1 Introduction

Les phases majeures de l'évolution du golfe de Corinthe ont été décrites par Ori [10] et se divisent en un protogolfe qui a pu débuter dès le Miocène et le rift actuel pliocène. De nombreux auteurs s'accordent désormais à penser que le rift actuel s'est développé depuis environ 1 Ma [9,12,14]. Pour cette dernière phase, les difficultés de datation des sédiments syn-rift ont laissé ouvertes certaines discussions sur la cinématique de la zone. Dans ce travail, nous avons employé la méthode de datation U/Th sur les ciments calcitiques des failles et fractures actives du golfe de Corinthe pour valider un modèle cinématique.

2 Échantillonnage, méthode et analyse

Sur l'ensemble des plans de faille affleurant dans le golfe de Corinthe, seul un petit nombre présente des ciments de calcite pure adaptée à la datation Th/U (Fig. 1). En effet, un certain nombre ne contient pas de remplissage de calcite et d'autres, comme la faille d'Helike, ont une gouge centrale principalement formée d'éléments de l'encaissant, dans laquelle aucune veine néoformée ne peut pratiquement être isolée. Nous avons néanmoins trouvé 23 échantillons apparemment susceptibles de mesures d'âges. Les résultats de l'analyse de 15 échantillons sont présentés ici, leur position est reportée sur la Fig. 1, leur description dans le Tableau 1 et les âges obtenus ainsi que les teneurs en radioéléments dans le Tableau 2. Le déséquilibre ²³⁰Th/²³⁴U/²³⁸U permet d'obtenir des âges valides dans un matériau contenant de l'uranium et pas de thorium au moment de sa cristallisation, ce qui est le cas des cristaux, et notamment de la calcite, formés à partir d'une solution, puisque l'uranium seul, de ces deux éléments, est soluble [6]. L'éventuelle présence de ²³²Th dans des calcites contenant des argiles, par exemple, signe la présence initiale de ²³⁰Th et nécessite une correction de l'âge calculé [2]. La mesure par TIMS (Thermo-Ionisation Mass Spectrometry) de ce déséquilibre permet de déterminer des âges jusqu'à 600 ka. Nous avons limité les analyses aux échantillons contenant plus de 50 ppb d'uranium, afin que la taille requise des échantillons reste faible (2 à 4 g) et puisse être obtenue en minimisant les risques de mélange entre calcite néoformée et encaissant. Les données U–Th et les âges obtenus sont présentés sur les Figs. 2 et 3.

Table 1

Faults and samples location and description

Description et localisation des échantillons et des failles

Sample #	Site description	Sample description
1, 2, 3	Road from Akrata to Tsivlos Lake	Well-crystallised pure calcite with large parallel crystals. Smaller crystals close to the conglomerate. No tectonism effect on crystallic fabric, no stylolitic fissure, breaks through some crystals never pass grain limits.
	37°56′–22°19′5″
	Veins in a syn-rift conglomeratic sequence.
	S0 N20–60° W
4	Road from Kato Potamia to Voutsimos	Same crystallisation as above.
	38°07′51″–22°15′27″
	Vein 170–10° E
5	Road from Kato Potamia to Voutsimos	Stylolitic fissures and many fluid inclusions are visible on a white well crystallised calcite.
	38°07′51″–22°15′27″
	Vein 70–80° S
6	Doumena Fault, southwest from Doumena Village	Well-crystallised white calcite showing thin laminar growing, looking like a flowstone passing to a thin and dense crystalline fabric.
	38°06′04″–22°07′42.3″
	Main fault plane contact Cretaceous carbonate/synrift
	N86–56° N Striae 84° E
9	Road from Platanos to Kalamias	Multi-layered cement symmetrically deposited on the fracture walls, poorly crystallised, detrital material abundant.
	Vertical open fracture reaching the present surface
10, 11	Main road one km North Kalabrita	Flowstone (10) grown in a fractured conglomerate and stalagmite (11) grown over it. Pure white calcite without tectonic marks.
12	Road Kalabrita–Diakofto, crossing road to Doumena	Small flowstone covering a fracture wall in a conglomerate. Pure white calcite without tectonic marks.
	Fracture in synrift conglomerate
	N110–60° N
13, 14, 15	Xylokastro Fault	Automorphous calcites on fault plane.
	Ano Loutro N80–67° N
16	Xylokastro Fault	Fibrous calcite typical of flowstones. Microstructure shown one stylolitic figure, normal to the long axis fibres, due to tectonism.
	About 500 m south from Ano Loutro
17	Fracture in the foot wall of Xylokastro Fault at a few metres from the main fault plane; same site as sample 16	Crystalline fabric is strongly affected by tectonism. Grain size is highly variable, stylolitic fissures are abundant and large, normally cutting the length axis of the major number of crystals.
18	Northern shore. Main fault plane N100 70° S	Poorly crystallised carbonaceous cement, multi-layered, containing detrital material.
	38°17′10.8″–22°33′53.9″
19	Same site 50 m southward	Pure white calcite of a single generation fabric, with smaller crystal surrounding breccia fragments.
23	Northern shore, near Itea	Highly variable size grain could reflect a multi-generation fabric of pure white calcite.
	Normal fault 110°55 S

