5G has been designed for blazing fast and low-latency communications. To do so, mm-wave frequencies were adopted and allowed unprecedently high radiated power densities by the FCC. Unknowingly, the architects of 5G have, thereby, created a wireless power grid capable of powering devices at ranges far exceeding the capabilities of any existing technologies. However, this potential could only be realized if a fundamental trade-off in wireless energy harvesting could be circumvented. Here, we propose a solution that breaks the usual paradigm, imprisoned in the trade-off between rectenna angular coverage and turn-on sensitivity. The concept relies on the implementation of a Rotman lens between the antennas and the rectifiers. The printed, flexible mm-wave lens allows robust and bending-resilient operation over more than 20 GHz of gain and angular bandwidths. Antenna sub-arrays, rectifiers and DC combiners are then added to the structure to demonstrate its combination of large angular coverage and turn-on sensitivity—in both planar and bent conditions—and a harvesting ability up to a distance of 2.83 m in its current configuration and exceeding 180 m using state-of-the-art rectifiers enabling the harvesting of several μW of DC power (around 6 μW at 180 m with 75 dBm EIRP).
Our era is witnessing a rapid development in the field of millimeter-wave (mm-wave) and Internet of Things (IoT) technologies with a projected 40 billion IoT devices to be installed by 20251. This could result in a huge number of batteries needing to be continuously charged and replaced. The design and realization of energy-autonomous, self-powered systems: the perpetual IoT, is therefore highly desirable. One potential way of satisfying these goals is through electromagnetic energy harvesting. A powerful source for electromagnetic scavenging is mm-wave energy, present in the fifth-generation (5G) of mobile communications bands (above 24 GHz), where the limits of allowable transmitted Effective Isotropic Radiated Power (EIRP) by the Federal Communications Commission (FCC) regulations are pushed beyond (reaches 75 dBm) that of their lower-frequency counterparts. Following the path loss model defined by the 3rd Generation Partnership Project Technical Report 3GPP TR 38.901 (release 16) in outdoor Urban Macro Line of Sight conditions (UMa LOS), the power density expected to be received at 28 GHz for a transmitted power of 75 dBm EIRP is 28 μW cm−2 at a distance of 100 m away from the transmitter. This demonstrates the ability of 5G to create a usable network of wireless power. In addition to the advantage of high transmitted power available at 5G, moving to mm-wave bands allows the realization of modular antennas arrays instead of single elements, thereby allowing a fine scaling of their antenna aperture, which can more than compensate for the high path loss at these frequencies through the addition of extremely-large gains2. However, one limitation accompanies large gain antennas: their inability to provide a large angular coverage. As the relative orientations of the sources and harvesters are generally unknown, the use of large aperture mm-wave harvesters may seem limiting and impossible. Individual rectennas, constituted of small antenna elements, can realistically be DC combined. However, this approach does not increase the turn-on sensitivity (lowest turn-on power) of the overall rectenna system: RF combination is needed.
Beamforming networks (BFNs) are used to effectively create simultaneous beam angular coverage with large-gain arrays, by mapping a set of directions to a set of feeding ports. An important class of these multiple networks is the microwave passive BFN that has been widely used in switched-beam antenna systems and applications. Hybrid combination techniques, based on Butler matrix networks, have been used in previous works for energy harvesting at lower frequencies3,4,—more specifically at 2.45 GHz—to achieve wider angular coverage harvesting. However, these Ultra-High Frequency (UHF) arrays are impractically large for IoT applications and the implementation of their Butler matrices at higher frequencies would necessitate costly high-resolution fabrication. While sub-optimal—because of its large size—in the UHF band, the Rotman lens becomes the BFN of choice in the realm of mm-wave energy harvesting. Compared to their lower frequencies counterpart, fewer implementations are presented in the literature targeting energy harvesting at higher frequencies, more specifically 24 GHz and above. However, these systems later displayed in the table of comparison5,6,7, suffer from a narrow angular coverage.
In this paper, the authors demonstrate a full implementation of an entirely flexible, bending-resilient and simultaneously high gain and large angular coverage system for 5G/mm-wave energy harvesting based on a Rotman lens. For IoT applications, there is a benefit to making extremely low-profile devices that can conformally fit onto any surface in the environment such as walls, bodies, vehicles, etc. Therefore, thanks to the use of mm-waves, antennas with such features can be readily designed and fabricated. A Rotman lens-based rectenna has been first proposed in8, where a preliminary prototype and approach were presented, resulting in a quasi-flexible system, 80° angular coverage and 21-fold increase in the harvested power compared to a non-Rotman-based system. Here, the previously-predicted potential of 5G-powered nodes for the IoT and long-range passive mm-wave Radio Frequency IDentification (RFID) devices, is further taken advantage of, and effectively demonstrated. In order to do so, a thorough analysis of the lens itself—a structure that was not revealed in8—is first presented, exposing its key design parameters and resulting measured broadband behavior tested in both planar and bent conditions over more than 20 GHz of bandwidth. In addition, a scalability study of the approach, outlining the optimal size of such a system is reported, thereby proving the extent of the capability of providing a combination of good array factor and wide beam coverage. The novelty of this system also lies in the realization of a fully-flexible 28 GHz Rotman-lens-based rectenna system, completed by the design of a new DC combiner on a flexible 125 μm-thin polyimide Kapton substrate. The new DC combiner uses a reduced number of bypass diodes and increases the angular coverage of the system by more than 30% compared to8. Furthermore, the frequency-broadband behavior enabled by the use of the Rotman lens makes the full rectenna system bending-resilient, a property now demonstrated through its characterizations in flexing and conformally-mounted configurations. Finally, the system’s potential for long-range mm-wave harvesting is expressed for the first time, by reporting an unprecedented harvesting range of 2.83 m.
