The internet already runs on light. Beneath the familiar language of packets and bandwidth, enormous volumes of information travel as pulses through glass fibres, and the devices that launch, steer and modulate those pulses determine how much data a network can carry. Engineers have long wanted faster, smaller optical modulators, components that encode electrical information onto a light beam. One material keeps appearing in that search: barium titanate. It responds strongly to an applied electric field, which is exactly what a modulator needs. Yet for decades, published measurements of how the crystal handles light have disagreed, leaving designers without a trustworthy reference. A team led by the Korea Advanced Institute of Science and Technology has now measured those properties across an unusually wide spectrum and worked out how to account for the crystal's internal complexity. The result, published in ACS Photonics, does not deliver a new device. It removes a long-standing uncertainty that has slowed the design of the devices themselves.

Why light-based data needs better materials

Modern telecommunications depends on converting electrical signals into optical ones and back again. A modulator sits at the heart of that conversion. It changes some property of a light beam, such as its phase or intensity, in step with an electrical input. The faster and more efficiently it can do that, the more information can be packed onto a single fibre.

The materials used in today's modulators are good but not perfect. Silicon photonics has scaled well, but silicon lacks a strong natural electro-optic response. Lithium niobate offers that response and is widely used in high-speed equipment, but integrating it into compact chips remains difficult. Researchers have therefore continued to look for alternatives that combine a strong effect with compatibility with small, manufacturable devices.

Barium titanate, often shortened to BTO, has attracted attention because it exhibits a strong Pockels effect: a change in refractive index proportional to an applied electric field. That means a modest voltage can alter how light moves through the material, which is precisely the behaviour a modulator exploits.

The problem is not the effect itself. It is the missing bookkeeping. To design an optical device, an engineer needs to know two numbers for every wavelength: the refractive index, which describes how light propagates, and the extinction coefficient, which describes how much light is absorbed. For BTO, those numbers have been surprisingly hard to pin down.

The KAIST-Caltech collaboration

The new work was led by Professor Min Seok Jang of KAIST's School of Electrical Engineering, working with Professor Harry A. Atwater's group at the California Institute of Technology. The Korea Research Institute of Standards and Science, known as KRISS, and the instrument maker J.A. Woollam also took part. KAIST announced the results on 8 November 2026, and ScienceDaily reported them on 20 November.

Jiwon Kang, a PhD candidate in Jang's group at KAIST, and Dr Nina Hong of J.A. Woollam served as co-first authors. Jang and Atwater were the corresponding authors. The paper, titled Broadband Optical Characterization of Tetragonal Barium Titanate by Mueller-Matrix Spectroscopic Ellipsometry, appeared online in ACS Photonics on 20 October 2026 with the DOI 10.1021/acsphotonics.6c01514.

That combination of partners is telling. The measurement problem was not only a matter of physics; it was also a matter of instrumentation and standards. KRISS brings metrology expertise, J.A. Woollam builds ellipsometers, and the Caltech group has deep experience in optical materials and nanophotonics. Solving a decades-old measurement puzzle required all of those strengths at once.

The funding reflects the international character of the effort. Support came from Korean programmes including the Culture, Sports and Tourism R&D Program administered by KOCCA, the Basic Science Research Program run by the National Research Foundation of Korea, and the Ministry of Science and ICT. The US Air Force Office of Scientific Research also contributed.

Why barium titanate is difficult to measure

On paper, measuring a crystal's optical constants sounds straightforward. Shine light of a known wavelength at a sample, observe what comes out, and infer the refractive index and extinction coefficient from the result. In practice, the answer depends on the sample. Barium titanate can contain multiple microscopic regions called domains, in which the crystal structure points in different directions.

Those domains are not cosmetic. Light travelling through a crystal with several orientations interacts differently with each one. A measurement taken on a whole sample can therefore combine several distinct optical responses, and the mixture depends on how the domains are arranged. Two laboratories measuring nominally similar BTO crystals can legitimately obtain different answers.

The issue has dogged the field for decades. Published values for BTO's refractive index and absorption have varied widely, which makes it hard for engineers to trust a simulation. If the input numbers are uncertain, a model of a waveguide or modulator cannot reliably predict how much light will be lost or how efficiently a voltage will change the beam.

