Large-Field Electron Imaging and X‑ray Elemental Mapping Unveil the Morphology, Structure, and Fractal Features of a Cretaceous Fossil at the Centimeter Scale

Large-Field Electron Imaging and X‑ray Elemental Mapping Unveilthe Morphology, Structure, and Fractal Features of a CretaceousFossil at the Centimeter ScaleNaiara C. Oliveira,† Joao H. Silva,‡ Olga A. Barros,§ Allysson P. Pinheiro,∥ William Santana,⊥

Anto nio A. F. Saraiva,§ Odair P. Ferreira,# Paulo T. C. Freire,∇ and Amauri J. Paula*,†

†Solid-Biological Interface Group (SolBIN), Departamento de Física, Universidade Federal do Ceara, P.O. Box 6030, 60455-900Fortaleza, Ceara, Brazil‡Universidade Federal do Cariri, Cidade Universitaria, 63048-080 Juazeiro do Norte, Ceara, Brazil§Laboratory of Paleontology, Departamento de Ciencias Físicas e Biologicas, and ∥Semiarid Crustaceans Laboratory, LACRUSE,Universidade Regional do Cariri, 63105-000 Crato, Ceara, Brazil⊥Sistematic Zoology Laboratory (LSZ), Pro-Reitoria de Pesquisa e Pos-Graduacao, Universidade Sagrado Coracao (USC), 17011-160Bauru, Sao Paulo, Brazil#Laboratory of Advanced-Functional Materials (LaMFA) and ∇Laboratory of Raman Spectroscopy, Departamento de Física,Universidade Federal do Ceara, 60455-900 Fortaleza, Ceara, Brazil

*S Supporting Information

ABSTRACT: We used here a scanning electron microscopyapproach that detected backscattered electrons (BSEs) and X-rays (from ionization processes) along a large-field (LF) scan,applied on a Cretaceous fossil of a shrimp (area ∼280 mm2)from the Araripe Sedimentary Basin. High-definition LFimages from BSEs and X-rays were essentially generated byassembling thousands of magnified images that covered thewhole area of the fossil, thus unveiling morphological andcompositional aspects at length scales from micrometers tocentimeters. Morphological features of the shrimp such aspleopods, pereopods, and antennae located at near-surfacelayers (undetected by photography techniques) were unveiledin detail by LF BSE images and in calcium and phosphoruselemental maps (mineralized as hydroxyapatite). LF elemental maps for zinc and sulfur indicated a rare fossilization eventobserved for the first time in fossils from the Araripe Sedimentary Basin: the mineralization of zinc sulfide interfacing tohydroxyapatite in the fossil. Finally, a dimensional analysis of the phosphorus map led to an important finding: the existence of afractal characteristic (D = 1.63) for the hydroxyapatite−matrix interface, a result of physical-geological events occurring withspatial scale invariance on the specimen, over millions of years.

Fossils are remains or vestiges of organisms that livedmillions of years ago, formed through several chemical

processes, many of them not perfectly understood. They are theresult of a process that starts when an organism or its bodyparts are trapped in sediments in an environment poor inbacteria and rich in minerals dissolved in fluvial flood or oceanicwater. They also represent extremely important sources ofinformation on both the environment (e.g., soil, diagenesis ofdeposits, weather, food availability) and the organisms (e.g.,species, territorial distribution, food supply, evolutionarycharacteristics) from the Earth of ancient times.1−5 The correctidentification of the fossil material is essential not only fortaxonomic studies, but also for studies on paleoecologicalinvestigations, population dynamics, biogeography, and evolu-tion. However, the lack of diagnostic characteristics due to thecommonly poor state of preservation of the material can

undermine the precise identification of the organisms. Newtechniques for detailed analysis of fossils have been increasinglyused and progressively contributed to the advance of thesestudies. Such techniques involve the use of computedtomography scan,2,6−8 synchrotron radiation,9−12 and severalmicroscopy approaches such as transmission electron micros-copy,13−15 scanning electron microscopy,16−19 infrared-lightmicroscopy,20,21 scanning transmission X-ray microscopy,22,23

and X-ray fluorescence microscopy.24−27 More recently,scanning electron microscopy (SEM) was applied to fossilswith areas up to ∼230 mm2 using assemblies of magnified

Received: July 22, 2015Accepted: September 5, 2015

Article

pubs.acs.org/ac

© XXXX American Chemical Society A DOI: 10.1021/acs.analchem.5b02815Anal. Chem. XXXX, XXX, XXX−XXX

electron images and X-ray elemental maps.18,19 This bottom-upassembly method was used to generate large, high-definitionmicrographs that provided detailed morphological informationon Cambrian arthropods. Here, we extend this analyticalapproach to a large-field scan which is capable of unveiling inhigh definition, and at length scales from micrometers tocentimeters, morphological, compositional, structural, anddimensional aspects at the near-surface layers of a shrimpfossil (area of about 280 mm2) from the Cretaceous periodfrom the Ipubi Formation.28

This geological formation belongs to the Araripe Basin,29,30

which is located at the geographic borders of the Ceara, Piaui,and Pernambuco states in Brazil and extends to about 10000km2. The Araripe Basin has a variety of fossils from theCretaceous period, including fish, ostracods, gastropods, turtles,crocodilians, pterosaurs, and dinosaurs. It presents a diversifiedstratigraphic structure consisting of limestones, sandstones,evaporites, shales, and concretions. Furthermore, as it holdspyrite, gypsum, and calcium carbonate in different geologicaldeposits, it is possible to find fossils produced by diversechemical processes.31 As a consequence, from a paleontologicalperspective, the Araripe Basin is considered an importantgeological formation not only because of the diversity andquality of specimens found, but also because of the occurrenceof several fossilization processes.32−34

The large-field (LF) imaging approach used here for theshrimp fossil comprises the assembly of thousands of magnifiedimages obtained in SEM from the detection of both electronsand X-rays emitted after the interaction between the incidentelectron beam and the sample. LF images acquired through thismethod contain a huge amount of information at several lengthscales, from a few tens of micrometers to centimeters, whichallows the interpretation of the fossil morphology and thegeological context. The method also provides insights regardingthe mineralization of the fossil through the elemental and phasecompositions and its fractal analysis, the latter being a powerfultool to quantitatively estimate the complex interface growthphenomenon that occurred during the fossilization process.

■ EXPERIMENTAL SECTION

The large-field scan was carried out in a Quanta-450 electronmicroscope (FEI) with a field-emission gun (FEG), a 100 mmstage, and an X-ray detector (model 150, Oxford) for energy-dispersive X-ray spectroscopy (EDS). The fossil material wasinserted into the microscope chamber without samplepreparation. The analyses were performed in a low vacuum(approximately 100 Pa in water vapor) to avoid samplecharging. Images were acquired at beam acceleration voltagesvarying from 5 to 20 kV and with a condenser aperture of 50μm. For 20 kV of acceleration, which was the best condition for

Figure 1. (a) Photograph of the fossil material from the shales of the Ipubi Formation, Araripe Sedimentary Basin (Ceara, Brazil). (b) Schemeshowing the ideal overlapping of grayscale value distributions for adjacent micrographs that were used to generate large-field (LF) images. Large-fieldscanning electron microscopy (LF-SEM) of the fossil. (c) Backscattered electron (BSE) image assembly with topographic information (BSE T). (d)BSE image assembly with composition information (BSE Z) obtained in low-contrast (LC) mode. (e) BSE Z image assembly obtained in high-contrast (HC) mode. The black scale bar corresponds to 10 mm.

Analytical Chemistry Article

DOI: 10.1021/acs.analchem.5b02815Anal. Chem. XXXX, XXX, XXX−XXX

B

acquiring the backscattered electron (BSE) images andelemental maps (see the Supporting Information for details),the beam current over the specimen was about 45 nA (valueprovided by the manufacturer considering the conditions usedin the column: condenser lens aperture, condenser lensconvergence angle, and accelerating voltage). A BSE detectorwas fixed at the end of the polar piece, and an X-ray detectorwas inserted at a collection angle of 55° with the column axis,positioned approximately at the end of the polar piece. In allanalyses, the working distance was set at 15 mm to maximizethe depth of focus, considering the sample roughness at amicrometer scale. For increasing the beam’s high-vacuum pathand minimizing spurious beam skirting, a gaseous analyticalcone (attached to the BSE detector) was used in all scans.To generate the large-field images, an overlapping of

marginal areas (a border which contains 20% of the imagearea) of adjacent images acquired independently afterdislocations of the microscope stage along the x and y axeswas performed. The largest constructive interference betweenthe two-dimensional distributions of grayscale values for theoverlapped adjacent images determines the best positions fortheir placement. The resulting images presented in this paperare assemblies of more than 3600 adjacent images acquired inindividual scans at 1000× magnification (horizontal and verticalfields of 0.41 and 0.29 mm, respectively; 512 × 368 pixels), andhave a definition of 5906 × 2119 pixels (600 pixel in.−1), with apixel size corresponding to ∼5 μm of the sample. Thedetermination of white pixels corresponding to the interface ofthe fossil (i.e., hydroxyapatite interface determined in thephosphorus elemental map) was performed by using theintegral image method35 programmed on Wolfram Mathema-tica (see the Supporting Information for details). Afterdetermination of the presence of white pixels inside each boxwith an increasing lateral size, the fractal dimension (D, orHausdorff dimension) was calculated. For detailed informationregarding the methods used for the LF scan, the imageprocessing, and the dimensional analysis, see the SupportingInformation.

■ RESULTS AND DISCUSSIONMorphological Assessment of the Fossil through LF

BSE Images. The fossil material studied here has a size ofabout 28 mm in length and 10 mm in width (see Figure 1a).Analyses of the material found in the shales of the IpubiFormation by light microscopy suggested, from a morpho-logical perspective, that it consists of a shrimp (decapodcrustacean). Clear morphological details that could enableaccurate classification could not be observed through this typeof analysis, and thus, we generically name the species here as A1(Figure 1a).To image the whole fossil area in high definition, a large-field

scan was performed in which the microscope acquiresthousands of images with a suitable magnification for allpositions covering the fossil area. Although commercialinstruments currently available possess a high-precisionpositioning of the stage, images acquired during this large-field scan present a positioning mismatch from approximately1% to 10% that prevents a coherent assembly of LF images. Inthis way, this critical process of image assembly can beautomated through an image overlapping algorithm thatperforms an appropriate matching between a pair of imagesin regard to their two-dimensional distributions of grayscalevalues (which vary from 0 to 1 at each point of the image plane;

see Figure 1b). Through this method, we were able to assemblemore than 3600 micrographs (with horizontal and vertical fieldsof 0.41 and 0.29 mm, respectively; 512 × 368 pixels each),covering the whole area of the fossil. The final assembledimages (5906 × 2119 pixels; 600 pixel in.−1) cover a samplearea of approximately 280 mm2 and have a pixel sizecorresponding to ∼5 μm of the sample. The minimum spatialresolution in the LF images was ∼25 μm, which was thesmallest length scale used in the dimensional analysis.Along the large-field scan, two signals manifested from the

electron beam−sample interaction were captured: one fromBSEs and another from the X-rays emitted. The signal fromBSEs was first manipulated to provide a contrast functionrelated to the topography of the sample (BSE T), which wasgenerated by subtracting the signal captured at half of the BSEannular detector from the signal captured at the other half (seeFigure 1c). By summing both signals (for the full annulus), acontrast function related to the composition of the sample(BSE Z) was acquired, originated from the relationshipbetween the BSE yield and the atomic mass (g mol−1) ofeach element present in the sample (see Figure 1d,e). Thesedifferent contrast functions used for topography (BSE T) andcomposition (BSE Z) can be compared through histograms ofgrayscale values extracted from the images (see the right panelsin Figure 1c−e). The topographic contrast (BSE T; see theright panel of Figure 1c) along the image does not substantiallyvary, as the fossil surface is rather flat, thus resulting in agrayscale histogram that has a progressive decrease in thenumber of pixels from black (zero intensity) to gray (around0.6) tones. In addition, for images associated with BSE Z (seeFigure 1d,e), the contrast is rather different and can also bealtered to obtain very distinct information regarding either themorphology or the composition of the shrimp.Low-contrast LF BSE images (BSE Z LC; see Figure 1d) are

suitable for providing differentiations in the chemicalcomposition along the fossil sample but fail to fully reveal thefossil morphological contours. These low-contrast imagespresent a grayscale histogram with a larger quantity of darkpixels (approximately 1 × 107 black pixels; see the right panel inFigure 1d). Regions with tones close to white (i.e., grayscalevalue 1) reveal areas of the fossil that have differentcompositions, and regions in gray tones partially reveal thefossil morphology imprinted in the matrix. However, as thesesamples commonly comprise a large quantity of elements, 16 inthis particular case, the interpretation of low-contrast BSE Zimages becomes difficult without the association with elementalmaps. These maps were obtained through the X-rays emittedfrom atomic ionization processes occurring through theelectron beam interaction. As further discussed here, thebrighter tones in Figure 1d are mainly associated with thepresence of Zn, the heaviest (highest atomic mass) among themajor elements present in the fossil, with the highest BSE yield.On the other hand, in high-contrast BSE images (BSE Z HC;

see Figure 1e), the quantity of brighter tones in the histogramswas increased, resulting in a balance of black and white pixelsalong the image (i.e., ∼3 × 106 pixels for both; see the rightpanel in Figure 1e). In these conditions, it is possible to fullyreveal the fossil contours due to the BSE yield of Ca. However,information regarding the composition along the fossil is lost bythe contrast increase. In this high-contrast image, the pleopodsand pereopods of the shrimp, features that are undetected inFigure 1a, were revealed in near-surface layers.

Analytical Chemistry Article

DOI: 10.1021/acs.analchem.5b02815Anal. Chem. XXXX, XXX, XXX−XXX

C

Compositional Assessment through LF ElementalMaps. The X-rays emitted from the specimen were detectedduring the LF scan along with the detection of backscatteredelectrons, thus allowing the assembly of the elemental maps.Considering the effect of the acceleration voltage on the large-field elemental maps produced, it was observed that scans withthe electron beam accelerated at 5 kV produced images with ahigher noise-to-signal ratio when compared to the imagesobtained at 20 kV for the same beam dwell time (60 μs perpixel). In addition, as no significant morphological differenceswere observed when comparing the maps generated at 5 and 20kV up to a scale of a few tens of micrometers (see Figure S2 inthe Supporting Information), elemental maps obtained at 20kV were preferably used due to the low noise-to-signal ratio(for detailed considerations on the use of this acceleratingvoltage, see the Supporting Information). The cumulative EDSspectrum containing the sum of all spectra (more than 3600)obtained through the LF scan is represented in the top panel ofFigure 2. Maps for elements present in the specimen with a

relative concentration above 1.0 wt % are represented in thebottom panels of the same figure. These maps were renderedby filtering the X-ray signal through an energy gap of 80 eV(spectral resolution ∼5 eV), centered at the energy associatedwith the peak maximum intensity for each element in the EDSspectrum (see Figure 2, top panel). Maps for elements withrelative concentration above 1.0 wt % were generated using Kαtransitions (i.e., electronic transitions from L to K shells), whichrepresent the most accurate signal that could be used for allelements. Although the carbon signal was excluded in the mapsand calculations due to the possible influence of contaminationduring the fossil handling, its exclusion did not impact thequantitative interpretations performed here, as they werecarried out by using just ratios between the elements’ relativemolar concentrations (in molar units). In the fossil, carbon ispresent mainly in carbonates (e.g., CaCO3).Fossils previously studied from the Ipubi Formation36 were

mainly imprints between dark shales that are rich in organicmatter and calcium sulfates. When P and Ca are traced in the

Figure 2. Top panel: total EDS spectrum for the whole fossil area. Bottom panels: elemental maps generated by using the signal of each ionizationpeak described. Elemental maps are represented from left to right, top to bottom, as a function of their relative concentrations (wt %) calculatedfrom the total energy-dispersive X-ray spectrum. Elements with quantities larger than 1.0 wt % are shown. The white scale bar corresponds to 10mm.