Table 2

U and Th data of calcite cements. All data are measured by TIMS, ages are calculated using Ludwig half-times. The isotopic dilution method was done using a spike calibrated with HU1 standard uraninite. Repeated analyses (N=10) of HU1 yield respective mean values of 1.0026 and 1.0015, with reproducibility (2σ) of 0.0050 and 0.0066 for ²³⁴U/²³⁸U and ²³⁰Th/²³⁴U activity ratios. When secular equilibrium is reached for both ratios (²³⁰Th/²³⁴U and ²³⁴U/²³⁸U), a lower limit of age equal to 1 Ma can be assumed (sample #23). If only ²³⁴U/²³⁸U is out of equilibrium, it means an age lower than the time necessary to reset the secular equilibrium for this ratio. The initial value for this ratio being unknown, an age calculation is impossible, but considering the ²³⁴U/²³⁸U values in this area, we suggest an age lower than 1 Ma (sample #17). Correction for Th-excess, using the hypothesis of an initial ²³⁰Th/²³²Th ratio equal to 1, was only needed for samples #9b and #12a (A₁ are corrected ages)

Données U–Th des ciments calcitiques. Les données sont mesurées par TIMS, les âges sont calculés en utilisant les périodes de Ludwig. La dilution isotopique a été réalisée avec un traceur calibré par rapport à l'uraninite standard HU1 supposée à l'équilibre. La réplication (N=10) des analyses de HU1 a donné des valeurs moyennes respectivement égales à 1,0026 (2σ=0,0050) et 1,0015 (2σ=0,0066) pour ²³⁴U/²³⁸U et ²³⁰Th/²³⁴U. Quand les deux rapports d'activité (²³⁰Th/²³⁴U et ²³⁴U/²³⁸U) sont à l'équilibre, on ne peut indiquer qu'une limite inférieure de l'âge (âge >1 Ma pour l'échantillon #23). Si ²³⁴U/²³⁸U est seul hors équilibre, l'âge est au-delà de la limite (>600 ka) et inférieur à la durée de remise à l'équilibre de ce rapport, soit un âge inférieur à 1 Ma, compte tenu des rapports obtenus dans les calcites du golfe de Corinthe (échantillon #17). Une correction du thorium initial, utilisant l'hypothèse d'un rapport initial ²³⁰Th/²³²Th égal à 1, a été nécessaire pour les échantillons #9b et #12a