The domain problem is especially awkward because it is intrinsic to the material. BTO is useful partly because its internal structure can be switched by an electric field, and that same structural flexibility produces the domains. A measurement method that simply ignores them will keep producing inconsistent results.

Domains inside a single crystal

The team's first move was to reduce the complexity rather than model it immediately. Instead of using crystals with an arbitrary mixture of orientations, they studied two types of single-crystal samples with relatively consistent domain orientations. That choice helped minimise the number of overlapping optical responses contributing to each measurement.

Working with well-characterised samples is a familiar strategy in materials science. It does not pretend that real devices will be so simple. It establishes a baseline: what are the optical properties of the material itself, before the complications introduced by domains and device geometry?

Once that baseline is known, the next task is to describe how domains modify it. The researchers treated the optical response of a multi-domain crystal as a combination of the responses of its constituent orientations. The proportion of the crystal occupied by each domain becomes part of the calculation.

This two-stage approach is one of the paper's most useful contributions. Earlier work often reported a number for a particular sample without making clear how much of that number belonged to the material and how much belonged to the domain pattern. Separating the two gives engineers a way to reason about crystals they have not yet measured.

Mueller matrix spectroscopic ellipsometry

The central measurement technique was Mueller matrix spectroscopic ellipsometry. Ellipsometry examines how the polarisation of light changes when it reflects from or passes through a material. Because the change depends on the material's optical constants, careful analysis can recover those constants without touching the sample.

The Mueller matrix is a way of describing the full polarisation transformation, including effects that simpler measurements cannot capture. That completeness matters for an anisotropic crystal such as BTO, where the response depends on direction. A method that records only one polarisation channel would miss important parts of the picture.

The researchers combined ellipsometry with separate measurements of light transmission. Transmission data are especially valuable in spectral regions where the material barely absorbs, because ellipsometry alone becomes less sensitive when absorption is extremely small. Using both methods together allowed the team to extract reliable values across the full range.

The technique is not new, but applying it systematically to tetragonal BTO across such a broad spectrum is. The result is a set of optical constants tied to specific crystal orientations, rather than a single ambiguous number for an unspecified sample.

A spectrum from ultraviolet to far-infrared

The measurements covered wavelengths from 190 nanometers in the ultraviolet to 40 micrometers in the far-infrared. That is an exceptionally broad span, and it was chosen deliberately. A material's optical behaviour changes with wavelength, so a value measured at one point says little about performance elsewhere.

For context, the visible spectrum occupied by human vision runs from roughly 400 to 700 nanometers. The study therefore reaches well beyond visible light in both directions. Telecommunications wavelengths used in fibre networks sit within the near-infrared portion of this range, while the far-infrared is relevant to other photonic applications.

Because the physics differs across such a wide span, the team divided the spectrum into three regions and applied a different analytical approach to each. This is a practical acknowledgement that one model cannot describe every interaction of light with a crystal. Different energies excite different responses, and the analysis has to respect those boundaries.

The result is a wavelength-by-wavelength picture of how BTO refracts and absorbs light, resolved by crystal orientation. For device designers, that is far more useful than a handful of representative numbers. It allows modelling of components that will operate at particular wavelengths rather than in an idealised generic band.

Measuring almost-zero absorption

One of the most striking findings concerns the range from 2 to 10 micrometers, where BTO allows nearly all incoming light to pass through. In this region the material's extinction coefficient is tiny. The team determined values of approximately 10 to the minus five, meaning that only a minute fraction of the light is absorbed.

Measuring such low absorption is difficult. When a sample is almost perfectly transparent, small experimental errors can dominate the result. That is why the researchers combined ellipsometry with direct transmission measurements rather than relying on a single method. The agreement between approaches gave them confidence in the tiny values.

Low absorption is exactly what optical device engineers want. Every bit of light lost inside a waveguide is signal that never reaches its destination, and in dense photonic circuits those losses accumulate. Knowing that BTO is exceptionally transparent in a useful wavelength band makes it easier to justify the material for real components.

The authors also emphasise what the numbers enable. With reliable refractive index and absorption values, engineers can model how light will travel through a component of a given shape and size before manufacturing it. That reduces the need for repeated prototypes, costly redesigns and extensive trial-and-error testing.

Predicting multi-domain behaviour

The study's second major advance addresses the domain problem directly. The researchers developed a way to predict the combined optical response of a crystal containing multiple domains, using the orientations of those domains and the proportion of the crystal each occupies as inputs.

