Effects of Spectralon absorption on reflectance spectra of ......Effects of Spectralon absorption on reflectance spectra of typical planetary surface analog materials Hao Zhang,1,*

Effects of Spectralon absorption on reflectance spectra of typical planetary surface analog

materials Hao Zhang,1,* Yazhou Yang,1 Weidong Jin,1 Chujian Liu,2 and Weibiao Hsu3

1Planetary Science Institute, School of Earth Sciences, China University of Geosciences, 388 Lumo Road, Hongshan District, Wuhan, 430074, China

2State Key Laboratory of Geological Process and Mineral Resources, China University of Geosciences, 388 Lumo Road, Hongshan District, Wuhan, 430074, China

3Key Laboratory of Planetary Sciences, Purple Mountain Observatory, Chinese Academy of Sciences, Nanjing, 210008, China

* [email protected]

Abstract: Acquiring accurate visible and near-infrared (VisNIR) reflectance values of atmosphereless celestial bodies is very important in inferring the physical and geological properties of their surficial materials. When a calibration target with inherent non-trivial absorption features is used, the calibrated reflectance would essentially always contain spurious spectral features and the spectroscopic data may easily be misinterpreted if the artifact is not properly taken care of. We demonstrate with laboratory reflectance measurements that the VisNIR spectra of three typical planetary surface analog materials, lunar simulant JSC-1A, olivine and pyroxene grains, have an artificial peak at 2.1 µm when Spectralon-type plaque made of polytetrafluoroethylene is used as the calibration target in the NIR region. The degree of severity of this artifact is dependent on the strength of the 2.0 µm absorption feature of the mineral. Empirical methods are proposed to remove this artifact to bring the spectra close to that calibrated by a gold mirror which does not have any conspicuous absorption features in the NIR region. The correction methods may be applied to reflectance data acquired by the VisNIR imaging spectrometer onboard the Yutu Rover of the Chinese Chang’E 3 lunar mission which employed an onboard Spectralon-type calibration target. ©2014 Optical Society of America OCIS codes: (120.0280) Remote sensing and sensors; (290.5850) Scattering, particles; (300.6340) Spectroscopy, infrared; (300.6170) Spectra; (300.6300) Spectroscopy, Fourier Transforms.

References and links 1. C. M. Pieters, “Strength of mineral absorption features in the transmitted component of near-infrared reflected

light: First results from RELAB,” J. Geophys. Res. 88(B11), 9534–9544 (1983). 2. J. W. Warren, K. Peacock, E. H. Darlington, S. L. Murchie, S. F. Oden, J. R. Hayes, J. F. Bell, S. J. Krein, and

A. Mastandrea, “Near Infrared Spectrometer for the Near Earth Asteroid Rendezvous Mission,” Space Sci. Rev. 82, 101–167 (1997).

3. N. R. Izenberg, J. F. Bell, J. W. Warren, S. L. Murchie, K. Peacock, E. H. Darlington, B. Carcich, C. Chapman, B. E. Clark, A. Harch, G. Heyler, J. Joseph, P. Martin, L. McFadden, B. Merline, M. Robinson, J. Veverka, and D. Wellnitz, “In-Flight calibration of the Near Earth Asteroid Rendezvous Mission's Near Infrared Spectrometer: I. Initial calibrations,” Icarus 148(2), 550–571 (2000).

4. M. Ohtake, C. M. Pieters, P. Isaacson, S. Besse, Y. Yokota, T. Matsunaga, J. Boardman, S. Yamomoto, J. Haruyam, M. Staid, Y. Mall, and R. O. Green, “One Moon, many measurements 3: Spectral reflectance,” Icarus 226(1), 364–374 (2013).

5. R. G. Burns, Mineralogical Applications of Crystal Field Theory, 2nd ed. (Cambridge U, 1993). 6. P. G. Lucey, “Model near-infrared optical constants of olivine and pyroxene as a function of iron content,” J.

Geophys. Res. 103(E1), 1703–1713 (1998). 7. J. M. Sunshine and C. M. Pieters, “Determining the composition of olivine from reflectance spectroscopy,” J.

Geophys. Res. 103(E6), 13675–13688 (1998).

#213461 - $15.00 USD Received 11 Jun 2014; revised 19 Aug 2014; accepted 19 Aug 2014; published 26 Aug 2014(C) 2014 OSA 8 September 2014 | Vol. 22, No. 18 | DOI:10.1364/OE.22.021280 | OPTICS EXPRESS 21280

8. R. N. Clark, “Spectroscopy of Rocks and Minerals, and Principles of Spectroscopy,” http://speclab.cr.usgs.gov/PAPERS.refl-mrs/

9. M. I. Mishchenko, L. D. Travis, and A. A. Lacis, Scattering, absorption and emission of light by small particles (Cambridge U, 2002).

