WATSON: In Situ Organic Detection in Subsurface Ice Using Deep-UV Fluorescence Spectroscopy

Abstract

Terrestrial icy environments have been found to preserve organic material and contain habitable niches for microbial life. The cryosphere of other planetary bodies may therefore also serve as an accessible location to search for signs of life. The Wireline Analysis Tool for the Subsurface Observation of Northern ice sheets (WATSON) is a compact deep-UV fluorescence spectrometer for nondestructive ice borehole analysis and spatial mapping of organics and microbes, intended to address the heterogeneity and low bulk densities of organics and microbial cells in ice. WATSON can be either operated standalone or integrated into a wireline drilling system. We present an overview of the WATSON instrument and results from laboratory experiments intended to determine (i) the sensitivity of WATSON to organic material in a water ice matrix and (ii) the ability to detect organic material under various thicknesses of ice. The results of these experiments show that in bubbled ice the instrument has a depth of penetration of 10 mm and a detection limit of fewer than 300 cells. WATSON incorporates a scanning system that can map the distribution of organics and microbes over a 75 by 25 mm area. WATSON demonstrates a sensitive fluorescence mapping technique for organic and microbial detection in icy environments including terrestrial glaciers and ice sheets, and planetary surfaces including Europa, Enceladus, or the martian polar caps.

1. Introduction

Once considered inhospitable to life, in recent decades icy environments on Earth have been found that preserve organic material and even harbor active microbial communities (Priscu et al., 1998, 1999; Siegert et al., 2001; Priscu, 2005; Santibáñez et al., 2005, 2018; Krembs et al., 2011; Uhlig et al., 2015). Glacial ice on Earth is known to contain organic material (Barnes et al., 2003; Barletta et al., 2012), including bacteria, algae, viruses, plant fragments, pollen grains, and black carbon. Glacial ice cores from Antarctica have been found to contain bulk cell concentrations of 10² to 10³ cells/mL (Abyzov et al., 1998, 2004; Christner et al., 2006), and microbes have been observed in 3590 m deep accretion ice above Lake Vostok (Priscu et al., 1999; Siegert et al., 2001). Higher cell concentrations, up to 10⁸ cells/mL, have been measured from silty regions on the Greenland ice sheet (Miteva et al., 2009), and the total number of bacterial cells in the Greenland and Antarctic ice sheets is estimated to be 9.61 × 10²⁵ cells (Priscu and Christner, 2004). Habitable spaces are known to persist in icy environments, including cryoconite holes on glacial surfaces that contain organic and mineral sediment and microenvironments within veins between ice grain boundaries (Maccario et al., 2015; Cook et al., 2016; Smith et al., 2016). Organic material, including cells, and geochemical compounds including sulfates and nitrates can become extruded into the fluid-filled veins and boundaries between grains during the recrystallization of snow into glacial ice (Barletta et al., 2012; Maccario et al., 2015), increasing the local concentration of these materials by a factor of 10⁴ to 10⁵ (Mader et al., 2006). In this concentrated liquid environment, active metabolic processes have been documented (Boetius et al., 2015), although at a slowed rate (Bakermans and Skidmore, 2011) compared to temperate conditions. While examples of these processes have been observed in glacial ices, the full extent to which microbes are adapted to these environments and their in situ activity remains largely elusive (Morgan-Kiss et al., 2006; Priscu et al., 2008). As the extent of active biology within glacial ice is not well understood, further work is required to constrain the chemical gradients, nutrients, and biosignatures and their distribution in ice. Understanding these aspects of icy habitats on Earth will inform the search for life throughout the Solar System, including the icy crusts of ocean worlds such as Europa and Enceladus as well as the polar caps of Mars (Jepsen et al., 2007; Priscu and Hand, 2012).

The study of organic material in ice through laser-induced fluorescence is well established (Bramall, 2007; Rohde and Price, 2007; Rohde, 2010; Pautler et al., 2012; D'Andrilli et al., 2017). Fluorescence techniques, including excitation emission matrices (EEMs) are capable of trace aromatic organic detection and characterization of dissolved organic matter (Bhartia et al., 2008; Barker et al., 2010; Johnson et al., 2011). However, in these and other established methods of ice analysis, bulk samples are typically melted and concentrated for analysis, resulting in a loss of the spatial context of organic deposition. In situ techniques, including Raman spectroscopy, have been demonstrated to enable detection of organic material in ice without melting the sample (Rohde et al., 2008; Böttger et al., 2017). Deep-UV (DUV) techniques have also been demonstrated to distinguish microbes and other organic compounds based on their fluorescence spectra (Bhartia et al., 2010; Eshelman et al., 2015; Eshelman et al., 2018; Shkolyar et al., 2018), and field DUV fluorescence instruments have been deployed to study deep marine biospheres, successfully detecting microbe-like signatures (Bhartia et al., 2010; Salas et al., 2015). The work described here builds on prior field instrumentation using DUV fluorescence to study bacteria in situ (Salas et al., 2015) and in icy environments (Rohde et al., 2008).

The Wireline Analysis Tool for Subsurface Observation of Northern ice sheets (WATSON) is an instrument under development at NASA's Jet Propulsion Laboratory in collaboration with The University of Southern California, Montana State University, and Honeybee Robotics. WATSON incorporates DUV fluorescence spectroscopy along with a fine-scale 2D mapping capability to search for and determine the spatial distribution of organic material. WATSON leverages technology developed for SHERLOC, the Mars 2020 DUV Raman and fluorescence spectrometer (Beegle et al., 2015), repackaged into a 101.6 mm diameter tube 1.2 m long that can be lowered into an ice borehole. Like SHERLOC, WATSON scans a pulsed 248.6 nm laser across a sample surface, collecting fluorescence emission in a backscatter geometry and producing a spectral map of fluorescent materials. The primary science objective of WATSON is to determine the spatial distribution of organic matter, including microbes, present in glacial ice. WATSON is intended to achieve this objective by in situ measurements of the ice borehole wall, obtaining spectral maps to spatially correlate organic material and biosignatures to inorganic minerals and the geochemical environment throughout the depth of the borehole.

The 248.6 nm excitation wavelength of the WATSON laser was selected for the ability to electronically excite the absorption bands of organic molecules containing aromatic rings, including those found in bacteria, resulting in strong laser-induced fluorescence in the 275–450 nm range (Bhartia et al., 2008). This technology is well suited for icy environments, as ice does not absorb significantly between 248.6 and 450 nm (Onaka and Takahashi, 1968). At the millimeter- to centimeter-scale depth of field of WATSON's spectrometer, bubble-free ice, analogous to blue ice found in a glacial ice environment, is effectively transparent, as demonstrated in Section 4.5. The primary spectral signature from ice, the Raman O-H stretching mode, presents as a narrow band between 268 and 274 nm (3000 and 3700 cm⁻¹) (Cross et al., 1937) and can provide morphological information about the physical makeup of the ice matrix (e.g., features including bubbles, cracks, and veins) while not overlapping with fluorescence signals from other materials.

Detection of organics, microbes, and potential biosignatures on Solar System bodies and their spatial distribution are fundamental capabilities required to meet NASA's strategic goals. This capability was highlighted by the Mars 2020 Science Definition Team report (Mustard et al., 2013) and subsequent Mars 2020 payload selection that incorporated instruments such as SHERLOC, PIXL, and SuperCam to use noncontact spectroscopic methods to assess the distribution of organics, minerals, key elements, and potential biosignatures. More recently, the Europa Lander Science Definition Team (Hand et al., 2017) highlighted the need for spectroscopic analysis of organics using SHERLOC as an example for the strawman payload. The primary goal of WATSON is in situ determination of the spatial distribution of microbial and organic material in a borehole. Subsurface in situ analysis is a necessary step for exploration of planetary environments including the northern polar cap on Mars, the icy surfaces of Europa, the deposited ice layers near Enceladus' plumes, and the organic and aqueous ices of Titan (National Research Council, 2011). In situ characterization of subsurface ice will lead to a better understanding of life in ice and constrain our understanding of how it can survive and be preserved in the icy regions of planetary bodies (e.g., Mars polar region, Europa, and Enceladus).

