Apparatus for controlled microwave exposure of aerosolized pathogens

Novel methods to inactivate potentially infectious aerosol-borne pathogens,^1,2 specifically those employing exposure to non-ionizing radiation,^3–8 have begun to receive greater focus from the research community largely due to the global pervasiveness of COVID-19. These aerosols can be generated by breathing,⁹ coughing,¹⁰ sneezing,¹¹ or talking.¹² Previous work involving inactivation of aerosolized pathogens through exposure to radio frequency (RF) electromagnetic radiation has typically been performed in the S-band at or around 2.45 GHz.^3–8 The apparatus used to perform these exposures has typically been comprised of a commercial microwave oven unit modified to include an aerosol flow tube that traverses the open cooking space within the oven.^3–8 While microwave ovens provide an inexpensive and easily acquired applicator for radio frequency (RF) exposure of aerosols, the complicated mode patterns within the cooking volume¹³ add substantial complexity in understanding the RF electric field dose incurred by the aerosol droplets.

To support future experiments designed to study microwave inactivation thresholds of airborne pathogens,^14,15 three apparatus configurations for exposing aerosols to controlled doses of microwave energy at frequencies ranging from the S-band (2 GHz–4 GHz) to the C-band (4 GHz–8 GHz) have been developed. Figure 1 provides a conceptual sketch of the apparatus. In the designed configuration, the RF is propagated through a single-mode waveguide. An aerosol flow tube, aligned along the center access of a single mode waveguide, allows movement of an aerosol stream along the direction of propagation of RF power within the waveguide. In single-mode operation, the electric field distribution within the waveguide can be readily calculated using available computational tools. Both the aerosol and the RF remain well-contained within the apparatus. Additionally, the apparatus itself can be made compact enough to fit within common bio-containment structures mandated for experimentation with aerosols containing pathogens requiring containment at elevated Bio-Safety Levels (BSL).

Four different frequencies (7.5 GHz, 5.6 GHz, 4.0 GHz, and 2.8 GHz), compatible with the three RF exposure apparatus configurations (based on WR-137, WR-187, and WR-284 waveguides) were selected for use in the future airborne pathogen inactivation experiments. These frequency selections were motivated by the need to utilize waveforms in the 2.45 GHz–8.2 GHz frequency range encompassing the previously published aerosol^3–8 and bulk fluid^16,17 pathogen inactivation studies.

Detailed electromagnetic analysis of the 7.5 GHz case (in the WR-137 RF exposure apparatus) will be provided in the main body of this paper, with major results of the other cases being summarized herein alongside the 7.5 GHz case. Detailed analyses of the three other cases (5.6 GHz in WR-187 GHz, 4.0 GHz in WR-187 GHz, and 2.8 GHz in WR-284) are provided in the supplementary material. A general description of the aerosol generation and collection system proposed for use with each apparatus is provided in  Appendix A.

It is important to comment on the potential range of mechanisms for inactivation of viruses. When reviewing the literature,^4,7,16–30 it becomes clear that there are a number of proposed inactivation mechanisms ranging from dipole coupling putting forces on the structure to electromagnetic-to-acoustic coupling to the virus structure. Additional mechanisms ranging from somewhat pedestrian (heating the surrounding medium) to more complex attacks such as impacting the hydration of the capsid structure or coupling to the DNA of the virus have been proposed. Despite this wide variety of mechanisms or perhaps because of this lack of consensus in the community, it is clear that the definitive mechanism for virus inactivation via electromagnetic fields remains an important goal of this research.

