In the Central Jilin Province, four litho-tectonic domains are recognized from west to east. They allow us to reconstruct the architecture of a convergent plate boundary (Fig. 1). To the west, the Late Palaeozoic series consists of shallow-water continental shelf deposits [6]. Cambrian and Ordovician limestone and argillite are overlain by Carboniferous and Permian limestone, sandstone, and shale.

Eastward, the second domain is characterized by metamorphic rocks (Hulan Group) and a mafic-ultramafic association (Hongqiling Group) [6,29,31]. The upper greenschist to amphibolite facies metamorphic rocks, namely biotite–garnet ± kyanite micaschists, quartzites, marbles and amphibolites enclose gabbro, diabase, serpentinite, ultramafic rocks, some of which contain economic Ni–Cu concentrations [30,32]. Although previously considered as post-orogenic intrusions (e.g., [28,35]), our structural observations show that these ultramafic bodies are underlain by amphibolites. As a whole, we interpret this mafic-ultramafic association as an ophiolitic suite surrounded by metamorphic rocks. Based on LA-ICP-MS dating of detrital zircons enclosed in the metapelites, an Early Permian age (ca. 290 Ma) is inferred for the age of the metamorphic rocks. Micaschists yield Rb–Sr whole-rock–muscovite–biotite ages of 233 ± 8 Ma and 253 ± 10 Ma, interpreted as that of the syntectonic metamorphism [29].

To the east, the sedimentary rocks assigned to the Permian exhibit typical features of a block-in-matrix formation. Well-bedded turbidite, lenticular masses of coarse-grain sandstone, slump deposit, and pebbly mudstone are common facies (Fig. 2). Blocks of various lithologies such as chert, serpentinite, acidic tuff, andesitic and basaltic breccia, with sizes ranging from the millimetre to the kilometre scale, are widespread in this chaotic formation. Carboniferous and Permian limestones form metre- to kilometre-size blocks included into the terrigenous matrix. The above-mentioned rocks are olistoliths deposited in a terrigenous matrix, and subsequently sheared. Based on the foraminifera found in limestone, the age of this chaotic formation is considered as Permian [21]. However, since limestones are blocks, the real age of the matrix remains unsettled yet, but it cannot be younger than the Late Triassic, which is the age of the terrestrial shallow-water conglomerate and sandstone that unconformably cover the chaotic formation [6] (Fig. 1). Therefore, a Late Permian–Early Triassic age for this mélange is proposed here. More to the east, but isolated from the previous domains by Late Mesozoic continental red beds and granitic plutons, several calc-alkaline plutons of gabbro, diorite and tonalite intrude into a metamorphic series developing a contact metamorphism. All these units are intruded by Jurassic quartz–diorite plutons [29].

These field observations led us to propose the following bulk architecture (Fig. 1). The Late Palaeozoic western sedimentary unit represents the passive continental margin of a western continental block. The mafic-ultramafic suite corresponds to a synmetamorphic, ductilely deformed, Permian ophiolitic series, tectonically inserted in metamorphic rocks. The metamorphic rocks and the ophiolitic series overthrust from east to west the continental margin series. The bloc-in-matrix series is interpreted as an accretionary complex formed in an active continental margin. Finally, to the east, the calc-alkaline dioritic-tonalitic plutonic rocks are the deep parts of a pre-Jurassic magmatic arc, emplaced in an eastern continental basement.

In spite of discontinuous exposures, a deformation gradient, increasing from west to east, can be documented (Fig. 1). The eastward-dipping formations are involved into a west-directed thrust system, locally refolded by kilometre-scale upright folds. In the western area, the ductile deformation is absent, but numerous brittle thrust faults or en-échelon tension gashes indicate a top-to-the-west displacement (Fig. 2). In the metamorphic rocks (Hulan group, e.g., [29]), the flat-lying mylonitic foliation contains an east–west-trending mineral and stretching lineation. In marbles, metre-scale marble boudins, also stretched in the east–west direction, are well developed (Fig. 1). The mafic rocks are often transformed into well-foliated and lineated amphibolites. Along the east–west-trending lineation, whatever the lithology, ductile shear criteria indicate a top-to-the-west motion. West-directed shearing is observed in the chaotic formation too. The block rims are foliated, and in the slaty mudstone matrix, pebbles are boudinated. Conversely, in the arc domain, the metamorphic rocks exhibit a north–south-trending lineation that is unrelated to this thrusting event since the plutons are not deformed (Fig. 1).

In the study area, stratigraphic and radiometric dates remain rare. Therefore, a ⁴⁰Ar/³⁹Ar study has been undertaken in order to assess the age of the synmetamorphic shearing. Samples, located in Fig. 1, are amphibolite from the base of the ophiolitic nappe: JL11, 12; marble: JL 17; and diorite from the magmatic arc: JL 28, JL 30.

