What’s in your water?

In August of 2014, four months after Governor Rick Snyder switched the water supply of Flint, Michigan, from Lake Huron to the heavily polluted Flint River, the city issued a mortifying warning: there might be fecal matter in Flint's water system. Though most notorious for causing widespread heavy metal poisoning as its corrosive water ate away at ancient lead plumbing, the Flint river also brought microscopic pathogens to residents' homes, including harmless bacteria that hinted at the possibility of more dangerous fecal matter contamination.

But identifying microbial threats in multiton reservoirs is a sensitive operation. Unlike nonliving microscopic contaminants, such as heavy metals, dangerous microbes like the bacteria Escherichia coli are too large and complex for tests based on their physical properties, such as centrifuging.

Instead, the current standard for identifying E. coli in water treatment systems is culturing, a labor-intensive process that can take days to complete. Routine sampling scans for total coliform bacteria, a broader category that includes E. coli as well as many similar, but harmless, bacterial strains; only if those tests return positive results are samples transferred to regional EPA laboratories for further testing. According to EPA regulations, technicians transfer filtered particles from testing sites to culture plates, where they are left to grow for 18 to 24 hours before a trained microbiologist manually counts the number of colony-forming units with a standard microscope.

Tests like these rely on time, training, and funding, all of which were in short supply in Flint at the time. Authorities issued multiple "boil water" advisories in response to the elevated E. coli levels that summer until, unable to locate or sequester the outbreak, they flooded the entire system with chlorine to disinfect it. The result was water that reportedly tasted like bleach and corroded metal components at the nearby General Motors plant. Later, unprecedented outbreaks of Legionnaire's disease in Flint during this period, as well as carcinogenic byproducts of the disinfectant, were also linked to the fluctuating chlorine levels.

Flint's water system during the city's water crisis led the bacterium Legionella pneumophila to proliferate, causing a deadly outbreak of Legionnaire's disease. Credit: Shutterstock

Bacterial blue-shift

For millennia, humans deployed their most cutting-edge chemical wisdom to defend their drinking water from invisible threats. Ancient Greek and Roman winemakers lined jugs with silver, relying on the noble metal's characteristic antimicrobial properties to reduce bacterial populations. Once silver is exposed to oxygen and humidity, its surface takes on a positive charge and interacts electrostatically with the negatively charged bacterial wall of "gram-negative" bacteria like E. coli. Silver ions then flow into the bacterial cell and kill the microbe.

Modern photonic approaches to biosensing for water safety are far more complex, reflecting the manifold increase in photonics expertise in the scientific community since antiquity. New approaches take advantage of photons' unique penetrative power and their sensitivity to electrochemical charge. Because different pathogens require such vastly different responses, modern photonic biosensors are most often tasked with detecting and identifying pathogens, rather than eliminating them.

Photonic devices are uniquely well-suited for this type of task for a number of reasons. Compared to electronic devices, biosensors powered by optical properties are sturdier, more sensitive, and far easier to manipulate at the nanoscale level, allowing customization for a wide variety of distinct targets. For example, the surface plasmon resonance (SPR) sensor, which measures the collective oscillations of surface plasmons on a metal, was one of the first such platforms to achieve wide recognition as part of fiber-optic biosensing technologies in the early 1990s.

"The functional mechanism of silver against bacteria is actually quite complex," says Giuseppe Maria Paternò, a chemist and material scientist based in Milan. But Paternò is most interested in the effect of the interaction on the silver, not the bacteria. His research group at the Center for Nanoscience and Technology within the Istituto Italiano di Tecnologia (IIT) is one of several currently incorporating principles of photonics into the realm of fluid biosensing. "We thought, ‘This effect must have an impact on the optical properties, on the plasmonic frequency, of the silver,'" recalls Paternò.

The question was how to detect it. "The plasmon absorption of silver lies in the UV region, where our eyes are not very sensitive. But you can engineer the photonic response in essentially any region you want," if you have the right tools, says Paternò.

Paternò's device, a handheld, one-dimensional photonic crystal biosensor, aims to make the SPR phenomenon detectable to the naked eye. Alternating layers of silicon or titanium dioxide and indium-tin oxide translate the blue shift of the silver-bacterial interaction into a colorimetric change within the visible spectrum, alerting operators to the presence of the pathogen—specifically, E. coli—by turning from pale to deep green upon exposure.

"Most of the systems that are used for counting bacteria require a special apparatus and a skilled person to use it," says Paternò. A biosensor with a simple color readout, requiring almost no training to operate, opens up the possibility of lay users and technicians testing water systems with little input needed from scientific and government institutions.

Advances in fabrication

Research into photonic applications in biofluid sensing has revealed a number of other mechanisms with potential for water quality testing. In her research at the University of Iowa's Department of Electrical and Computer Engineering, Fatima Toor relies on the extreme customizability of nanophotonic structures to build devices that maximize the surface area of the sensor interface.

Though manufacturing of silicon nanowires was once prohibitively costly, Toor's team has developed an original technique called the metal-assisted chemical etch (MACE) that is inexpensive and scalable to field applications. The new etching process creates a solar cell based on an array of silicon nanowires, bookended by source and drain electrodes, all resting on a dielectric-coated silicon wafer. The device infers the presence of gram-negative contaminants by measuring how well the fluid in the system conducts electricity.