Here we present the instrument design and operations of WATSON and discuss the results of laboratory experiments intended to determine the sensitivity to organic material and the ability to detect organic material under various depths of ice. Interrogation of the ice with a single laser pulse is shown to result in sufficient emission to characterize the morphology of the ice and detect microbial cells in analogs to terrestrial glacial ice. These experiments demonstrate the utility of a field-deployable compact DUV borehole instrument for the detection and mapping of organic material in a natural glacial ice system. The work described here was performed in anticipation of coupling the WATSON instrument to a Honeybee Robotics Planetary Deep Drill (Zacny et al., 2016) for a 2019 deployment to the Greenland ice sheet.

2. WATSON Instrument

2.1. Instrument overview

A schematic of WATSON is shown in Fig. 1. Excitation is provided by a 248.6 nm NeCu hollow-cathode laser, along with a custom laser power supply (LPS) capable of operating up to 160 Hz. Laser pulses are directed and focused onto the ice wall through a fused silica UV-transparent window (Fig. 1-W). The optical path is described in detail in Section 2.2. Linear and rotational motors (Fig. 1-C) move a carriage assembly containing the focusing optic and a rotating mirror, resulting in the ability to raster scan the laser over a 25 × 75 mm area with a spatial resolution up to 100 μm. Details of the carriage assembly are presented in Fig. 1-C. The carriage also includes an outward-facing camera that provides visible context images of the ice surface with a field of view of approximately 50 × 50 mm. Fluorescence emission is collected in a backscatter geometry and sent into the spectrometer and detector (Fig. 1-S). An onboard CPU, electronics, and power package coordinate the firing of the laser with the motorized stages and communication with the detector. Power and Ethernet are provided through a tether to a surface computer, through which commands are sent and data is returned. Figure 2A shows an image of the WATSON instrument, with Fig. 2B, 2C depicting the spectrometer and carriage shown in the schematic insets, respectively. The instrument specifications and typical acquisition parameters are listed in Table 1. A block diagram of the instrument subsystems is presented in Fig. 3.

FIG. 1.

WATSON optical schematic in the expected field configuration. Exiting the laser, the 248.6 nm laser line is isolated using two laser edge filters (F1, F2). Laser line mirrors (M1–M4) direct the beam to a third laser edge filter (F3) and into the optical path of the spectrometer. A triplet and aluminum mirror (L1 and M5) focus and direct the beam out of the instrument window (W) and onto the target. Fluorescence emission is collected in a backscatter geometry and focused (L2) into the spectrometer. Inset (S) shows (a) the focusing optic into the spectrometer, (b) 90 degree off-axis parabolic mirror, (c) 1800 lines/mm grating, (d) 100 mm focal length concave mirror, (e) PMT array detector and control electronics package. Inset (C) depicts (f) the focusing optic, (g) outward-facing context camera, (h) field of view of the context camera at the sample plane, (i) rotational stepper motor, and (j) focused laser spot. By moving the assembly depicted in (C) on a linear stage, the laser beam can raster across the sample surface.

FIG. 2.

(A) WATSON instrument in the lab alongside the instrument tube. (B) Miniature spectrometer, 32-channel PMT array detector, and detector electronics (see also Fig. 1–S for the spectrometer concept). (C) Motorized carriage on a linear stage containing the focusing objective, context camera, and rotational mirror for rastering the laser across the sample surface. Figure 1-C notates the carriage components. (D) WATSON deployment near Kangerlussuaq, Greenland. (E) The 14 cm × 4.5 m borehole used for WATSON testing. (F) Borehole with WATSON instrument.

FIG. 3.

WATSON block diagram presenting the subsystems of the instrument. A surface power supply and computer are connected to WATSON through an Ethernet and power tether. Inside WATSON, a CPU and power distribution unit (PDU) control and power the 32-channel PMT detector, linear and rotational motors for rastering the laser across the sample, and the LPS. The carriage subassembly block is shown in Fig. 2-C and is noted by a dashed box in Fig. 1-C.

Table 1.

WATSON Instrument Parameters

Subsystem	Property	Value
Laser	Excitation wavelength	248.6	nm
	Pulse energy	<5	μJ
	Pulse width	35	μs
	Laser repetition rate	40	Hz
	Spot size at sample	∼100	μm
Spectroscopy	Working distance	0.125	in.
	Spectral range	265–440	nm
	Spectral resolution	5.5	nm
	PMT high voltage	800	V
Instrument	WATSON diameter	101.6	mm
	WATSON length	1.2	m

2.2. Optical concept

The optical path through the instrument is shown in Fig. 1. Exiting the laser, the 248.6 nm line is isolated; and other NeCu transitions, primarily a 252.9 nm line, are removed by two custom edge filters (F1, F2). The edge filters transmit light above 249 nm and reflect >99.5% of the 248.6 nm emission. Following isolation of the 248.6 nm line, a series of mirrors (M1–M4 in Fig. 1) direct the light to an additional edge filter (F3), which co-boresights the laser with the optical path of the spectrometer. The three edge filters were developed from commercially available filters to have a steep 16° angle of incidence (AOI). The steep incidence angle allows the optical path to have a compact geometry and reduces the length and volume of the instrument. The laser beam is then focused onto the ice wall, located 3.175 mm from the outer surface of the instrument, with a 90 mm EFL triplet lens assembly. The focused spot at the target is approximately 100 μm in diameter. An additional mirror (M5), placed after the triplet and fixed to the rotating stepper motor, is used in conjunction with the linear stage to raster the laser spot across the sample surface. Fluorescence emission from the sample is collected by the focusing triplet (L1) and directed onto the spectrometer entrance slit by a second triplet with a 45 mm EFL (L2).

The instrument optics allow for collection of emitted light not just from the sample surface but at millimeter- to centimeter-scale depths within the ice, demonstrated in Section 4.5 below. This results in an in situ measurement capability, where organic material can be observed without any alteration to the local environment.

2.3. Spectroscopy

The spectrometer design is shown as an inset in Fig. 1-S. Entering the spectrometer, fluorescence emission is collimated by using a 50 mm focal length 90 degree off-axis parabolic mirror (Fig. 1-b). An 1800 lines/mm grating (Fig. 1-c) and a 100 mm focal length concave circular mirror (Fig. 1-d) are used to achieve a dispersion across the detector from 270 to 440 nm. A Hamamatsu H7260-106 32-channel photomultiplier tube (PMT) array connected to an electronics assembly developed by Photon Systems, Inc. (Fig. 1-e) is used as a detector. Each of the 32 PMTs, arranged in a linear array, has dimensions of 0.8 by 7.0 mm, resulting in a combined sensor array size of 31.8 by 7.0 mm. This includes a 200 μm gap between adjacent anodes. An onboard computer communicates with the electronics assembly and triggers the PMT array in coordination with the laser pulses. Background spectra are acquired by integrating the signal on the PMT array between laser pulses. This capability also allows the instrument to serve as a passive UV spectrometer, measuring ambient light within the ice.

The spectral range of the spectrometer, 270–440 nm, encompasses laser-induced fluorescence generated by a wide range of astrobiologically relevant organic compounds, including aromatic amino acids found in proteins, single-ring and multi-ring aromatic hydrocarbons, and aromatic heterocycles. Additionally, the higher-energy region of this range contains the Raman O-H stretching band, a strong spectral feature found in liquid water and water ice. The overall size of the spectrometer was kept to a minimum, with a length of about 140 mm. The magnification of the laser spot at the detector was kept close to 1 × to obtain adequate spectral resolution. Much of the optical path travels through the center axis of the instrument, aiding alignment of the system as the motorized carriage moves along its z-range. A PMT array was selected as a detector to avoid the thermal management (i.e., active thermoelectric cooling or passive liquid N₂) requirements associated with a CCD. While CCDs can provide a higher spectral resolution than a PMT array, achieving adequate sensitivity to measure weak fluorescence signals would require significant cooling of the detector and would be challenging in the hermetically sealed tube.

