Broadband High-Brightness Sources for Spectroscopy

January 1, 2019

n the near past, discharge lamps, dye lasers, and optical parametric oscillators were the only useful sources for spectroscopy. New broadband sources, such as supercontinuum lasers, laser-driven plasma sources, and high-brightness light-emitting diodes (LEDs), are now available. We look at what these options offer for spectroscopy.

Once the sole domain of a variety of discharge lamps, and later dye lasers and optical parametric oscillators (OPOs), a variety of stable light sources are now available for spectroscopy. Here we focus on the characteristics of supercontinuum lasers, laser-driven plasma sources, and high-brightness light-emitting diodes (LEDs), examining each as potential sources for light generation over a broad range of wavelengths.

Spectroscopy systems depend on light sources, and thus new light sources bring new opportunities for spectroscopists. For much of the last century, broadband light was only available from discharge lamps, plasma sources, heated glowbars, or the sun. For ultraviolet (UV), visible, and near infrared (NIR) spectroscopy, xenon, deuterium, and tungsten halogen lamps have been for a long time the mainstay sources for the laboratory, and have long been employed in atomic absorption spectrometry. Typical laboratory deuterium lamps (with 200 to 400 nm emission) are 30 W, but water-cooled versions of 150 W and more are available. The water-cooled versions feature improved stability and power, as they are able to operate at higher temperatures. Tungsten halogen lamps, which emit roughly from 350 to 2500 nm, are available at a variety of powers and color temperatures.

Until the advent of the laser, one of the only ways to obtain narrowband line radiation was to use a low-pressure gas discharge lamp, such as a mercury or sodium lamp. Smaller "pen" lamps are usually used for calibration of spectrometers. At higher pressures, such lamps can provide continuous radiation in some bands. Dye lasers and OPOs, of course, have filled this niche nicely for roughly the past 45 years. However, these systems require a pump laser, alignment, and tuning. They are large enough and sensitive enough that they are only useful in the laboratory.

Now, however, there are a set of new, relatively rugged sources earning their places in the spectroscopist's toolkit. Supercontinuum lasers, laser-driven plasma sources, and high brightness LEDs in a range of colors are all being used in spectroscopic applications. In this article, we will explore each of them briefly, in turn.

Supercontinuum Lasers

A supercontinuum is generated when one or more nonlinear processes broaden the frequency spectrum of a laser pulse. There are numerous ways to generate a supercontinuum, several of which have been observed since the 1960s and 1970s. One of the early drivers for discovery of supercontinuum sources was to improve on the use of flashlamps (too weak when designed to have short pulses), laser plasmas (too unstable), and laser-pumped dye emission (too weak) for nanosecond-scale probing of matter. For example, a short-duration broadband source could give an instantaneous, single-shot absorption spectrum, or the combination of a laser probe plus a broadband source could be used for a pump-probe experiment to excite and measure subsequent absorption bands, or to generate an inverse Raman absorption. Inverse Raman occurs when a monochromatic source with a high electric (E) field (such as a laser) is coincident in a medium with a broadband source of equal or shorter duration. Absorption is generated at frequencies equal to the sum of the laser frequency and corresponding Raman frequencies in the medium. Spectroscopy such as these demanded new sources that were more stable, broadband, and brighter than existing sources.

The processes behind supercontinuum generation are varied and complex, depending on the medium in which the effect is occurring. In fiber, the combination of stimulated Raman scattering and soliton effects can take a spectrally-broad pump beam and generate a supercontinuum to the red of the pump beam. In the early 1980s, investigations began into using femtosecond pulses to generate the continuum, and further research improved the physical understanding of the process. This eventually has led to the present day, in which fiber-based ultrafast lasers are used to pump fiber-based supercontinuum. The last decade has seen an explosion of sources and device characteristics, from continuous wave, nanosecond, picosecond, and femtosecond sources, using fibers with specific designs and tailored material characteristics.

The unique features of the supercontinuum light sources are their single-mode beam quality, laser-like pointing stability, and high brightness. These sources are like broadband lamps turned into lasers. The output of these devices is used in spectroscopy in many ways, such as developing novel setups for mid-IR absorption spectroscopy, optical coherence tomography, and visible cavity-enhanced spectroscopy.

Recent work from Kilgus and associates coupled an infrared supercontinuum laser with mid-infrared (IR) Fabry-Pérot (FP) detectors, utilizing a chopper wheel and lock-in detection. The authors note that the supercontinuum source competes with quantum cascade lasers (QCLs) in the mid-IR, but can fill the so-called "QCL gap" where OH and CH stretching vibrations occur. This is key, as many of the important gases are detected based on these features. Also, they note that the spectral coverage of the supercontinuum source is much greater than that of the QCLs.