Experiments, results and discussions
Rotman lens scalability study for harvesting applications
The Rotman lens, introduced in the 1960s, constitutes one of the most common and cost-effective designs for BFNs and is commonly utilized to enable multibeam phased array system9 and wide-band operation, thanks to its implementation of true-time-delays10. By properly tuning the shape of the lens according to the geometrical optics approximation with the goal of focalizing plane waves impinging on the antenna side of the lens to different focal points on the beam-ports side of the lens, one achieves a lens-shaped structure with two angles of curvatures: one on the beam-ports side, and the other on the antenna side11. Because the lens is capable of focusing the energy coming from a given direction into its geometrically-associated beam port, the proposed scheme loads each of these ports with a rectifier, thereby channeling the energy coming from any direction to one of the rectifiers as shown in Fig. 1a. This subsection investigates the effect of varying the number of antenna ports Na and beam ports Nb in the Rotman lens on its maximum array factor and angular coverage. The (Na, Nb) set, resulting in the best combination, will define the Rotman lens design parameters used for this work. Structures of varying sizes were designed using Antenna Magus and identical material parameters (substrate, conductors) as the ones of the presented design, before being simulated in CST STUDIO SUITE 2018. The simulated data was then processed in MATLAB to output the array factors created by the respective lens structures using a modified version of Eq. (1)12, presented next in Eq. (2):
where AF, n, Na, k, d, and are, respectively, the lossless array factor, the antenna number, the total number of antenna ports, the wave vector, the spacing between the elements, the direction of radiation and the difference in phase excitation between the elements. Since this formula describes a lossless array with a single antenna port, we introduced the following equation that takes into account the losses induced by the feeding network as well as the introduction of multiple feeding ports.
where and are, respectively, the array factor for beam port j and the S parameters between antenna ports n and beam ports j. The maximum value of the array factors as well as their total (accounting for the aggregated coverage of all ports) 3 dB beamwidths where then tabulated. The five simulated lenses had the following (Na, Nb) combinations: (4,3), (8,6) representing the system implemented in this work, (16,12), (32,24) and (64,48). Figure 1b shows the increase in the array factor until reaching a peak of around 7.8 dB for a lens surrounded by 16 antennas and 12 beam ports, after which the array factor starts dropping, down to approximately 5.2 dB for a 64 antennas structure with 48 beam ports. The array factor reduction is explained by the increased losses within the lens accompanied by the increase of complexity and internal reflections, as the lens grows in electrical size. The same plot shows the decrease in angular coverage from 180° with 4 antennas down to 80° with 64 antennas. This study shows that the combination composed of eight antennas and six beam ports, offers a nearly optimal compromise, with these materials, between a high array factor of 5.95 dB and a 120° total angular coverage, while maintaining a reasonable number of antennas and beam ports. It should be noted that the choice of the number of beam ports is related to the 3dB-beamwidth of the individual antennas, the reason for which will be detailed later.
Flexible broadband Rotman lens design
After setting the number of antenna ports and beam ports, the design was printed on flexible copper-clad Liquid Crystal Polymer (LCP) substrate ( and ) using an inkjet-printed masking technique followed by etching, resulting in the structure shown in Fig. 1c. It should be noted that the use of impedance-matched dummy ports is common with Rotman lenses13,14,15,16. Nevertheless, the goal in the implementation hereby described is not (as is usually the case) the generation of clean beam patterns with low side-lobe levels. Here, the lens’ properties are used for harvesting. Consequently, as long as the presence of the side lobes does not significantly interfere with the level of the array factor at broadside, side lobes are of no concern. Such a structure, including eight antenna ports and six beam ports—and, therefore, six radiating directions—was designed, simulated, and tuned. The structure, shown in Fig. 1c, with the antenna ports connected to matched loads, was then tested in planar and bent configurations—cylinders with different bending radii ranging from 1.5 to 2.5 in. radii—to assess the effect of bending on the S parameters behavior. Figure 2a shows the measured reflection coefficient of the Rotman lens at beam port 4 for four different scenarios, in comparison with the simulated structure in a planar position. The results reveal the Rotman lens’ ability to be mounted on curved surfaces down to a radius R = 1.5″, while maintaining a stable matching and minuscule losses compared to being held in a planar position.
The gain and angular bandwidths of this structure—defined by the frequency range in which the maximum array factor and angular direction per beam are stable within 3 dB and 5° respectively,—are studied next. The ultimate assessment of these properties involves calculating the beams’ magnitude and angular directions over a wide range of frequencies17, in order to ascertain their stability or lack thereof. For this purpose, the maximum array factors were calculated and the beams’ angular directions were extracted and plotted in Fig. 2b for the first, third and fifth beam ports, P1, P3 and P5, representing the edge, secondary and central beams in this symmetrical structure. These plots prove the unique capabilities offered by the Rotman lens; although the Rotman lens is designed at a specific frequency—28 GHz in this work—this analysis proves that both the magnitude and the angular direction of the beams remain relatively stable over a very wide frequency range. In Fig. 2b, three plots refer to the maximum array factors of the three beam ports, where minor fluctuations between 4 and 7 dB are observed over the range from 10 to 43 GHz for ports P3 and P5 and similar fluctuations over a fairly reduced frequency range for the extreme edge beam P1. On the same graph, three plots present the angular direction’s stability of P1, P3 and P5 beams, where P3 (in particular) preserves its angular direction over 33 GHz of bandwidth. The lens’ angular coverage resides between ports 1 and 6 and can be extracted from Fig. 2b. Knowing that the structure is symmetrical and that beam port P1 is at around , the overall structure covers an angle larger than 100° in front of the lens, a result further detailed in the next subsection. It should be noted that such a beamwidth is maintained over a large angular bandwidth exceeding 20 GHz, as shown in Fig. 2b. This study demonstrates the stability and robustness of a low-cost, printed and flexible mm-wave Rotman lens structure, tested with respect to bending and frequency, and supports the choice of such an architecture at the heart of the harvesting system proposed in this work.