2.4. Telemetry

To spatially define a sample position within an ice borehole, both the instrument depth and roll angle must be recorded. Depth information is provided external to the instrument, for example by an encoder attached to the instrument tether, and received by the surface computer during data acquisition. The instrument roll angle is measured by a combination of an internal gyroscope, magnetometer, and accelerometer. In WATSON, data from the internal gyroscope, magnetometer, and accelerometer sensors are continuously recorded at 20 Hz as spectral data is acquired. Temperature sensors placed at the LPS and near the onboard computer measure the air temperature and are monitored through a LabVIEW interface to determine if the LPS is at a risk of overheating. To avoid condensation inside WATSON, a humidity sensor combined with temperature readings provides dew point measurements using worst-case assumptions that the instrument tube temperature is at the ambient borehole temperature.

2.5. Operations

WATSON has two primary modes of operation: extended linear sidewall mapping of the ice borehole as the instrument is lowered or raised from the borehole, and fine-scale x/y spectral mapping over a 75 × 25 mm region when the instrument is stationary. These operations, described in the remainder of this section, were successfully tested in a −10 °C cold freezer during the experiments described in Section 4. In that environment, the instrument was cold-soaked for periods lasting up to 5 days. The internal temperature of the instrument during continuous operation, with repeated scans each lasting 45 min, plateaued below +30 °C, well below the +50 °C upper operational limit. This indicates that WATSON could likely operate continuously in cold environments such as terrestrial ice boreholes at a stable and safe internal temperature.

2.5.1. Ice borehole mapping

The total area of the interior of a 108 mm (4 in.) diameter × 100 m deep ice borehole exceeds 65 m², while the focused laser spot of WATSON is under 150 μm. As the distribution of organic and biological material in glacial ice is expected to be both low in concentration and spatially localized, it is necessary to have a method to locate regions in the ice that contain organic material. As WATSON is raised or lowered through an ice borehole, the instrument can fire the laser and collect data continuously, correlating spectra with depth and roll data from the sensor suite. Over multiple trips up and down, the data collected would populate a point-cloud with spectral information, giving broad fluorescence trends that can be used to down-select regions within the ice where detailed mapping would provide a more high-resolution investigation of organic material that is present.

2.5.2. Fine-scale spectral mapping

The linear and rotational motors within WATSON are used to raster the laser over a 25 × 75 mm sample surface, generating a fluorescence spectral map. The motor speed and laser repetition rate can be coordinated to define the spatial distance between sampling points, with typical parameters resulting in a 100–200 μm distance between points. Similar to the operations described in Section 2.5.1, measurements of depth and roll can locate the high-resolution spectral map within the borehole. Spectral maps are overlaid on visible imagery provided by the context camera to help interpret features observed in the spectra. Typically, the laser fires as the motors continuously move, resulting in a single laser pulse per spatial sampling point. If weak fluorescence was observed in a natural system during a field deployment, multiple pulses per point (up to 10,000) could be obtained to increase the signal-to-noise ratio. Spatial binning of the resulting spectral map can also increase the signal to noise, for example by averaging all spectra correlated with a morphological feature present in the visible context image, such as a crack, bubble, or vein within the ice. A 25 × 75 mm scan with a spatial resolution of 200 μm contains 46,875 points. Assuming a conservative interrogation depth into the ice (i.e., 1 cm), averaging these spectra can be considered a bulk measurement of approximately 2 cm³ and could reveal weak fluorescence signals not detectable in a single sampling point. A spectral map of this size with the instrument parameters listed in Table 1 can be obtained in 30 min.

3. Materials and Methods

Laboratory ice samples containing organic compounds were prepared for two experiments, with the goals of determining the sensitivity of WATSON to organic material in ice and the ability of WATSON to detect organic material at increasing ice thickness. Three organic materials were selected for study: aqueous solutions of the aromatic amino acids dl-tyrosine and dl-tryptophan (Sigma-Aldrich), and a suspended culture of Escherichia coli (K-12 MG1655). Experiments were performed in a −10 °C walk-in cold room. For all measurements, the WATSON instrument was oriented with the laser and spectrometer direction of observation pointing downward onto a motorized XY platform where ice samples were placed. After obtaining a visible context image of the sample surface, the WATSON laser was scanned across the sample surface while emitting laser pulses at 40 Hz. The detector electronics were triggered to the laser pulse, generating one spectrum for each pulse. The speed of the motorized stages resulted in a 200 μm spatial separation between points on the sample. The detector was triggered between laser pulses to obtain a dark spectrum used for background subtraction. Other instrument parameters used for these experiments are listed in Table 1. This procedure is similar to that anticipated for operations inside a borehole. Spectral maps generated from these scans were used for the subsequent data analysis in Section 4. All spectral maps presented here were obtained with a single laser pulse per point. The subsequent analyses and detection limits that are discussed are therefore based on these single-pulse measurements.

3.1. Sample preparation

All glassware and metal apparatus used during the sample preparation and scanning processes were cleaned with soap, washed with deionized water, and baked for 12 h at 550 °C to remove organic contamination. Initial concentrations for each of the aromatic amino acids were 0.98 mM dl-tryptophan and 1.1 mM dl-tyrosine. To grow the E. coli, a frozen glycerol bead stock (K12 MG1655) was inoculated into 1 mL of defined minimal media (M9 broth: ATCC medium 2511) supplemented with 0.4% glucose and 0.1 mg/mL thiamine and incubated (37 °C) with shaking (225 rpm) overnight. After 24 h the initial stock was transferred to 9 mL of M9 and incubated overnight under the same conditions. Three milliliters of this culture was transferred to 27 mL of fresh M9 media and harvested in stationary phase. Cells were fixed in 4% paraformaldehyde (PFA) at room temperature for 60 min, pelleted and washed three times in phosphate-buffered saline (PBS: 137 mM NaCl, 10 mM phosphate, 2.7 mM KCl, pH 7.4.), and stored in PBS at 4 °C. An aqueous E. coli suspension was generated by successively resuspending the cells at PBS-to-water ratios of 75:25, 50:50, 25:75, and 0:100 to minimize osmotic shock and possible cell rupture. After washing, the cells were suspended in deionized water to a concentration of 1.1 × 10⁹ cells/mL, and the cell density was determined by measuring the optical density at 600 nm (OD600). OD600 was calibrated to E. coli grown in M9 medium and compared to microscopic counts. Dilution series of the aromatic amino acids and E. coli were then made in triplicate at relative concentrations of 100%, 50%, 25%, 5%, 2.5%, and 0.5%. Following suspension in deionized water, the cells were observed under a light microscope to confirm that the cell integrity survived the cleaning process. A deionized water control dilution series was prepared by using the same transfer techniques as those used during dilution to verify that the sample preparation procedure did not introduce any organic contamination at levels that could be detected by WATSON.

3.1.1. Sample preparation to determine sensitivity to organic material in ice

In a −10 °C cold room, a series of 6.5 μL drops of each dilution were deposited on a cleaned aluminum block in a 6 × 4 grid, with a 4 mm spacing between drops. The drops were pipetted onto the prechilled aluminum block inside the −10 °C cold chamber in order to freeze and scan the drops as quickly as possible, thereby minimizing sublimation. The first three columns contained the two amino acid standards selected for study plus the E. coli, while the fourth column contained a water control. Drops placed on the aluminum typically reached a supercooled liquid state and were frozen by touching the base of the drop with the tip of a pipette that had been in contact with a pure ice control drop. Three of these blocks were prepared, each using one of the triplicate dilution series. WATSON was then used to obtain a spectral map over the sample surface, using the instrument parameters described in Table 1. Images of the ice drops on the aluminum surface are shown in Fig. 4.

FIG. 4.

Ice drops with varying quantities of dl-tryptophan, dl-tyrosine, and E. coli were scanned to investigate the ability of WATSON to detect organic material in ice. In each of the three sections, (A) presents a context image of the 6.5 μL ice drops; (B) presents a fluorescence map of the drops, generated by averaging 64 points around the center of each drop. The total amount of each organic material deposited in each drop is annotated on the left axis. (C) Presents average spectra of the sampled points within each drop. The average spectrum of a pure water control drop is shown as a gray dashed line. Vertical bars indicate the spectral region that was integrated to generate the fluorescence maps in (B). An asterisk indicates the position of the Raman O-H stretching mode, visible as a peak between 268 and 279 nm. The calculated mass is for the amount of test material deposited in the ice droplet.