After showing that the supercontinuum source could approach the noise characteristics of extended-cavity QCLs, below 1% root-mean-square pulse-to-pulse fluctuations, Kilgus and his team proceeded to make measurements. One of the most interesting measurements was the evaporation of 2-propanol at five meters, measuring transflected light with the FP detector coupled to a lead selenide (PbSe) detector, which could sweep the roughly 300 cm-1 range in 20 ms. This is shown in Figure 1. This absorbance measurement over the "QCL gap" illustrates an important feature of supercontinuum sources, which, coupled with their standoff capability, could provide new, sensitive detectors for gases at range or in fenceline monitoring situations.

Figure 1: Evaporation of 2-propanol measured using a swept Fabry-Pérot detector and a supercontinuum source. From reference (4), used with permission.

Optical coherence tomography (OCT) has also benefitted greatly from the development of supercontinuum devices. In OCT, incident light is split into a reference beam and an interrogation beam. Reflected light from the sample interacts with the reference beam, allowing determination of distance and mapping the surface quality of the sample. At the same time, interactions of the incident light with the sample, in the form of absorption or other interaction, can provide spectroscopic information. In this way, OCT can provide 3-D spatially-sensitive spectral information.

Recently, Yi and associates showed that they could measure retinal oxygenation during progressive hypoxic challenge (systemic oxygen starvation). They did this noninvasively in rat eyes, using an OCT system based on a supercontinuum source that had been bandpass-filtered to transmit supercontinuum light from 520 to 630 nm. Their method, called vis-OCT, centered for this application on spectral bands around 585 nm. As they explain, one of the strengths of vis-OCT is the ability to extract spectral information from a specific depth, which allows rejection of scattered light that would otherwise be typical in imaging-based methods. Figure 2 illustrates the validation of the measurement method using pulse oximetry, and the measured incoming (arterial) and outgoing (venous) oxygen saturation rate measured in situ as the subject rats were undergoing increasing oxygen starvation.

Figure 2: Left: verification of the visible–optical coherence tomography (vis–OCT) spectroscopic method via pulse oximetry. Right: in situ measurements of retinal arterial and venous oxygen saturation rate measured in rats with increasing degrees of hypoxia. From reference (5), under the Creative Commons, by license.

Another excellent example of the application of supercontinuum sources can be found in the work of J.M. Langridge and colleagues, who were the first to use supercontinuum radiation combined with cavity-enhancement to measure absorption spectra of multiple gases. These authors injected a spectrally-filtered bandpass of approximately 100 nm from a supercontinuum source into a commercial absorption cavity, such as used in cavity-ringdown spectroscopy. They were able to measure H2O, O2, and O2–O2 (a collision-induced absorption observable in pure oxygen). In addition, they recorded quantitative spectra of NO2 and NO3, with an estimated detection limit of 3 parts-per-trillion of NO3 achieved in a 2-second measurement time. These achievements were made possible due to the spatially-coherent, high-brightness supercontinuum source.


Laser-Driven Light Sources

As mentioned above, there are historical uses of laser plasmas as broadband light sources for spectroscopy, but these have traditionally been found too unstable or too weak to be useful for analytical spectroscopy. Recently, a substantial amount of work into laser-driven plasma sources has been motivated by the search for extreme ultraviolet sources for nanolithography. This has spilled over into stable, laser-driven sources useful for spectroscopy. Recently (September, 2017), Hamamatsu purchased Energetiq Technology, a manufacturer of laser-driven plasma sources, for $42 million.

The technology behind Energetiq's approach is to focus a continuous wave (CW) laser inside a bulb filled with a noble gas, such as xenon. In a traditional lamp, the electrodes act as a heat sink, limiting the temperature of the lamp. In addition, during operation the electrodes continuously degrade and sputter electrode material on the surface of the lamp, which necessitates frequent replacement of the bulb. Using the CW laser to excite the bulb allows higher operating temperatures (10,000 to 20,000 K versus 5000 to 7000 K for a typical Xenon arc lamp), by eliminating the heat sink, and lengthens the lifetime of the source substantially. This higher temperature improves the spectral performance in the UV, and increases the overall brightness.