Flexible, high-gain and wide-angular-coverage mm-wave Rotman-lens-based antenna array
Eight of the linear antenna sub-arrays introduced in8 were then added to the antenna ports of the array, and its beam-ports were extended by microstrip lines to enable their connection to end-launch connectors. The antenna sub-array consists of five serially-fed patch antenna elements, providing an operation centered at 28.55 GHz with a reflection coefficient lower than dB within this range. Their E-plane beamwidth of about (provided by the five antennas) is appropriate for most use cases, where environments expand mostly horizontally. Its simulations showed a gain of 13 dBi and a H-plane beamwidth of 80° in the plane perpendicular to the linear array. In this configuration, six beams were chosen to intersect at angles providing 3dB lower gain than broadside. Eight antennas provide a 3dB-beamwidth of 15°, which covers a total of in front of the array. The design was then also printed on flexible LCP substrate, resulting in the structure shown in Fig. 3a, mounted on a 1.5″ radius cylinder. The radiation properties of the lens-based antenna system were simulated using the time-domain solver of CST STUDIO SUITE 2018, resulting in the six gain plots shown in Fig. 3b. The gain of the Rotman lens at every port was also accurately measured using a 20 dBi transmitter horn antenna and by terminating all five remaining ports with a load for every port measurement to guarantee the proper operation of the lens. Both simulated and measured radiation patterns (shown in Fig. 3b) display a remarkable similarity with a measured gain of approximately 17 dBi, and an angular coverage of around 110°, thereby validating the operation of the antenna array. The gains on the first three ports were also measured for the bent structure over a curvature of 1.5″ radius, shown in Fig. 3a and compared to the measured results on a planar surface. The previous subsection in addition to previous works18,19 have demonstrated that the performance of the Rotman lens is not deteriorated by wrapping or folding the structure compared to its conventional planar counterpart. However, after adding the antenna arrays, bending the structure can indeed have effects on its phase response, especially if the structure is large and the bending is severe. Figure 3c shows the gains of P1, P2 and P3 for the two scenarios (three ports only because the structure is symmetrical), demonstrating again the ability of the lens in maintaining a stable gain (especially over the center beams) upon bending. The beam located at the edge, however, suffers additional deterioration in received power under bending, because of the shift of the source away from the broadside of the bent antenna arrays.
Fully-flexible 28 GHz Rotman lens-based system
In this section, the fully-flexible rectenna system—based on the Rotman lens and a new DC combiner network—is presented. This architecture, shown in Fig. 4a, consists of a series of eight antenna sub-arrays attached to the Rotman lens from one side, facing six rectifiers at the opposite side where DC serial combination is implemented. The basic rectenna elements, that are the antenna and the rectifier, are presented in details in8. The diode used in this work is the MA4E2038 Schottky barrier diode from Macom. The Rotman-based rectenna was first characterized as a function of its received power density. The system was positioned at a specific harvesting angle (approximately ) and illuminated with a horn antenna with a gain of 20 dBi, placed at a distance of 52 cm away from the rectenna array, within the far field region starting at 23 cm, and outputting powers ranging from 18 to 25 dBm, corresponding to an RF input power sweep from around − 9 dBm to − 2 dBm. The array was loaded with its optimal load impedance of 1 , corresponding to the optimal load of a single rectifier—since only one rectifier will be “ON” at a time, given that the Rotman lens focalizes all the power to one beam port depending on the direction of the incoming wave—as detailed earlier. The results of this experiment are shown in Fig. 4b, where the harvested voltages and powers of the array are shown. It can be observed that, at low powers, the Rotman-based rectenna effortlessly produces an output. The Rotman-based rectenna turns on well below − 6 dBm cm−2, which compares quite favorably to the literature6. The output voltage of the rectenna was also measured over its operating frequency range. Like in the first experiment, the system was positioned at the same harvesting angle, at a range of 25 cm away from the source’s horn antenna. The output voltages under open load conditions were recorded and plotted, as shown in Fig. 4c for the Rotman lens-based rectenna, for , and incident power densities. The plots present a wide frequency coverage—from 27.8 to 29.6 GHz.
Flexible DC combining network
Power summation is very critical when it comes to the unbalanced rectification outputs produced from realistic RF sources, and can be implemented differently depending on its costs and benefits20.
This paper does not rely on a direct voltage summation topology (i.e. back-to-back RF diodes); however, it introduces a minimalist architecture relying on a total of bypass diodes, where N is the number of RF or rectifying diodes. Equipped with a low turn-on voltage of 0.1 V, the Toshiba 1SS384TE85LF bypass diodes used in the DC combiner design create a low resistance current path around all other rectifiers that received very low or close to zero RF power. This topology is optimal when only one diode is turned on, which can be assumed if a single, dominant source of power irradiates this particular design from a given direction. This new combiner circuit is shown in the schematic of Fig. 5a. This simplified schematic—shown for four rectifying diodes—uses different colors to highlight the paths that the current will take for every case where an RF diode turning “ON” while the serially-connected diodes are “OFF”. This DC combiner was then fabricated on a flexible -thin polyimide Kapton substrate and connected to the Rotman lens-based rectenna through a series of single connectors to make the entire system fully flexible and bendable. The harvested power under a load of 1 versus the angle of incidence of the mm-wave energy source for the Rotman-lens-based rectenna is compared for both rigid (presented in8, and relying on bypass diodes) and flexible new DC combiners. For this experiment, a horn transmitter antenna was used to send 25 dBm of RF power at 28.5 GHz to the lens placed 70 cm away, as shown in Fig. 5b, while the array was precisely rotated in angular increments of 5°. Figure 6a shows that the new DC combiner, with a reduced number of diodes, was able to provide a complete angular coverage of almost 110° over the entire lens spectrum as presented in Fig. 3b, thus solving the voltage nulling occurring at the first and last ports, using the rigid DC combiner adopted previously in8. The new DC combiner offers therefore, an increase of more than 30% in the system’s spatial angular in addition to enabling a fully-bendable structure due to the unique fabrication on flexible Kapton substrate and connection to the rectenna using individual interconnects.