This lack of a definitive signature for the mechanism motivated our experimental design in a fundamental way. While our initial design for this experimental apparatus focused on driving mechanical resonances (Ref. 16), we decided to add the experimental range to the apparatus to handle a wide variety of mechanisms from heating and resonant coupling to the structure to simple forcing of the charge in the virus against the inertial mass of the surrounding fluid. Additionally, given the prevalence of Brillouin scattering as a diagnostic in the biomedical field, we also considered the potential that we have interaction of phonons with virus. In this, we recall that Brillouin scattering was first demonstrated in water.³¹ Thus, we scoped this experimental design to provide for significant variation in both peak and average electric field exposures due to the power handling capability of both the waveguide system and the electromagnetic sources used to drive the experiment. Additionally, we provide for a range of energy deposition due to our ability to vary the residence time for the droplets in the waveguide system by varying either the length of the system and/or the flow rate of the aerosol media. Finally, as discussed below, we allow our system to be deployed in different waveguides and different RF sources, thereby allowing experimental control over the frequency.

In summary, as we will demonstrate below, we have designed an experimental apparatus with the capability to present electromagnetic waves to an aerosol mixture of water and virus with the experimental controls to vary power, energy, and frequency. In this way, we believe that our experimental design is capable of a fundamental investigation of a wide variety of inactivation mechanisms. This range of capability is especially important given the wide variety of potential interaction mechanisms found in the literature. Specific estimates for a given mechanism will be presented in other manuscripts that will be published as this work proceeds.

Photos of the three RF exposure hardware configurations based on WR-137, WR-187, and WR-284 waveguides are provided in Figs. 2(a)–2(c), respectively. All configurations include the same general set of RF hardware subject to minor differences due to the size and availability of specific vendor versions of the components. Power is injected via a standard “right angle” coaxial line-to-waveguide RF input adapter. An in-line RF isolator is used to protect the RF source, a traveling wave tube (TWT), from damage due to reflections. The RF power then flows through an E-plane bend modified to include a port for the aerosol tube. A linear waveguide section, containing three directional couplers, allow for calibration and monitoring of the RF environment during aerosol exposures. Finally, a second modified E-plane bend allows for exit of the aerosol flow tube while routing the power to an RF load.

All three apparatus configurations were designed for use with 1.6 cm outer diameter (OD), 1.3 cm inner diameter (ID), and polycarbonate aerosol flow tubes. Glass or ceramic tubes may also be used, but the subsequent electromagnetic analysis of Sec. III would need to be reperformed using appropriate dielectric property data for the chosen material. The modified E-plane waveguide bends, a photograph of which is provided in Fig. 3, utilize a 1.7 cm ID copper tube as an RF choke (10.3 GHz cutoff frequency) to limit RF leakage during operation. This general class of waveguide E-plane bend with the feedthrough choke has been used successfully for optical³² and electron beam³³ feedthroughs in other high power RF systems and was evaluated to be a compact and low-risk option for use in the present application.

In order to quantify and describe the RF electric field exposure given to a volume of aerosol, the RF electric field environment must be calculated and correlated with a measurable value within the experiment, such as directional coupler power measurements. Frequency domain simulations, performed using CST Microwave Studio, were used to model the RF electric field environment within the aerosol flow tube during RF exposures.

Figure 4(a) depicts the geometry used for simulations of the 7.5 GHz, WR-137 RF exposure apparatus. For reference, the axial locations of the C1 (input-side forward power coupler), C2 (reverse power coupler), and C3 (load-side forward power coupler) in each apparatus depicted in Fig. 2, referenced to their respective simulation geometry, are provided in Table I. The linear waveguide section coaxial with the aerosol flow tube extends from z = 0 cm to z = 72.1 cm. The RF power (0.5 W, average) is injected from the left waveguide port, resulting in an average power flow, P_ave, (including both forward and backward wave contributions) of 476 mW at C1.

The polycarbonate aerosol flow tube within the WR-137 apparatus was modeled with a relative permittivity of 2.78 and a loss tangent (tan δ) of 0.0035, which is consistent with the range of values given in the literature for room temperature operation frequencies near the X-band.^34,35 Waveguide walls were modeled using the default CST loss model for copper. These simulations were performed under the assumption of a low density aerosol such that the change in bulk permittivity within the volume of the flow tube due to the presence of expected aerosol compositions is negligible for the purposes of calculating the RF electric field exposure of the bulk aerosol. An estimation of the aerosol dielectric properties is provided in  Appendix B.