10. Y. G. Shkuratov, V. Kaydash, Y. Velikodsky, N. Opanasenko, and G. Videen, “Optical measurements of the Moon as a tool to study its surface,” Planet. Space Sci. 59(13), 1326–1371 (2011).

11. V. R. Weidner and J. J. Hsia, “Reflection properties of pressed polytetrafluoroethylene powder,” J. Opt. Soc. Am. 71(7), 856–861 (1981).

12. R. O. Green, C. Pieters, P. Mouroulis, M. Eastwood, J. Boardman, T. Glavich, P. Isaacson, M. Annadurai, S. Besse, D. Barr, B. Buratti, D. Cate, A. Chatterjee, R. Clark, L. Cheek, J. Combe, D. Dhingra, V. Essandoh, S. Geier, J. N. Goswami, R. Green, V. Haemmerle, J. Head, L. Hovland, S. Hyman, R. Klima, T. Koch, G. Kramer, A. S. K. Kumar, K. Lee, S. Lundeen, E. Malaret, T. McCord, S. McLaughlin, J. Mustard, J. Nettles, N. Petro, K. Plourde, C. Racho, J. Rodriquez, C. Runyon, G. Sellar, C. Smith, H. Sobel, M. Staid, J. Sunshine, L. Taylor, K. Thaisen, S. Tompkins, H. Tseng, G. Vane, P. Varanasi, M. White, and D. Wilson, “The Moon Mineralogy Mapper (M3) imaging spectrometer for lunar science: Instrument description, calibration, on-orbit measurements, science data calibration and on-orbit validation,” J. Geophys. Res. 116, E00G19 (2011), doi:10.1029/2011JE003797.

13. A. Springsteen, “Standards for the measurement of diffuse reflectance – an overview of available materials and measurement laboratories,” Anal. Chim. Acta 380(2-3), 379–390 (1999).

14. K. J. Voss and H. Zhang, “Bidirectional reflectance of dry and submerged Labsphere Spectralon plaque,” Appl. Opt. 45(30), 7924–7927 (2006).

15. O. Svensen, M. Kildemo, J. Maria, J. J. Stamnes, and Ø. Frette, “Mueller matrix measurements and modeling pertaining to Spectralon white reflectance standards,” Opt. Express 20(14), 15045–15053 (2012).

16. A. Ferrero, A. M. Rabal, J. Campos, A. Pons, and M. L. Hernanz, “Spectral and geometrical variation of the bidirectional reflectance distribution function of diffuse reflectance standards,” Appl. Opt. 51(36), 8535–8540 (2012).

17. C. J. Bruegge, A. E. Stiegman, R. A. Rainen, and A. W. Springsteen, “Use of Spectralon as a diffuse reflectance standard for in-flight calibration of earth-orbiting sensors,” Opt. Eng. 32(4), 805–814 (1993).

18. Report of Scientific Validations of Visible-Near-infrared Imaging Spectrometer of the Chang’E 3 Mission, Document # CE3-GRAS-CSSY-003–F3, ver. 1.0, Released May 16 2012.

19. B. Liu, J. Z. Liu, G. L. Zhang, Z. C. Ling, J. Zhang, Z. P. He, B. Y. Yang, and Y. L. Zou, “Reflectance conversion methods for the VIS/NIR imaging spectrometer aboard the Chang'E-3 lunar rover: based on ground validation experiment data,” Res. Astron. Astrophys. 13, 862–874 (2013).

20. V. Orofino, R. Politi, A. Blanco, and S. Fonti, “Diffuse reflectance of altered olivine grains: Remote sensing detection and implications for Mars studies,” Planet. Space Sci. 54(8), 784–793 (2006).

21. T. Hiroi and C. M. Pieters, “Estimation of grain sizes and mixing ratios offine powder mixtures of common geologic minerals,” J. Geophys. Res. 99(E5), 10867–10879 (1994).

22. C. Carli, M. Ciarniello, F. Capaccioni, G. Serventi, and M. Sgavetti, “Spectral variability of plagioclase-mafic mixtures (2): Investigation of the optical constant and retrievaed mineral abundance dependence on particle size distribution,” Icarus 235, 207–219 (2014).