As mentioned earlier, the DC combiner is mainly used with the Rotman-lens-based rectenna to automatically direct the active rectifier’s output to a single DC common port, independent of which port this might be. An alternative to the DC combiner in the Rotman lens-based system, would be to manually connect to the active port if the location of the source were known. To study the effect of the implemented DC combiner on the turn-on sensitivity of the system, the output voltage of the rectenna was measured for a specific source location with and without the combiner over a range of RF transmitted power and load variations; the direction was chosen such that the non-DC-combined rectifier would output its maximum power. Figure 6b shows eight different plots where three of them represent the harvested power with a direct connection to the active rectifier for 1 , 10 and 100 conditions. Plotted with the same colors are the other three, representing the harvested power with the addition of the DC combiner for the same load values. The last two plots display the measured voltages with and without the combiner under open load conditions. The rectenna was placed 61 cm away from the transmitter horn antenna and the power was swept from 10 to 25 dBm. The results show the performance superiority in all considered load conditions when the contact is made directly to the rectifier and not through the DC combiner. The lens-based system is able to achieve a turn-on power as low as in this case. This behavior is explained by the voltage drop introduced by the bypass diodes present in the combiner—that consistently decrease the expected output voltage by 0.1 to 0.2 V—when one or two diodes are, respectively, added to the current path. The variation of load values also shows that the rectenna can achieve better efficiencies at lower loads. More importantly, the reduction in the turn-on sensitivity—the minimum power density required output 10 mV—induced by the combiner is only of about 2 dB in loaded conditions, while the combiner enables an increase in the angular coverage of the rectenna system from about 18° to 110°. The remarkable angular and high-power turn-on sensitivity offered by the Rotman-lens-based rectenna are finally benchmarked using the following table for comparison with several state-of-the-art works, as presented in literature. In Table 1, the striking performance of the proposed system is displayed, highlighted by its flexibility and ability of achieving an angular coverage as large as 110° at extremely high turn-on sensitivity, thereby allowing mm-wave long-range harvesting in ad-hoc and conformal-mounting implementations.
Rectenna system performance under bending
This section displays the operation of the Rotman-lens-based system under different bending scenarios. This and previous work18,19 show that the lens is able to maintain an efficient electromagnetic energy distribution across the output ports under convex and concave flexing conditions. The lens-based rectenna was placed on cylinders with different curvatures, 70 cm away from the transmitter sending 25 dBm of power at 28.5 GHz, as shown on Fig. 7a. The voltage was collected using a load of 1 for the planar and three bent conditions and plotted in Fig. 7b with respect to the source’s angle of incidence. The graph shows an unprecedented consistency and stability in the system’s scavenging and rectification abilities, knowing that several sub-systems are exposed to warping and the pressures of bending: the antenna sub-arrays, the Rotman lens and the rectifiers. Slight attenuation can be observed at the edges, but the system otherwise performs unimpeded by the bending. This remarkable property qualifies this system as a perfect candidate for use in wearables, smart phones and ubiquitous, conformal 5G energy harvesters for IoT nodes.
As described earlier, one of the main appeals of the proposed approach is its ability to use the high EIRPs allowed for 5G base-stations while guaranteeing an extended beam angular coverage, which is a necessary feature for ad-hoc ubiquitous harvesting implementations. In order to demonstrate the lens based-rectenna for longer-distance harvesting and detect that maximum range, a high-performance antenna system—comprised of a 19 dBi conical horn antenna and a 300 mm-diameter PTFE dielectric lens (for high directivity) providing an additional 10 dB of gain—was used as shown in Fig. 7c. With a transmitted power of 25 dBm (and an associated EIRP of approximately 54 dBm), corresponding to an incident power density of approximately − 6 dBm cm−2, the lens-based rectenna displayed an extended range of 2.83 m under open load conditions, with an output voltage around 10 mV, thereby demonstrating (to our knowledge) the longest-ranging rectenna demonstration at mm-wave frequencies. With a transmitter emitting the allowable 75 dBm EIRP, the theoretical maximum reading range of this rectenna could extend to 16 m. In addition, the use of advanced diodes—designed for applications within the 5G bands and enabling rectifiers’ sensitivities similar to that common at lower (UHF) frequencies—are showing a potential path towards achieving a turn-on sensitivity of the rectifiers as low as − 30 dBm21,22. If this were practically applied to the Rotman lens system presented in this work, the harvesting range could be extended beyond 180 m (where the received power density for a transmitted power of 75 dBm is ), which is only slightly smaller than the recommended cell size of 5G networks23. This observation enables the striking idea that future 5G networks could be used not only for tremendously-rapid communications, but also as a ubiquitous wireless power grid for IoT devices.
Through the use of the Rotman lens, this paper demonstrates that the usual paradigm constrained by the (often considered fundamental) trade-off between the angular coverage and the turn-on sensitivity of a wireless harvesting system can be broken. Using the reported architecture, one can design and fabricate flexible mm-wave harvesters that can cover wide areas of space while being electrically large and benefit from the associated improvements in link budget (from source to harvester) and, more importantly, turn-on sensitivity. The approach has been shown, however, to only be scalable up to the degree where the additional incremental losses introduced by the growing lens counterbalance the increase in the aperture of the rectenna. Nevertheless, this inflection point only appears (in the particular context considered in this paper) after the arraying of 16 elements, or up to a scale of . In the 5G Frequency Range 2 (FR2), this translates to harvesters of 4.5 cm to 9.6 cm in size, which are perfectly suited for wearable and ubiquitous IoT implementations. With the advent of 5G networks and their associated high allowed EIRPs and the availability of diodes with high turn-on sensitivities at 5G frequencies, several of DC power (around 6 with 75 dBm EIRP) can be harvested at 180 m. Such properties may trigger the emergence of 5G-powered nodes for the IoT and, combined with the long-range capabilities of mm-wave ultra-low-power backscatterers24, of long-range passive mm-wave RFIDs.