A color map of the complex electric field vector magnitude (E₀) for locations in the x–z plane for an RF input of 0.5 W at 7.5 GHz is shown in Fig. 4(b). The presence of the waveguide discontinuities represented by the RF choke and penetration of the aerosol flow tube into the waveguide result in the formation of the standing wave pattern observable in Fig. 4(b). E₀ for locations along the axis of the aerosol flow tube is plotted in Fig. 4(c).

As is evident from the plot in Fig. 4(c), there is an overall reduction of the electric field amplitude as the RF propagates in the waveguide due to both ohmic loss in the waveguide walls and power dissipation in the polycarbonate tube. Performing an exponential fit of E₀ along the axis of the aerosol flow tube within the linear section of the waveguide yields an attenuation coefficient, α, of 0.108 Np/m. The rate of reduction in the electric field magnitude along the RF path is greatest in the 7.5 GHz WR-137 apparatus case and becomes less severe with decreasing frequency and increasing waveguide size (as comparable in Table II).

In situations where additional discontinuities may be present, such as in the experiment, the standing wave amplitude, evident in Figs. 4(c) and 5, may be substantially different than that calculated from the simulation data. Because the simulated RF apparatus does not include components such as the RF isolator, input couplers, and real RF loads, a process to estimate corrections to the predicted on-axis E₀ exposure range using measured experimental parameters (forward and reverse power in the vicinity of C1) is implemented.

The amplitude of the standing wave pattern can be bounded by curves E_0,max(z) and E_0,min(z), described by Eqs. (1) and (2), respectively, where P_fwd is the power carried by the forward-going wave at C1 and P_rev is the power carried by the reverse-going wave at C1. The attenuation coefficient, α, was previously calculated to be 0.108 Np/m⁻¹ [based on the exponential curve fit shown in Fig. 4(c)] and the constant, A, which will be determined later, is related to the waveguide impedance,

In the simulation, E_0,max and E_0,min have values of 1303 V/m and 1165 V/m, respectively, in the vicinity of C1. Using equivalent definitions of the voltage standing wave ratio [VSWR, Eq. (3)] and the average power value, P_ave, at C1 in the simulation (476 mW), the calculated values for VSWR, P_fwd, and P_rev are 1.12, 477 mW, and 2 mW, respectively. These results, in turn, allow us to solve for A using Eqs. (1) and (2) at the location of C1 (z = 8.9 cm)—A = 1803 V/m,

Having a complete description of E_0,max(z) and E_0,min(z), the described curves are plotted in Fig. 5. The lower bound of the on-axis E₀ exposure range is then defined as the value of E₀ at which E_0,min(z) and E₀(z) cross (1073 V/m). As noted in Fig. 5, the high electric field exposure region is defined as the axial length over which E₀(z) exceeds the lower bound of the on-axis E₀ exposure range (from z = −3.1 cm to z = 74.5 cm). Outside of this high electric field exposure region, RF exposures of the aerosol are ignored. Finally, the upper bound of the on-axis E₀ exposure range is defined as the value of E_0,max(z) at the lower bound of the high electric field exposure region (E_0,max (z) = 1321 V/m at z = −3.1).

It is noted that the defined procedure for finding the on-axis E₀ exposure range may slightly over- or underestimate the E₀ range bound values. For example, in the 7.5 GHz WR-137 apparatus case, the defined upper bound (1321 V/m) exceeds the actual maximum simulated value of 1315 V/m by ∼0.5%. Additional minor discrepancies may be introduced in the experiment by subtle differences in the polycarbonate dielectric properties and additional reflections caused by small perturbations including joints, couplers, and misalignment.