Analytical Chemistry Article

DOI: 10.1021/acs.analchem.5b02815Anal. Chem. XXXX, XXX, XXX−XXX

D

e lementa l maps (minera l i zed as hydroxyapat i t e[Ca5(PO4)3OH]; see Figure 2), it is possible to observe inbetter detail the characteristics of the abdomen, the partiallypreserved carapace, parts of the pleopods, and other cephalicappendages (i.e., antenna and antennule). Maps also makepossible the improvement of the description of othercharacteristics such as the thoracic appendages (pereopods3−5), the abdominal pleura of somites 4−5, and newmorphological details of the pleopods, uropods, and telson.The rostral spines were not preserved, and parts of theantennae and antennules were curved upward, above thecephalothorax.An appropriate interpretation of the fossil composition is

better performed in association with elemental maps (seeFigure 3), as there are 16 elements identified through EDS (seeTables S1 and S2 for details regarding the energy lines used fordetection and the elements’ relative concentrations, respec-tively). On the elemental composition of the fossil, it wasobserved that there are three characteristic regions: the matrix(region A) containing mainly O, Si, Ca, S, Al, Mg, and Fe, aphosphorus-rich region of the fossil (region B) containing O,Ca, P, F, S, and Na, and finally a region in the fossil containingmostly S and Zn (region C).The calculation of the relative concentration for all elements

identified in the cumulative EDS spectrum (see Figure 2) aswell as in spectra for regions A, B, and C (see Figure 3) isrepresented in Table S2. Data from the cumulative EDS

spectrum for the whole area of the fossil are represented inTable S2 in an ascending order in regard to the atomic mass (gmol−1) of elements detected. Surprisingly, the peaks of traceelements present at relative concentrations smaller than 1.0 wt% could be detected in the cumulative EDS spectrum thoughtheir relative concentrations (wt %) could not be considered inquantitative interpretations. This is the case for Cl, K, Ti, Ce,and Yb.The fluorine map indicates that the presence of this element

is limited to the boundaries of the fossil, largely overlapping thephosphorus map. This fact suggests the possibility of a partialsubstitution of the hydroxyl group of hydroxyapatite by fluorine[Ca5(PO4)3(OH(1−x)Fx)]. The total substitution can lead to theformation of fluoroapatite [i.e., Ca5(PO4)3F], which canprovide to the fossil more chemical stability and resistance todissolution.37,38 Another major conclusion was made consider-ing the overlapping of zinc and sulfur elemental maps in Figure2, and also considering their normalized relative molarconcentrations (NRMCs, in molar units) calculated throughspectra from region C (approximate ratio of 97.5 mol of sulfurto 99.8 mol of Zn; see Table S2), which confirms the presenceof zinc sulfide (ZnS) in the fossil. The presence of (Zn1−xFex)S(wurtzite or sphalerite with a high amount of iron [x value])and FeS2 (pyrite) must be considered to a lesser extent, becausethere is an overlapping of Fe, Zn, and S elemental maps insidethe fossil contours, but the Fe normalized relative molarconcentration is very low in regard to Zn and S (ratio of

Figure 3. EDS spectra from specific regions of the fossil associated with the areas shown in the LF BSE Z LC image (top panel). Regions defined arethe (A) matrix, (B) phosphorus-rich area of the fossil, and (C) zinc- and sulfur-rich area of the fossil. Bottom panels: resulting spectra from the sumof at least five spectra acquired in each region described.

Analytical Chemistry Article

DOI: 10.1021/acs.analchem.5b02815Anal. Chem. XXXX, XXX, XXX−XXX

E

approximately 1:100; see Table S2). Although the presence ofsulfur in other compositions such as sulfates (e.g., CaSO4·xH2O) must also be considered, EDS spectra for this regionshow an approximate sulfur:oxygen molar ratio (around 2:1)that does not indicate the presence of sulfates in highconcentrations, thus supporting the conclusion regarding thelarger mineralization of ZnS or (Zn1−xFex)S, with a low x value.Furthermore, the oxygen concentration (wt %) value alsocomprises signals from impurities on the fossil surface and fromthe water vapor used in the chamber (to attain 100 Pa), whichleads to the conclusion that the sulfur:oxygen molar ratio onregion C is in fact larger than 2:1. Fossilization through themineralization of pyrite in the Araripe Sedimentary Basin waspreviously observed.36 However, this is the first observation ofthe formation of zinc sulfide in fossils from this basin.The presence of ZnS in fossils, appearing as sphalerite or

wurtzite, was observed in biogenic remains in the form ofinternal casts, as infillings of voids in skeletons, and also asreplacement of skeleton materials.39,40 More specifically, thiscompound has been observed in fossils from deep-sea hotsprings at hydrothermal vent fields present along crests ofvolcanic spreading ridges.41−43 Fossils in these areas wereformed from fauna commonly comprising worms, crabs, and avariety of fish.44,45 The mineralization of ZnS in these fossils isrelated to the acidification of the solution and the presence ofhydrogen sulfide,46 which consequently induce the precip-itation of zinc sulfide.44,45