This work was supported by the Air Force Research Laboratory and the NSF-EFRI. The work was performed in part at the Georgia Tech Institute for Electronics and Nanotechnology, a member of the National Nanotechnology Coordinated Infrastructure (NNCI), which is supported by the National Science Foundation (Grant ECCS-1542174).
The authors declare no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
About this article
Cite this article
Eid, A., Hester, J.G.D. & Tentzeris, M.M. 5G as a wireless power grid. Sci Rep 11, 636 (2021). https://doi.org/10.1038/s41598-020-79500-x
ANDOVER, Mass. — The liquid that many hope could help end the Covid-19 pandemic is stored in a nondescript metal tank in a manufacturing complex owned by Pfizer, one of the world’s biggest drug companies. There is nothing remarkable about the container, which could fit in a walk-in closet, except that its contents could end up in the world’s first authorized Covid-19 vaccine.
Pfizer, a 171-year-old Fortune 500 powerhouse, has made a billion-dollar bet on that dream. So has a brash, young rival just 23 miles away in Cambridge, Mass. Moderna, a 10-year-old biotech company with billions in market valuation but no approved products, is racing forward with a vaccine of its own. Its new sprawling drug-making facility nearby is hiring workers at a fast clip in the hopes of making history — and a lot of money.
In many ways, the companies and their leaders couldn’t be more different. Pfizer, working with a little-known German biotech called BioNTech, has taken pains for much of the year to manage expectations. Moderna has made nearly as much news for its stream of upbeat press releases, executives’ stock sales, and spectacular rounds of funding as for its science.
Each is well-aware of the other in the race to be first.
But what the companies share may be bigger than their differences: Both are banking on a genetic technology that has long held huge promise but has so far run into biological roadblocks. It is called synthetic messenger RNA, an ingenious variation on the natural substance that directs protein production in cells throughout the body. Its prospects have swung billions of dollars on the stock market, made and imperiled scientific careers, and fueled hopes that it could be a breakthrough that allows society to return to normalcy after months living in fear.
Both companies have been frequently name-checked by President Trump. Pfizer reported strong, but preliminary, data on Monday, and Moderna is expected to follow suit soon with a glimpse of its data. Both firms hope these preliminary results will allow an emergency deployment of their vaccines — millions of doses likely targeted to frontline medical workers and others most at risk of Covid-19.
There are about a dozen experimental vaccines in late-stage clinical trials globally, but the ones being tested by Pfizer and Moderna are the only two that rely on messenger RNA.
For decades, scientists have dreamed about the seemingly endless possibilities of custom-made messenger RNA, or mRNA.
Researchers understood its role as a recipe book for the body’s trillions of cells, but their efforts to expand the menu have come in fits and starts. The concept: By making precise tweaks to synthetic mRNA and injecting people with it, any cell in the body could be transformed into an on-demand drug factory.
But turning scientific promise into medical reality has been more difficult than many assumed. Although relatively easy and quick to produce compared to traditional vaccine-making, no mRNA vaccine or drug has ever won approval.
Even now, as Moderna and Pfizer test their vaccines on roughly 74,000 volunteers in pivotal vaccine studies, many experts question whether the technology is ready for prime time.
“I worry about innovation at the expense of practicality,” Peter Hotez, dean of the National School of Tropical Medicine at Baylor College of Medicine and an authority on vaccines, said recently. The U.S. government’s Operation Warp Speed program, which has underwritten the development of Moderna’s vaccine and pledged to buy Pfizer’s vaccine if it works, is “weighted toward technology platforms that have never made it to licensure before.”
Whether mRNA vaccines succeed or not, their path from a gleam in a scientist’s eye to the brink of government approval has been a tale of personal perseverance, eureka moments in the lab, soaring expectations — and an unprecedented flow of cash into the biotech industry.
It is a story that began three decades ago, with a little-known scientist who refused to quit.
Before messenger RNA was a multibillion-dollar idea, it was a scientific backwater. And for the Hungarian-born scientist behind a key mRNA discovery, it was a career dead-end.
Katalin Karikó spent the 1990s collecting rejections. Her work, attempting to harness the power of mRNA to fight disease, was too far-fetched for government grants, corporate funding, and even support from her own colleagues.
It all made sense on paper. In the natural world, the body relies on millions of tiny proteins to keep itself alive and healthy, and it uses mRNA to tell cells which proteins to make. If you could design your own mRNA, you could, in theory, hijack that process and create any protein you might desire — antibodies to vaccinate against infection, enzymes to reverse a rare disease, or growth agents to mend damaged heart tissue.
In 1990, researchers at the University of Wisconsin managed to make it work in mice. Karikó wanted to go further.
The problem, she knew, was that synthetic RNA was notoriously vulnerable to the body’s natural defenses, meaning it would likely be destroyed before reaching its target cells. And, worse, the resulting biological havoc might stir up an immune response that could make the therapy a health risk for some patients.
It was a real obstacle, and still may be, but Karikó was convinced it was one she could work around. Few shared her confidence.
“Every night I was working: grant, grant, grant,” Karikó remembered, referring to her efforts to obtain funding. “And it came back always no, no, no.”
