It took thousands of years, following the first questioning gaze at the sky, to understand why the clear daytime sky is blue [1]. The theory of Rayleigh scattering, however, does not explain the entire palette of sky colors that delight us at dawn or dusk. It was only realized in the middle of the last century [2]that the unstable oxygen molecule ozone (O3) has a profound impact on twilight colors.

It is well known that the ozone layer protects the surface of our planet from the damaging effects of ultraviolet radiation but few people appreciate that, without ozone, the color of the zenith (overhead)sky at twilight would be a pale green/straw yellow rather than the deep, steely blue that we observe. The electronic Chappuis absorption band of ozone is intrinsically weak and has little effect on the color of the daytime sky. This band, extending from 450 to 850 nm, only becomes significant when the pathlength of sunlight through the atmosphere is dramatically increased around sunrise and sunset. At these times, the Chappuis band becomes by far the strongest feature in the visible spectrum of the sky or the setting sun.

In preparation for future ‘transit spectroscopy’ (analysis of the light transmitted by the atmosphere of a planet transiting its parent star) we have performed relative spectrophotometry of the sky during twilight. This can be compared with the observations of the eclipsed Moon [3] which examine light that has grazed the Earth’s atmosphere during a lunar eclipse. The eclipse geometry results in a much stronger influence of Rayleigh and aerosol extinction in the light reflected from the eclipsed Moon than in scattered light from the horizon sky at twilight, resulting in strong suppression of the blue end of the spectrum.

The twilight spectra appearing in this report were obtained with an Ocean Optics JAZ spectrometer covering 350–1000 nm and using a single optical fiber input pointed about 10° above the western sky in overcast conditions. The altitude of the observing site was 560 m at a latitude of +47.8°.

Figure 1 The experimental setup for the twilight observations showing the JAZ spectrometer, the fiber feed and the data-taking laptop. These photographs, taken around the time of sunset with a ‘daylight’ white balance setting on a Canon 5D MkII, illustrate the high color temperature of the ambient light at this time and caused predominantly by the ozone absorption (see Fig 2).

Figure 2 The variation of sky brightness at 700 nm, in units of counts per 8192 ms, and color temperature, measured with a Gossen 2F COLORMASTER.

Figure 3 A sequence of ‘Relative Irradiance’ spectra of a cloudy western sky at approximately 10 min intervals from an hour before, to 15min after, sunset. Also plotted is the Lunar eclipse spectrum from [3]. The rapid development of the ozone Chappuis absorption, centered at 600 nm, is apparent together with the dramatic bluening of the sky color during this period. The somewhat irregular behavior of the blue end of the spectrum is due to the variable cloud thickness during these measurements. The reference spectrum is with a solar altitude of+13° and the final spectrum of the sequence is with an altitude of -3°.

Figure 4 The observed ratio (dark blue), with a normalized continuum, of a spectrum taken with a solar altitude of-3° to one at +13°. Atmospheric models [4], using the HITRAN database [5], with a similar but pure transmission geometry, are over plotted (red: pure ozone absorption; light blue: ozone + O2 + H2O). The models have been scaled in intensity by a factor of 1.7 as a way of accounting for the simplification of the model geometry.

This simple experiment clearly illustrates the profound influence of the ozone Chappuis absorption on the color of the twilight sky (the color temperature of the western sky varied from 6,000K to14,000K during the course of the observations) and the very strong visible spectral signature that could be expected in transit spectroscopy of an oxygen-rich exoearth atmosphere.

The absence of strong signatures of O2 and H2O in the spectral ratios probably indicates that most of the light entering the spectrometer has passed above the cloud layer, in the stratosphere (~25km), where ozone dominates the absorption.

An interesting exercise with these data is to invert them to derive the ozone Chappuis absorption cross-section times the column density of the atmospheric pathlength that is characteristic of these twilight observations.

Figure 5 The inversion of the observation (ln[observed ratio] * const) compared with the data in the radiative transfer code we used for the modeling, remembering that we have neglected the water and diatomic oxygen contributions which probably accounts for some of the deviations in the band wings.

This project, designed for high-school education, will go on to explore different illumination geometries involving long absorption pathlength through the atmosphere.

Acknowledgments: We thank Enric Pallé for providing the Lunar eclipse digital data.

