z) and the electric field magnitude (E) in obtaining the magnetic field density (H). That is, since an E-field open-ended probe was used in Figure 22, the H-field measured values were approximated from the E-field.

The first implementation of the proposed CRHA has been realized and is fully operational for dosimetry studies at the Electronics and Telecommunications Research Institute (ETRI), South Korea [10]. In this, a more detailed look into the compact 1 m × 1 m chamber is seen with the RF applicator shown in the inset photo overlooking a Petri dish holder. Under the Petri dish support is a probe and a positioner for xyz offsets. More details are shown in the communication from [10].

6. Conclusions

In this communication, a detailed outline of the design process for realizing a choke ring horn antenna have been presented for use as an RF applicator in a bioelectromagnetic dosimetry chamber. In line with the stringent electrical requirement to be met, the novel use of electromagnetic slots within the choke structure have been introduced to reduce the cross-polarization effect that is characteristic of metal-based choke ring horn antennas. This proposal achieved excellent results in both simulation and measurement, which have both been presented. Onward measurement and post-processing techniques have been presented to show the validity of the proposed RF applicator for dosimetry studies at 28 GHz.