The pivotal point, however, is that the presence of sulfur andzinc in fossils is a limited and rare event. In the case ofgastropods,39 the organisms accumulate a large quantity of Znduring their lives, allowing this chemical element to be availablein the environment after their deaths to form compounds suchas ZnS. Regarding the shrimp A1, the presence of Zn can alsobe attributed to an accumulation process during life, although itis known through modern toxicological studies that shrimps arenot an efficient trap for Zn47,48 as are gastropods,39,49−51 andare able to regulate their total body levels of Zn.52 Obviously,we cannot exclude a different origin for Zn, such as itsintroduction by water with a high concentration of zinc.The process of ZnS formation itself can be very complex, as

pointed out in the case of gastropod fossils. In fact, at least twopossibilities were found: a direct infilling of the carbonate shellsby ZnS, or the infilling of the carbonate shells by calcite and/orpyrite, with the replacement of these materials by sphalerite orwurtzite.39 Related to the direct infilling by ZnS, it was shownthat the presence of sulfate-reducing bacteria cultured in aZnSO4/FeSO4 medium leads to ZnS precipitation, but ironsulfides such as pyrite are not precipitated.53 This is animportant finding that can account for the formation of zincsulfide in the shrimp fossil studied here because, as our datashow, only about 0.5 wt % Fe was found in the region of thefossil where ZnS is present.On the other hand, one cannot discard a more complex

mechanism involving the precipitation of ZnS, calcite, andpyrite. The fossilization process of some fossils from the IpubiFormation occurs through pyritization in several stages, aspreviously shown by our group.36 In this way, the hypothesis offormation of ZnS through different stages is another possibility,similarly to the process occurring for gastropods.39 First, theanimal is infilled by iron sulfides and calcite, and finallysphalerite or wurtzite substitutes for the calcite casts.Dimensional Analysis of the Fossil Hydroxyapatite

Interface. Although it was observed in LF images that the

fossil was preserved from a macroscopic perspective, thusallowing paleontological and morphological interpretation, atsmall length scales (micrometers), the contours of the fossil−matrix interface are irregular and complex, a result of thetaphonomic processes. In this context, a suitable approach forobtaining quantitative information about these processes is thedimensional analysis of the fossil interfaces, more specificallythe contours that define the interfaces between the fossil andthe matrix. This quantitative analysis could be performed atdifferent length scales because LF images possess a contourdefinition from micrometers to centimeters. The dimensionwas assessed through a box-counting method that divides theimage into a determined number of square areas (or boxes;N( )) which is dependent on the lateral size ( ), thus coveringthe whole image (for details, see the Experimental Section andthe Supporting Information). As seen in Figure 4a−c, the

fossil−matrix interface dimension was assessed by determiningthe presence of white pixels inside these boxes after thephosphorus elemental map was binarized (i.e., converted to ablack and white scale, with 0 and 1 intensities, respectively),and the contours were revealed by applying a differentiationfunction on the resulting binarized contrast function (i.e., edge-finding process).Considering the approach introduced by Mandelbrot for the

dimensional evaluation of natural events,54 the edge-finding

Figure 4. (a) Phosphorus elemental map from Kα electronic transitionassociated with the presence of hydroxyapatite. (b) Example of thebinarization and edge-finding processes sequentially applied to adetermined area of the phosphorus map. (c) Hydroxyapatite−matrixinterface revealed by image processing performed for the wholephosphorus map. (d) Logarithms of the number of boxes used tocover the two-dimensional hydroxyapatite interface (log[N( )]) andthe inverse of the box size (log[1/ ]) found in the dimensionalanalysis. Linear fitting of the points is shown as an inset.

Analytical Chemistry Article

DOI: 10.1021/acs.analchem.5b02815Anal. Chem. XXXX, XXX, XXX−XXX

F

process used here reveals what would be analogous to “islandsand lakes”, the boundaries of which correspond to the fossil−matrix interface. Therefore, boundaries (i.e., contours) ofindividual islands and lakes do not contact each other. Thehydroxyapatite interface was analyzed through the boundariesrevealed in the phosphorus elemental map. The dimensionalanalysis for the P elemental map (see Figure 4d) shows theexistence of a linear relationship between the logarithms of thenumber of boxes used to cover the two-dimensionalphosphorus interface (log[N( )]) and the inverse of the boxsize (log[1/ ]), confirming the fractal spatial arrangement ofthis interface, a result related to the hydroxyapatite mineraliza-tion process. The linear fit precisely provided a fractaldimension D (Hausdorff dimension) of 1.63 for the interface.The fractal dimension in this case demonstrates the spatial

scale invariance and self-similarity of the hydroxyapatite−matrixinterface, and it is a number specifically related to the fossilcontours as well as to its fragmentation attained over time, bothruled by diffusion phenomena that occurred along and acrossthe P elemental map plane analyzed (see Figure 4a). Anillustrative comparison can be made with the dimensionalfeatures of continents and islands. The fractal dimension of theAustralian coastline is 1.13.55 This number rises if (i) thecontinent is fragmented in the sea (the Norway coastline has D= 1.52) and/or if (ii) the contours of the lakes and rivers insidethe continent are considered along with the contours of thecoastline (i.e., water−crust interface seen from above). Morespecifically considering fossils, values of D close to thatobtained (1.63) were observed for coral corallites from theUpper Jurassic,56 graphoglyptid trace fossils,57,58 and rangeo-morph fronds from the Ediacaran Period.59 However, it mustbe mentioned that similar fractal dimensions can correspond toradically different morphologies, so that other morphometricanalyses must be performed to interpret them. Finally, theresults from the box-counting method are influenced by thedefinition of the images used. In this way, imaging methodssuch as that presented here, which allows the obtainment ofimages in ultrahigh definition, can substantially improve theprecision of the calculation of the fractal dimension.From a macroscopic perspective, the morphology of the