By assuming that P_ave at C1 is held fixed, the on-axis E₀ exposure range and high electric field exposure region length can be found for a range of VSWR values at C1 using Eqs. (1)(2)(3)(4). Figure 6 provides plots of the on-axis E₀ exposure range upper and lower bounds [Fig. 6(a)] and high electric field exposure region length [Fig. 6(b)] for aerosols traversing the WR-137 RF exposure apparatus, calculated for C1 VSWR values from 1.1 to 1.6, for a fixed P_ave value of 476 mW (at C1).

While all portions of the aerosol must traverse the full axial length of RF exposure apparatus, the individual droplets comprising the aerosol will be distributed across the cross-sectional area of the flow tube. Uncertainties in the range of E₀ exposure must be accounted for due to nonuniformities in E₀ in the x and y directions within the aerosol flow tube. Figures 7(a) and 7(b) show a cross section of the waveguide and aerosol flow tube and a color map plot of the E₀ in the x–y plane, respectively, at the location of the input-side forward power coupler, C1, for the WR-137 RF exposure apparatus (z = 8.9 cm).

Plots of relative change in E₀ (in %) moving along the x and y axes in the plane located at forward power coupler, C1, are provided in Figs. 8(a) and 8(b) (x-axis and y-axis, respectively). The changes in E₀ as a function of the position within the aerosol flow tube are greatest in the 7.5 GHz case due to both the relative volume of the flow tube polycarbonate walls (compared to the waveguide volume) and the comparatively smaller wavelength at 7.5 GHz with respect to the tube wall thickness. Including the maximum variation from the centerline E₀ values along the x- and y-axes (+13.5% and −12.2%, respectively), the on-axis E₀ exposure range experienced by aerosol is estimated to be 1073 V/m–1321 V/m. Thus, the full range of E₀ experienced by the aerosol when traveling through the primary RF exposure region of the simulated WR-137 apparatus is estimated to be 1073 V/m–1321 V/m +13.5%/−12.2%.

A summary of simulation results for each of the four frequency–waveguide combinations is provided in Table II. Scaling of predicted E₀ ranges for changes in C1 P_ave is performed using the following equation:

Values for the E₀₁ and P_ave1 values are taken from Table II.

High power testing of the three waveguide hardware setups (Fig. 2) at their respective frequencies was performed using a kW-class RF source system based on a traveling wave tube (TWT) amplifier. RF pulses with a 2 µs square wave envelope were used during high-power testing of each apparatus. The required waveform was readily generated using the RF source configuration represented by the schematic provided in Fig. 9.

Following the schematic depiction in Fig. 9, a continuous wave (CW) RF input signal at the desired frequency is generated by using the local oscillator (Hewlett Packard 83620A). The RF input signal is blocked by the switch (American Microwave Corporation SWNND-2184-1DT) except when a voltage pulse is supplied by the trigger generator (Stanford Research Systems DG535) to gate open the switch. For the high frequency RF operation (7.5 GHz and 5.6 GHz), the gated RF input signal is amplified by an Applied Systems Engineering MODEL 147/C TWT amplifier (4–8 GHz). For the low frequency RF operation (4.0 GHz and 2.8 GHz), the gated RF input signal is amplified by an Applied Systems Engineering 147/S TWT amplifier (2 GHz –4 GHz).

Raw signals from high power testing of the WR-137 RF exposure apparatus at 7.5 GHz from the C1, C2, and C3 couplers were sampled using a (Tektronix DPO71254) oscilloscope (after being further attenuated to protect the oscilloscope). Coupler calibration and attenuator measurement data are provided in the supplementary material. A plot of the C1, C2, and C3 raw signals is presented in Figs. 10(a)–10(c), respectively. Recall that locations for these couplers within each apparatus are provided in Table I and that the C1 and C3 signals (plotted in Fig. 10) correspond to input side and load side forward power flow measurements, respectively, while the C2 signal corresponds to the input-side reverse power flow measurement.