23. R. N. Clark, T. V. V. King, M. Klejwa, G. A. Swayze, and N. Vergo, “High spectral resolution reflectance spectroscopy of minerals,” J. Geophys. Res. 95(B8), 12653–12680 (1990).

24. Labsphere Reflectance and Transmittance Measurement Integrating Spheres, http://woodall.ece.ucdavis.edu/pdf/labsphere_reflectance_manual.pdf

25. A Guide to Reflectance Materials and Coatings, http://www.labsphere.com/uploads/technical-guides/a-guide-to-reflectance-materials-and-coatings.pdf

26. H. Zhang, Y. L. Liu, K. Zhu, G. G. Siu, Y. H. Xiong, and C. S. Xiong, “Fourier transform infrared characterization of nanometre SnO2,” J. Phys. Condens. Matter 10(48), 11121–11127 (1998).

27. H. Zhang and K. J. Voss, “Comparisons of Bi-directional reflectance measurement of closely packed particulate layers with radiative transfer models,” Appl. Opt. 44, 597–610 (2005).

28. H. Zhang and K. J. Voss, “Bi-directional reflectance measurement of closely packed natural and prepared particulate layers,” in Light Scattering Reviews III, A. Kokhanovsky, ed. (Springer-Praxis, 2008).

29. H. Zhang and K. J. Voss, “On Hapke photometric model predictions on closely packed particulate layers,” Icarus 215, 27–33 (2011).

30. H. Zhang, W. D. Jin, W. B. Xu, Z. W. Wang, Y. Z. Yang, Y. Yuan, and H. Sun, “A goniometric system to measure the incomplete mueller matrices of packed layers.” presented at the 45th Lunar and Planetary Science Conference, The Woodlands, TX, Mar. 17–21, 2014.

31. Refractive index of Au, Gold, http://www.filmetrics.com/refractive-index-database/Au/Gold

1. Introduction

To accurately extract the surficial mineralogical, compositional and physical information of atmosphereless celestial bodies using visible and near-infrared (VisNIR) reflectance spectroscopy in space missions, space-borne and in situ spectrometers should be rigorously calibrated to obtain the reliable reflectance of the surface materials (e.g., [1–4]). For example, the band-depth, band-position and relative intensities of the electronic transition bands around


1 µm and 2 µm appearing in reflectance spectra of olivine and pyroxene, two of the most common minerals found on lunar and asteroidal surfaces, have been found to strongly depend on their mineral morphologies, chemical compositions and particle size distributions [5–7]. As a result, the geological and mineralogical information retrieved is critically dependent on the accuracy of the reflectance values measured. Despite of the rapid increase of sensor numbers in the space exploration era, reflectance calibration remains a difficult task as the spectral, radiometric and photometric properties of the calibration targets must be very well understood and must remain stable to provide long-term reliability of the spectra taken during the mission. Any imperfections or spurious spectral features of the calibration target itself should be well compensated during the calibration process or spurious spectral features of the measured sample may result [8].

Reflectance spectra measured by most space mission sensors are bi-directional reflectance r in nature,

,IrF

= (1)

where I is the sensor-measured radiance reflected by planetary surface under study and F is the solar irradiance reaching the surface [9,10]. In many practices, however, F is not directly measured and the sample reflectance rs is obtained by ratioing I from the sample (Is) to that from a calibration target (Ic) with near-Lambertian behavior and known reflectance values (rc) within the spectral range of interest,

,ss c s cc

Ir r R r

I= ≡ (2)

where Rs is the relative reflectance of the sample to the calibration target measured under the same illumination and viewing geometries as rc.

Two NIST-traceable and commercially available calibration targets manufactured by Labsphere Corp. (North Sutton, NH), Spectralon plaque and InfraGold plate, are the most extensively used reflectance standards in remote sensing applications [11–16]. Spectralon is a sintered polytetrafluoroethylene (PTFE) plaque and InfraGold is a rough metal plate with gold coating. The different materials that make up the reflectance standards make them suitable in different spectral ranges and different environments. For example, InfraGold plate was chosen as the calibration target over Spectralon in the Near-Infrared Spectrometer (NIS) onboard the Near Earth Asteroid Rendezvous (NEAR) spacecraft for its resistance to solar irradiation damages [17] to ensure the long-term stability of measured spectra (0.8-2.7 µm) [2,3]. The Moon Mineralogy Mapper imaging spectrometer onboard the Chandrayaan-1 spacecraft, on the other hand, used a Spectralon plaque in spectral range 0.4-1.7 µm and an InfraGold plate in 1.7-3.0 µm, respectively [12]. The most recent lunar exploration mission, Chinese Chang’E-3, which had its lander successfully landed on the Moon on December 15, 2013 and deployed a lunar rover Yutu to roam in the Mare Imbrium region, had an imaging spectrometer (0.45-2.4 µm) with a Spectralon-type calibration target made of PTFE onboard the rover [18] and performed the first in situ optical spectroscopic measurement of the Moon [19].