living animal was distributed along the space (Euclidean)through an auto-organized arrangement as a result of emergentproperties from its intrinsic complex nature. As observed in theLF images, its outer morphological contours had macroscopicdimensional characteristics that are partially preserved in thefossil. In addition, at the micrometer scale, the fossilization ofthe animal is a result of long-term physical-geological processesthat took place correlated with the morphology of the livinganimal: mineral nucleation and growth were limited to someextent by the animal contours. In a first analysis, it must beconsidered that geometrically well-defined mathematicalmodels can generate fractals with fragmentation anddimensions close to those observed for this fossil (for instance,fractals based on generalized Koch curves can attain D ≈1.61).54 However, the hydroxyapatite−matrix interface wasformed through mineralization events that can be dimensionallybest correlated with randomly generated fractals (e.g., Brownfractals).60

■ CONCLUSIONSAn LF scanning electron microscopy approach was used herefor analyzing a Cretaceous fossil of a shrimp from the AraripeSedimentary Basin (Brazil) by imaging through the detection of

BSEs and X-rays from ionization processes. The resulting LFimages revealed morphological features of the fossil at near-surface layers, which could not be detected through lightmicroscopy. The morphological details of the pleopods,pereopods, antennae, antennules, and somites of the shrimpwere better visualized in the LF BSE images and also in LFcalcium and phosphorus elemental maps, which are present inthe fossil mineralized as hydroxyapatite. Furthermore, thecorrelation of LF elemental maps with EDS spectra of specificregions of the fossil revealed important aspects of thefossilization process. In this particular case, it was observedhere that, along with the fossilization of the animal through theformation of hydroxyapatite, ZnS was also formed, interfacingwith the hydroxyapatite. This is a rather rare process offossilization, for the first time observed in the AraripeSedimentary Basin. As the LF images possess preciseinformation at several length scales, ranging from a few tensof micrometers to centimeters, this imaging approach alsoallowed a dimensional analysis of the specimen, morespecifically of the interface formed between hydroxyapatiteand the matrix. This numerical analysis surprisingly revealed theexistence of a fractal arrangement of this interface (Hausdorffdimension of 1.63), a result of physical-geological events thatoccurred over millions of years, with spatial scale invariance.Finally, it must also be mentioned that this image processingapproach can be extended to other scientific contexts, forinstance, to polymers, metals, rocks, and organic tissues, inwhich morphological, structural, compositional, and dimen-sional characterization must be carried out along a wide rangeof length scales, thus providing unique information for eachcontext.

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.anal-chem.5b02815.

Detailed information on the methods used, scanningelectron micrographs, and X-ray energy-dispersive spec-troscopy tables (PDF)

■ AUTHOR INFORMATIONCorresponding Author*Phone: +55 85 3366 9270. E-mail: [email protected].

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSWe gratefully thank Central Analitica-UFC/CT-INFRA/MCTI-SISNANO/Pro-Equipamentos CAPES for providingthe FEG-SEM instrument, as well as CNPq and FUNCAPfor funding this research (Grant 446800/2014-7).

■ REFERENCES(1) Thomas, D. B.; Nascimbene, P. C.; Dove, C. J.; Grimaldi, D. A.;James, H. F. Sci. Rep. 2014, 4, 5226.(2) Castanhinha, R.; Araujo, R.; Junior, L. C.; Angielczyk, K. D.;Martins, G. G.; Martins, R. M. S.; Chaouiya, C.; Beckmann, F.; Wilde,F. PLoS One 2013, 8, e80974.(3) Zabini, C.; Schiffbauer, J. D.; Xiao, S.; Kowalewski, M.Palaeogeogr., Palaeoclimatol., Palaeoecol. 2012, 326−328, 118−127.

Analytical Chemistry Article

DOI: 10.1021/acs.analchem.5b02815Anal. Chem. XXXX, XXX, XXX−XXX

G

(4) Bennett, C. E.; Williams, M.; Leng, M. J.; Siveter, D. J.; Davies, S.J.; Sloane, H. J.; Wilkinson, I. P. Palaeogeogr., Palaeoclimatol.,Palaeoecol. 2011, 305, 150−161.(5) Maisey, J. G. Environ. Biol. Fishes 1994, 40, 1−22.(6) Ketcham, R. A.; Carlson, W. D. Comput. Geosci. 2001, 27, 381−400.(7) Spoor, F.; Wood, B.; Zonneveld, F. Nature 1994, 369, 645−648.(8) Sereno, P. C.; Wilson, J. a.; Witmer, L. M.; Whitlock, J. a.; Maga,A.; Ide, O.; Rowe, T. a. PLoS One 2007, 2, e1230.(9) Houssaye, A.; Xu, F.; Helfen, L.; De Buffrenil, V.; Baumbach, T.;Tafforeau, P. J. Vertebr. Paleontol. 2011, 31, 2−7.(10) Bergmann, U.; Morton, R. W.; Manning, P. L.; Sellers, W. I.;Farrar, S.; Huntley, K. G.; Wogelius, R. a; Larson, P. Proc. Natl. Acad.Sci. U. S. A. 2010, 107, 9060−9065.(11) Muscente, A. D.; Marc Michel, F.; Dale, J. G.; Xiao, S.Precambrian Res. 2015, 263, 142−156.(12) Anne, J.; Edwards, N. P.; Wogelius, R. A.; Tumarkin-Deratzian,A. R.; Sellers, W. I.; van Veelen, A.; Bergmann, U.; Sokaras, D.;Alonso-Mori, R.; Ignatyev, K.; Egerton, V. M.; Manning, P. L. J. R. Soc.,Interface 2014, 11, 20140277.(13) Javaux, E. J.; Knoll, A. H.; Walter, M. R. Geobiology 2004, 2,121−132.(14) Li, J.; Bernard, S.; Benzerara, K.; Beyssac, O.; Allard, T.;Cosmidis, J.; Moussou, J. Earth Planet. Sci. Lett. 2014, 400, 113−122.(15) Brasier, M. D.; Antcliffe, J.; Saunders, M.; Wacey, D. Proc. Natl.Acad. Sci. U. S. A. 2015, 112, 4859−4864.(16) Garcia-Ruiz, J. M.; Hyde, S. T.; Carnerup, A. M.; Christy, A. G.;Van Kranendonk, M. J.; Welham, N. J. Science 2003, 302, 1194−1197.(17) Matzke-karasz, R.; Neil, J. V.; Smith, R. J.; Symonova, R.;Morkovsky, L.; Archer, M.; Hand, S. J.; Cloetens, P.; Tafforeau, P.Proc. R. Soc. London, Ser. B 2014, 281, 20140394.(18) Ma, X.; Cong, P.; Hou, X.; Edgecombe, G. D.; Strausfeld, N. J.Nat. Commun. 2014, 5, 3560.(19) Ma, X.; Hou, X.; Edgecombe, G. D.; Strausfeld, N. J. Nature2012, 490, 258−261.(20) Riquelme, F.; Northrup, P.; Ruvalcaba-Sil, J. L.; Stojanoff, V.;Peter Siddons, D.; Alvarado-Ortega, J. Appl. Phys. A: Mater. Sci. Process.2014, 116, 97−109.(21) Edwards, N. P.; Barden, H. E.; van Dongen, B. E.; Manning, P.L.; Larson, P. L.; Bergmann, U.; Sellers, W. I.; Wogelius, R. a. Proc. R.Soc. London, Ser. B 2011, 278, 3209−3218.(22) Benzerara, K.; Yoon, T. H.; Tyliszczak, T.; Constantz, B.;Spormann, A. M.; Brown, G. E. Geobiology 2004, 2, 249−259.(23) Cosmidis, J.; Benzerara, K.; Gheerbrant, E.; Esteve, I.; Bouya, B.;Amaghzaz, M. Geobiology 2013, 11, 139−153.(24) Schweitzer, M. H.; Zheng, W.; Cleland, T. P.; Goodwin, M. B.;Boatman, E.; Theil, E.; Marcus, M. A.; Fakra, S. C. Proc. R. Soc. London,Ser. B 2014, 281, 20132741.(25) Edwards, N. P.; Manning, P. L.; Bergmann, U.; Larson, P. L.;van Dongen, B. E.; Sellers, W. I.; Webb, S. M.; Sokaras, D.; Alonso-Mori, R.; Ignatyev, K.; Barden, H. E.; van Veelen, A.; Anne, J.; Egerton,V. M.; Wogelius, R. a. Metallomics 2014, 6, 774−782.(26) Ryan, C. G.; Siddons, D. P.; Kirkham, R.; Li, Z. Y.; de Jonge, M.D.; Paterson, D. J.; Kuczewski, A.; Howard, D. J.; Dunn, P. A.;Falkenberg, G.; Boesenberg, U.; de Geronimo, G.; Fisher, L. A.;Halfpenny, A.; Lintern, M. J.; Lombi, E.; Dyl, K. A.; Jensen, M.;Moorhead, G. F.; Cleverley, J. S.; Hough, R. M.; Godel, B.; Barnes, S.J.; James, S. A.; Spiers, K. M.; Alfeld, M.; Wellenreuther, G.;Vukmanovic, Z.; Borg, S. J. Phys. Conf. Ser. 2014, 499, 012002−012012.(27) Wogelius, R. A.; Manning, P. L.; Barden, H. E.; Edwards, N. P.;Webb, S. M.; Sellers, W. I.; Taylor, K. G.; Larson, P. L.; Dodson, P.;You, H.; Da-qing, L.; Bergmann, U. Science 2011, 333, 1622−1626.(28) Da Silva, J. H.; de Sousa Filho, F. E.; Saraiva, A. A. F.; Andrade,N. A.; Viana, B. C.; Sayao, J. M.; Abagaro, B. T. D. O.; Freire, P. D. T.C.; Saraiva, G. D. J. Spectrosc. 2013, 2013, 1−7.(29) Antonietto, L. S.; Gobbo, S. R.; Do Carmo, D. A.; Assine, M. L.;Fernandes, M. A. M. C. C.; Lima e Silva, J. E. L. E. J. Paleontol. 2012,86, 659−668.

(30) Martill, D. M. Cretaceous Res. 2007, 28, 895−920.(31) Freire, P. T. C.; Silva, J. H.; Sousa-Filho, F. E.; Abagaro, B. T.O.; Viana, B. C.; Saraiva, G. D.; Batista, T. A.; Barros, O. A.; Saraiva, A.A. F. J. Raman Spectrosc. 2014, 45, 1225−1229.(32) Pinheiro, A. P.; Saraiva, A. A. F.; Santana, W. An. Acad. Bras.Cienc. 2014, 86, 663−670.(33) Fara, E.; Saraiva, A. A. F.; de Almeida Campos, D.; Moreira, J. K.R.; de Carvalho Siebra, D.; Kellner, A. W. A. Palaeogeogr.,Palaeoclimatol., Palaeoecol. 2005, 218, 145−160.(34) Menon, F.; Makarkin, V. N. Palaeontology 2008, 51, 149−162.(35) Viola, P.; Jones, M. Proc. 2001 IEEE Comput. Soc. Conf. Comput.Vis. Pattern Recognition. CVPR 2001 2001, 1, I511−I518.(36) Freire, P. T. C.; Abagaro, B. T. O.; Souza Filho, F. E.; Silva, J.H.; Saraiva, A. A. F.; Brito, D. D. S.; Viana, B. C. In Pyrite: Synthesis,Characterization and Uses; Whitley, N., Vinsen, P. T., Eds.; NovaScience Publishers: Hauppauge, NY, 2013; pp 123−140.(37) Snioszek, M.; Telesinski, A.; Musik, D.; Zakrzewska, H. J. Elem.2008, 13, 675−684.(38) Grynpas, M. D. J. Bone Miner. Res. 1990, 5, S169−S175.(39) Szczepanik, P.; Sawlowicz, Z. Naturwissenschaften 2008, 95,869−876.(40) Zbinden, M.; Martinez, I.; Guyot, F.; Cambon-Bonavita, M.-A.;Gaill, F. Eur. J. Mineral. 2001, 13, 653−658.(41) Halliday, I.; Blackwell, A. T.; Griffin, A. A. Science 1984, 223,1405−1407.(42) Grassle, J. F. Science 1985, 229, 713−717.(43) Jannasch, H. W.; Mottl, M. J. Science 1985, 229, 717−725.(44) Corliss, J. B.; Dymond, J.; Gordon, L. I.; Edmond, J. M.; vonHerzen, R. P.; Ballard, R. D.; Green, K.; Williams, D.; Bainbridge, A.;Crane, K.; van Andel, T. H. Science 1979, 203, 1073−1083.(45) Spiess, F. N.; Macdonald, K. C.; Atwater, T.; Ballard, R.;Carranza, A.; Cordoba, D.; Cox, C.; Garcia, V. M. D.; Francheteau, J.;Guerrero, J.; Hawkins, J.; Haymon, R.; Hessler, R.; Juteau, T.; Kastner,M.; Larson, R.; Luyendyk, B.; Macdougall, J. D.; Miller, S.; Normark,W.; Orcutt, J.; Rangin, C. Science 1980, 207, 1421−1433.(46) Edmond, J. M.; Von Damm, K. L.; Mcduff, R. E.; Measures, C. I.Nature 1982, 297, 187−191.(47) Ahsanullah, M.; Negilski, D. S.; Mobley, M. C. Mar. Biol. 1981,64, 299−304.(48) Nunez-Nogueira, G.; Rainbow, P. S.; Smith, B. D. J. Exp. Mar.Biol. Ecol. 2006, 332, 75−83.(49) Romeo, M.; Gharbi-Bouraoui, S.; Gnassia-Barelli, M.; Dellali,M.; Aïssa, P. Sci. Total Environ. 2006, 359, 135−144.(50) Kulikova, N. N.; Maksimova, N. V.; Suturin, A. N.; Paradina, L.F.; Sitnikova, T. Y.; Timoshkin, O. A.; Saibatalova, E. V.; Khanaev, I. V.Geochem. Int. 2007, 45, 478−489.(51) Gundacker, C. Environ. Pollut. 2000, 110, 61−71.(52) White, S. L.; Rainbow, P. S. Mar. Ecol.: Prog. Ser. 1982, 8, 95−101.(53) Labrenz, M.; Druschel, G. K.; Thomsen-Ebert, T.; Gilbert, B.;Welch, S. A.; Kemner, K. M.; Logan, G. A.; Summons, R. E.; De Stasio,G.; Bond, P. L.; Lai, B.; Kelly, S. D.; Banfield, J. F. Science 2000, 290,1744−1747.(54) Mandelbrot, B. B. The Fractal Geometry of Nature, 3rd ed.;Freeman: New York, 1983.(55) Mandelbrot, B. Science 1967, 156, 636−638.(56) Martin-Garin, B.; Lathuiliere, B.; Verrecchia, E. P.; Geister, J.Coral Reefs 2007, 26, 541−550.(57) Lehane, J. R.; Ekdale, A. A. Palaios 2013, 28, 23−32.(58) Lehane, J. R.; Ekdale, A. A. J. Paleontol. 2014, 88, 747−759.(59) Cuthill, J. F. H.; Morris, S. C. Proc. Natl. Acad. Sci. U. S. A. 2014,111, 13122−13126.(60) Mandelbrot, B. B. Proc. Natl. Acad. Sci. U. S. A. 1975, 72, 3825−3828.

Analytical Chemistry Article

DOI: 10.1021/acs.analchem.5b02815Anal. Chem. XXXX, XXX, XXX−XXX

H