By 1995, after six years on the faculty at the University of Pennsylvania, Karikó got demoted. She had been on the path to full professorship, but with no money coming in to support her work on mRNA, her bosses saw no point in pressing on.
She was back to the lower rungs of the scientific academy.
“Usually, at that point, people just say goodbye and leave because it’s so horrible,” Karikó said.
There’s no opportune time for demotion, but 1995 had already been uncommonly difficult. Karikó had recently endured a cancer scare, and her husband was stuck in Hungary sorting out a visa issue. Now the work to which she’d devoted countless hours was slipping through her fingers.
“I thought of going somewhere else, or doing something else,” Karikó said. “I also thought maybe I’m not good enough, not smart enough. I tried to imagine: Everything is here, and I just have to do better experiments.”
In time, those better experiments came together. After a decade of trial and error, Karikó and her longtime collaborator at Penn — Drew Weissman, an immunologist with a medical degree and Ph.D. from Boston University — discovered a remedy for mRNA’s Achilles’ heel.
The stumbling block, as Karikó’s many grant rejections pointed out, was that injecting synthetic mRNA typically led to that vexing immune response; the body sensed a chemical intruder, and went to war. The solution, Karikó and Weissman discovered, was the biological equivalent of swapping out a tire.
Every strand of mRNA is made up of four molecular building blocks called nucleosides. But in its altered, synthetic form, one of those building blocks, like a misaligned wheel on a car, was throwing everything off by signaling the immune system. So Karikó and Weissman simply subbed it out for a slightly tweaked version, creating a hybrid mRNA that could sneak its way into cells without alerting the body’s defenses.
“That was a key discovery,” said Norbert Pardi, an assistant professor of medicine at Penn and frequent collaborator. “Karikó and Weissman figured out that if you incorporate modified nucleosides into mRNA, you can kill two birds with one stone.”
That discovery, described in a series of scientific papers starting in 2005, largely flew under the radar at first, said Weissman, but it offered absolution to the mRNA researchers who had kept the faith during the technology’s lean years. And it was the starter pistol for the vaccine sprint to come.
And even though the studies by Karikó and Weissman went unnoticed by some, they caught the attention of two key scientists — one in the United States, another abroad — who would later help found Moderna and Pfizer’s future partner, BioNTech.
Derrick Rossi, a native of Toronto who rooted for the Maple Leafs and sported a soul patch, was a 39-year-old postdoctoral fellow in stem cell biology at Stanford University in 2005 when he read the first paper. Not only did he recognize it as groundbreaking, he now says Karikó and Weissman deserve the Nobel Prize in chemistry.
“If anyone asks me whom to vote for some day down the line, I would put them front and center,” he said. “That fundamental discovery is going to go into medicines that help the world.”
But Rossi didn’t have vaccines on his mind when he set out to build on their findings in 2007 as a new assistant professor at Harvard Medical School running his own lab.
He wondered whether modified messenger RNA might hold the key to obtaining something else researchers desperately wanted: a new source of embryonic stem cells.
In a feat of biological alchemy, embryonic stem cells can turn into any type of cell in the body. That gives them the potential to treat a dizzying array of conditions, from Parkinson’s disease to spinal cord injuries.
But using those cells for research had created an ethical firestorm because they are harvested from discarded embryos.
Rossi thought he might be able to sidestep the controversy. He would use modified messenger molecules to reprogram adult cells so that they acted like embryonic stem cells.
He asked a postdoctoral fellow in his lab to explore the idea. In 2009, after more than a year of work, the postdoc waved Rossi over to a microscope. Rossi peered through the lens and saw something extraordinary: a plate full of the very cells he had hoped to create.
Rossi excitedly informed his colleague Timothy Springer, another professor at Harvard Medical School and a biotech entrepreneur. Recognizing the commercial potential, Springer contacted Robert Langer, the prolific inventor and biomedical engineering professor at the Massachusetts Institute of Technology.
On a May afternoon in 2010, Rossi and Springer visited Langer at his laboratory in Cambridge. What happened at the two-hour meeting and in the days that followed has become the stuff of legend — and an ego-bruising squabble.
Langer is a towering figure in biotechnology and an expert on drug-delivery technology. At least 400 drug and medical device companies have licensed his patents. His office walls display many of his 250 major awards, including the Charles Stark Draper Prize, considered the equivalent of the Nobel Prize for engineers.
As he listened to Rossi describe his use of modified mRNA, Langer recalled, he realized the young professor had discovered something far bigger than a novel way to create stem cells. Cloaking mRNA so it could slip into cells to produce proteins had a staggering number of applications, Langer thought, and might even save millions of lives.
“I think you can do a lot better than that,” Langer recalled telling Rossi, referring to stem cells. “I think you could make new drugs, new vaccines — everything.”
Langer could barely contain his excitement when he got home to his wife.
“This could be the most successful company in history,” he remembered telling her, even though no company existed yet.
Three days later Rossi made another presentation, to the leaders of Flagship Ventures. Founded and run by Noubar Afeyan, a swaggering entrepreneur, the Cambridge venture capital firm has created dozens of biotech startups. Afeyan had the same enthusiastic reaction as Langer, saying in a 2015 article in Nature that Rossi’s innovation “was intriguing instantaneously.”
Within several months, Rossi, Langer, Afeyan, and another physician-researcher at Harvard formed the firm Moderna — a new word combining modified and RNA.
Springer was the first investor to pledge money, Rossi said. In a 2012 Moderna news release, Afeyan said the firm’s “promise rivals that of the earliest biotechnology companies over 30 years ago — adding an entirely new drug category to the pharmaceutical arsenal.”
But although Moderna has made each of the founders hundreds of millions of dollars — even before the company had produced a single product — Rossi’s account is marked by bitterness. In interviews with the Globe in October, he accused Langer and Afeyan of propagating a condescending myth that he didn’t understand his discovery’s full potential until they pointed it out to him.