Sample #	238U (ppm)	2σ	234U/238U AR	2σ	230Th/234U AR	2σ	230Th/232Th AR	2σ	Age (ka)	2σ	U (ppb) ICPMS
1a	0.0862	0.0001	1.351	0.012	1.061	0.013	237	3	380	+45/–32	55–96
2	0.0691	0.0003	1.445	0.011	1.018	0.011	170	2	289	+15/–13	87
3	0.0846	0.0003	1.436	0.010	1.002	0.026	123	3	274	+31/–25	55
4	0.0430	0.0001	1.318	0.020	1.093	0.019	38	0.3	561	+643/–133	121
6a	0.0613	0.0001	1.537	0.011	0.719	0.013	281	4	124.3	±3	55
6b	0.0703	0.0001	1.512	0.013	0.762	0.067	1326	100	138.3	+23/–19	72
9b	0.6235	0.0012	1.039	0.007	0.436	0.005	31	0.4	62(A0) 50(A1)	±0.7	407
12a	0.0485	0.0002	1.231	0.012	0.335	0.034	3.5	0.7	43(A0) 35(A1)	±5	45
16a	0.887	0.006	2.123	0.004	0.698	0.002	140	1	112.3	±0.4	871
16b	0.549	0.001	2.131	0.009	0.681	0.005	208	2	108.2	±0.9	642
17b	0.0425	0.0002	1.023	0.009	1.001	0.016	57	1	$⩾600\&<1000$	nd	37
17c	0.0880	0.0004	1.021	0.008	1.041	0.031	333	10	$⩾600\&<1000$	nd	47
18	0.2576	0.0002	1.303	0.005	1.124	0.011	58	1	U leaching	nd	230–319
19	0.0497	0.0001	1.068	0.009	0.985	0.016	210	4	360	+67/–41	49
23	0.1018	0.0005	1.001	0.010	1.013	0.031	59	2	>1000	nd	93

3 Discussion et conclusion

The evolution of the Gulf of Corinth is often described as a two-phase process with a slow first phase and a fast second one from Late Pliocene to now [10]. Other authors propose a simple tectonic history of the Corinth Rift as a continuous sequence from 900 ka to present time with a progressive migration of the tectonic activity from south to north [12]. New field observations and dating of calcite cements collected from various faults allow us to present a more complex scenario. Our field study is focussed on the western part of the Corinth Rift from Xylocastro to Aigion (Fig. 1). The normal faults, which have a general east–west trending, are dipping to the north and are present from Kalabrita to the centre of the gulf [9]. The first synrift deposits are now uplifted in the northern Peloponnese, which underwent a strong uplift during the last 1 Myr.

Classically, the fault appearance is dated by the first syn-tectonic deposit, but in the Gulf of Corinth a large part of the syn-rift deposit is poorly dated; therefore, we decided to approach the dating of the fault by measurement of the Th/U disequilibrium of calcite cements collected on plane faults. Thermo-Ionisation Mass Spectrometry (TIMS) measurements were performed on calcite containing at least 50 ppb of uranium in order to limit the sample size (2 to 4 g) to ensure proper selection of pure cement and avoid calcareous fragments of host rocks.

Among plane faults, only few contain pure calcite cements suitable for Th/U dating. Indeed, a large part of the fault zones shows no cement at all, whereas others show calcareous breccias dominantly composed of fragments of host rocks with very thin and discontinuous calcite veins that are practically impossible to isolate, and the last unsuitable category shows a complete mixture of new cement and detrital material coming from host rock.

In order to get a more precise picture of the kinematics of the area rather than the time lapse of a specific fault activity, we sampled as many faults as possible in the western part of the Gulf, from both southern and northern shores. The map in Fig. 1 shows the location of the analysed samples.

Faults and samples description and location are summarised in Table 1.

Samples could be divided into three groups:

• (i) major faults, Doumena and Xylokastro (samples #6, and #13 to #17),
• (ii) minor faults (on the southern shore, samples #4, #5, on the northern shore, samples #18, #19, #23),
• (iii) fractures and minor faults intra synrift deposits (samples, #1 to #3, #9 and #10 to #12).

Microstructures of all thin sections were observed in order to check crystalline fabric of calcite, eventual mixing of different phases, stylolitic figures indicating pressure effect related to tectonism. Except two samples (#9 and #18) chosen in spite of their poor crystalline appearance, but considered as highly significant for tectonic studies, calcite cements were generally made of pure white calcite sometimes automorphous. Only a few of them have clearly shown tectonic influence (#5, #16, #17) marked by stylolitic fabric and/or very variable size grain.

Because Th is practically insoluble under natural conditions and U is highly soluble, a time-dependent disequilibrium may form between these two elements, which makes some samples datable by means of measuring ²³⁰Th/²³⁴U/²³⁸U disequilibria. To obtain a valid date with these data, some prerequisite conditions have to be satisfied:

• (i) at crystallisation time (zero time or initial time) the sample contains only uranium and no thorium,
• (ii) the geochemical system remains closed after crystallisation, i.e., no migration of uranium has occurred after crystallisation (see [6] for a review).

In such a favourable case, there is no ²³²Th, and U content, as well as the ²³⁴U/²³⁸U ratio, are constant in one geological unit. For calcite cements collected from fault planes of the studied area, these two conditions were satisfied when we analysed only true calcareous cement, crystallised from a liquid phase, no matter if they were of surficial or deep origin. While doing it, we avoided mixture with detrital material from host rocks, and selected calcite cements characterised by a high degree of crystallisation resulting in a non-porous calcite, making the secondary percolation by a fluid phase impossible. As fault planes are impermeable, the assumption of closed system conditions with respect to uranium for calcite cements seems reasonable. An additional proof may be found through consistent results obtained for sub-samples cut from the same calcite cement.

The presence of ²³²Th would mean a mixture with detrital material containing all U-series isotopes and particularly initial ²³⁰Th, not produced in situ. For such samples, the age calculation needs a correction for this so-called excess thorium, through the knowledge of initial ²³⁰Th/²³²Th [2]. Corrected so, the ages are assumed valid if the correction effect is limited, for ²³⁰Th/²³²Th ratios greater than 20, or if the correction introduces age differences lower than the error bar, expressed, as usually, by the 2σ counting statistics [6].

From 18 fault planes sampled for Th/U dating (Fig. 1 and Table 1) when their calcite cements look suitable for this dating method, i.e., apparently made of pure white and non-porous calcite, we obtained only 13 measurable ages for various reasons (Table 2).

The major reason is the U content of calcite. The first U–Th measurements on calcite cements from the Corinth Rift faults have shown large variations for U contents: around 1 ppm to a few ppb. These latter samples were unsuitable for dating, because the size sample needed to ensure a precise selection of the pure cement would be too large. Another reason is the existence of excess thorium, only interpretable as a result of U leaching after crystallisation. Additionally, some samples are found to be close to, or have reached, the limit of the method.

Considering that three grams are a minimum weight to obtain ages as young as 40 ka for a sample containing 50 ppb of uranium, it would be impossible to control cleanliness in terms of pure calcite of large enough samples to get an age with a 5 ppb U content calcite. In order to avoid unusable tedious chemical preparations, we first checked U content through ICPMS measurements. When measured by TIMS, ICPMS measurements were generally found in agreement (Table 2). We observed a continuous variation from 5 ppb up to 800 ppb. We retained 40 ppb as a practical threshold of sample suitability for Th/U dating. The lowest U contents were found for samples containing noticeable Th amounts. Consequently, these samples showed Th/U mass ratios greater than 1 (samples #10 to #15). When two or three sub-samples were isolated from calcite cement, they showed comparable values arguing for a homogeneous composition and a probable single generation. The time gap of around 4 ka for sub-samples a and b of sample #16 does not argue for distinct generations.

On the northern coast, sample #18 did not give an age, because ²³⁰Th/²³⁴U showed a default of uranium due to a probable U leaching after deposition despite a high residual U content. This could explain one of the rare U content variations (230 and 319 ppb) shown by ICPMS analyses (Table 2, Fig. 2). This sample was made of a multi-layered cement not well crystallised. On the same fault area, sample #19, made of white calcite cementing a fault breccia, gave an age of about 360 ka (+67/–41 ka as 2σ). Close to Itea, one sample, made of an ‘apparently’ young calcite, showed that the ²³⁰Th/²³⁴U and ²³⁴U/²³⁸U activity ratios were equilibrated, which demonstrates that the age is above the upper limit of the method, i.e., greater than the time necessary to reset equilibrium between ²³⁴U and ²³⁸U, which means more than one million years.

On the southern coast, breaks cutting conglomerates of synrift deposits, south of the Pirgaki Fault, are filled with large calcite crystals samples (#1–3). Sample #1 gave an old age comprised between 350 and 425 ka, within a 2σ interval. Samples #2 and #3 gave also old ages respectively attributed to 276–304 ka and 269–305 ka within 2σ interval. The same conclusion of an old age may be applied to sample #4 in spite of a low precision of the age determination. We propose to retain a minimum age of 430 ka. This calcite cement filled a minor fault in calcareous sediments parallel to the major Pirgaki Fault.

Two samples collected on the Xylokastro Fault gave highly different ages, respectively old (⩾600 ka but <1000 ka) and young (around 110 ka). Two sub-samples #17 showed ²³⁰Th/²³⁴U activity ratios at secular equilibrium, taking into account the error bars, and ²³⁴U/²³⁸U activity ratios characterised by an excess of daughter isotope. ²³⁴U excess is not important, but it is clearly out of equilibrium. These results demonstrate that ages are comprised between the durations necessary to reach the respective secular equilibrium of these two ratios. We retain ages comprised between 600 and 1000 ka. On the same fault, a multi-layered sample, made of a fibrous calcite around 4-cm thick covering a fracture in calcareous host rock (sample #16) was also analysed twice. The U content was the most elevated in this area, as well as ²³⁴U/²³⁸U activity ratios. Only a minor correction has to be made for Th-excess; the two layers analysed gave coherent ages around 110 ka with a time lag of about 4 ka between the two analysed layers. This sample also reported by Flotté et al. [4] was classified as post-tectonism by these authors. But microstructures observed through thin sections for all samples rarely shown stylolitic figures, except for the two samples collected on this fault.

Another sample undoubtedly crystallised very soon after a tectonic phase on the Doumena Fault (sample #6), made of pure white calcite filling a 2 cm void between fault walls, gave also a young age from two analyses of two sub-samples. The U content is not very high (less than 100 ppb) and the ²³⁴U/²³⁸U activity ratios are rather elevated, around the same value close to 1.5. The age determination around 124 ka (±3 ka, as 2σ) obtained for sub-sample #6a is confirmed also with a lower precision by the result given by sub-sample #6b.

A multi-layered calcareous deposit fills a fracture affecting syn-rift deposits close to Helike Fault. We analysed the sub-sample (#9b) apparently more suitable for the U–Th measurements: the elevated U content and low detrital material (pure white layer). The calculated age is equal to 62 ka and 50 ka after correction for Th-excess. These data have to be considered as indicative of a young age, without more precision because of a large correction due to ²³²Th abundance meaning that this carbonate is not pure, but mixed with detrital material, inferring a questionable real amount of in situ ²³⁰Th. It should also be noticed that this layer is apparently porous as well as other layers, poorly arguing for closed system conditions.

The youngest sample of this sampling is dated around 35 ka. This age determination is not precise, as a lot of detrital material mixed with the calcite resulted in a high ²³²Th amount and a low ²³⁰Th/²³²Th ratio equal to 3.5, implying a non-negligible correction for excess thorium. Nevertheless, this sample is obviously young, 50 ka being an upper age limit and 35 ka is not very far from the true age of this calcite deposited in a recently opened fracture.

The precise kinematic of the Gulf of Corinth is still widely debated due to the lack of dates. Our measurements, therefore, allow us to precise the central/western gulf evolution. The age distribution of the analysed samples is presented in Fig. 3.

The two samples on the northern shore indicate (1) an age older than 1 Myr for a fault with a large offset located at the level of Itea and (2) 3–400 000 yr for the normal fault on the Peninsula southeast from the previous fault. This second fault has an offset of more than 50 m. These two dates indicate that these faults are synchronous of the faults on the southern shore. They constitute another evidence of the fault activity along the northern shore of the Gulf of Corinth already emphasised by marine data [9,13] and by archaeological evidences [11], although minimised by some authors.

On the southern shore, two ages have been determined for the Xylocastro Fault: one between 0.6 and 1 Myr and the other at about 110 000 yr (this second dating has also been published in [4]). One may note that in the Mamousia–Pirgaki Fault, located westward but more or less in the prolongation of the Xylocastro Fault, Micarelli et al. [8] have evidenced a polyphasic evolution by means of structural analyses. The first phase cannot be dated yet on the Pirgaki Fault, but is ‘old’, since at least 3 km of sediments have been eroded from this time. The second phase is contemporaneous of the Gilbert type fan delta deposit (between 1.2 and 0.9 Myr, [9]) and the 2001 seismic event proved than the system is still (or again) active [15]. The veins in the fault where we collected sample #4, a fault parallel to the Mamousi–Pirgaki fault located just northward of the main fault plane, have been dated at 430 000 yr as a minimum age. This result suggests that the Pirgaki Fault System had undergone a continuous activity during the last Myr.

The Doumena Fault is located southward from the Pirgaki–Mamousia fault system. The outcrop, south of the Doumena town, exposes an impressive east–west fault plane formed by numerous shear planes with variable dips (from 45° to 65° N). The footwall consists of the Mesozoic carbonates and the hangingwall of synrift deposits. These deposits are part of a fan delta that could be an internal part of the delta situated northward from the Pirgaki Fault. On the opposite, the Pirgaki–Mamousia System clearly affects the synrift deposits, the Gilbert delta growth above the through created by the Pirgaki fault offset. The Doumena Fault does not affect the sedimentation of the exposed layers on its hanging wall in term of grain size and pebble content. On the opposite, the existence of structural features, such as drag and hangingwall syncline, also suggests that the fault has affected the synrift sequence and their deposits, and thus postdates it. The date (125 ka) given by our results confirms this out-of-sequence position of the Doumena Fault, as compared to a simple northward-propagation model proposed by [3,12]. At this stage of our knowledge we cannot precise the appearance of the Doumena Fault, the fan delta is not directly dated, but we consider it as contemporaneous of the Gilbert type delta. In this hypothesis, the Doumena Fault is younger than 0.9 Myr and our new dates have proven that it was active 125 000 years ago. It means that during the same period 120 000±10 000 yr, the Helike Fault appeared, the Xylocastro Fault has been reactivated and the Doumena Fault has been active. All these data suggest an increase of the tectonic activity in the Gulf during the last 125 000 yr. Based on the analysis of onshore and offshore data, Moretti et al. [9] and Lykousis et al. [7] have highlighted this activity and proposed the identification of a third tectonic phase in the Gulf of Corinth's evolution. They consider that the main feature of this phase is the uplift of the Peloponnese that triggers reactivation of oldest faults and the appearance of new ones.

The other two sites of dating were in minor faults, fractures and veins through the synrift deposits. Samples #1, #2 and #3 gave ages between 0.3 and 0.4 Myr. The structural importance of these intra-post rift sediment faults is weak, but the calcite clearly post-dated the sediments, which are therefore older than 400 000 years.

In the location where we collected sample #9, the veins affect the syn-rift sediment uplifted on the footwall of the Helike Fault. The outcrop is located just a few hundred metres southward from the Helike Fault. The veins have been dated at around 50 ka. This age is the same as that of the Aigion fault appearance [8] and confirms the continuation of the current kinematic phase.