3.1.2. Samples prepared to determine the ability to detect organic material at varying depths in clear and opaque ice

Blocks of clear ice and opaque ice with small bubble inclusions with approximate dimensions of 100 × 150 × 50 mm were obtained from a commercial ice vendor. These ices served as analogs to clear glacial ice and glacial ice with morphological features that could scatter incident radiation. The structure of the opaque ice was visually inspected by using a CANON EOS 70D camera equipped with a lens providing a 13 μm/pixel resolution. The ice was determined to have a size distribution that included ∼500 μm diameter bubbles with a number density of approximately 30/cm³, 50–200 μm diameter bubbles with a number density of approximately 5000/cm³, and a dense homogenous distribution of <50 μm diameter bubbles at or below the camera resolution. These ice blocks were formed into a wedge with dimensions shown in Fig. 6 using a clean aluminum surface on a hot plate. The purpose of the wedge was to provide an angled surface on which to deposit organic material in order to observe organics under varying thicknesses of ice. The blocks were oriented with the angled surface of the wedge as an overhang, and organics were observed top-down through the level surface of the wedge block. A wedge angle of 35° was used, and a grid of 4 × 6 6.5 μL drops was deposited into melted pits on the wedged surface. The pits were produced by pressing a 4 × 6 grid of 1.5 mm diameter heated steel dowel pins onto the ice surface. This angle and grid spacing resulted in a maximum organic deposition depth of 2.4 cm. This geometry is shown in detail in Fig. 6E.

4. Results

4.1. Sensitivity to organic material in ice

The ice drops with varying concentrations of organic material, described in Section 3.1.1, were scanned with WATSON, resulting in the spectral maps shown in Fig. 4. The three sections of Fig. 4 each contain a visible context image of the 6.5 μL drops (Fig. 4A), a spectral map with a 200 μm spatial resolution (Fig. 4B), and a plot of the average spectra of 64 sampled points around the center of each drop (Fig. 4C). Intensity values of the spectral maps were calculated by integrating the wavelength region denoted by the vertical lines in the spectral plots. The intensities of the spectral maps and plots are presented on a logarithmic scale to accommodate the dynamic range of the PMT detectors. An asterisk denotes the position of the Raman O-H stretching mode of the ice, which is observed as a peak between 268 and 274 nm (Cross et al., 1937). The spectrum of the water control droplet is displayed as a dashed line on each spectral plot.

The fluorescence intensities measured from the drops decrease as the concentration of organic material in the drops decreases. The quantity of organic material deposited within each drop is noted on the axis between the context images (Fig. 4A) and the spectral plots (Fig. 4B). Properties of the fluorescence signatures of the aromatic amino acids, including their spectral profiles and peak emission wavelengths, agree with the known profiles of these compounds (Teale and Weber, 1957). The fluorescence spectrum of the E. coli agrees with previously reported results (Seaver et al., 1998). Fluorescence from the E. coli is primarily due to the presence of aromatic amino acids, which are present in the cell proteins. The bacteria (E. coli) fluorescence spectrum is a combination of tryptophan and tyrosine amino acids in proteins, while the spectrum of bacterial spores is dominated by tyrosine-like residues (Pandey and Aronson, 1979; Leblanc and Dufour, 2002; Ammor, 2007; Bhartia et al., 2008). While the aromatic amino acids in the bacterial cells and spores are the l enantiomer, the magnitudes of the electronic transitions are similar for both d and l isomers, and as such, the fluorescence spectra for both isomers are identical.

4.2. Modeled instrument performance compared to measured calibration curves

To determine whether the instrument met its design sensitivity requirements, a model of the instrument performance was applied that included the known fluorescence cross sections of the compounds selected for study, as well as the properties of the optics and detector electronics. This model is plotted in Fig. 5 as a dashed gray line and is described in the remainder of this section. The number of fluorescence photons that reach a single PMT element in the 32-PMT array following illumination by a single laser pulse is given by Eq. 1 \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*}{ \phi _ { { \rm { fluor } } } } = G \; { \epsilon _ { { \rm { ice } } } } \; \gamma { \rm { \; \sigma } } \frac { { \; { F_ { \rm { w } } } \; { \rm { \Omega \; } } { N_ { \rm { m } } } } } { A } \tag { 1 } \end{align*} \end{document}

FIG. 5.

Calibration curves for dl-tryptophan, dl-tyrosine, and E. coli at their respective λ_max. Points show observed values, including those excluded from the curve fitting. The solid blue line shows the fitted power function line, and the dashed gray line shows the expected instrument performance from the sensitivity model described in Section 4.2. The dashed blue lines are the limits of detection (LOD) and quantitation (LOQ) at the emission maximum. Error bars represent the standard deviation of the 64 points sampled in each drop.

where \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$${ \phi _{{ \rm{fluor}}}}$$ \end{document} is the number of fluorescence photons generated by a single laser pulse that reach the detector in the desired band, G is the combined throughput of the collection optics, \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$${ \epsilon _{{ \rm{ice}}}}$$ \end{document} is a sample scattering factor, \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$$\gamma$$ \end{document} is the number of photons generated by the laser at the sample, \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$${ \rm{ \sigma }}$$ \end{document} is the fluorescence cross section of the organic material at the desired wavelength, \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$${F_{ \rm{w}}}$$ \end{document} is the spectral width of a single PMT (5.5 nm), \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$${ \rm{ \Omega }}$$ \end{document} is the collection solid angle, \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$${N_{ \rm{m}}}$$ \end{document} is the number of molecules illuminated, and A is the area of the spot illuminated by the laser. The sample scattering factor \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$${ \epsilon _{{ \rm{ice}}}}$$ \end{document} was included to account for the scattering properties of the ice drops and was determined by measuring the attenuation of the laser beam through a control drop containing pure water. This model assumes that there is no photobleaching of the fluorescing material within the sampling volume, which would, if present, result in a lower measured fluorescence intensity compared to the model. This assumption is likely valid, as the laser is rastered continuously across the sample surface and only a single laser pulse is used to generate a spectral “pixel” in the map. The fluorescence cross section of dl-tyrosine was extrapolated from the absorptivity at 248.6 nm, the quantum yield of the compound, and published values of the dl-tryptophan cross section (Faris et al., 1997; Lakowicz, 2006). The cross section of E. coli was determined from published values (Faris et al., 1997; Seaver et al., 1998) and is listed in Table 2. \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$${ \phi _{{ \rm{fluor}}}}$$ \end{document} can be converted to counts read out by the analog to digital converter (ADC) and directly compared to experimental measurements by considering the properties of the PMT and detector electronics, described in Eq. 2. \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*}{ I_ { { \rm { ADC } } } } = \left( { { \phi _ { { \rm { fluor } } } } + { \phi _ { { \rm { noise } } } } } \right) { \frac { \; { Q_ { { \rm { PMT } } } } \;G \; { K_ { I = { I_ { { \rm { anode } } } } } } } { \eta } } \tag { 2 } \end{align*}\end{document}

Table 2.

Calibration Curve Fitting Parameters and Determined LOD/LOQ for Each Material

Material	λ_max (nm)	m^a	b^a	LOD	LOQ	Cross section
Material	λ_max (nm)	m^a	b^a	LOD	LOQ	(cm² nm⁻¹ mol⁻¹ sr⁻¹)
E. coli ^a	332	4.75E-5	0.96219	4.72E6 (± 1.20E6)	8.89E6 (± 2.26E6)	1.00E-13
E. coli ^a	332	(± 4.58E-5)	(± 0.0373)	cells/mL	cells/mL	1.00E-13
dl-tryptophan	349	5.41E7	0.99164	2.07E-6 (± 1.36E-6) M	3.89E-6 (± 2.55E-6) M	1.00E-21
dl-tryptophan	349	(± 4.71E7)	(± 0.0663)	2.07E-6 (± 1.36E-6) M	3.89E-6 (± 2.55E-6) M	1.00E-21
dl-tyrosine	306	6.44E7	1.2971	2.09E-5 (± 3.98E-6) M	3.29E-5 (± 6.26E-6) M	1.00E-22
dl-tyrosine	306	(± 2.23E7)	(± 0.0320)	2.09E-5 (± 3.98E-6) M	3.29E-5 (± 6.26E-6) M	1.00E-22

For cell units of reported values and parameters given in cell density (cells per cm⁻³). For molecules, given in moles per liter (M).

\documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$${I_{{ \rm{ADC}}}}$$ \end{document} is the intensity measured in ADC counts, \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$${ \phi _{{ \rm{noise}}}}$$ \end{document} is the dark current of the PMT, represented as equivalent number of photons at the cathode, \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$${Q_{{ \rm{PMT}}}}$$ \end{document} is the quantum efficiency of the PMT with units of photoelectrons produced per incident photon, G is the gain between the PMT cathode and anode at the selected supply voltage, \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$$\eta$$ \end{document} is the smallest measurable charge by the detector electronics (in units of coulombs per count), and K is the linearity of the PMT response at the corresponding anode current. Equations 2 and 3 are first-order models intended to determine whether the instrument is significantly under- or overperforming the design expectations.

To validate the instrument performed according to the model, calibration curves based on experimental data were determined by examining the measured signal intensities of the water ice drops detailed in Section 3.1, shown in Fig. 5. The individual drops were roughly of uniform size and thickness and appeared homogeneous on visual inspection. Around the center of each drop, 8 × 8 spectra (64 points) were averaged to obtain the mean spectrum. The three spectral maps resulting from the triplicate dilution series of ice drops were used to generate three values for each of the six concentrations prepared. The measured intensity at the maximum emission wavelength for each material at varying concentrations was plotted to generate the calibration curve. For tyrosine, the peak emission wavelength (λ_max) was 306 ± 3 nm, for dl-tryptophan λ_max was 349 ± 3 nm, and for E. coli λ_max was 332 ± 3 nm. The limit of detection ( \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$${I_{{ \rm{LOD}}}}$$ \end{document} ) in units of counts was determined at each wavelength of interest by taking the signal background level ( \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$${I_{{ \rm{Background}}}}$$ \end{document} ) plus 3.3 times the signal standard deviation ( \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$$\sigma$$ \end{document} ) from frozen pure H₂O drops, shown in Eq. 3. \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*}{I_{{ \rm{LOD}}}} = {I_{{ \rm{Background}}}} + 3.3 \sigma \tag{3} \end{align*} \end{document}

The limit of quantitation ( \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$${I_{{ \rm{LOQ}}}}$$ \end{document} ) in units of counts was determined at each wavelength of interest by taking the signal background level ( \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$${I_{{ \rm{Background}}}}$$ \end{document} ) plus 10 times the signal standard deviation ( \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$$\sigma$$ \end{document} ) from frozen pure H₂O drops, shown in Eq. 4. \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*}{I_{{ \rm{LOQ}}}} = {I_{{ \rm{Background}}}} + 10 \sigma \tag{4} \end{align*} \end{document}

For dl-tyrosine, the measured intensity signals from the lowest concentration examined were deemed below the LOQ and were not used in the calibration curve fit. For E. coli, the suspension at a cell density of 1.1 × 10⁹ cells/mL was noted to be visibly cloudy prior to freezing, and these measurements were also not used in the calibration curve fit. Concentration points above 1 mM were discarded when fitting, as they represent unrealistic values for a natural system and the signal intensities were close to the detector saturation point. The resulting points were fit to a power law curve of the form shown in Eq. 5 \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} y = A{x^b} \tag{5} \end{align*} \end{document}

where y is the measured signal intensity, x is the sample concentration/density in M (for molecules) or cells/mL (for cells), and A and b are fitted values. To determine A and b, a linear fit of plotted log₁₀ values of measured signal and concentration was generated from the 15 E. coli, dl-tyrosine, and dl-tryptophan measurements. The error of the slope and intercept from the fit to 1 standard deviation was converted to high and low formula values and compared to the formula using the derived parameters without added error. The higher of the two returned difference values was propagated and was used in range reporting of the predicted parameter values. It was noted that the relative residuals of the predicted versus observed values were well distributed for the fitted calibration curves. Figure 5 shows the experimental values and fitted curves for E. coli, dl-tryptophan, and dl-tyrosine. The curves are based on the observed measurements over the concentration range tested; while no obvious inflection points were noted, it is likely that subranges of the curves may fit a linear function. Curves for dl-tryptophan and E. coli have exponents near 1, close to linear within the concentration ranges used to derive the curve fit. A future refinement of the calibration curve for these molecules could reveal a linear/nonlinearity break (thus inflection) at lower signal counts, although for this manuscript a power function was used. Figure 5 presents a good agreement between measured and modeled intensities versus concentration, indicating that the instrument performed as expected. Differences between the performance model and experimental values are likely due to factors not accounted for in the model, such as scattering and stray light in the spectrometer, or, in the case of E. coli, due to a cross section determined from the literature and not directly measured.

4.3. Determination of LOD and LOQ concentrations for organic materials and E. coli

The calibration curves shown in Fig. 5 can be extrapolated to where they intersect the limit of quantitation (LOQ) and limit of detection (LOD) signal levels for the appropriate wavelength to determine the LOD/LOQ concentration limits. These values are shown in Table 2. The error range for the LOD/LOQ concentrations is based on the % RSE (relative standard error) for the calibration curve (Edgerley, 1998). The calibration curves show a unique correspondence that spans at least 2 orders of magnitude for the substances tested. This indicates consistency and predictability (within error limits) when comparing frozen samples of unknown concentrations/cell densities of the measured amino acids/E. coli cells that were prepared with the same techniques to generate drops with similar morphological parameters (e.g., shape across the 8 × 8 area, thickness of the frozen drop).

4.4. Absolute detection quantities for various molecules and E. coli

It is possible to convert the concentration or cell density values in Table 2 into absolute cell numbers or absolute number of molecules given an assumed interrogation volume into the ice drops. Our assumed interrogation volume is in the shape of a cylinder 150 μm in diameter with a height of 1.5 mm. This assumption is based on a 150 μm laser beam diameter, which is an average of the beam diameter focused through the center of a 1.5 mm volume, accounting for the objective focal length, the laser beam diameter at the objective, and the M2 value of the laser. The 1.5 mm cylinder height is the estimated height of the center of the 6.5 μL drops. The 8 × 8 grid of sample points was located near the center of each drop and therefore had a close to uniform thickness. This results in a cylindrical volume of 2.65 × 10⁻² μL, which is 0.41% of the amount of the total 6.5 μL drop volume. Given the LOD and LOQ values listed in Table 2, the absolute number (in units of moles, cells) in the estimated interrogation volume of 2.65 × 10⁻⁵ mL is listed in Table 3.

Table 3.

Absolute LOD and LOQ Quantities of Material in the 150 Micron × 1.5 mm Cylindrical Interrogation Volume Assuming Uniform Distribution of Materials throughout Sample

Material	E. coli	dl-tryptophan	dl-tyrosine
LOQ	236 (± 60) cells	6.2E10 (± 4E10) molecules	5E11 (± 1E11) molecules
LOD	125 (± 32) cells	3E10 (± 2E10) molecules	3.3E11 (± 6E10) molecules

4.5 Detecting organic material at various depths under clear and opaque ice

To determine the ability of WATSON to detect organic material under increasing thicknesses of ice, organic material was deposited on clear and opaque ice wedges, shown in Fig. 6, using the method described in Section 3.1.2. The clear and opaque ice serve as analogs to clear glacial ice and ice with morphological features such as bubbles resulting in significant scattering of light. Spectral maps were obtained in the geometry shown in Fig. 6E. Figure 7 presents an analysis of the rows of drops containing E. coli cells. Figure 7A, 7C shows visible top-down context images of the regions of the ice wedges highlighted in Fig. 6C, 6D, where the depth of the E. coli under the ice increases toward the right. Figure 7B, 7D presents spectral maps obtained by WATSON over this region with a 200 μm spatial resolution. Figure 7E presents the average fluorescence intensity of the signal measured within each drop as a function of the drop depth within the ice. The six data points, corresponding to the six drops, were normalized to the intensity of the drop at the depth nearest to the ice surface. The fluorescence intensity was measured at the emission maximum. WATSON was able to detect E. coli at the maximum depth tested, 2.4 ± 0.14 cm, without significant attenuation of the signal in the clear ice wedge. In the opaque ice wedge, the fluorescence signal from the E. coli was attenuated significantly, but the signal was still detectable at a 2.4 cm depth. Solid traces on Fig. 7E present the intensity of the Raman O-H stretching mode of the ice as a function of the ice depth. As the ice thickness increased, the Raman intensity from the clear ice increased, and the intensity from the opaque ice plateaued, indicating that the scattering properties of the ice matrix are likely the cause of the attenuation of the organic signal. Cracks, bubbles, or other morphological features in natural ice would also result in similar scattering of incident radiation. Even in a highly scattering environment, WATSON was able to detect E. coli at a 1 cm depth with less than an order of magnitude attenuation in the fluorescence intensity.

FIG. 6.

Images of clear and opaque ice wedges with 6.5 μL drops of organic material deposited in a grid pattern. The organic material was deposited on the angled undersurface of the wedges shown in (A) and (B); (C) and (D) display top-down views of the wedges, with the organic deposits visible. The diagram in (E) shows the illumination angle through the ice block and the angle and total depth of the ice block. Six hundred fifty nanograms of dl-tyrosine and dl-tryptophan were deposited in each pit in rows 1 and 2, respectively, 3.6 × 10⁶ E. coli cells were deposited in each pit in row 3, and the pits in row 4 contained pure water controls. The uncertainty in depth of a measurement of a point within a sample deposit was taken to be the range in depth of the ice pit that the sample was placed into (± 1.4 mm). White boxes in (C) and (D) indicated the scanned regions presented in Fig. 7.

FIG. 7.

Drops measuring 6.5 μL and containing 3.6 × 10⁶ E. coli cells each were deposited on the ice wedges shown in Fig. 6 and scanned to determine WATSON's capability to detect microbes under various thicknesses of ice. The context images and fluorescence scans shown in (A–D) are oriented similarly to the images in Fig. 6, with the depth of organic material increasing to the right. (E) The average signal intensity of each drop at λ_max = 332 ± 3 nm as a function of the depth of the drop in the ice wedge, in clear and opaque ice matrices. Vertical error bars represent the standard deviation of the points measured in each drop, while horizontal error bars represent an uncertainty in the depth due to the geometrical footprint of each drop, shown in detail in Fig. 6E. A smoothed trace of the Raman O-H stretch is overlaid as a measure of the strength of the signal originating from the ice.

5. Discussion

5.1. Implications for detecting cells in natural glacial systems

Bulk measurements of cell levels have previously been obtained in natural glacial ice systems and were determined to be between 10² and 10⁸ cells/mL (Mader et al., 2006; Miteva et al., 2009; Santibáñez et al., 2018). However, what is unaccounted for in these measurements is the submillimeter- to centimeter-scale spatial heterogeneity in the distribution of cells within the ice. In a natural system, microbes are likely not homogenously distributed, but rather extruded and concentrated into bubbles, veins, nodes, and grain boundaries present in the ice. Studies by Mader et al. have demonstrated in lab-grown ice that the local cell concentration within these regions could increase by a factor of 10⁴ to 10⁵ (Mader et al., 2006). The analysis presented in this work demonstrates that WATSON has sufficient sensitivity to detect cells in ice with a single laser pulse, with a limit of detection of 125 (± 32) cells in a sampling volume, corresponding to an average value of 4.72E6 (± 1.20E6) cells/mL.

The sampling volume of WATSON was designed to allow the instrument to detect material at a millimeter- to centimeter-scale depth within ice. Therefore, an ice vein, node, or grain boundary containing cells will occupy only a portion of the sampling volume of the instrument. The absolute LOD values in numbers of cells presented in Table 3 can be compared against a simplified model of a natural system, for example, when a 100 μm diameter ice node is in the laser beam path. In this case, if lower bounds are taken for the bulk cell load of the ice (10² cells/mL) and concentration factor into the node (10⁴), the resulting cell concentration in the node will be 10⁶ cells/mL. The number of cells in the node ( \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$${N_{{ \rm{cells}}}}$$ \end{document} ), assuming a homogeneous distribution of cells throughout the fluid network of the ice, can be calculated as the node volume × cell concentration (C), following Eq. 6. \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*}{ N_ { { \rm { cells } } } } = \frac { 4 } { 3 } \pi { r^3 } \times C \tag { 6 }\end{align*} \end{document}

For the lower bound values, the number of cells present in a node in this simplified model is less than one cell (with \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$$r = 50 \;{ \rm{ \mu m}}$$ \end{document} , \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$$C = {10^6}{ \rm{ \;cells}} / { \rm{mL}}$$ \end{document} ); and for upper bound values, the number of cells is 2094 (with \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$$r = 50 \;{ \rm{ \mu m}}$$ \end{document} , \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$$C = {10^{13}}{ \rm{ \;cells}} / { \rm{mL}}$$ \end{document} ). By obtaining spectral maps of a natural ice surface, WATSON would be able to scan and detect these locally concentrated regions of microbes, if present in quantities above the determined limit of detection.

For spatial investigations of ices with cell loads on the order of 10² cells/mL, similar to the cleaner ices measured terrestrially, it would be desirable for WATSON to have an improved limit of detection. A primary outcome of this work is a benchmarking of the instrument requirements needed to advance the detection limit to a level where cells can be observed in situ at the lower bounds of expected concentration. We identify three key areas that can be addressed. (1) The detector electronics used in this work exhibited a time-dependent oscillation in the voltage supplied to the PMT array. This introduced a sawtooth-like signal to the acquired data that accounted for approximately 20% of the standard deviation in the signal used to calculate the limits of detection and quantitation. (2) The f-number of the collection optics can be halved to increase the amount of light collected by a factor of 4. While this would increase the magnification at the spectrometer, spectral resolution would still be limited by the PMT anode dimensions. Coupled with a factor of 5 increase in laser output (realized from improved mirrors in the DUV laser), the instrument would collect up to 20 times more light generated from the laser interaction with the sample compared to the current configuration. (3) The sample preparation procedure employed to determine the instrument sensitivity, described in Section 3.1.1, generated small frozen drops of organic-containing material. When these drops were frozen, they became significantly opaque due in part to gasses present in the liquid. The drops therefore are a poor analog to clear glacial ice and are a highly scattering environment. While this effect was measured and accounted for in the sensitivity model described in Section 4.2 (with the sample scattering factor \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\usepackage{upgreek}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}$${ \epsilon _{{ \rm{ice}}}}$$ \end{document} ), a closer analog to clear glacial ice could result in up to an order of magnitude improvement in sensitivity for cells in glacial ice.

6. Conclusions

In the configuration described herein, WATSON provides DUV fluorescence spectroscopy in a compact instrument package capable of obtaining measurements in situ within an ice borehole. WATSON was developed to detect the spatial distribution of organic material containing aromatic rings, including microbes, and correlate these signatures to their environmental context. The results of laboratory experiments indicate that WATSON is sufficiently sensitive to identify cells with a single laser pulse in a glacial ice analog with a limit of detection of 125 (± 32) cells in a sampling volume and that WATSON can observe organic material at millimeter- to centimeter-scale depths in both clear ice and ice with morphological impurities that scatter the incoming radiation. WATSON demonstrates instrumentation for organic detection in terrestrial icy environments and supports the development of technology for future landed missions to the icy bodies of Europa, Enceladus, and the polar caps of Mars. WATSON represents the first step in the development of technology and measurement methodologies that provide the necessary information to infer past climate, distribution of organics, and habitability of both terrestrial extreme environments and icy planetary surfaces. By increasing our understanding of the limits and constraints on life in extreme environments such as the Greenland ice sheet, we gain a better understanding of how to detect, identify, and characterize life and life-related chemistry that may exist or may have existed on other Solar System bodies.