“It’s total malarkey,” said Rossi, who ended his affiliation with Moderna in 2014. “I’m embarrassed for them. Everybody in the know actually just shakes their heads.”
Rossi said that the slide decks he used in his presentation to Flagship noted that his discovery could lead to new medicines. “That’s the thing Noubar has used to turn Flagship into a big company, and he says it was totally his idea,” Rossi said.
Afeyan, the chair of Moderna, recently credited Rossi with advancing the work of the Penn scientists. But, he said, that only spurred Afeyan and Langer “to ask the question, ‘Could you think of a code molecule that helps you make anything you want within the body?’”
Langer, for his part, told STAT and the Globe that Rossi “made an important finding” but had focused almost entirely “on the stem cell thing.”
Despite the squabbling that followed the birth of Moderna, other scientists also saw messenger RNA as potentially revolutionary.
In Mainz, Germany, situated on the left bank of the Rhine, another new company was being formed by a married team of researchers who would also see the vast potential for the technology, though vaccines for infectious diseases weren’t on top of their list then.
A native of Turkey, Ugur Sahin moved to Germany after his father got a job at a Ford factory in Cologne. His wife, Özlem Türeci had, as a child, followed her father, a surgeon, on his rounds at a Catholic hospital. She and Sahin are physicians who met in 1990 working at a hospital in Saarland.
The couple have long been interested in immunotherapy, which harnesses the immune system to fight cancer and has become one of the most exciting innovations in medicine in recent decades. In particular, they were tantalized by the possibility of creating personalized vaccines that teach the immune system to eliminate cancer cells.
Both see themselves as scientists first and foremost. But they are also formidable entrepreneurs. After they co-founded another biotech, the couple persuaded twin brothers who had invested in that firm, Thomas and Andreas Strungmann, to spin out a new company that would develop cancer vaccines that relied on mRNA.
That became BioNTech, another blended name, derived from Biopharmaceutical New Technologies. Its U.S. headquarters is in Cambridge. Sahin is the CEO, Türeci the chief medical officer.
“We are one of the leaders in messenger RNA, but we don’t consider ourselves a messenger RNA company,” said Sahin, also a professor at the Mainz University Medical Center. “We consider ourselves an immunotherapy company.”
Like Moderna, BioNTech licensed technology developed by the Pennsylvania scientist whose work was long ignored, Karikó, and her collaborator, Weissman. In fact, in 2013, the company hired Karikó as senior vice president to help oversee its mRNA work.
But in their early years, the two biotechs operated in very different ways.
In 2011, Moderna hired the CEO who would personify its brash approach to the business of biotech.
Stéphane Bancel was a rising star in the life sciences, a chemical engineer with a Harvard MBA who was known as a businessman, not a scientist. At just 34, he became CEO of the French diagnostics firm BioMérieux in 2007 but was wooed away to Moderna four years later by Afeyan.
Moderna made a splash in 2012 with the announcement that it had raised $40 million from venture capitalists despite being years away from testing its science in humans. Four months later, the British pharmaceutical giant AstraZeneca agreed to pay Moderna a staggering $240 million for the rights to dozens of mRNA drugs that did not yet exist.
The biotech had no scientific publications to its name and hadn’t shared a shred of data publicly. Yet it somehow convinced investors and multinational drug makers that its scientific findings and expertise were destined to change the world. Under Bancel’s leadership, Moderna would raise more than $1 billion in investments and partnership funds over the next five years.
Moderna’s promise — and the more than $2 billion it raised before going public in 2018 — hinged on creating a fleet of mRNA medicines that could be safely dosed over and over. But behind the scenes the company’s scientists were running into a familiar problem. In animal studies, the ideal dose of their leading mRNA therapy was triggering dangerous immune reactions — the kind for which Karikó had improvised a major workaround under some conditions — but a lower dose had proved too weak to show any benefits.
Moderna had to pivot. If repeated doses of mRNA were too toxic to test in human beings, the company would have to rely on something that takes only one or two injections to show an effect. Gradually, biotech’s self-proclaimed disruptor became a vaccines company, putting its experimental drugs on the back burner and talking up the potential of a field long considered a loss-leader by the drug industry.
Meanwhile BioNTech has often acted like the anti-Moderna, garnering far less attention.
In part, that was by design, said Sahin. For the first five years, the firm operated in what Sahin called “submarine mode,” issuing no news releases, and focusing on scientific research, much of it originating in his university lab. Unlike Moderna, the firm has published its research from the start, including about 150 scientific papers in just the past eight years.
In 2013, the firm began disclosing its ambitions to transform the treatment of cancer and soon announced a series of eight partnerships with major drug makers. BioNTech has 13 compounds in clinical trials for a variety of illnesses but, like Moderna, has yet to get a product approved.
When BioNTech went public last October, it raised $150 million, and closed with a market value of $3.4 billion — less than half of Moderna’s when it went public in 2018.
Despite his role as CEO, Sahin has largely maintained the air of an academic. He still uses his university email address and rides a 20-year-old mountain bicycle from his home to the office because he doesn’t have a driver’s license.
Then, late last year, the world changed.
Shortly before midnight, on Dec. 30, the International Society for Infectious Diseases, a Massachusetts-based nonprofit, posted an alarming report online. A number of people in Wuhan, a city of more than 11 million people in central China, had been diagnosed with “unexplained pneumonia.”
Chinese researchers soon identified 41 hospitalized patients with the disease. Most had visited the Wuhan South China Seafood Market. Vendors sold live wild animals, from bamboo rats to ostriches, in crowded stalls. That raised concerns that the virus might have leaped from an animal, possibly a bat, to humans.
After isolating the virus from patients, Chinese scientists on Jan. 10 posted online its genetic sequence. Because companies that work with messenger RNA don’t need the virus itself to create a vaccine, just a computer that tells scientists what chemicals to put together and in what order, researchers at Moderna, BioNTech, and other companies got to work.