Hornblende (JL 11, JL12, JL28, JL30: 450 to 350 m), biotite (JL28: 450 to 350 m) and muscovite (JL17: 450 to 350 m) were concentrated using a Frantz magnetic separator and a vibration table. Several grains free of alteration were selected under the binocular for each sample. Ten to fifteen grains for each mineral phase were loaded in an Al disk and irradiated for 12 h in the CLICIT facility of the TRIGA reactor at the Oregon State University. After irradiation, single grains are degassed with a defocused CO₂ Synrad^© laser, then analyzed with a MAP 215C mass spectrometer using the methods described by Renne et al. [15]. Neutron fluence (J) was monitored by a Fish Canyon sanidine (FCS: 28.02 Ma) loaded in the same disks. The J value (0.003103 ± 0.000007) is determined from individual analyses of six separate FCS grains fused with the CO₂ laser and is calculated with the total decay constant of Steiger and Jäger [22]. Mass discrimination, monitored by automated analysis of air pipettes, was 1.00374 ± 0.000248 (n = 28; 2 σ error) per mass atomic unit. Nucleogenic production ratios from K and Ca were (⁴⁰Ar/³⁹Ar)_K = (7.3 ± 0.9) × 10⁻⁴, (³⁹Ar/³⁷Ar)_Ca = (7.60 ± 0.09) × 10⁻⁴ and (³⁶Ar/³⁷Ar)_Ca = (2.70 ± 0.02) × 10⁻⁴ (2 σ error). Plateau ages are defined as comprising three continuous steps corresponding to a least 70% of the total ³⁹Ar released. The individual fraction ages should agree within 2 σ with the integrated age of the plateau segment.

All uncertainties are given at 2 σ (95% confidence). The biotite single-grain spectrum (JL28, Fig. 3) is slightly discordant, the first 15–20% of the spectrum displaying younger apparent ages. The plateau defined by 78% of the total ³⁹Ar yields an age of 187.0 ± 0.6 Ma. Such a weakly discordant behaviour can be explained by the spurious presence of chlorite interlayer or a partial loss of ⁴⁰Ar in the outer part of the crystal [13]. Muscovite (JL 17, Fig. 2) displays a flat spectrum with a plateau age of 187.9 ± 1.2 Ma (100% of ³⁹Ar). Hornblende spectra (JL11, JL12, JL28, JL 30, Fig. 2) are also flat, with young apparent ages at low temperatures (first 10–20% of total ³⁹Ar). These young age steps, positively correlated with lower Ca/K ratios, suggest that the K-rich alteration phase degassed early in the experiment. The plateau age of ca 190 Ma obtained on hornblende JL 28 is slightly older than the one of 187 Ma, obtained on the biotite from the same sample, suggesting a fast cooling.

The Late Triassic unconformity recognized in the field implies that the metamorphic rocks cannot be younger than 225 Ma, according to the time scale [4]. Even the older dates of 218 and 208 Ma for hornblende in the amphibolites JL 11, JL 12 do not comply with this constraint. Furthermore, muscovite from marble (JL 17) and biotite and amphibole from the diorites (JL 28, JL 30) yield radiometric ages around 190 Ma, that is, Early Jurassic. Such a discrepancy can be explained by a complete thermal reset of the K–Ar system during the emplacement of the Jurassic granites [29], whereas it suggests excess temperatures of 500–550 °C. The numerous Mesozoic plutons that crop out in the area might be also responsible for fluid circulation and recrystallization of the host rocks, as well as for the opening of the K–Ar chronometer. Another possibility will be a ⁴⁰Ar loss during deformation [11]; however, this hypothesis is less probable. It is worth noting that our results are close to the 216–217-Ma Rb/Sr age yielded by biotite micaschist [29]. Moreover, the ⁴⁰Ar/³⁹Ar ages of ca 225 Ma from biotite and muscovite obtained in the micaschist associated with the mafic-ultramafic rocks comply with an Early Triassic age for the thrusting [30]. South of the Fushun-Mishan fault (Fig. 2), calc-alkaline metavolcanites similar to those of the above-described magmatic arc yield 260 ± 2-Ma zircon SHRIMP ages that are interpreted as the age of the magmatism, in good agreement with our own conclusion. Finally, sericite from the hydrothermalized gold-bearing vein yields a ⁴⁰Ar/³⁹Ar age of 204 ± 0.5 Ma, which suggests an important Late Triassic magmatic episode associated with a hydrothermal event [10].

These new petrological, structural, and geochronological data of the Jilin area allow us to propose that the study area is a suture zone separating two continental blocks. The eastward dip of the subduction is deduced from the magmatic arc in the eastern block (Fig. 1). The blocky formation is the accretionary complex developed in the subduction zone. The ophiolite nappe was transported westward in metamorphic conditions during the Late Permian–Early Triassic collision. As discussed below, the western and eastern blocks can be respectively correlated with the northeastern part of the NCB (here named Liaoji) and Xilinhot Blocks, which are recognized in the western part of the CAOB (e.g., [17,32]).