References

1              Pesic, P., “Sky in a bottle”, The MIT Press

2              Hulbert, E. O., 1953, “Explanation of the Brightness and Color of the Sky, Particularly the Twilight Sky”, Journal of the Optical Society of America, 43, 113-118

3              Pallé, E. et al. 2009, “Earth’s transmission spectrum from lunar eclipse observations”, NATURE, 459, 814-816

4              Clough, S.A., et al. 2005, “Atmospheric radiative transfer modeling: a summary of the AERcodes”, Journal of Quantitative Spectroscopy and Radiative Transfer, , 91, 233

5              Rothman L.S., et al. 2009, “The HITRAN 2008 molecular spectroscopic database”, Journal of Quantitative Spectroscopy and Radiative Transfer, 110, 533

Robert (Bob) Fosbury [a] & Andreas Seifahrt [b]

a. Space Telescope – European Coordinating Facility, Garching bei München, 85748, Germany(rfosbury@eso.org)

b. Physics Department, Univ. of California, Davis, CA 95616, USA (seifahrt@physics.ucdavis.edu)

A 2-meter 35?m bifurcated borosilicate bundle was used inside a T-300 sleeve, and an LS-1 tungsten halogen light source was used. All components were placed inside an environmentally controlled chamber, along with a second beaker of seawater being monitored by a Ross pH electrode. The beakers were sealed with Parafilm®, and the probes kept a tight seal with the film to prevent evaporation. The experimental setup is pictured below:

Galvanic corrosion became apparent with the use of the stainless steel sleeve, most likely due to improper sealing of the mirror’s metal from the steel. This led to accumulation of rust and sediment on the mirror, steadily causing a continued tilt in the absorbance curve. The typical pH algorithm looks at two wavelengths, an analytical wavelength and a baseline correction to account for vertical offset.

Though, a tilt in the absorbance spectra skews these values and yields incorrect measurements. To correct for this, an accompanying algorithm was used to eliminate this tilt effect, bringing the absorbance spectra back to its appropriate position, knowing that 750nm and 509nm (isosbestic point) should be pegged at zero absorbance. The algorithm is shown below, followed by the plot showing the corrected absorbance curve:

Knowing that it is required to monitor three wavelengths in order to account for the tilt, a long-term experiment was run to monitor pH patch stability. Analysis was performed that calculated the continuous pH assuming the original algorithm alone using a 750nm baseline correction, as well as a 509nm (isosbestic point) correction, and also using the tilt correction algorithm that looks at all three wavelengths. The plot of these three trends is shown below:

Clearly the original approach of using a single wavelength for baseline correction is not feasible for long-term measurement; rather, we see that the tilt algorithm is able to account for these absorbance curve distortions and produces a very coherent trend. The probe originally reads sea water as being roughly pH 8.2, and then drifts downwards to just above pH 6, where it equilibrates. This drift is most likely due to galvanic corrosion reactions occurring, which alter the pH of the system.

Ocean Optics also makes a transmissive probe of the same design using chemically inert peek material. This experiment was repeated using this alternate model, the TP-300, again with the Ross pH electrode running in parallel.

The accumulation of sediments and particles on the mirror and/or lens led to the absorbance curve tilt phenomenon previously seen, though this occurs much more slowly than the stainless steel probe. The system equilibrates after roughly 20 hours, and remains at a constant pH for the following 30 hours, showing the potential for strong long-term stability. A comparative plot shows the readings of the optical pH probe and the Ross electrode, with the electrode’s values being recorded manually at five discrete times:

The drift of the Ross pH electrode is slightly over 0.03 pH unit per day, which is roughly 15 times more drift than the specification reported in the user manual. While the Ocean Optics pH probe showed enhanced stability over the Ross electrode, the measurement was cut short due to light source failure. The LS-1 light source bulbs are offered in 900 hour and 10,000 hour lifespans; the longer-use bulb is recommended for long-term measurements. LED(s) may also be used with these sensors as another alternative. This study has been vital in determining a number of aspects and limitations for this optical pH probe system, some notable points include:

- The peek TP-300 probe should be used over the stainless steel version for pH measurement in seawater or high salinity environments; corrosion occurs extremely quickly with the stainless steel and sediment accumulates on the mirror.

- As sediments form on the mirror and/or lens, the absorbance curve may show a tilting distortion; this can be corrected via a dynamic algorithm that looks at three wavelengths to eliminate the effect.

- The light source should be relatively new and stable when performing long-term measurements. As the light source begins to die, there is a non-uniform intensity decay that causes an incorrect pH reading drift, followed by the light source finally dying completely. LED’s may be used as an alternative.

Algorithms Used:

pH Calculation:

Absorbance Spectra “Tilt” Correction:

A reflective pH patch was affixed to the inner wall of a small 7mL tube, and was held in place using a ring stand and clamp. A 600um bifurcated reflective probe was connected to an LS-1 light source with a blue filter, as well as a USB2000 spectrometer with grating #1, 200um slit, and no lens installed. The first feasibility test used clear buffers instead of the muddied buffers, in order to observe the performance. The experimental setup is pictured below, along with the plot of the resulting titration.