The new fabrication technique helps to simplify the process of functionalization, in which the surface of the sensor is treated with a biochemical substance that will bond to the target molecule. Crucially, says Toor, this functionalization sidesteps a common, but more costly, chemical assay technique. Instead of coating the sensor surface so that more targets stick to it, labeling involves priming the sample with labels: easily detectable molecules, such as fluorescent proteins, that pair up with targets to make the latter visible to fluorescent scans and other detection methods.

Labels make target microbes like E. coli visible to assays based on their physical properties, but they also present an additional step in the diagnostic process. "Label-based sensor technologies are often labor- and cost-intensive, as well as time-consuming," explained Toor. Label-free methods, in contrast, "utilize intrinsic physical properties of the analytes, such as molecular weight, size, charge, electrical impedance, dielectric permittivity, or refractive index, to detect their presence in a sample."

Moreover, label-free biosensing methods can be more easily integrated into lab-on-a-chip platforms, which can monitor the concentration of target analytes in real time—an essential feature of water supply testing, where rapid diagnostics are of the utmost importance.

One of the first use cases for photonic biosensing was for nitrate, a surface water pollutant from fertilizer runoff that disturbs marine ecosystems, resulting in algae blooms and other dangerous effects. "There are already well-known, proven optical techniques to detect nitrate, which absorbs in the UV range, but these are very costly, high-end probes," explains Marcel Zevenbergen, a physicist and program manager of liquid sensing solutions at Imec's Holst Center in the Netherlands. "Now we think, with miniaturization, the price per test can decrease greatly," even to the point of incorporation into commercial products.

Automating the microbiologist

The potential for a biosensor with greater processing power recently motivated Aydogan Ozcan, who leads the Bio- and Nano-Photonics Laboratory at the UCLA School of Engineering, to develop a photonic biofluidic sensor that recently beat EPA-recommended methods for time to detection, returning test results in roughly nine hours. The new system relies on holography, another analytical technique made possible by the photonic properties of light, to replicate the EPA's growth-based method by identifying colony formation over time. "It's the colonies that make you sick. If the bacteria cannot replicate, cannot form a colony, then it cannot do anything to you," explains Ozcan.

Components of Ozcan's bacterial colony growth detection and classification system. Credit: Aydogan Ozcan

Focusing on growth eliminates the possibility of false positives that plagues genetics-based assays like ELISA, which cannot differentiate between live and dead bacteria. But unlike the EPA standard, Ozcan's "almost-handheld sensor" replaces the expensive wisdom of a microbiologist with the relatively low-cost judgment of AI. The device captures periodic holographic images of the filtered material from the sample and feeds the data into a neural network, which registers colony growth and then identifies each bacterial species by its characteristic shape and growth pattern.

Each holographic scan of a 25 mm² agar plate takes place in roughly one minute. "It's a very high-throughput imaging system, which is important, because this way we can actually capture information of the entire imaging plane." And unlike a human microbiologist, the field-portable sensor does not need breaks, but "can rapidly scan large areas of the culture longitudinally, over several hours," enabling constant monitoring of a treatment system. When filters inevitably rupture and leak, "you need systems that will constantly monitor bacterial content," says Ozcan, to catch growths before they become dangerous.

An interdisciplinary challenge

The proliferation of photonic devices in biofluid sensing is a good sign for the future of the subfield, says Olivier Henry, project manager of life sciences and medical device technologies at Imec and a colleague of Zevenbergen. "You see more and more photonics systems at the research level. But the integration of devices into fully automated systems, which fully utilize the advantages of photonics, is not so trivial." One factor is the difficulty of getting reliable measurements in fluid settings, whose relevant properties, such as temperature, are difficult to control and maintain.

The complexity of integrated devices makes them vulnerable to the whims of the system. "Fluidic sensing is a very harsh environment," notes Zevenbergen. "Fouling of the sensor is one of the major issues. How do you protect the parts of your sensor that are exposed? The interface, the electro-optical connection, they all have to be protected," from water, debris, and any corrosive or harmful substances in the fluid, including the target analyte itself.

But for Zevenbergen and Henry, the main obstacle to widespread adoption of photonic biosensors in water safety applications may be institutional. Fully integrated devices for water system testing require expertise in a wide range of physical and biological sciences. That can present serious logistical challenges for the design team, but also in terms of human expertise. "Optical physics, optics, surface modification and chemical engineering, physical chemistry—all these [backgrounds] need to be in-house to create one of these complex systems," explains Zevenbergen.

Henry agrees, noting that the interdisciplinary collaboration is worth the extra effort. "If you can combine these different sensing approaches in a single platform, especially for bioprocessing and bioreactors, where we now see tremendous traction [for photonic biosensing projects]"—you can have a perfect marriage of all different analytical techniques—optical, electrical, electrochemical, material—and balance the pros and cons of each type."

For underfunded cities like Flint, Michigan, the most important pros and cons are not technical, but budgetary. Flint made the switch from river water back to lake water in 2015, amidst felony negligence lawsuits against water plant officials and other administrators. Four years later, Mayor Karen Weaver issued a statement in response to an EPA official's claim that Flint's water was once again safe to drink. "The medical community and scientific community will both have to be in agreement, after a period of testing over time, that the water is safe to drink before I ever declare it safe."