Because some widely used calibration targets such as Spectralon and Halon (see Section 2 below) have reflectance values very close to 1 in the VisNIR, often times the quantity Rs is also dubbed as “reflectance” especially in comparative studies. For example, the pre-flight calibration of the NIS onboard the NEAR spacecraft performed measurements on many types of minerals relative to a Halon standard and the data are presented in terms of “reflectance” (e.g., see Fig. 38 in [2]) to be compared with earlier reflectance data calibrated by Halon. However, if the calibration target contains absorption features that are not calibrated out, rs and Rs are not interchangeable over the wavelengths of the absorption.

In this paper, we use both a Spectralon plaque and a gold mirror as reflectance targets (or backgrounds) to obtain the reflectance spectra of three common planetary surface analog


materials: lunar simulant JSC-1A, olivine and pyroxene grains. We demonstrate that the absorption feature at 2.14 µm of the Spectralon plaque incurs spurious spectral features with varying degrees of severity in sample spectra calibrated by Spectralon plaque in the NIR region. Empirical correction procedures are proposed to remove this artifact to bring the spectra close to that recorded with a gold mirror in the same spectral region.

2. The absorption feature of Spectralon at 2.14 µm

We first notice that laboratory reflectance (Rs) spectra of pure olivine grains obtained using Labsphere Spectralon material as the calibration target show a peak at 2.1 µm (see, e.g., Figs. 4 and 5 in [20]). Since olivine ((Mg/Fe)2SiO4) is one of the dominant rock-forming minerals of atmosphereless celestial bodies such as the Moon and many asteroids, the reflectance spectra of olivine particles with varying chemical compositions (Mg/(Mg + Fe) values), impurities, morphologies and particle size distributions (from microns to bulk) are very well documented and a peak at 2.1 µm is rarely seen in well-calibrated reflectance (e.g., [1,3,4,6] [7,10,21]). Moreover, if calibrated using a Spectralon (-type) plaque in this spectral region, essentially all types of minerals measured have shown such a peak (see e.g., Figs. 6 and 7 in [19]; Fig. 1 in [22]; and many others). The appearance of a 2.1 µm peak would easily confuse and complicate mineral identifications as it apparently splits an otherwise flat 1.6-2.4 µm region into two separate dips which could be easily misinterpreted as a mixture of clinopyroxene (with an absorption at ~2.2-2.3 µm) and orthopyroxene (with an absorption at ~1.9 µm). Of course these apparent bands could also be misinterpreted to other origins such as aqueous alteration of the olivine grains [20].

This spectral artifact can be qualitatively understood by looking at the spectral properties of the Spectralon plaque in the visible and NIR regions. In Fig. 1, we plot the 8°-hemispherical reflectance (8°-HR) of a Spectralon plaque with 99% nominal reflectance provided by Labsphere (Reflectance Standard SRS-99-020, Serial # 7A15A-1426) in the spectral range of 0.4-2.5 µm. For most regions in the visible, the reflectance remains at 99%. Longer than 1.8 µm or so, the reflectance starts to drop below 98% and reaches a “local minimum” value of 94% at 2.14 µm. This absorption feature (a dip of ~5%) must have caused the spurious peak located at 2.1 µm of the sample Rs spectra when radiance of the sample is ratioed by radiance from the plaque during the calibration process. Indeed, the peaks at 2.1 µm in reflectance spectra of some mineral grains (see Section 4) have similar features of the 2.1 µm absorption dip of the Spectralon plaque: the slope is steeper at long wavelength side and less steep in the short wavelength side. Halon, a once popular “white” reflectance target (also made of PTFE) which began to become obsolete in the late 1990’s, has a similar absorption feature near 2.1 µm [23]. For comparison purposes, reflectance data of InfraGold from two different sources [24,25], both provided by Labsphere, are also plotted in Fig. 1. The theoretical gold reflection for the configuration of the EasiDiff device used in this study (see Sections 3 and 4) is also shown. Obviously in the NIR region above 1 µm the gold reflectance is smoother than that of Spectralon.