Recently, the Ultraviolet Spectral Comparator Facility at the U.S. National Institute of Standards (NIST) replaced their traditional commercially-available argon mini-arc source with a laser-driven light source (LDLS). In the paper published on the transition, the authors note that not only did the substitution dramatically improve the optical power available in the ultraviolet (UV), but stability of the signal increased to approximately 0.1% RMS deviation, as compared with an average of about 0.3% for the argon mini-arc. A spectral comparison is available online, similar to that which is discussed in reference (7).

This unique light source is improving spectroscopy from the near-IR to the UV. For example, scattering-type scanning near-field optical microscopy (s-SNOM) is often used to probe nanoscale photonic phenomena or chemical compositions with a 10 to 20 nm spatial resolution. Using infrared light, this is several orders of magnitude below the diffraction limit. However, as explained by Wagner and colleagues in a recent study, the application of s-SNOM has been limited by the availability of affordable broadband infrared sources. Tunable continuous-wave quantum cascade lasers are expensive and only of limited tunability, and other laser sources such as difference-frequency generation (DFG) femtosecond laser setups are only more expensive and have their own drawbacks. Traditional Globar sources are broadband, but inherently weak.

In reference (9), Wagner and associates used an LDLS in an s-SNOM setup to obtain data on microstructure and composition of several interesting features, including common polymers and a 20 nm thin boron nitride nanotube. An example of the spectrum of the boron nitride nanotube and a spatial map are shown in Figure 3. They report that, in the infrared region of interest, the spectral intensity of the LDLS is approximately 40 times higher than a Globar source, allowing for much greater sensitivity for the s-SNOM instrument.

Figure 3: Left: spectrum of boron nitride nanotube using the scattering scanning nearfield optical microscopy (s-SNOM) instrument. Right: corresponding spatial map of spectrum. The width of the blue region is approximately 50 nm. Adapted with permission from reference (9). Copyright 2018 American Chemical Society.

In the UV, the LDLS have enabled numerous advances, as anticipated by the work in reference (7). For example, application to ultraviolet microscopy has allowed quantitative spectral analysis of biomolecules. The UV wavelength allows sub-cellular spatial resolution. An example of this work is the emergence of ultraviolet hyperspectral interferometric (UHI) microscopy, reported this year. Using a high-brightness LDLS in conjunction with a modified Mach-Zehnder interferometer yields information over a large spectral range (roughly 200 nm in the UV), corrected for chromatic aberration, with minimal fluorescence interference and high spatial resolution.

Figure 4 illustrates some of the interesting results from reference (10). In particular, the topology of a single, live red blood cell is measured with sub-micron accuracy. Second, the molar extinction coefficient of the live red blood cell is measured, along with error bars in grey; it compares very well to the extinction coefficient of bulk hemoglobin solution, shown with the red dotted lines. These results are averaged from spectral data obtained across the entire cell. Overall, these examples suggest the usefulness of the LDLS as a high-brightness source from the UV to the infrared.

Figure 4: Left, topology of a live red blood cell measured with UHI microscopy. Right: the molar extinction coefficient of a single red blood cell (uncertainty bands in grey) compared with hemoglobin (Hb) solution. From reference (10) under the Creative Commons by license.

High-Brightness LEDs

The newest of the light sources discussed in this article are high-brightness light-emitting diodes (LEDs), which have only recently been available across much of the spectral range. Although the effect had been studied for some time, and the first patent on a practical LED was issued based on a filing by Texas Instruments in 1962, it took many years of evolution to bring LEDs to the point that they are today. Based on photon emission resulting from an energetic electron filling a hole in a diode one-way valve (p-n junction), the color of the light is determined by the bandgap of the semiconductor. While not monochromatic, based on the distribution of energies of the electrons, most LEDs emit in a narrow band of light. Various colors are generated by different semiconductor chemistries.

For commercial LEDs used in lighting, the main problem has centered around how to produce both high-power, high-efficiency LEDs, and how to generate the right color blend to be pleasing to the human eye. For spectroscopy, the challenge is different. Spectroscopists would like particular colors (ideally a source would be tunable), with high spectral purity or spectral reproducibility, and high stability. A particular challenge for LEDs is stabilization of the output as a function of temperature. Obtaining sources in the UV has been a particular desire and challenge, to access electronic transitions of environmentally-important gas molecules.