Footnotes

Acknowledgments

The authors would like to thank the reviewers of this work, including Chris McKay, whose feedback significantly improved the paper and in particular the discussion of the data presented. We would also like to thank Jim Wilcox, Ray Reid, Bill Hug, Prashant Oswal at Photon Systems Inc. for developing and supporting the PMT and detector electronics. Thanks to Rufus Simon, Matt James, and Vatche Vorperian at JPL for developing the laser power supply and providing integration assistance. Thanks to Haley Sapers and Ellie Hara for discussions and assistance with the experimental work presented here. Funding: This work was carried out in part at the Jet Propulsion Laboratory, California Institute of Technology under the NASA PSTAR program (NNH14ZDA001N-PSTAR). Government sponsorship is acknowledged. Copyright 2018.

Author Disclosure Statement

No competing financial interests exist.

Abbreviations Used

References

abyzov

S.S.

, Mitskevich

I.N.

, Poglazova

M.N.

, Barkov

N.I.

, Lipenkov

V.Y.

, Bobin

N.E.

, Koudryashov

B.B.

, and Pashkevish

V.M.

(1998) Antarctic ice sheet as a model in search of life on other planets. Adv Space Res, 22:363–368.

Abyzov

S.S.

, Hoover

R.B.

, Imura

, Mitskevich

I.N.

, Naganuma

, Poglazova

M.N.

, and Ivanov

M.V.

(2004) Use of different methods for discovery of ice-entrapped microorganisms in ancient layers of the Antarctic glacier. Adv Space Res, 33:1222–1230.

Ammor

M.S.

(2007) Recent advances in the use of intrinsic fluorescence for bacterial identification and characterization. J Fluoresc, 17:455–459.

Bakermans

and Skidmore

(2011) Microbial respiration in ice at subzero temperatures (−4°C to −33°C). Environ Microbiol Rep, 3:774–782.

Barker

, Klassen

, Sharp

, Fitzimons

, and Turner

(2010) Detecting biogeochemical activity in basal ice using fluorescence spectroscopy. Annals of Glaciology, 51:47–55.

Barletta

R.E.

, Priscu

J.C.

, Mader

H.M.

, Jones

, and Jones

(2012) Chemical analysis of ice vein microenvironments: II. Analysis of glacial samples from Greenland and Antarctica. Journal of Glaciology, 58:1109–1118.

Barnes

, Wolff

, Mallard

, and Mader

(2003) SEM studies of the morphology and chemistry of polar ice. Microsc Res Tech, 62:62–69.

Beegle

, Bhartia

, White

, DeFlores

, Abbey

, Wu

Y.-H.

, Moore

, Fries

, Burton

, Edgett

K.S.

, Ravine

M.A.

, Hug

, Reid

, Nelson

, Clegg

, Wiens

, Asher

, and Sobron

(2015) SHERLOC: Scanning Habitable Environments with Raman & Luminescence for Organics & Chemicals. In 2015 IEEE Aerospace Conference, IEEE, Piscataway, NJ, doi:10.1109/AERO.2015.7119105.

Bhartia

, Hug

, Salas

, Reid

, Sijapati

, Tsapin

, Abbey

, Nealson

K.H.

, Lane

A.L.

, and Conrad

P.G.

(2008) Classification of organic and biological materials with deep ultraviolet excitation. Appl Spectrosc, 62:1070–1077.

10.

Bhartia

, Salas

, Hug

, Reid

, Lane

, Edwards

, and Nealson

(2010) Label-free bacterial imaging with deep-UV-laser-induced native fluorescence. Appl Environ Microbiol, 76:7231–7237.

11.

Boetius

, Anesio

A.M.

, Deming

J.W.

, Mikucki

J.A.

, and Rapp

J.Z.

(2015) Microbial ecology of the cryosphere: sea ice and glacial habitats. Nat Rev Microbiol, 13:677–690.

12.

Böttger

, Bulat

S.A.

, Hanke

, Pavlov

S.G.

, Greiner-Bär

, and Hübers

H.-W.

(2017) Identification of inorganic and organic inclusions in the subglacial antarctic Lake Vostok ice with Raman spectroscopy. J Raman Spectrosc, 48:1097–4555.

13.

Bramall

N.E.

(2007) The Remote Sensing of Microorganisms, PhD dissertation, University of California, Berkely, CA, ProQuest Dissertations Publication Number AAI3275351.

14.

Christner

B.C.

, Royston-Bishop

, Foreman

C.M.

, Arnold

B.R.

, Tranter

, Welch

K.A.

, Lyons

W.B.

, Tsapin

A.I.

, Studinger

, and Priscu

J.C.

(2006) Limnological conditions in subglacial Lake Vostok, Antarctica. Limnol Oceanogr, 51:2485–2501.

15.

Cook

, Edwards

, Takeuchi

, and Irvine-Fynn

(2016) Cryoconite: the dark biological secret of the cryosphere. Progress in Physical Geography: Earth and Environment, 40:66–111.

16.

Cross

, Burnham

, and Leighton

P.A.

(1937) The Raman spectrum and the structure of water. J Am Chem Soc, 59:1134–1147.

17.

D'Andrilli

, Foreman

C.M.

, Sigl

, Priscu

J.C.

, and McConnell

J.R.

(2017) A 21000-year record of fluorescent organic matter markers in the WAIS Divide ice core. Climate of the Past, 13:533–544.

18.

Edgerley

(1998) Techniques for improving the accuracy of calibration in the environmental laboratory. In WTQA ‘98—14^th Annual Waste Testing & Quality Assurance Symposium, pp 181–187.

19.

Eshelman

, Daly

M.G.

, Slater

, and Cloutis

(2015) Time-resolved detection of aromatic compounds on planetary surfaces by ultraviolet laser induced fluorescence and Raman spectroscopy. Planet Space Sci, 119:200–207.

20.

Eshelman

, Daly

M.G.

, Slater

, and Cloutis

(2018) Detecting aromatic compounds on planetary surfaces using ultraviolet time-resolved fluorescence spectroscopy. Planet Space Sci, 151, doi:10.1016/j.pss.2017.09.003.

21.

Faris

G.W.

, Copeland

R.A.

, Mortelmans

, and Bronk

B.V.

(1997) Spectrally resolved absolute fluorescence cross sections for bacillus spores. Appl Opt, 36:958–967.

22.

Hand

K.P.

, Murray

A.E.

, Garvin

J.B.

, Brinckerhoff

W.B.

, Christner

B.C.

, Edgett

K.S.

, Ehlmann

B.L.

, German

C.R.

, Hayes

A.G.

, Hoehler

T.M.

, Horst

S.M.

, Lunine

J.I.

, Nealson

K.H.

, Paranicas

, Schmidt

B.E.

, Smith

D.E.

, Rhoden

A.R.

, Russell

M.J.

, Templeton

A.S.

, Willis

P.A.

, Yingst

R.A.

, Phillips

C.B.

, Cable

M.L.

, Craft

K.L.

, Hofmann

A.E.

, Nordheim

T.A.

, Pappalardo

R.P.

, and the Project Engineering Team. (2017) Report of the Europa Lander Science Definition Team, JPL D-97667, NASA, Washington, DC.

23.

Jepsen

, Priscu

, Grimm

R.E.

, and Bullock

(2007) The potential for lithoautotrophic life on Mars: application to shallow interfacial-water environments. Astrobiology, 7:342–354.

24.

Johnson

, Hodyss

, Bolser

, Bhartia

, Lane

, and Kanik

(2011) Ultraviolet-stimulated fluorescence and phosphorescence of aromatic hydrocarbons in water ice. Astrobiology, 11:151–156.

25.

Krembs

, Eicken

, and Deming

(2011) Exopolymer alteration of physical properties of sea ice and implications for ice habitability and biogeochemistry in a warmer Arctic. Proc Natl Acad Sci USA, 108:3653–3658.

26.

Lakowicz

J.R.

(2006) Principles of Fluorescence Spectroscopy, Springer, New York.

27.

Leblanc

and Dufour

(2002) Monitoring the identity of bacteria using their intrinsic fluorescence. FEMS Microbiol Lett, 211:147–153.

28.

Maccario

, Sanguino

, Vogel

, and Larose

(2015) Snow and ice ecosystems: not so extreme. Res Microbiol, 166:782–795.

29.

Mader

H.M.

, Pettitt

M.E.

, Wadham

J.L.

, Wolff

E.W.

, and Parkes

R.J.

(2006) Subsurface ice as a microbial habitat. Geology, 34:169–172.

30.

Miteva

, Teacher

, Sowers

, and Brenchley

(2009) Comparison of the microbial diversity at different depths of the GISP2 Greenland ice core in relationship to deposition climates. Environ Microbiol, 11:640–656.

31.

Morgan-Kiss

, Priscu

, Pocock

, Gudynaite-Savitch

, and Huner

(2006) Adaptation and acclimation of photosynthetic microorganisms to permanently cold environments. Microbiol Mol Biol Rev, 70:222–252.

32.

Mustard

J.F.

, Adler

, Allwood

, Bass

D.S.

, Beaty

D.W.

, Bell

J.F.

III , Brinckerhoff

W.G.

, Carr

, Des Marais

D.J.

, Drake

, Edgett

K.S.

, Eigenbrode

, Elkins-Tanton

L.T.

, Grant

J.A.

, Milkovish

S.M.

, Ming

, Moore

, Murchie

, Onstott

T.C.

, Ruff

S.W.

, Septon

M.A.

, Stelle

, and Trieman

(2013) Report of the Mars 2020 Science Definition Team, posted July 2013 by the Mars Exploration Program Analysis Group (MEPAG) at mepag.jpl.nasa.gov/reports/MEP/Mars_2020_SDT_Report_Final.pdf

33.

National Research Council. (2011) Vision and Voyages for Planetary Science in the Decade 2013–2022, The National Academies Press, Washington, DC.

34.

Onaka

and Takahashi

(1968) Vacuum UV absorption spectra of liquid water and ice. Journal of the Physical Society of Japan, 24:548–550.

35.

Pandey

N.K.

and Aronson

A.I.

(1979). Properties of the Bacillus subtilis spore coat. J Bacteriol, 137:1208–1218.

36.

Pautler

B.G.

, Woods

G.C.

, Dubnick

, Simpson

A.J.

, Sharp

M.J.

, Fitzsimons

S.J.

, and Simpson

M.J.

(2012) Molecular characterization of dissolved organic matter in glacial ice: coupling natural abundance ¹H NMR and fluorescence spectroscopy. Environ Sci Technol, 46:3753–3761.

37.

Priscu

and Christner

(2004) Earth's icy biosphere. In Microbial Diversity and Bioprospecting, edited by Bull

A.T.

, ASM Press, Washington, DC, pp 130–145.

38.

Priscu

and Hand

K.O.

(2012) The microbial habitability of extraterrestrial icy worlds: a view from Earth. Microbe, 7:167–172.

39.

Priscu

, Adams

E.E.

, Lyons

W.B.

, Voytek

M.A.

, Mogk

D.W.

, Brown

R.L.

, McKay

C.P.

, Takacs

C.D.

, Welch

K.A.

, Wolf

C.F.

, Kirshtein

J.D.

, and Avci

(1999) Geomicrobiology of subglacial ice above Lake Vostok, Antarctica. Science, 286:2141–2144.

40.

Priscu

, Tulaczyk

, Studinger

, Kennicutt

M.C.

II , Christner

, and Foreman

(2008) Antarctic subglacial water: origin, evolution and ecology. In Polar Lakes and Rivers, Oxford University Press, Oxford, UK, pp 119–135.

41.

Priscu

J.C.

(2005) Perennial Antarctic Lake Ice: A Refuge for Cyanobacteria in an Extreme Environment, Princeton University Press, Princeton, NJ.

42.

Priscu

J.C.

, Fritsen

C.H.

, Adams

E.E.

, Giovannoni

S.J.

, Paerl

H.W.

, McKay

C.P.

, Doran

P.T.

, Gordon

D.A.

, Lanoil

B.D.

, and Pinckney

J.L.

(1998) Perennial Antarctic lake ice: an oasis for life in a polar desert. Science, 280:2095–2098.

43.

Rohde

, Price

P.B.

, Bay

R.C.

, and Bramall

N.E.

(2008) In situ microbial metabolism as a cause of gas anomalies in ice. Proc Natl Acad Sci USA, 105:8667–8672.

44.

Rohde

R.A.

(2010) The Development and Use of the Berkeley Fluorescence Spectrometer to Characterize Microbial Content and Detect Volcanic Ash in Glacial Ice, PhD dissertation, University of California, Berkely, CA, ProQuest Dissertations Publication Number AAI3402647.

45.

Rohde

R.A.

and Price

P.B.

(2007) Diffusion-controlled metabolism for long-term survival of single isolated microorganisms trapped within ice crystals. Proc Natl Acad Sci USA, 104:16592–16597.

46.

Salas

E.C.

, Bhartia

, Anderson

, Hug

W.F.

, Reid

R.D.

, Iturrino

, and Edwards

K.J.

(2015) In situ detection of microbial life in the deep biosphere in igneous ocean crust. Front Microbiol, 6, doi:10.3389/fmicb.2015.01260.

47.

Santibáñez

, McConnell

, and Priscu

(2005) A flow cytometric method to measure prokaryotic records in ice cores: an example from the West Antarctic Ice Sheet Divide Drilling Site. Journal of Glaciology: Instruments and Methods, 62:655–673.

48.

Santibáñez

, Maselli

, Greenwood

, Grieman

, Saltzman

, McConnell

, and Priscu

(2018) Prokaryotes in the WAIS Divide ice core reflect source and transport changes between Last Glacial Maximum and the early Holocene. Global Change Biology, 24:2182–2197.

49.

Seaver

, Roselle

D.C.

, Pinto

J.F.

, and Eversole

J.D.

(1998) Absolute emission spectra from Bacillus subtilis and Escherichia coli vegetative cells in solution. Appl Opt, 37:5344–5347.

50.

Shkolyar

, Eshelman

, Farmer

, Hamilton

, Daly

M.G.

, and Youngbull

(2018) Detecting kerogen as a biosignature using colocated UV time-gated Raman and fluorescence spectroscopy. Astrobiology, 18:431–453.

51.

Siegert

, Ellis-Evans

, Tranter

, Mayer

, Petit

J.-R.

, Salamatin

, and Priscu

(2001) Physical, chemical and biological processes in Lake Vostok and other Antarctic subglacial lakes. Nature, 414:603–609.

52.

Smith

H.J.

, Schmit

, Foster

, Littman

, Kuypers

M.M.

, and Foreman

C.M.

(2016) Biofilms on glacial surfaces: hotspots for biological activity. NPJ Biofilms Microbiomes, 2, doi:10.1038/npjbiofilms.2016.8.

53.

Teale

F.W.

and Weber

(1957) Ultraviolet fluorescence of the aromatic amino acids. Biochem J, 65:476–482.

54.

Uhlig

, Kilpert

, Frickenhaus

, Kegel

J.U.

, Krell

, Mock

, Valentin

, and Beszteri

(2015) In situ expression of eukaryotic ice-binding proteins in microbial communities of Arctic and Antarctic sea ice. IMSE J, 9:2537–2540.

55.

Zacny

, Shara

, Paulsen

, Mellerowicz

, Spring

, Ridilla

, Hedlund

, Sharpe

, Bowsher

, Hoisington

, Gorevan

, Abrashkin

, Cubrich

, and Reichenbach

(2016) Development of a planetary deep drill. In Earth and Space 2016: Engineering for Extreme Environments, edited by Malla

R.B.

, Agui

J.H.

, and van Susante

P.J.

, American Society of Civil Engineers, Reston, VA, pp 256–266.