3. Laboratory measurements

To further understand the plaque absorption effects and devise a possible fix, we carried out reflectance measurements on powders of lunar simulant JSC-1A, olivine and pyroxene grains. The JSC-1A was purchased from OrbiTech and its most grains are much smaller than 1 mm. Olivines and pyroxenes are nearly pure minerals collected from Hebei Province, China and were ground and sieved into particle sizes from 0 to 45 µm. These samples were chosen not only because they are among the most typical planetary surface analogs, but also for their various strengths of the 2.0 µm absorption feature so we may evaluate the impact of the Spectralon 2.14 µm absorption. Measurements were carried out on a Bruker Optics Vertex 70 Fourier transform infrared (FTIR) spectrometer with an aluminum mirror system to enhance the throughput in the VisNIR region. To measure the reflectance of the powdery surfaces, EasiDiff reflectance accessories from Pike Technologies were used. The EasiDiff measures the biconical reflectance at incident zenith angles range from 30 to 65°, the same angular


spans of the viewing zenith and a viewing azimuth of 180°. Although the spectral region of interest of this study is 0.5-2.4 µm, the typical region of VisNIR remote sensing of atmosphereless solar system bodies, we performed measurements in 3 overlapping spectral ranges: 0.5-1.1 µm, 0.83-2.85 µm and 1.33-38 µm to double-check the quality of the spectra obtained. The light sources, beamsplitters, detectors and diffuse accessories for these 3 overlapping spectral ranges from visible to mid-infrared (MIR) are listed in Table 1. To record spectra using Spectralon as the background signal, a circular portion of the plaque was carefully cut out of the original 2-inch diameter piece (SRS-99-020, serial #7A15A-1426) to fit into the sample holder inside the EasiDiff accessory. Great care was taken to make sure that the surface was not contaminated or damaged during the machining process. Because an InfraGold plate is difficult to be modified to fit into our diffuse reflectance accessory, a gold mirror with nominal 97% reflectance that can be fitted into the EasiDiff sample cup slide, provided by Pike Technologies along with other parts that came with the EasiDiff device, was used to compare with the Spectralon. It is noted that a gold mirror or thin gold film is widely used in measuring relative reflectance in spectroscopic studies of granular materials (e.g [26].). Sample cups and calibration targets are shown in Fig. 2. With a sample layer thickness of 3 mm, radiative transfer computations [27–29] show that for the wavelength spans in this study these particulate layers have optical thickness values large enough to have their reflectance not affected by any substrate retro-reflections. All measured data presented here

Fig. 1. Visible and near-infrared reflectance values of Labsphere Spectralon (red), InfraGold (yellow and green) and a theoretical gold reflection curve (blue, see Section 4). The two data sets for InfraGold materials, InfraGold 1 and InfraGold 2, are taken from [23] and [24], respectively.

are reflectance relative to Spectralon with nominal 99% reflectance, or Rs in Eq. (2), except indicated individually.

Figure 3 displays the original VisNIR spectra of pyroxene grains (0-45 μm) obtained using Spectralon plaque as the calibration target. It is seen from this plot that the differences between spectra taken using different accessories (Table 1) within the overlapping regions are within a few percent and the spectral shapes can match each other very well. Although in principle the more accurate absolute reflectance factors at three visible wavelengths, 0.473, 0.532 and 0.633 μm, can be obtained by using our custom-built goniometric bi-directional reflectance system [30] and used to calibrate all spectra, this work concentrates on the possible spurious spectral features incurred by calibration target absorption and thus the absolute reflectance value is not critically important, we simply scaled the NIR region to connect to the visible. For the case shown in Fig. 3, for example, the NIR curve (in green) was brought up by 2.3% to be connected with the VisNIR one (in blue). In this work the MIR


spectra were only used to cross-check the NIR data quality and thus are not shown in the rest part of this paper.

Fig. 2. The background targets and 2 out of the 3 mineral samples used in this work mounted in Pike EasiDiff sample cup sliders: the smaller circular Spectralon cut from the original larger Spectralon plaque (SRS-99-020) (top left), Pike gold mirror (top right), olivine grains (0-45 μm) (bottom left) and lunar simulant JSC-1A (bottom right). Photo taken by lead author of this work (Hao Zhang).

Table 1. Accessories used to obtain the spectra of visible to mid-infrared spectral ranges.