The idea of using LEDs for differential optical absorption spectroscopy (DOAS) measurements was considered as early as 2006, if not before. Recently, as the power and available wavelength selection of LEDs has soared, several groups have begun to capitalize on the use of UV LEDs as sources for gas sensing. Michel and Kapit report detection of SO2 using balanced detection and a deep UV LED. Their detection limit was on the order of 1 part-per-million (ppm), and the sensor worked for concentrations as high as 50 ppm. Similarly, Li and coworkers demonstrated a breath detector with a 0.19-L multipass cell, a UV LED near 285 nm, and a balanced detection method. Their demonstrated detection limit was 0.7 ppm, with a precision of 0.4 ppm. The acetone analyzer could be useful for measurement of acetone in the human breath, which can be an indicator for metabolic conditions and/or diabetic conditions.

Quite recently, a new research instrument using long-path DOAS was constructed with three diodes, intended for atmospheric SO2, NO2, and O3 detection. LEDs between 275 to 377 nm optimized for each of the gases were coupled into 200 µm fiber optics, and transmitted via a fiber bundle to a launch point. The launched LED light was collimated by a spherical mirror and transmitted 350 m to a retroreflector, and then returned, making a total pathlength of 700 m. Return light was captured by a collection fiber in the center of the launch fiber.

In a research study of air pollution in Heifei China, the LED-based system was operated side-by-side with a commercial DOAS system for slightly more than 1 week. Over the measurement period, the range of SO2 was from 0.3 to 19.8 parts-per-billion (ppb), O3 was 3.5 to 80 ppb, and NO2 was 1.5 to 50 ppb. Agreement between the commercial analyzer and the LED-based analyzer was excellent.



Broadband sources are no longer confined to lamps, traditional plasmas, globars, or the Sun. A variety of new, high brightness sources are increasingly being used in successful spectroscopy applications, from microscopy to gas sensing, and from industrial to biological applications. These applications span from the UV to the infrared. Table 1 illustrates some of the characteristics of these sources. Next time you are considering a new application, consider one of these new sources. Between the combination of brightness, collimation, spectral range, and affordability, a new source may be in your spectroscopic future.

Table I: Some characteristics of new, broadband, high-brightness sources


(1) C. Lin and R.H. Stolen, Appl. Phys. Lett. 28, 216 (1976); doi: 10.1063/1.88702.

(2) R.A. McLaren and B.P. Stoicheff, Appl. Phys. Lett. 16, 140 (1970); doi: 10.1063/1.1653129.

(3) R. L. Fork, C. V. Shank, C. Hirlimann, R. Yen, and W. J. Tomlinson, Opt. Lett. 8, 1-3 (1983).

(4) J. Kilgus, K. Duswald, G. Langer, and M. Brandstetter, Appl. Spectrosc. 72, 634 (2018).

(5) J. Yi, W. Liu, S. Chen, V. Backman, N. Sheibani, C.M. Sorenson, A.A. Fawzi, R.A. Linsenmeier, and H.F. Zhang, Light: Science & Applications 4, e334, (2015); doi:10.1038/lsa.2015.107.

(6) J.M. Langridge, T. Laurila, R.S. Watt, R.L. Jones, C.F. Kaminski, and J. Hult, Opt. Express 16, 10178 (2008).

(7) U. Arp, R. Vest, J. Houston, T. Lucatorto, Appl Opt. 53, 1089 (2014).


(9) M. Wagner, D.S. Jakob, S. Horne, H. Mittel, S. Osechinskiy, C. Phillips, G.C. Walker, C. Su, and X.G. Xu, ACS Photonics, 5, 1467 (2018).

(10) A. Ojaghi, M.E. Fay, W.A. Lam, and F. E. Robles, Scientific Reports 8, 9913 (2018). doi:10.1038/s41598-018-28208-0.

(11) C. Kern, S. Trick, B. Rippel, U. Platt, Appl. Opt. 45, 2077 (2006).

(12) A.P.M. Michel and J. Kapit, Appl. Spectrosc. 71, 996 (2017).

(13) J. Li, T.M. Smeeton, M. Zanola, J. Barrett, V. Berryman-Bousquet, Sensors and Actuators B: Chemical 273, 76 (2018).

(14) N. Zheng, K.L. Chan, P. Xie, M. Qin, L. Ling, F. Wu, and R. Hu, Atmos. Pollution Res. 9, 379 (2018).

Steven G. Buckley, PhD, is the CEO of Flash Photonics, an affiliate associate professor at the University of Washington, and has started and advised numerous companies in spectroscopy and in applications of machine learning. He has approximately 40 peer-reviewed publications and 6 patents. His work in practical optical spectroscopy, such as LIBS, Raman, and TDL spectroscopy, dovetails with the coverage in this column, which reviews methods (new and old) in laser-based spectroscopy and optical sensing. Direct correspondence to: