Post-Paleozoic evolution of the northern Ardenne Massif constrained by apatite fission-track thermochronology and geological data

– The exhumation history of basement areas is poorly constrained because of large gaps in the sedimentary record. Indirect methods including low temperature thermochronology may be used to estimate exhumation but these require an inverse modeling procedure to interpret the data. Solutions from such modeling are not always satisfactory as they may be too broad or may con ﬂ ict with independent geological data. This study shows that the input of geological constraints is necessary to obtain a valuable and re ﬁ ned exhumation history and to identify the presence of a former sedimentary cover presently completely eroded. Apatite ﬁ ssion-track (AFT) data have been acquired on the northern part of the Ardenne Massif close to the Variscan front and in the southern Brabant, in particular for the Visean ash-beds. Apatite ﬁ ssion-track ages for surface samples range between 140 ± 13 and 261 ± 33 Ma and con ﬁ ned tracks lengths are ranging between 12.6 ± 0.2 and 13.8 ± 0.2 m m. Thermal inversion has been realized assuming that (1) samples were close to the surface (20 – 40 ° C) during Triassic times, this is supported by remnants of detrital Upper Permian – Triassic sediments preserved in the south of the Ardenne and in the east (border of the Roer Graben and Malmédy Graben), and (2) terrestrial conditions prevailed during the Early Cretaceous for the Ardenne Massif, as indicated by radiometric ages on paleoweathering products. Inversion of the AFT data characterizes higher temperatures than surface temperatures during most of the Jurassic. Temperature range is wide but is compatible with the deposition on the northern Ardenne of a signi ﬁ cant sedimentary cover, which has been later eroded during the Late Jurassic and/or the Early Cretaceous. Despite the presence of small outliers of Late Cretaceous (Hautes Fagnes area), no evidence is recorded by the ﬁ ssion-track data for the deposition of a signi ﬁ cant chalk cover as highlighted in different parts of western Europe. These results question the existence of the London-Brabant Massif as a permanent positive structure during the Mesozoic.


Introduction
The post-chain evolution of intracontinental basement areas is difficult to reconstruct because these domains are often devoid of younger sedimentary rocks. These basement areas have long been considered stable as they are remote from places of intense tectonic activity. However, recent studies have pointed up the occurrence of intraplate lithospheric deformation associated with plate boundary dynamics (review in Cloetingh and Burov, 2011) or mantle dynamics (Braun, 2010). The former sedimentary cover of these domains and their relationship with the neighboring basins are poorly constrained although they closely control paleogeography and lithospheric dynamics. These features are particularly difficult to ascertain as sedimentary infilling and exhumation are of low amplitude and cannot be detected by most standard methods of reconstructing erosion. Apatite fission-track (AFT) thermochronology has been successfully applied to the Mesosoic and Cenozoic evolution of the Caledonian and Variscan basements of Europe (Green, 1986;Larson et al., 1995;Wagner et al., 1997;Barbarand et al., 2001Barbarand et al., , 2013 and these studies have revealed that significant thicknesses of Mesozoic sediments (mainly Cretaceous in age) were deposited and then eroded in these areas. Although this method is appropriate due to its lowtemperature sensitivity domain (60-110°C, Green et al., 1989), data have to be interpreted using an inversion model for thermal histories to match with the data. While such modeling is easy in domains where vertical movements are large, lowamplitude deformation requires the introduction of robust geological constraints to decipher plausible thermal evolution.
A fission-track study on apatite crystals has been undertaken in the northern Ardenne and Brabant Massifs to test the effect of robust geological constraints on the interpretation of apatite fission-track data. The Ardenne Massif has been located in an intracontinental setting since the end of the Variscan orogeny, although it may have experienced regional deformations (opening of the North Atlantic Ocean, uplift and opening of the North Sea rift, collision between the European and African plates, opening of the Cenozoic grabens; see Ziegler (1990), Faerseth (1996), Rosenbaum et al. (2002), Dèzes et al. (2004)). The post-Variscan evolution of the Ardenne Massif is still poorly understood given the potentially large impact of geodynamic events affecting the northwestern European margin. So far the vertical evolution of the Ardenne Massif during the Mesozoic remains an open question. The first AFT study was carried out in the northern part of the Ardenne and Brabant Massifs ( Fig. 1) by van den Haute and Vercoutere (1989). Subsequent AFT analyses were conducted on the northern edge of the Ardenne and in the western part of the Dinant Paraautochthonous zone but without thermal inverse modeling (Vercoutere and Van den Haute, 1993). Glasmacher et al. (1998) also made an AFT study of the Stavelot Massif ( Fig. 1) in the eastern part of the Ardenne and in the close Rhenish Massif. The Rhenish Massif and the Ruhr Carboniferous Basin have been studied by Karg et al. (2005). The latest fission-track study (Xu et al., 2009) covers a NE-SW axis between the Variscan front and the Givonne Massif. These studies conclude there has been either continuous slow cooling from the Late Carboniferous to the Tertiary (Xu et al., 2009) or episodic faster cooling (Jurassic for Vercoutere and Van den Haute, 1993 or mid-Cretaceous for Glasmacher et al., 1998) with significant erosion of Paleozoic rocks. Analysis of the sedimentary infilling of the adjacent basins (Paris Basin to the SW and West Netherlands Basin to the NE) reveals also the presence of regional unconformities since the Permian and especially during the Middle Jurassic, Early Cretaceous and the Cenozoic (e.g. Ziegler, 1990;Guillocheau et al., 2000). The meaning of these unconformities and their link with potential erosion phases observed in nearby basements need to be addressed.
The objective of this work is to characterize the vertical evolution of the northern Ardenne and the Brabant massifs considering new fission-track data, using robust geological constraints, in order to investigate relationships between these domains and the nearby basins.
2 Geological context 2.1 Variscan framework The pre-Permian basement of the Ardenne contains Paleozoic sedimentary rocks which experienced the Caledonian (only for Cambrian to Silurian) and Variscan orogenies (e.g. Fielitz and Mansy, 1999). This basement belongs to the main Rhenish Massif, which is well exposed in Germany. These massifs were structured from North to South into three units during the Variscan orogeny (mainly Late Carboniferous): the Brabant Para-autochthon, the tectonic wedges of the Paraautochthonous (also called "the Haine-Sambre-Meuse Overturned Thrust Sheets" or HSM-OTS), and the Ardenne Allochthon (Lacquement et al., 1999;Mansy et al., 2003;Belanger et al., 2012). The Ardenne Allochthon overthrusts the tectonic wedges of the Para-autochthonous through the complex detachment of the "Midi" Fault and its eastern equivalent, the Eifelian Fault.
The tectonic wedges of the Para-autochthonous are Devonian to Carboniferous units organized into a succession of synclines and anticlines, which have been partly affected by weaker Variscan metamorphism than the Ardenne Allochthon (Larangé, 2002).
The Brabant massif is formed by Cambrian to Silurian siliciclastic sedimentary rocks that experienced greenschist facies metamorphism and were deformed by the Acadian orogeny during the Early Devonian . It is part of a larger domain, the London-Brabant Massif. These rocks are largely hidden to the south by the mildly deformed cover of Devonian-Carboniferous sedimentary rocks of the tectonic wedges of the Para-autochthonous, and by the Mesozoic-Cenozoic cover. The Lower Paleozoic basement crops out only in river valleys.

Mesozoic-Cenozoic geological history
Geological history since the Permian is partly recorded by the sedimentary infillings of the two sedimentary basins bordering the Ardenne Massif: the Paris Basin to the southwest and the West Netherlands Basin and the Roer Valley Graben to the northeast (Fig. 1). These basins result from the breakup of Pangea during Permian times. Their sedimentary infilling is composed mainly of calcareous or clayey units characteristic of shallow marine environments with the exception of Lower Jurassic marls and the Upper Cretaceous chalk, which were deposited in deeper environments (Ziegler, 1990;Guillocheau et al., 2000;Dusar et al., 2001). Sandstones are also observed during Triassic and Early Cretaceous. In the Paris Basin, the series is up to 3 km thick and includes the Germanic Triassic trilogy (Bundsandstein sandstones, Muschelkalk limestones and Keuper evaporites), the Jurassic calcareous units, the sandy, clayey and calcareous units of the Early Cretaceous, the chalk of the Late Cretaceous, and the Cenozoic with its changing facies and depositional environments. The main subsidence occurred during the Jurassic for both basins with a secondary peak during the Cenozoic for the West Netherlands Basin (Brunet and Le Pichon, 1982;Duin et al., 2006). The rocks found today at the surface of the Ardenne and Brabant Massifs do not record the same sedimentary history. The Brabant Para-autochthonous is overlain by a relatively thick Upper Cretaceous cover and locally by Paleogene and Neogene deposits (Legrand, 1968), whereas the Ardenne Massif has only thin isolated relicts of Mesozoic-Cenozoic strata. Nevertheless, the earliest literature (19th-20th century) reports rather numerous occurrences of residual Cretaceous, Paleogene and Neogene remnants often preserved in paleokarst. Triassic and Jurassic series are absent from the Ardenne Massif except for its southern border (Gaume or Belgian "Lorraine") where these rocks form an extension of the Paris Basin described as the "Luxembourg Gulf" (Boulvain et al., 2001;Schintgen and Förster, 2013) and of the Stavelot Massif (Malmédy Graben; Bultynck et al., 2001). In the N-S Eifel Depression, close to the eastern border of the Ardenne, Triassic infill is up to 650 m thick in its southern part (Mondorf; Schintgen and Förster, 2013) and ∼400 m in the northern part (Mechernich area; Knapp, 1980). This subsidence area extended westward at the expense of the emerged Ardenne area over time (Lucius, 1948;Mader, 1985). The unconformity between Paleozoic rocks and Triassic sandstones is tilted southwards in Belgian "Lorraine" and has been described as a pre-Triassic peneplain. In the North of the Ardenne Massif, 450 m-thick Lower Jurassic marls deposited in open marine environments occur as subcrops in the Roer Valley Graben, Campine (Dusar et al., 2001). Demyttenaere and Laga (1988)  Upper Cretaceous facies and thickness differ substantially in Belgium from region to region but characterize a general progressive onlapping from the West Netherlands Basin towards the south (Felder, 1994;Dusar and Lagrou, 2007;Demoulin et al., 2010). Upper Cretaceous deposits are observed on the Ardenne, in the Hautes-Fagnes area (Stavelot Massif), and contain Late Campanian to Maastrichtian glauconitic sands, flints and silicified chalk, related to the decalcification of chalk (Bless and Felder, 1989;Bless et al., 1991;Felder, 1994;Robaszynski, 2006).
Periodically, during Cenozoic times, several transgressions reached the Brabant and some areas of the Ardenne Massif (Demoulin, 1995;Vandenberghe et al., 2004). On the western edge of the Ardenne, in Avesnois, Thanetian glauconiferous flint conglomerates indicate both flint-bearing chalk erosion and marine transgression on the Paleozoic basement . "Sparnacian" fluvial flint gravels, sands and organic clay are evidence of terrestrial depositional environments, before another marine sequence with siliciclastic units during the Early Ypresian ("Silts et Sablons de l'Avesnois", "Sables de Trélon", etc., . Bruxellian (Lower Lutetian) calcareous sands and nummulitic limestones covered the "Entre Sambre et Meuse" (ESEM) (Briart, 1888;Gulink, 1963). Fragments of this limestone are sometimes reworked in the Miocene sands concealed in the giant paleo-karsts scattered on the ESEM plateau. Uppermost Eocene marine sands were trapped and preserved by karst development during the Neogene and currently lie at the base of their infilling (Ertus, 1990;Dupuis and Ertus, 1994;Dupuis et al., 2003). The sea overflowed the Ardenne for the last time during the Oligocene (e.g., Demoulin, 1995;Boulvain and Vandenberghe, 2018;Demoulin et al., 2018) covering its eastern part, between the ESEM and the Hautes Fagnes area (Fig. 1) as a result of tilting of the basement towards the Rhine graben area.
The geological history of the area can be also reconstructed from the study and the dating of weathering profiles capping the Ardenne Massif (Fig. 2). The central Ardenne locally displays large thicknesses of weathered rocks (Dupuis, 1992;Dupuis et al., 1996;Yans, 2003a, b;Thiry et al., 2006;Demoulin et al., 2018). The residual kaolin deposits of Transinne have been dated by several techniques (K/Ar on hollandite, 40 Ar/ 39 Ar on cryptomelane, paleomagnetism) and three main episodes of weathering can be recognized (Yans, 2003a, b;Thiry et al., 2006): (1) Early Cretaceous (∼130 Ma) with paleomagnetic and K/Ar ages on hollandite ranging from 126 ± 10 to 135 ± 10 Ma; (2) early LateCretaceous with K/Ar ages on hollandite ranging from 89 ± 3 to 95 ± 6 Ma; and (3) Early Miocene with 40 Ar/ 39 Ar ages on cryptomelane of 21 ± 0.4 Ma. Recent Ar/Ar dating on cryptomelane has also been performed for the Bihain quarry (Stavelot Massif) confirming an Early Miocene age (20 ± 0.2 Ma-22 ± 0.6 Ma; Demoulin et al., 2018). Potential other phases of weathering are locally recognized in Belgium by Demoulin et al. (2018): 1) at the Late Permian -Early Triassic interval in the Malmedy area, in the Belgian "Lorraine" and in the Ardenne, but these phases remain debatable (see discussion below), and 2) at the Paleocene-Eocene transition in the ESEM area and in the northern part of the Mons Basin (Barbier, 2012).

Sampling and analytical method
Samples have been collected during several field surveys to cover the northern Ardenne and the Brabant massifs. The Caledonian basement, Devonian and Carboniferous ash-beds and various Paleozoic and Mesozoic sandy facies were sampled to recover apatite crystals. Of these samples, 19 yield apatite crystals in suitable amounts for apatite fission-track dating (Tab. 1). All the rocks analyzed were outcrop samples except BEL41 which was taken from a drillcore in the Brabant Massif.
AFT analysis was carried out in the GEOPS laboratory (Université Paris-Sud). Apatite concentrates were isolated using the conventional techniques of grinding, density and magnetic separation. Spontaneous tracks were revealed by a solution of nitric acid (HNO 3 5M) during 20 ± 1 seconds for Page 4 of 16 a temperature of 20 ± 1°C. The external detector technique was used in this study (Gleadow, 1981). Samples were covered by muscovite sheets as external detectors and three glass dosimeters CN-5 as well as two apatite standards (Durango and Fish Canyon Tuff; Hurford, 1990) were irradiated by thermal neutrons. Irradiation was made in the channel P1 of the ORPHEE reactor (Pierre Süe laboratory, CEA, Saclay, France) with a requested fluence of 5.10 15 neutrons/cm 2 . Induced tracks were etched by a 40% fluorhydric acid solution at 20°C for 20 minutes. Track counting and measurement were made using a Leica optical microscope at a magnification of Â1000. Tracks were measured in accordance with the recommendations of Laslett et al. (1984) using a digitized tablet associated with a computer. AFT ages are central ages at ± 1s (Galbraith and Laslett, 1993). AFT ages were calculated by the zeta calibration approach (Hurford and Green, 1983) where the zeta value is determined by multiple analyses of apatite standards (Durango, Fish Canyon Tuff) following the recommendations of Hurford (1990). Eleven standard samples resulting from six different irradiations with their associated dosimeters were measured to determine a zeta value (IB) of 325 ± 4. Dpar measurements were used to characterize the kinetic properties of individual apatite crystals (Burtner et al., 1994;Barbarand et al., 2003b). Thermal modeling was realized using the track-annealing model developed by Ketcham et al. (2007) and the AFTSolve software (Ketcham et al., 2000;Ketcham, 2005) using 50 000 iterations.
Apatite composition was determined for five samples covering a large range of Dpar values using a Cameca SX100 electron probe with a wavelength dispersive system and a counting time of 30 seconds.

Apatite fission-track data
Fission-track ages for outcrop samples range from 140 ± 13 to 261 ± 33 Ma (Fig. 3); sample BEL41 recovered in drillhole at a depth of 745 m is slightly younger (114 ± 8 Ma). All samples pass the x 2 test with values higher than the 5% level and thus characterize a single-age population. Relatively large age variation exists and is partly correlated to the sample position: ages increase southwards and younger ages are measured mainly in the Brabant Massif (Fig. 4).
Because of the low U content of apatite crystals (below 20 ppm for most of the grains), the probability of finding confined tracks is low and a sufficient number of tracks has been found for only a few samples. Considering samples with more than 20 confined tracks measured, mean confined track length is homogeneous and ranges from 12.6 ± 0.2 (± 1 standard error of the mean) to 13.7 ± 0.2 mm with a standard deviation of 0.9 to 1.4 mm (Figs. 3 and 4). With respect to the limited amount of measured confined track length, track length distributions appear unimodal and negatively skewed.
Dpar values are homogeneous within each sample but vary from 1 to 2.2 mm between samples. This wide distribution reproduces the large variety of samples analyzed: apatites from crystalline rocks show Dpar in the 1-1.2 mm range and are Fapatites whereas apatite crystals from ash-levels have higher Cl content and higher Dpar (1.3 to 2.2 mm).

Thermal history modeling
The thermal history was modeled for all the samples where a minimum of 20 tracks has been measured. This relatively low   r: track density (Â105 tracks/cm 2 ); N: number of tracks counted; P(x2): probability of obtaining Chi-square value (x2) for n degrees of freedom (where n = number of crystals -1); Age ± 1s -central age ± 1 standard error (Galbraith and Laslett, 1993); MTL: mean track length (mm); SD: standard deviation of track length distribution (mm); N(L): number of horizontal confined tracks measured; Dpar: average etch pit diameter parallel to c. Ages were calculated using the zeta calibration method (Hurford and Green, 1983), glass dosimeters CN-5, and a zeta value of 325 ± 4 (IB).
Page 6 of 16 number of counted tracks might detract from our conclusions, but our track length data are coherent with data already published. The modeling strategy was to input constraints progressively to get a good agreement between data and model (Fig. 5). These constraints correspond to stratigraphic, geodynamic or weathering evidences of age: the Visean age (345 Ma) of volcanic ash-beds, when these samples are analyzed, and their presence at the surface at the time of their deposition (Delcambre, 1996); the Ardenne Paleozoic rocks were close to the surface (20-40°C) during the Late Permian-Triassic after the post-Carboniferous inversion. Late Permian-Triassic sedimentary rocks still crop out in Belgian "Lorraine" and in the Malmédy Graben (Fig. 1). This hypothesis is debatable, though, and will be discussed later in the text; the presence of the Ardenne basement at surface during the Early Cretaceous (140-120 Ma). This assumption is supported by dating of weathering profiles.

Present-day average temperature of 10°C for the outcropping samples
Robust modelings are presented in Figure 6 for samples with the measurement of more than 85 tracks. Moreover, assuming that the track length distribution is unimodal, Rahn and Seward (2000) and Barbarand et al. (2003a) have shown that track length data do not evolve from 50 to 100 tracks. We therefore consider that modelings presented in Figure 6 may be used with confidence.
Modeling results using these constraints are good (pink envelope) or acceptable (green envelope) with a value of the Kolmogorov-Smirnov test (KS) > 0.80 or 0.5 respectively (Press et al., 1988;Ketcham, 2005). The history can be summarized in two cycles of temperature rise and fall: (1) Late Carboniferous/Permian and beginning of the Triassic, and (2) Early and Middle Jurassic/Late Jurassic and Early Cretaceous (Fig. 5).

Discussion
Thermal history modeling suggests for the Late Paleozoic and the Mesozoic two episodes of cooling: Permian -Early Triassic and Late Jurassic -Early Cretaceous and two events of heating (Carboniferous and Early/ Middle Jurassic). These conclusions are supported by apatite fission-track data but also by independent geological data.
A default history can be reconstructed for the Late Paleozoic as temperatures higher than those investigated by the FT system in apatite crystals existed. The temperature increase at the end of the Paleozoic is well-established and corresponds to thrusting and thickening occurring during the Variscan orogeny leading to Carboniferous sedimentation in the coal foreland basin (Paproth et al., 1983). Synorogenic burial metamorphism was dated between 336 and 298 Ma by the K/ Ar method for the Devonian and Carboniferous units of the Ardenne Allochton (Piqué et al., 1984). Paleotemperatures determined by illite crystallinity (Larangé, 2002), conodont coloration index (Helsen, 1995) and zircon fission tracks (Brix, 2002) suggest much higher temperatures than the track stability in apatite. These data suggest then that apatite fissiontrack data have been totally reset at that time. This flysch cover crops out today in the Para-autochthonous tectonic wedges (it can be up to 900 m thick in the centre of HSM-OTS, Fielitz and Mansy, 1999) but it probably previously covered a part of the Brabant. Paleotemperature estimates of this event using our data represent probably a minimum.

Mesozoic thermal history
Significant cooling during Permian-Triassic times is recorded by the samples which display a FT age higher than 200 Ma (samples BEL37, BEL36, BEL35, BEL34, BEL28, BEL27). For younger samples, modeling of this cooling is not robust as the pre-Triassic history cannot be reconstructed as fission tracks have been further severely annealed. This Permian-Triassic cooling is also constrained by the hypothesis that samples were close to the surface (20-40°C) during Triassic times. These temperatures correspond to the thermal domain in which fission-track are not affected by temperatures. Assuming a paleogradient of 20-40°C/km and a surface temperature of ∼20°C, samples present today at surface might have been buried at ∼500-1000 meter depths. The following elements are proposed to justify this hypothesis. Morphology of the Ardenne and Brabant Massif is difficult to establish at the end of the Paleozoic but it might correspond to a peneplain with a minor relief where locally Triassic sedimentation occurred. Triassic deposits are lacking on top of the Brabant and in the extreme north of France. On the western side of the Ardenne, south of the Rocroi High, Triassic deposits correspond to playa deposits associated with low relief (Debrabant et al., 1992). Triassic deposits are preserved locally in the north east of the Paris Basin, in the Luxembourg gulf, in the Eifel Depression, in eastern Ardenne and in the West Netherlands Basin (see Mesozoic-Cenozoic geological history). During early Triassic, paleogeography of western Europe at this period envisages the Paris Basin as the western end of the German Basin (Péron et al., 2005). Paleocurrent directions obtained from fluvial facies indicate a mainly eastward flow, the source being the Armorican Massif (Durand, 1978). Facies maps built for the Bundsandstein show that the more proximal facies represented by conglomerates are deposited on the western edge and evolve laterally eastward to sandstone deposits (Bourquin et al., 2006). Ardenne Massif did not represent at that time the main source of detrital materials deposited in the Paris and German Basins. As soon as Muschelkalk, marine influences are recorded with very low input of detrital materials. The former Triassic surface cannot be deduced directly from these Triassic outcrops but onlapping of the Lower Jurassic deposits in the Luxembourg gulf characterizes a very flat morphology of the Ardenne Massif (Schintgen and Förster, 2013), which is not compatible with a contrasted evolution of the borders and the centre of the massif during Triassic times. We considered then that these observations attest for minor erosion (< 1000 m) of the Paleozoic cover after Triassic times.
This Permian-Triassic cooling is followed by a Lower and Middle Jurassic temperature increase suggested by the modeling. Maximum temperature is varying between samples but is generally low (∼50-80°C). For sample BEL 39, the best model of thermal history corresponds to higher temperature than the other samples during the Jurassic. The same observation was done by Glasmacher et al. (1998) but for samples close to ore deposits. This younger sample is observed close to the "Midi" Fault where fluid circulation at a temperature of 110°C, forming Mississippi Valley Type Pb-Zn deposits, is observed (Verviers area; Heijlen et al., 2001). Similar deposits in the nearby Eifel have been dated at 170 ± 4 Ma by Rb-Sr on sphalerites (Schneider et al., 1999). Thus, the modeling of sample BEL 39 involves a high temperature during the Early Jurassic, which is compatible with such local hot fluid circulation. In general, inverse modeling requires a maximum temperature increase of 50-80°C depending on the samples. Although this temperature corresponds to the lower limit of the apatite fission-track annealing domain (60-110°C), we considered it significant as it is observed for all the samples. The present-day geothermal gradient is averaged in the Netherlands and French sedimentary basins (mostly Paris Basin) at 29-30°C/km (Ramaekers, 1991;Bonté et al., 2010) heat flux, thermal conductivity of vanished rocks and surface temperature. Other mechanisms than erosion might be also considered but cannot be applied to all the samples. Large heat flux increase is difficult to envisage as no remnant of high volcanic activity is observed. As discussed earlier, hot fluids may have affected the fission-track system but only locally, in particular for samples close to the Variscan front.
Our new thermal history is slightly different from the models proposed by previous thermochronological data on the Ardenne Massif (Fig. 7). Based on Late Palaeozoic samples covering the Ardenne Massif, Xu et al. (2009) suggested a Permian-Triassic cooling episode but followed by an unique slow cooling up to the Eocene. Glasmacher et al. (1998) studying Devonian sandstone samples from the eastern Ardenne Massif (Linksrheinisches Schiefergebirge) pointed out that the area had recorded high temperatures (> 110°C) during the Carboniferous followed by a Mesozoic cooling with two stages: slow cooling up to 120 Ma and faster cooling between 120 and 80 Ma. The study of a Permian sample from the Malmedy Graben suggested an alternative local history with rapid heating during the Permian and Triassic followed by a similar cooling history during the rest of the Mesozoic First step considers only that sample is at surface during the Visean and today. Second step introduced that sample is close to the surface (20-40°C) during the Triassic. Last step considers that sample is close to the surface also during the Early Cretaceous. Numbers correspond to the measured (in italic) and modeled FT age and mean track length. GOF (goodness of fit) is an indicator of the match between the measured data and the data in agreement with the modeled thermal history (for perfect match, GOF = 1). Histograms correspond to the measured mean confined track length distributions; green lines attached to these histograms correspond to modeled distributions.
Page 9 of 16 (Glasmacher et al., 1998). Van den Haute and Vercoutere (1989) concluded that the southern border of the Brabant Massif records a cooling phase during Jurassic times from temperatures higher than 100°C to ambient temperatures. Vercoutere and Van den Haute (1993) extended the studied area and identified that this event was also recorded for the Hercynian basement to the south. Bour (2010) proposed a preliminary fission-track study based on the same samples than those used in this study. This author suggested a Jurassic to Upper Cretaceous cover on the Ardenne.