Spectral coverage Source Beamsplitter Detector EasiDiff

VisNIR (0.5-1.1 µm) NIR Quartz Silicon diode Aluminum

NIR (0.83-2.85 µm) NIR Quartz InGaAs diode Aluminum

MIR (1.33-38 µm) GloBar KBr LaDTGS Gold

Typical reflectance spectra from 0.5 to 2.4 μm for 3 powdery samples are shown in Fig. 4. Because FT-IR technique is less efficient in the shorter wavelength region as compared to the long wavelength region, spectra below 0.7 μm are noisy. The spike at 0.63 μm was caused by the He-Ne laser used to perform the alignment in the Vertex-70 system. Other than these two regions, the spectra have very high signal-to-noise ratios as each spectrum was obtained by averaging 1000 scans.

4. Discussions

Two of the three spectra in Fig. 4 (JSC-1A in Fig. 4(a) and Olivine in Fig. 4(b)) show a conspicuous asymmetric peak (“hump”), starting from ~1.9 μm, ending at ~2.2 μm and centered at 2.14 μm with similar but complimentary spectral features as the 2.14 μm absorption dip in the Spectralon reflectance spectra shown in Fig. 1: both the humps and the dip have a more gradual slope in the short wavelength side (1.9-2.14 μm) and are steeper in the long wavelength side (2.14-2.2 μm) of the peak (compare Figs. 4 and 5 in [20] and Figs. 6 and 7 in [19] with this work’s Fig. 4(b)). The Pyroxene spectrum (Fig. 4(c)), although does not show such a conspicuous asymmetric peak, does have a shoulder-like structure at 2.14 μm. Later we will show that this is also the spurious feature caused by Spectralon’s 2.14 μm absorption, though this artifact is diminished somehow for pyroxene (see Section 4).


Fig. 3. Original reflectance spectra Rs of pyroxene grains (0-45μm), relative to a Spectralon plaque with 99% nominal reflectance, taken at 3 overlapping spectral ranges. Accessories used for each spectral range are listed in Table 1.

As the quantity Rs presented in Fig. 4 contains the spurious 2.1 μm peak and the absolute reflectance of target rc (Fig. 1) contains the 2.1 μm dip, it would be interesting to see if this artifact could be cancelled out in sample absolute reflectance rs through Eq. (2). Note the absolute reflectance of Spectralon shown in Fig. 1 is the 8°-HR and the biconical reflectance measured under the EasiDiff geometry should be closer to a 47°-incidence and 47°-viewing bidirectional reflectance. These two quantities should have different reflectance values and band depths because of the directional effects of the Spectralon especially under oblique incidences (e.g [28].). Because few goniometers can accurately measure the absolute bi-directional reflectance above 1.5 μm and thus the real rc data containing the 2.1 μm peak under the EasiDiff geometry are not available. Nevertheless, we attempted the following empirical corrections in an attempt to remove the 2.1 μm feature in reflectance spectra.

Method 1. We simply multiplied the sample Rs spectra by the 8°-HR of the Spectralon plaque (as shown in Fig. 1), rc, to obtain the sample reflectance rs as given by Eq. (1).


Fig. 4. Reflectance spectra Rs of (a) lunar simulant JSC-1A, (b) olivine grains (0-45 μm) and (c) pyroxene grains (0-45 μm). Reflectance values are relative to a Spectralon plaque with 99% nominal reflectance.

Method 2. Same as Method 1, but the sample reflectance rs was obtained by multiplying Rs by the Spectralon reflectance _Spectralon goldcr evaluated under the EasiDiff geometry as follows. The relative reflectance of the Spectralon to the Pike gold mirror, _Spectralon goldcr , was measured on the EasiDiff. Since the mirror exhibited a much stronger specular peak than the Spectralon, this relative reflectance has very low values. Using the widely available data of the real and imaginary refractive indices of gold (Au) (e.g., [31]), the average of the theoretical gold Fresnel reflections from 30° to 65° incidences (corresponding to the EasiDiff configuration) at 1° step was evaluated and shown in Fig. 1 (blue curve). The quantity _Spectralon goldcr was then multiplied by this Fresnel correction factor in the study spectral range. In principle, if the averaged Fresnel correction factor is the true absolute reflectance of the gold mirror under the EasiDiff configuration, the corrected _Spectralon goldcr would be the absolute Spectralon reflectance under the EasiDiff configuration. In reality, however, the mirror has a diffuse component and the true reflectance value is different from the theoretical


curve. Hence the quantity _Spectralon goldcr was scaled to have the same 2.14 μm absorption center value of the 8°-HR and is shown in Fig. 5 (orange circles).