RMS voltages calculated for the portions of the measured signal consisting of the amplified output signal + noise and for the noise with no amplified output signal are noted in Fig. 10. A FFT magnitude plot for the C1 raw signal is provided in Fig. 11. Intermodulation peaks, expected from the output of TWTs, are observed at amplitudes ∼20 dB lower than the amplified input signal peak.

Power values provided in Table III are calculated from the square of the C1, C2, and C3 signal V_RMS values divided by the 50 Ω channel impedance of the oscilloscope, after which corrections for the coupler attenuation and in-line attenuator measured values (from the Appendix of the supplementary material) were then applied. Average power flow (P_ave) values at C1 for each frequency are calculated by subtracting the magnitude of the reverse power (P_rev) measured at the C2 directional coupler (adjacent to C1) from the forward power (P_fwd) and that measured at the C1 directional coupler. Calculation of VSWR at C1 is performed using the C1 P_fwd and C2 P_rev values from Table III in Eq. (3).

Comparing the VSWR values calculated from the experimental data (Table III) with those calculated from the simulations (Table II), the experimental VSWR values exceed those of the simulations by amounts greater than those that can be accounted for by the calculated error for the 7.5 GHz, 4.0 GHz, and 2.8 GHz cases. These differences are not unexpected as imperfections in the experimental hardware geometry (introduced during manufacturing and modification of the waveguide components), impedance mismatches in RF loads, and non-ideal material properties not captured in the model may all contribute to increased reflections, which lead to more severe standing wave patterns within the apparatus.

By correcting for the upper and lower bounds of the E₀ exposure range using the process described in Sec. III and scaling the results using Eq. (5), the expected E₀ exposure ranges can be calculated for C1 average power values from 100 mW to 1 MW provided in Table IV. These data are plotted for the 7.5 GHz case in Fig. 12. Analogous plots for the remaining waveguide-frequency combinations are provided as the supplementary material. For the C1 P_ave value for 7.5 GHz in Table III, the predicted E₀ exposure range is 45.8–71.6 kV/m +13.5%/−12.2%.

A set of three apparatus enabling RF exposure of aerosolized pathogens at four chosen frequencies (2.8 GHz, 4.0 GHz, 5.6 GHz, and 7.5 GHz) has been designed, simulated, fabricated, and tested. Each apparatus was intended to operate at high power without leakage of RF into the local environment and to be compact enough to fit entirely within biocontainment enclosures required for elevated Bio-Safety Levels (BSL). Figure 13 shows a photograph of the largest of the three RF exposure apparatus (WR-284) during setup within a SterilGARD III® Advance class II biosafety cabinet (The Baker Company, Sanford, ME).

A combination of simulations and analytic scaling was used to calculate predictions for the range of RF electric field exposure, represented by the complex electric field vector magnitude (E₀), that an aerosol stream would be expected to encounter while passing through the apparatus. These E₀ ranges were correlated with the experimentally measurable values from power couplers built into each apparatus. The hardware and analysis described herein will be used to enable upcoming experiments designed to further understand inactivation thresholds of aerosolized pathogens under microwave irradiation.

See the supplementary material for the data and analysis for the WR-187 RF exposure apparatus (at 5.6 GHz and 4.0 GHz) and for the WR-284 RF exposure apparatus (at 2.8 GHz) provided online.

B.W.H., J.W.M., and Z.W.C. contributed equally to this work.

The data that support the findings of this study are available within the article and its supplementary material.

This work was funded by the Air Force Research Laboratory and the Air Force Office of Scientific Research portfolio for Aerospace Materials for Extreme Environments under AFOSR (Grant No. LRIR 20RDCOR022). The authors thank the VIDER Senior Leadership Panel (K. Hammett, K. Geiss, S. Welsh, D. Shiffler, R. Naik, P. Roach, S. Miller, and M. Ewy) for providing guidance and the other VIDER team leads (B. Ibey, J. Payne, R. Thomas, and I. Echchgadda) for helpful technical discussions. The authors acknowledge additional helpful discussions with D. French, P. Mardahl, J. MagGillivray, B. Jawdat, B. R. Jayan, and P. Kuehl. Cleared for public release (Reference No. OPS-20-39011, Case No. AFRL-2020-0138).

A schematic of the aerosol system designed for use with the previously described RF exposure apparatus is provided in Fig. 14. In this configuration, the fluid with the suspended pathogen is located within the reservoir of a three-jet Collison nebulizer³⁶ (CH Technologies, Westwood, NJ, USA), “Collison Nebulizer” in Fig. 14. Filtered compressed air is used to drive the nebulization process and to provide the volumetric flow through the aerosol flow tube. The aerosol flow tube transports the aerosol through the RF exposure apparatus.

At the collection side of the system (right-hand side of the figure), the aerosol flow tube connects with a second compressed air stream as it is directed to the input of the biosampler (SKC BioSampler,³⁷ SKC, Inc., 84, PA, USA). The second nebulizer (Heart Continuous Medication Nebulizer, Westmed, Inc., Tuscon, AZ), “Heart Nebulizer” in Fig. 14, is optional and contains purified (no pathogen) water used to humidify the additional air flow sent to the biosampler. The added humidity has been observed to reduce the rate of buildup on the biosampler nozzles (compared to dry air) when sampling aerosols generated from viral growth media. The biosampler initially contains a reservoir of the purified (no-pathogen) fluid that begins to accumulate pathogen as the aerosol stream is merged into it.

One common cell culture media used for the production of viral pathogens is Minimum Essential Medium (MEM) (Life Technologies Corporation, Carlsbad, CA). The commercial formulation is based on the culture media originally described by Eagle³⁸ and generally consists of aqueous inorganic salts with lesser amounts of amino acids and vitamin compounds. The Minimum Essential Medium (MEM) is the primary constituent of the fluid droplets in the generated aerosol and would, therefore, be the primary contributor to differences in complex permittivity of the bulk aerosol, compared to air.

Measurements of the bulk MEM were performed using the open ended coaxial probe technique³⁹ with an Agilent N5245A PNA-X and a 85070E dielectric probe kit with a high temperature-capable probe. Real and imaginary components of the frequency dependent relative permittivity measured for the Gibco MEM are plotted in Fig. 15.

Estimation of the complex permittivity of the MEM aerosol is performed using the Maxwell Garnet effective medium approximation⁴⁰ [Eq. (B1)]. Here, $ϵ$_m (the background matrix material) is air, ϵ_i (the inclusion material) is the MEM, and ϵ_eff is the effective permittivity of the bulk aerosol. As shown in Eqs. (B2), (B3), and (B4), ϵ_m, ϵ_i, and ϵ_f, may all be complex, although, in the present situation, ϵ_m, being air, is assumed to be equal to ϵ₀, the permittivity of free space. The constant F is the volume fraction of inclusion to matrix constituents of the aerosol. The constant α is the loss coefficient [defined by Eq. (B5)], f is the RF frequency, and μ₀ is the permeability of free space. A subscript “r” denotes a relative value (e.g., relative permittivity),

As described by May,³⁶ when using fluids with viscosity values similar to water (which, for the purposes of this estimation, MEM is assumed to be), Collison nebulizers (as described in  Appendix A and depicted in Fig. 14) generate aerosols with a fluid-to-air volume fraction of F = 10⁻⁵. Calculated effective relative permittivity values and resulting attenuation coefficients at each of the operating frequencies are provided in Table V. The authors note that measurement uncertainties indicated in Fig. 15 were neglected in the calculations of the values in Table V.

Effective permittivity values in Table V are very close to that of free space, and the calculated attenuation coefficients are multiple orders of magnitude below those calculated from the simulations due to only the effects of the aerosol flow tube material provided in Table II. The authors, thus, conclude that it is reasonable to assume that the change in bulk permittivity within the volume of the flow tube due to the presence of expected aerosol compositions is negligible for the purposes of calculating the RF electric field exposure of the bulk aerosol within each RF exposure apparatus.