Although the curves showed a slight distortion, they followed the expected trend and produced a linear calibration plot. Dirt was obtained and mixed in with the buffer solutions; the experimental setup is shown below.

This yielded a very interesting titration plot, unlike any that had been seen before:

Typically for reflective pH patches, the peak is observed at 620nm and the baseline correction is done at 509nm. In this case, however, we see the peak had shifted closer to 635nm, and the 509nm region was distorted completely. As a result, 750nm was used as the baseline correction wavelength, which produced an incredibly linear calibration plot, as seen below.

This application has been critical in determining the effect of turbidity on the absorbance curves for the reflective pH patches, and has concluded that the correct wavelengths for acquisition and baseline need to be chosen based on the nature of the analyte solution.

Hardware Used:
•    Phase Fluorometer Electronics (MFPF)
•    Bifurcated Fiber Optic Cable
•    HiOXY-R Sensor.

Abstract:
This application note explains how partial pressure of oxygen is measured even in the presence of gases like Methane.

Background:
Methane is a chemical compound with the chemical formula CH4. It is the simplest alkane, and the principal component of natural gas. Burning methane in the presence of oxygen produces carbon dioxide and water. The relative abundance of methane makes it an attractive fuel. Several oil companies have approached Ocean Optics to provide a solution of measuring oxygen in the presence of Methane.

The HiOXY oxygen sensor design by Ocean Optics was tested for this application.

Feasibility Test Procedure:
The MFPF electronics is an instrument platform for measurement of fluorescence lifetime and phase. This frequency domain electronics uses a blue LED excitation and a photodiode for detection. A fluorescence method is used to measure the partial pressure of dissolved or gaseous oxygen. A bifurcated optical fiber carries excitation light produced by the blue LED to the thin-film coating of the Hioxy-R sensor. Fluorescence generated at the tip of the probe is collected by the probe and carried by the optical fiber to the detector of the MFPF. The phase shift between the blue LED excitation and emission signal of fluorescence is used to calculate the lifetime.  The Lifetime is an essential parameter for the calculation of oxygen.

A short term feasibility test has been performed using HiOXY sensor in Methane gas.

The chart below first shows the response of the HiOXY sensor in Nitrgoen and Air first.  The sensor is again exposed to nitrogen and a small flow of methane is introduced. There is no observed change in the lifetime due to the introduction of methane in the nitrogen flow.
The sensor is now exposed back to air to see if there are any effects or changes caused in the lifetime due exposing the sensor to Methane gas. It is observed that there is no change in the lifetime in Air. A small flow of Methane is introduced in the air stream and a slight increase in the lifetime caused from the methane.

When only methane is exposed to the sensor, it behaves in the same way as nitrogen as the sensor does not see any oxygen.

In conclusion, the HiOXY sensor can be used to measure oxygen in the presence of Methane gas without affect the intensity or the lifetime of the sensor.

Method:
pH Sensing

Hardware Used:
USB2000 Spectrometer: Grating #1, 200um slit, no lens installed
LS-1 Light Source
Bifurcated Reflective Probe, 600um
SpectraSuite Software

Measurement Mode:
Absorbance

Notes:

There was an interest in non-intrusively monitoring pH in tubes containing wet soil samples. This was significant for a number of reasons; this was the first time the reflective patches were to be tested in something other than a cuvette, on a curved surface rather, and this was also the first time an analyte solution was observed that contained sediment or turbidity.

A reflective pH patch was affixed to the inner wall of a small 7mL tube, and was held in place using a ring stand and clamp. A 600um bifurcated reflective probe was connected to an LS-1 light source with a blue filter, as well as a USB2000 spectrometer with grating #1, 200um slit, and no lens installed. The first feasibility test used clear buffers instead of the muddied buffers, in order to observe the performance. The experimental setup is pictured below, along with the plot of the resulting titration.

Although the curves showed a slight distortion, they followed the expected trend and produced a linear calibration plot. Dirt was obtained and mixed in with the buffer solutions; the experimental setup is shown below.

This yielded a very interesting titration plot, unlike any that had been seen before:

Typically for reflective pH patches, the peak is observed at 620nm and the baseline correction is done at 509nm. In this case, however, we see the peak had shifted closer to 635nm, and the 509nm region was distorted completely. As a result, 750nm was used as the baseline correction wavelength, which produced an incredibly linear calibration plot, as seen below.

This application has been critical in determining the effect of turbidity on the absorbance curves for the reflective pH patches, and has concluded that the correct wavelengths for acquisition and baseline need to be chosen based on the nature of the analyte solution.

Algorithm Used: