Pushing Raman spectroscopy over the edge: purported signatures of organic molecules in fossil animals are instrumental artefacts

Widespread preservation of fossilized biomolecules in many fossil animals has recently been reported in six studies, based on Raman microspectroscopy. Here, we show that the putative Raman signatures of organic compounds in these fossils are actually instrumental artefacts resulting from intense background luminescence. Raman spectroscopy is based on the detection of photons scattered inelastically by matter upon its interaction with a laser beam. For many natural materials, this interaction also generates a luminescence signal that is often orders of magnitude more intense than the light produced by Raman scattering. Such luminescence, coupled with the transmission properties of the spectrometer, induced quasi‐periodic ripples in the measured spectra that have been incorrectly interpreted as Raman signatures of organic molecules. Although several analytical strategies have been developed to overcome this common issue, Raman microspectroscopy as used in the studies questioned here cannot be used to identify fossil biomolecules.


Introduction 31
Remnants or derivatives of ancient biomolecules are preserved in exceptional cases 32 in fossils, providing unique information to document the evolutionary history of life 33 during geological time. They can be used, for example, to clarify the phylogenetic 34 affinities of enigmatic fossils [1,2] , or to reconstruct the coloration of extinct organisms 35 such as invertebrates, feathered dinosaurs, and mammals [3] . 36 The search for such fossil biomolecules often requires combining as many 37 techniques as available [2] . Fossilized organic pigments were identified using a suite of 38 mass spectroscopy techniques such as gas chromatography-mass spectrometry (GC-39 MS) and time of flight secondary ion mass spectroscopy (ToF SIMS) [3] . Fourier-40 transform infrared (FTIR) spectroscopy of 1-billion-year-old microfossils was combined 41 with morphological and ultrastructural observations by transmission electron 42 microscopy (TEM) to interpret them as the earliest fungi [4] . Advanced synchrotron 43 spectroscopic techniques made it possible to highlight that a range of organic 44 (bio)molecules can sometimes experience only partial degradation during diagenesis 45 and even metamorphism, and be identified in various taxa including bacteria, plants 46 and animals [5][6][7][8][9][10][11][12][13][14] . Recently, it was suggested that conventional Raman spectroscopy 47 (i.e. equipped with a 532 nm laser as the excitation source under continuous 48 illumination) can be added to the list of techniques previously mentioned, and be used 49 alone to identify organic compounds in fossils [15][16][17][18][19][20] . 50 In the latter studies, spectroscopic data were interpreted as evidence for the 51 preservation of diverse organic degradation products of biomolecules in more than a 52 hundred different metazoan fossils, such as organic pigments in eumaniraptoran 53 dinosaur eggshells [15] and in a non-avian dinosaur skin [18] , as well as of protein, lipid 54 and/or sugar fossilization products in fossil bones [16] , dinosaur eggshells [20] , and 55 vertebrate and invertebrate soft-tissues [17,19] . Unfortunately, the purported claims of 56 biomolecules in these fossils are not well supported by the data provided, which 57 actually result from instrumental artefacts and data processing. In this paper, we 58 outline the limitations of Raman spectroscopy with respect to the identification of 59 biomolecules in fossil materials, and then describe in detail the origin of the 60 misinterpreted signal.

Raman spectroscopy has important limitations in the study of organic fossils 62
Raman spectroscopy is widely used in geosciences because it probes the vibration 63 modes of chemical bonds in both solids, liquids, and gases, with minimal sample 64 preparation [21] . Yet, there are several limitations in terms of the sensitivity and 65 accessibility of chemical fingerprints with the technique as used in the studies 66 questioned here. First, excitation with green 514-or 532-nm lasers mostly provides 67 specific information on C-C bonds --and not about other covalent linkages --in 68 diagenetically altered organic materials such as fossils [22] . As a result, Raman spectra 69 of organic materials preserved in (meta)sedimentary rocks are dominated by the so-70 called graphite (G) and defect (D1-D4) bands, which provide information about the 71 structural organization of the aromatic skeleton [23] . Consistently, Raman spectra of 72 geologically altered organic materials can be similar even when they have significantly 73 different elemental and molecular compositions [13,14,[24][25][26] . Second, under continuous 74 illumination, luminescence occurs concurrently with Stokes Raman scattering and 75 generates a signal that overlaps with the Raman spectral window [21,27,28] . Cross 76 sections of Raman (the probability that Raman scattering takes place) are typically 8 77 to 10 orders of magnitude smaller than that of luminescence. As a result, a number of 78 precautions are often necessary to be able to detect and interpret Raman spectral 79 features among a number of other spectral variations. 80

The reported periodic broad bands are not Raman signals 81
In all the studies questioned here [15][16][17][18][19][20] , the spectral bands assigned to organic 82 molecules are broader than the bands usually associated with Raman scattering, and 83 appear quasi-periodic, in contrast to the non-periodic spectral features typically 84 attributed to Raman scattering. 85 We investigated the periodicity of these bands using wavelet transform (Fig. 1), 86 an effective signal processing technique that is used to decompose a distorted signal 87 into different frequency scales at various resolution levels. Unlike classical Fourier 88 spectral analyses, wavelet transform analyses are advantageous in describing non-89 stationarities, i.e. localized variations in frequency or magnitude, and providing a direct 90 visualization of the changing statistical properties. It has become a common tool for 91 analysing localized variations within a time series [29,30] , but also for spike removal, 92 denoising and background elimination of Raman spectra [31,32] . We selected one 93 spectrum from each of the two studies for which data were made available [15,19] . For the wavelet analysis displayed in Fig. 1a,b, we selected the spectrum corresponding 95 to the eggshell of the extant flightless bird Rhea americana [15] , because it is more likely 96 that a pigment is preserved in a modern sample rather than in a fossil. For the wavelet 97 analysis displayed in Fig. 1c,d, we selected the spectrum collected from the crustacean 98 Acanthotelson stimpsoni specimen YPM52348 [19] , because the chitin-protein complex 99 of crustacean cuticles has a high preservation potential [8,33] , and this specimen appears 100 to be one of the best preserved (see fig. 1f in [19] ) --the spectrum clearly having been 101 measured from the specimen (unlike for the specimen shown in fig. 1d of [19] ). Note 102 that these two spectra, as well as all other reported ones, were provided by the original 103 authors as baseline-subtracted spectra, not as raw data. 104 Both spectra display numerous broad bands for which our wavelet transform 105 analysis reveals clear high-frequency periodicities of ~64-96 cm -1 for wavenumber 106 shifts <1000-1200 cm -1 , and of ~128 cm -1 for higher wavenumber shifts (Fig. 1a,c). 107 Similar high-frequencies of 130.9 cm -1 are obtained by Fast Fourier Transform. Note 108 that the same frequencies are found for all spectra provided by the authors. The 1086 109 cm -1 carbonate Raman peak present in the R. americana spectrum reflects the 110 calcified composition of the eggshell, in contrast to all the other (broader) bands, which 111 are best described as a superposition of quasi-periodic wavelets (Fig. 1b,d). These 112 broad, quasi-periodic bands are not the consequence of any Raman effect, but rather 113 result from physical and instrumental artefacts. Indeed, when a sample is illuminated 114 by the laser, the presence of structural defects and inorganic/organic components can 115 generate significant luminescence, often overwhelming the weak Raman signal [21,27] . 116 When this background luminescence is intense, the transmission properties of the 117 interferometric edge filter used to reject the Rayleigh line induce quasi-periodic 118 "ripples" in the measured spectrum [34] . 119 To further illustrate this point, we performed a wavelet analysis on a 120 transmission spectrum of a 532 nm RazorEdge ® ultrasteep long-pass edge filter, 121 provided by the manufacturer (Semrock), that is designed to be used as an ultrawide 122 and low-ripple passband edge filter for Raman spectroscopy (Fig. 2). The transmission 123 spectrum of the filter exhibits the aforementioned ripples ( Fig. 2a,b). Our wavelet 124 analysis highlights high-frequency periodicities of 64-96 cm -1 for low wavenumbers, 125 and of 128 cm -1 for higher wavenumbers (Fig. 2b, c), similar to the results reported in 126 the studies questioned herein [15][16][17][18][19][20] . Such edge filter-related instrumental artefacts actually explain the presence of most, if not all, of the broad bands that were attributed 128 to organic molecules. 129

Sample composition does not affect the position of ripples but impacts the 130
shape of the background 131 The transmission properties of the edge filter induce ripples on the measured spectra 132 when luminescence is intense, making it challenging to identify Raman features 133 without appropriate data processing for background subtraction [34] . The data provided 134 in the publications questioned here [15][16][17][18][19][20] are only the baseline-subtracted spectra, not 135 the raw data, which makes it impossible to precisely assess the impact of non-Raman 136 processes and sample composition on the corrected spectra from which the presence 137 of organic molecules was inferred. To address these issues, we collected Raman 138 microspectroscopy data on modern and fossil crustaceans in analytical conditions 139 similar to those of the aforementioned studies (for details, see Material and Methods 140 in SI). 141 We reproduced the experiment performed by McCoy et al. [19] using a specimen 142 of the crustacean Peachocaris strongi (Fig. 3a) from the same fossil locality (Mazon 143 Creek, Carboniferous, USA). As with other fossils from Mazon Creek, this specimen is 144 preserved as aluminosilicates and calcite in a sideritic concretion (Fig. S1). In order to 145 further assess the impact of the sample's chemical composition on the measured 146 spectra, we also performed Raman spectroscopy on (i) a specimen of the penaeid 147 shrimp Cretapenaeus berberus from the Cretaceous of Morocco (Fig. 3b) preserved 148 as a mixture of calcium phosphates and iron oxides in an illite mudstone ( Fig. S1; see 149 also Gueriau et al. [35] and references therein), and (ii) a specimen of the modern shrimp 150 Neocaridina davidii (Fig. 3c) dried after death and still rich in organic carbon, likely in 151 the form of chitin (Fig. S1). Whether or not organic carbon is present, and whatever 152 the mineralogical composition of the specimen or its mineral matrix, all the measured 153 spectra (Fig. 3d, solid lines) display broad bands, which all occur at the same 154 wavenumber shifts and add up to a significant background (Fig. 3d, dotted lines). Yet, 155 the shape of the background differs significantly from one measurement to another, 156 and the more intense it is, the more the ripples are expressed. In the baseline-157 subtracted spectra, the differences in the relative intensity between bands from one 158 measurement to another (Fig. 3e) only result from distinct background profiles of the 159 measurements. A wavelet transform analysis reveals high-frequency periodicities of 64-128 cm -1 (Fig. 3f), as was the case for the spectra questioned in the previous 161 section [15][16][17][18][19][20] . Finally, other than the presence of sharp peaks around 964 and 1086 cm -162 1 (Raman peaks of fluorapatite and calcite, respectively), as well as one unidentified 163 peak at 1156 cm -1 in the modern shrimp (possibly carotenoids), which are all three still 164 expressed after subtraction of the frequency components (Fig. 4), spectral differences 165 are limited to relative variations in the ripple band intensity that result from the shape 166 and quality of the baseline fit. 167 In short, the ripples observed in the Raman microspectroscopy data questioned 168 here represent remnant instrumental signals that result from confounding broad 169 luminescence and inappropriate data processing. The broad luminescence transmitted 170 by the edge filter induced the ripple-shape features above the cut-off wavelength on 171 the raw spectrum. Background correction did not eliminate the ripple-shape distortions 172 induced, and instead accentuated them to give the appearance of putative signatures 173 of organic molecules. 174

Conclusion and Outlook 175
Broad bands interpreted to be Raman signatures of diverse organic molecule 176 degradation products in various metazoan fossils [15][16][17][18][19][20] are artefactual quasi-periodic 177 ripples induced by the edge filter due to intense luminescence, and there is no 178 evidence for Raman signal of organic molecules. Unfortunately, conventional Raman 179 microspectroscopy does not provide direct information on fossil biomolecules [22] . 180

Conventional Raman spectroscopy remains an important paleontological tool 181
providing crucial information on the mineralogical composition of fossils and the degree 182 of crystallization of the carbonaceous remains they preserve, which is often used to 183 quantify the peak temperature organics reached during geological burial [23] . Extracting 184 and interpreting the data, however, requires robust and optimized analytical strategies 185 and/or data processing. Several methods have been developed to remove, a 186 posteriori, the undesired contribution of luminescence and ripples in Raman 187 spectra [34,36] . Note that in the papers discussed here [15][16][17][18][19][20] , such processing would leave 188 no signal other than the mineral peaks. Distinct excitation wavelengths, such as near-189 infrared and ultraviolet, can also be used to significantly limit luminescence [37,38] . 190 Alternatively, non-conventional time-resolved Raman spectroscopy offers new ways to 191 limit or exploit luminescence signals, while techniques based on coherent anti-Stokes ultraviolet resonance Raman spectroscopy, allow the Raman signal to be considerably 194 enhanced (see Beyssac [27] for review). Furthermore, synchrotron-based X-ray Raman 195 scattering can probe the chemical speciation of light elements such as carbon, in 196 heterogeneous materials usually encountered in life, earth, environmental and 197 materials sciences [39,40] . 198 The search for biomolecules in fossils is a very exciting field of research, offering 199 critical knowledge on both evolutionary events and fossilization processes, yet 200 conventional Raman spectroscopy alone cannot be used to identify fossil 201 biomolecules. Instead, non-conventional Raman spectroscopy, mass spectrometry 202 and infrared and X-ray absorption spectroscopy techniques, are successfully used by 203 paleontologists to identify fossil biomolecules in the geological record [2,41] . 204

Supporting Information 205
Supporting Information, including details on materials and methods and a supporting 206