Method 3. The sample relative reflectance Rs is corrected to remove the 2.1 μm feature. First Rs is divided by the 8°-HR with the 2.1 μm dip removed, or smoothcr , then multiplied by the original 8°-HR rc to obtain the corrected relative reflectance corrsR ,

.corr ss csmoothc

RR r

r= • (3)

Specifically, we first removed the reflectance values between 1.98 and 2.2 μm in Spectralon 8°-HR where reflectance values drop down significantly due to the 2.14 μm absorption, as shown in Fig. 5. Then a 4th-order polynomial was fit to the 8°-HR between 1.94 and 2.4 μm and the fitting results between 1.98 and 2.2 μm were used to replace the removed data. The interval of the data points being removed before fitting [1.98, 2.2μm], the order of the polynomial (4th) and the fitting interval [1.94, 2.4 μm] were found on a trial and error basis to yield a reasonably shaped baseline of the 2.14 μm dip, a smooth connection of the fitted data points at 1.98 and 2.2 μm and at the same time a corrsR spectrum that is not too much “over-corrected” (see discussions on Fig. 6 below). The quantity smoothcr used in Eq. (3) was obtained by combining the dip-removed 8°-HR (blue trace in Fig. 5) and the fitted 8°-HR (red trace in Fig. 5), as shown in Fig. 5.

Fig. 5. Original Spectralon 8°-hemispherical reflectance cr with the 2.14 μm absorption dip (green) removed (blue), the 4th-order polynomial fit to the dip-removed area (red) and the scaled relative reflectance of Spectralon to gold mirror measured on EasiDiff ( _Spectralon goldcr in blue). The blue and red data were pieced together to form the quantity smoothcr used in Eq. (3).

Method 4. This approach performs curve fitting to sample spectra without dealing with the calibration target spectra. Specifically, we first removed the sample reflectance values between 1.95 and 2.23 μm where the spurious 2.1 μm peak dominates, forming a gap in between, then fitted reflectance between 1.6 and 2.4 μm (1.6 to 2.3 μm for pyroxene grains as this sample has an absorption structure at 2.32 μm) using a 5-term polynomial. The fitted data between 1.95 and 2.23 μm were then used to fill the gap, producing the smooth sample spectra.


Figure 6 displays comparisons of the original mineral spectra taken with the Spectralon (Rs in red), the above 4 method corrected spectra (either Rs or rs) and the spectra recorded using gold mirror as background (Rs in orange) for all 3 minerals measured. We only plot the NIR region above 0.9 μm to compare the results because gold reflectance decreases sharply below 0.8 μm, as shown in Fig. 1. We scaled the spectra corrected by Method 2 (purples in Fig. 6) and that recorded with the gold mirror (Rs in orange) by constants to better compare with other spectra for a number of reasons. First, these two quantities are unlikely to be the true reflectance values because the gold mirror has a strong specular reflection component which resulted much lower reflectance values as compared with that taken with the Spectrlon. Second, the Pike gold mirror used here is not a NIST-traceable calibration target and thus the resultant spectra do not have any meaningful absolute values. Third, the main purpose of using the gold mirror as reflectance background here is to show that the gold mirror calibrated spectra do not have any spurious features near 2.1 μm, in contrast to that recorded with the Spectralon. Subwindows show the details around the 2.1 μm region. Obviously one can see from Fig. 6 that the Rs spectra of all 3 samples taken with the gold mirror are all free of the 2.1 μm artifact appearing in every spectrum calibrated with the Spectralon.

The spectra corrected by Method 1 (curves in light blue in Fig. 6) decreases quicker above 1.5 μm than the Rs spectra as the Spectralon has a declining absolute reflectance in that region. After multiplying the Spectralon reflectance with the 2.1 μm dip, the artifact transformed into a dip from a peak. It is very likely that the 8°-HR (with a near-normal incidence) has a deeper band-depth than reflectance taken using the EasiDiff device which has a much more oblique incidence. For Method 2, when the relative reflectance of Spectralon over gold is multiplied by the sample Rs spectra over Spectralon, the resultant spectra (the purple curves in Fig. 6) are nearly identical to the sample spectra recorded with the gold mirror except for a scaling constant. This is not a surprise because of the algebra relationship in deriving the spectra. Spectra corrected by Method 3 yielded nearly identical results with Method 4 except for olivine grains which suffered a “over-correction” with an obvious dip produced at 2.1 μm (Fig. 6(b)) (~1% maximum difference at 2.1 μm). Although Method 4 produced the most smooth spectra that are comparable to spectra recorded with the gold mirror for all 3 spectra, real planetary mission data are much more scattered than the smooth laboratory data presented here and thus Method 4 might not work well all the time. A correction method that removes the spurious feature reasonably well and at the same time distort the raw data as little as possible should be found and used to analyze the field data.