Mesozoic geological history: Jurassic deposition
The presence of a Lower and Middle Jurassic cover on the Brabant and the northern Ardenne massifs is coherent with independent sedimentological arguments. Marine Liassic rocks close to the Ardenne Massif in the Paris Basin correspond to a significant bathymetry, without any particular detrital contribution nor any littoral evidence (Garcia et al., 1996;Thiry-Bastien et al., 2000). Clastics influx is only observed during the Hettangian and Sinemurian with the deposition of the Luxembourg Sandstone Formation in a tidal influenced deltaic environment (Van den Bril and Swennen, 2009). The Dogger platform (Bajocian-Bathonian) located directly close to the Ardenne Massif (from Hirson to Sedan) is characterized by oolithic, shell-rich and sub-chalky limestones which are related to environments representing inner and outer carbonate platform (Fischer, 1979). This platform follows the Liassic series after the break of the Aalenian unconformity. The thickness of this platform is decreasing towards the NW in the Boulonnais area where it is deposited directly on the Palaeozoic basement (Vidier et al., 1995). In the Boulonnais area, the absence of Liassic strata may be due either to nondeposition or to erosion before the Middle Jurassic (Mansy et al., 2003). The borehole A901 located close to Charleville-Mézières goes through a thick series of Jurassic strata with a thickness of 870 m (∼600 m of Lias and Dogger) (Debrabant et al., 1992). A geological cross-section across the northwestern part of the Paris Basin supports these observations and shows that the thicknesses of the Jurassic series, in particular   (Ketcham, 2005) carried out on the samples with more than 85 confined tracks counted. Modelling results are displayed in a time-temperature diagram (left) and frequency distribution (right) of measured confined track length. Results in the t-T diagram are indicated by two different reliability levels: green envelopes indicate acceptable fit; pink envelopes indicate good fit; APAZapatite partial annealing zone; GOF is an indicator of the goodness of fit: statistical comparison of the measured input data (age and fission-track length) and modelled output data, where a "good" result corresponds to a value of 0.5 or higher, "the best" result corresponds to a value of 1. The squares represent the constraints applied to the modeling and are discussed in the text. Blue lines are weighted mean paths averaged at each time.
Page 10 of 16 for the Liassic and Dogger, do not reduce towards the Ardenne Massif and do not display any pinch-out (Fig. 8). Their limits rather correspond to erosion truncations interrupting the original continuity of the series.
On the other side of the massifs, in the Campine area, thick (> 450 m) Jurassic deposits are also observed in drill-holes (Dusar et al., 2001). In the West Netherlands Basin, a significant Jurassic (Altena Group, Lias and Dogger) exists with a mean thickness of ∼1 km (Duin et al., 2006). In the British Isles, facies succession during the Jurassic shows a similar evolution (Hesselbo, 2008). Especially for the Early Jurassic and early Middle Jurassic, successions show some strong similarities at the scale of ammonite zones between widely separated basins (Hesselbo, 2008). The Jurassic is characterized mostly by fully marine environments, especially in the South, (London Platform and Wessex Basin which are the closest sedimentation areas from the Ardenne and Brabant massifs). Detrital inputs are recorded with sandy facies during several intervals, for example the Late Pliensbachian (Macquaker and Taylor, 1996; sand-rich mudstone with 5% of sand), the Late Toarcian (Morris et al., 2006) and the Oxfordian, but always in marine conditions as illustrated by shoreface sands during the Late Toarcian (Hesselbo, 2008). High thickness may have been deposited in the Weald Basin as well, where more than 1500 m of Jurassic are preserved (Chadwick, 1986).
The situation may have changed during the Late Jurassic and various factors argue for the existence of terrestrial conditions or inputs into the marine realm: (1) the origin of the detrital particles accumulated in the Callovian-Oxfordian clays in the Bure area of the eastern Paris Basin has been determined as NNW-SSE from magnetic susceptibility data (Esteban et al., 2006) and (2) sandy limestones are observed in the Malm of the eastern border of the Paris Basin as well as on the edge of the West Netherlands Basin.
Based on these data, it may be interpreted that the Paris Basin and the West Netherlands Basin were much more widespread than their current geometry suggests. The geological contacts between the Mesozoic sedimentary cover and the Ardenne Massif thus correspond to a limit of erosion.

Stability of the Ardenne and Brabant massifs revisited
The Brabant and Ardenne massifs have been depicted as a positive structure which was clear of sediments during most of the Mesozoic and was considered as stable throughout the Mesozoic (Ziegler, 1990;Dercourt et al., 2000). The Brabant Massif is part of a larger massif, the Anglo-Brabant Massif or the London-Brabant Massif (LBM), which has formed a positive structural element since the mid-Devonian (Rijkers et al., 1993). As a consequence, the Caledonian basement lies at relatively shallow depth across the massif, encompassing a large area of southern Britain, the southern North Sea, southern Netherlands, Belgium and northern France. Based on geophysical studies, the onlapping pattern of sediments of Late Paleozoic, Mesozoic and Cenozoic ages shows that the LBM has been tectonically stable since Devonian times (Rijkers et al., 1993;Rijkers and Duin, 1994).
This study shows however that significant, although relatively minor, vertical movements are recorded for this domain during the Mesozoic and questions its relative stability (Fig. 9). Since the breakup of Pangea, the European continent has undergone several episodes of uplift (Ziegler, 1987). Most of the constraining data are coming from the sedimentary records in inverted basins. Four main episodes are well characterized by erosion surface and/or non-deposition and unconformity. A regional event known as the Middle Jurassic thermal dome or mid-Cimmerian event is recorded at the transition between Toarcian and Aalenian and characterized a large domain centered in the North Sea where erosion occurred (Ziegler, 1990;Underhill and Partington, 1993). Unconformities at the transition between the Jurassic and the Cretaceous and in the Berriasian (Late Cimmerian phases) are evidenced in the Paris Basin (Guillocheau et al., 2000), in the West Netherlands Basin (Nelskamp et al., 2008), in sub-basins from the North Sea (Broad Fourteens, Van Wijhe, 1987) and at larger scale in the Central European Basin System (Mazur and Schek-Wenderoth, 2005). The dominant uplift event (Laramide phase) affecting Europe occurred during the Late Cretaceous and Middle Paleocene when many Paleozoic and Mesozoic rifts and basin structures in the interior of the European continent underwent inversion (Ziegler, 1987). A last main inversion phase generally referred as the Pyrenean phase took place during Bartonian-Priabonian times and is recorded in the southern border of the North Sea area and in Belgium (Deckers et al., 2016;Parrish et al., 2018;Deckers and Matthijs, 2017).
Our FT data combined with geological constraints illustrate than the main phase of inversion recorded in the Brabant and the Ardenne massifs corresponds to the Late-Cimmerian phase. This conclusion agrees with the previous studies of Vercoutere and van den Haute (1993). This Upper Jurassic uplift of the London-Brabant Massif has already been suggested by Ziegler (1990). On the contrary, the mid-Cimmerian event is recorded in the Paris Basin close to the studied area (Debrabant et al., 1992;Mansy et al., 2003) and at distance (Guillocheau et al., 2000) but is not recorded by the thermal data. This is probably due to the amount of erosion associated, which is far below the resolution of FT method. Inversion at the limit between the Jurassic and the Cretaceous is the main event detected by our data. The Laramide and Pyrenean phases are also not recorded by our data, indicating that for the Brabant and the Ardenne massifs, vertical movements are minor and not detected. After this inversion, during the Early Cretaceous, terrestrial conditions prevail in the Ardenne Massif as indicated by deep weathering (Dupuis, 1992;Demoulin, 1995;Yans, 2003a, b;Demoulin et al., 2018) but also in the Paris Basin and its other bordering basements (Thiry et al., 2006). The Borne de Fer ferricrete is observed on top of the Marnes de Longwy (Early Bajocian) close to Luxembourg (Quesnel et al., 2003;Théveniaut et al., 2007). It corresponds to an indurated saprolite of highly siliceous origin, which has been dated using paleomagnetism as Early Cretaceous (Théveniaut et al., 2007). Detailed studies of Dogger sedimentary units from the eastern Paris Basin also show that circulation of meteoric fluid is associated with an episode of terrestrial hydrological refill (Vincent et al., 2007;Brigaud et al., 2009). In the West Netherlands Basin, this interval is characterized by a main unconformity at the base of the Valanginian (Rijnland Group) (Duin et al., 2006). At a global scale, Western Europe is characterized during the Early Cretaceous by terrestrial conditions (Thiry et al., 2006).

Conclusion
The apatite fission-track data of this study suggest a new post-Variscan exhumation scenario of the Ardenne and Brabant Massifs assuming geological, sedimentological and paleoweathering constraints. This scenario is strictly controlled by these constraints which are required limits for thermochronological data interpretation.
After burial of the pre-Carboniferous basement under a Carboniferous flysch estimated to be 3 km thick, the basement was eroded following a generalized uplift of the Variscan chain. The Ardenne Massif was then a terrestrial environment with erosion and the probable deposition of Triassic fluvial sediments similar to those observed in the northeastern Paris Basin. After the Hettangian, the area experienced marine   Fig. 9. Schematic evolution of the Ardenne massif considering sedimentation, erosion and weathering periods.
conditions with the deposition of Lower and Middle Jurassic strata. This cover may be estimated between 650 and 2000 m thick depending on the paleothermal hypotheses and sample location. Preserved Jurassic series in the nearby sedimentary basins (Paris Basin, West Netherlands Basin, Wessex Basin) are consistent with the low thickness value of this cover. The area has been then eroded from the Late Jurassic to the Early Cretaceous and terrestrial conditions prevail again in the Ardenne and Brabant massifs with the formation of thick kaolin deposits but also in the Paris Basin with the deposition of the mainly fluvial Wealden facies and the development of laterites and ferricretes. This inversion event has been recorded in the nearby Paris Basin and West Netherlands Basin by a regional unconformity capped by Valanginian deposits. Further evolution of the area cannot be studied using paleotemperature data which remain low (< 50°C) during the Late Cretaceous and the Tertiary. Our results question the relative stability of the Paleozoic massifs and the connections between sedimentary basins. They moreover illustrate that these massifs shared a similar history with nearby sedimentary basins during almost all the Mesozoic.