This is a piece of modelling that turns a nuisance into a calculable quantity. Rather than discarding multi-domain samples as too complicated, an engineer can describe their internal structure and estimate the resulting optical behaviour. The prediction can then be checked against measurement.

The method also clarifies why historical results differed. If two crystals had different domain mixtures, their measured optical constants would naturally differ even if the underlying material were identical. Recognising that removes some of the apparent contradiction in the literature.

For future work, the ability to predict domain-averaged behaviour could guide materials growth. If a particular domain pattern gives a more favourable optical response, crystal growers can aim for that pattern rather than discovering its benefits by accident.

What the numbers mean for device engineers

The practical value of the paper lies in its reliability. A designer working on a BTO modulator needs to know how much light the material will absorb at the operating wavelength and how strongly the beam will be confined. The new constants supply both pieces of information with an accuracy that earlier reports could not consistently provide.

Better models mean fewer prototypes. In photonics, fabricating a device and testing it can take weeks, and a wrong assumption about material properties can send a project back to the drawing board. A trustworthy reference shortens that loop and makes simulation a more dependable guide.

The work also helps quantify a key trade-off. A strong electro-optic effect is attractive, but only if it does not come with unacceptable optical loss. By measuring both quantities across a broad spectrum, the team gives engineers the data needed to judge whether BTO wins that trade-off for a particular application and wavelength.

Professor Jang summarised the contribution directly: the study reveals an optical response that had long remained unclear and provides a reliable reference for how the material interacts with light. That is a modest description of what could become a standard reference dataset.

The ScienceDaily report framed the result in terms of faster internet, and optical communications are the most obvious application. Modulators that waste less light and switch more efficiently could help networks carry more data without increasing power consumption or signal degradation.

The authors point to other possibilities as well. BTO could be used in devices that generate light at different wavelengths, which are important for sensing, spectroscopy and integrated photonics. Such sources depend on careful management of optical properties across a spectrum, precisely the information the study supplies.

Quantum photonics is another area where precision matters. Circuits that process quantum information using light require exceptionally tight control over how photons propagate and interact. Unclear material constants are a serious obstacle when a device must behave predictably at the level of individual photons.

None of these applications follows automatically from a measurement paper. Each requires engineering work, fabrication development and system integration. The study's role is to make that work less speculative by replacing uncertain numbers with measured ones.

What still stands in the way

It is important not to oversell the result. Measuring a material's optical constants is not the same as building a better modulator. BTO still has to be integrated into manufacturing processes, and engineers must solve problems of electrode design, optical confinement and compatibility with other chip components.

The study also focuses on tetragonal BTO in single-crystal form. Real devices may use thin films, different growth conditions or compositions adjusted for particular properties. Each variation could change the optical response, and the new measurements provide a reference point rather than a universal answer for every form of the material.

There are also practical questions about domains in operating devices. The Pockels effect relies on switching the crystal's internal order with an electric field, so the domain structure is not static. Understanding how the optical constants evolve during switching would be a natural extension of this work.

Even with those caveats, the paper marks a genuine advance. A material that has been discussed for decades as a promising optical platform now has a much clearer specification. That may not produce a headline device, but it is the kind of groundwork that makes devices possible.

A clearer reference for optical design

The broader lesson is that progress in technology often depends on unglamorous measurements. Barium titanate did not become interesting because of a new discovery in this study. It was already interesting. What changed is that engineers can now model it with confidence.

The team's combination of broad spectral coverage, careful sample selection and two complementary measurement methods produced something the field lacked: a consistent account of how tetragonal BTO handles light from the ultraviolet to the far-infrared. The added method for estimating multi-domain behaviour makes that account useful beyond the idealised samples used in the lab.

For the communications industry, the payoff would be faster optical links that lose less light. For quantum photonics, it could be a set of materials that can finally be simulated before they are fabricated. Both outcomes remain works in progress.

For now, the result is a reference. It closes a chapter of conflicting measurements that began decades ago and gives the next generation of optical device designers a firmer place to start. In a field where a few nanometers and a few parts in a hundred thousand can decide whether a device works, that is a substantial contribution.


Corrections and updates

Nexuswild welcomes factual corrections. Email contact@nexuswild.com with evidence and the article URL.