It is interesting to note that for pyroxene spectra shown in Fig. 6(c), although the comparison of the gold mirror calibrated spectrum (in orange) and the Spectralon calibrated spectrum (in red) unambiguously identifies that the shoulder-like structure at 2.1 μm comes from Spectralon absorption, this peak is minimal among the three samples, possibly because the 2.1 μm Spectralon absorption (~5%) is overshadowed by the much more stronger 2.0 μm pyroxene absorption (~20%) and thus the artifact is greatly reduced in magnitude. For planetary missions that employed a Spectralon-type calibration plaque for the NIR region like Chang’E 3′s VisNIR spectrometer, a detailed study on the impact of plaque’s 2.1 μm feature on various sample information retrievals would be necessary.

Finally, we notice that in addition to the 2.14 μm absorption of the Spectralon, the continuum drop of its reflectance toward the longer wavelengths could also potentially distort reflectance features. However, this drop has a much gentle gradient as compared with the 2.1 μm sharp feature and thus its role in potentially confusing mineral identifications should be minimal. Furthermore, because of the presence of space weathering effects, most atmosphereless planetary surface spectra are “reddened” (increased reflectance with increasing wavelength) and exhibit a much steeper sloped feature. For this reason, to compare with laboratory measurements or spectral library data, some continuum removal procedures need to be applied to measured planetary spectra. As a result, the effects of continuum drop of the calibration target are largely removed, but not the artifact incurred by the 2.1 micron absorption feature.


Fig. 6. Comparisons of the original spectra obtained using Spectralon and gold as background targets (red and orange) and the spectra corrected by the 4 methods described in text in the NIR region for (a) JSC-1A, (b) olivine grains and (c) pyroxene grains. The 2.1 μm region for each sample is magnified and shown in subwindows.

5. Conclusions

Although the possible spurious feature at 2.14 μm incurred by using a Spectralon-type calibration target in the near-infrared was pointed out by R. Clark and colleagues ([8,23]) and others as early as more than 20 years ago, it seems that reflectance spectra containing this artifact floating around in the literature have not diminished substantially. By performing reflectance measurements of 3 common planetary surface analog materials with differing 2 μm intrinsic absorption strengths, using both the Spectralon plaque and a gold mirror as the calibration targets, we demonstrated unequivocally that the use of Spectralon (PTFE) in calibrating NIR spectra can produce a spurious peak at 2.1 μm, caused by the absorption feature of the Spectralon (PTFE) at 2.14 μm. The severity of this artifact in the spectra is dependent on the absorption strength of the 2 μm peak of the mineral itself. An intrinsically very strong 2 μm absorption peak present in the mineral itself could minimize this artifact but for minerals lacking any 2 μm features this artifact is significant and may easily cause


confusions in mineral identifications. By performing 4 empirical corrections this artifact could be reduced to a lower level but except for Method 4 (fitting sample spectra), all 3 more or less have produced a dip at 2.1 μm (over-correction). The results presented here should caution researchers, both in laboratory measurement and payload development, that calibration targets with conspicuous absorption features should be avoided or carefully taken care of before doing data interpretations, or erroneous conclusions may result. The correction methods (except for Method 2 which requires additional measurements) will be used and evaluated in our analysis of the VisNIR spectra taken with the in situ VisNIR imaging spectrometer onboard the Chinese Chang’E 3′s Yutu Rover which carried a Spectralon-type plaque (PTFE) as the onboard calibration target in the NIR spectral region.

Acknowledgments

This study was initiated from a discussion with Carle M. Pieters who pointed out the possible artifacts when Spectralon plaque is used as calibration target in the NIR region. We thank Wang Ziwei, Yuan Ye and Sun Hui for help in sample preparations, Zhou Jian from Bruker Optics China and Ann M. Woys from Pike Technologies for technical support. Helps in laboratory setup from Ming Houli and Huang Dinghua are also greatly appreciated. The constructive and critical comments from three anonymous reviwers improved the quality of the manuscript. This work was supported in part by the National Natural Science Foundation of China through grants 41071229, 41173076, 41273079 and 41276180, and by the Minor Planet Foundation of China.


Documents

Effects of Spectralon absorption on reflectance spectra of ......Effects of Spectralon absorption on reflectance spectra of typical planetary surface analog materials Hao Zhang,1,*