A pandemic loomed. The companies’ focus on vaccines could not have been more fortuitous.
Moderna and BioNTech each designed a tiny snip of genetic code that could be deployed into cells to stimulate a coronavirus immune response. The two vaccines differ in their chemical structures, how the substances are made, and how they deliver mRNA into cells. Both vaccines require two shots a few weeks apart.
The biotechs were competing against dozens of other groups that employed varying vaccine-making approaches, including the traditional, more time-consuming method of using an inactivated virus to produce an immune response.
Moderna was especially well-positioned for this moment.
Forty-two days after the genetic code was released, Moderna’s CEO Bancel opened an email on Feb. 24 on his cellphone and smiled, as he recalled to the Globe. Up popped a photograph of a box placed inside a refrigerated truck at the Norwood plant and bound for the National Institute of Allergy and Infectious Diseases in Bethesda, Md. The package held a few hundred vials, each containing the experimental vaccine.
Moderna was the first drug maker to deliver a potential vaccine for clinical trials. Soon, its vaccine became the first to undergo testing on humans, in a small early-stage trial. And on July 28, it became the first to start getting tested in a late-stage trial in a scene that reflected the firm’s receptiveness to press coverage.
The first volunteer to get a shot in Moderna’s late-stage trial was a television anchor at the CNN affiliate in Savannah, Ga., a move that raised eyebrows at rival vaccine makers.
Along with those achievements, Moderna has repeatedly stirred controversy.
On May 18, Moderna issued a press release trumpeting “positive interim clinical data.” The firm said its vaccine had generated neutralizing antibodies in the first eight volunteers in the early-phase study, a tiny sample.
But Moderna didn’t provide any backup data, making it hard to assess how encouraging the results were. Nonetheless, Moderna’s share price rose 20% that day.
In addition, some critics have said the government has given Moderna a sweetheart deal by bankrolling the costs for developing the vaccine and pledging to buy at least 100 million doses, all for $2.48 billion.
That works out to roughly $25 a dose, which Moderna acknowledges includes a profit.
In contrast, the government has pledged more than $1 billion to Johnson & Johnson to manufacture and provide at least 100 million doses of its vaccine, which uses different technology than mRNA. But J&J, which collaborated with Beth Israel Deaconess Medical Center’s Center for Virology and Vaccine Research and is also in a late-stage trial, has promised not to profit off sales of the vaccine during the pandemic.
Over in Germany, Sahin, the head of BioNTech, said a Lancet article in January about the outbreak in Wuhan, an international hub, galvanized him.
“We understood that this would become a pandemic,” he said.
The next day, he met with his leadership team.
“I told them that we have to deal with a pandemic which is coming to Germany,” Sahin recalled.
He also realized he needed a strong partner to manufacture the vaccine and thought of Pfizer. The two companies had worked together before to try to develop mRNA influenza vaccines. In March, he called Pfizer’s top vaccine expert, Kathrin Jansen.
“I asked her if Pfizer was interested in teaming up with us, and she, without any discussion, said, ‘Yes, we would love to do that,’” Sahin recalled.
Philip Dormitzer, chief scientific officer for viral vaccines at Pfizer, said developing a coronavirus vaccine is “very much in Pfizer’s comfort zone as a vaccine company with multiple vaccine products.”
Pfizer has about 2,400 employees in Massachusetts, including about 1,400 at its Andover plant, one of three making the vaccine for the New York-based company in the U.S.
Pfizer, through its partnership with BioNTech, isn’t taking any money upfront from the government. Rather, the federal government will pay the partners $1.95 billion for at least 100 million doses if the vaccine gets approved.
Pfizer CEO Albert Bourla, who rose through the ranks after more than 25 years with the company, said in a September interview with “Face the Nation” that if the Pfizer-BioNTech vaccine fails, his company will absorb the financial loss. He said Pfizer opted not to take government funding up front to shield the drug giant from politics.
“I wanted to liberate our scientists from any bureaucracy,” he said. “When you get money from someone, that always comes with strings.”
Top executives at Pfizer also have sold far less stock compared to Moderna since the pandemic began.
BioNTech executives haven’t sold any shares since the company went public last year, according to Securities and Exchange Commission records. Still, the soaring share prices of BioNTech and Moderna have made both Sahin and Bancel billionaires, according to Forbes.
Some experts worry about injecting the first vaccine of this kind into hundreds of million of people so quickly.
“You have all these odd clinical and pathological changes caused by this novel bat coronavirus, and you’re about to meet it with all of these vaccines with which you have no experience,” said Paul Offit, an infectious disease expert at Children’s Hospital of Philadelphia and an authority on vaccines.
Several other drug makers have also developed experimental mRNA vaccines for the coronavirus, but are not as far along, including CureVac, another German biotech, and Translate Bio, which has partnered with the French vaccine giant Sanofi Pasteur.
Pfizer began its late-stage trial on July 27 — the same day as Moderna — with the first volunteers receiving injections at the University of Rochester. It announced its promising early results from that trial on Monday, and hopes to have sufficient data this month to seek emergency use authorization of the vaccine for at least some high-risk people.
Moderna may not be far behind. Its spokesperson Ray Jordan said Monday that executives suspected Pfizer would release some preliminary late-stage trial data before Moderna, in part because of the dosing schedule of the rival vaccines. Recipients of Pfizer’s vaccine get two doses three weeks apart, while recipients of Moderna’s get two doses four weeks apart.
Striking a magnanimous note, he described Pfizer’s news as “an important step for mRNA medicine.”
“We’ve said that the world needs more than one Covid-19 vaccine,” Jordan said. “We remain on track.”
Mark Arsenault of the Globe staff contributed reporting.