Study on Late Cretaceous-Cenozoic exhumation of the Yanji area,NE China:insights from low-temperature thermochronology

Xiaoming Li

Abstract The Yanji area,northeastern China,a part of the orogenic collage between the North China Block in the south and the Jiamusi-Khanka Massifs in the northeast,is the most likely location where the Pacific Plate subductionrelated magmatic activities and subsequent exhumation processes occurred.Here,we report new low-temperature thermochronology of apatite and zircon data from the granitoid samples in the Yanji area.The exhumation rates of Tianfozhishan, Yanji area, were ～0.049 and ～0.073 mm/year,interpreted from the elevations and apatite and zircon fission track ages, respectively. The exhumation,integrated with the geological setting,suggested that the paleogeothermal gradient of the Tianfozhishan,even extending to the Yanji area,was possibly to be greater than 35°C/km in the Late Cretaceous.The thermal history modeling of the data indicates a basically similar pattern,but the various timing for different samples between the Oligocene-Early Miocene and the Middle Miocene in the Yanji area.We hence conclude that a fourstages of cooling,from ～6.7°C/Ma(during the Late Cretaceous),to ～0.8°C/Ma(during the Late Cretaceous to the Oligocene-Early Miocene),then to ～2–3°C/Ma with varied styles(between the Oligocene-Early Miocene and the Middle Miocene), and finally to ＜0.2°C/Ma(since the Middle Miocene),has taken place through the exhumation of the Yanji area.The maximum exhumation is ＞3 km under a reasonable paleogeothermal gradient(＞35°C/km),speculated from the possible exhumation rate of Tianfozhishan.Combined with the tectonic setting,this exhumation,including two stages of pronounced tectonic uplift and denudation and two stages of weak exhumation driven by the low regional erosion rate,is possibly related to the subduction of the Pacific Plate beneath the Eurasian Plate since the Late Cretaceous.This study used more robust evidence to propose higher paleogeothermal gradient(＞35°C/km),reflecting exhumation of ＞3 km in the Yanji area since the Late Cretaceous.

Keywords Low-temperature thermochronology·Exhumation·Pacific Plate subduction·Yanji area·Late Cretaceous-Cenozoic

1 Introduction

Northeastern(NE)China is geologically located in the eastern segment of the Central Asian Orogenic Belt(CAOB),intersected in the Siberian Craton,North China Block(NCB)and Western Pacific Plate(Fig.1a).NE China was formed by the continental margin accretion and continent–continent collisions(Zhao et al.1990,1994;Se¨ngor et al.1993;Van der Voo et al.1999;Li 2006).Since the Paleozoic,it has undergone the evolution of the Paleo-Asian Ocean,the Mongol-Okhotsk Sea and the circum-Pacific tectonic domain.Alternatively,the subduction of the Palaeo-Asian Ocean from the early Paleozoic to the early Mesozoic forms the CAOB, then the Mongol-Okhotsk Ocean closed diachronously in a scissor-like fashion from west to east during the Middle Jurassic and the Early Cretaceous,signifying the completion of the accretion of the East Asian Continent,and finally,the NE China and the adjacent region has become part of the circum-Pacific tectonomagmatic belt since the Late Mesozoic(Zoneshain et al.1990;Ma and Yang 1993;Jia et al.2004).Accordingly,the voluminous granitoids were emplaced during the Paleozoic and Mesozoic eras in the NE China and along the northern margin of NCB (Wu et al.2000,2011;Hong et al.2004;Zhang et al.2004).However,it is still controversial on the relationship between the extensive magmatic activities in the NE China(probably throughout eastern China)and the(Paleo-)Pacific Plate subduction(Lapierre et al.1997;Xu et al.1999;Li 2000;Yin et al.2000;Zhou and Li 2000;Wu et al.2005;Sun et al.2007).Moreover,the cooling and exhumation history of the granitoids after the collision is poorly understood(Li et al.2010;Li and Gong 2011).

Fig.1 Maps of the regional geology and tectonics. a Tectonic divisions of NE Asia(Modified after Li 2006)and b tectonic divisions of NE China(Modified after Zhang et al.2004)

The Yanji area,located at the border of China,Russia,and Korea and about 100 km west of the Japan Sea,was considered as a part of the orogenic collage between NCB in the south and the Jiamusi-Khanka Massifs in the northeast(Fig.1b).To the east is the Japan Sea back-arc basin,which is considered to have formed in response to the subduction of the Pacific Plate during the Oligocene-Early Miocene to the Middle Miocene(Nakamura et al.1990; Yamaji 1990; Jolivet et al. 1994; Taria 2001).Consequently,the Yanji area is the most likely location where subduction-related magmatic activities and subsequent exhumation processes would occur.The Phanerozoic granitoids have been extensively exposed in the Yanji area(Fig.2a),suggesting that the rocks overlying these granitoids have largely been removed.The total exhumation since the Late Cretaceous was on the kilometer scale.Li et al.(2010)obtained a preliminary conclusion of threestage of cooling,and inferred exhumation of granitoids in the Yanji area,just by six apatite fission-track(AFT)and six zircon FT(ZFT)data and thermal modeling results,under an assumed steady-state geothermal gradient of 35°C/km.

The closure temperatures of ZFT and AFT are generally considered to be 210±40 and 100±20°C,respectively(Wagner and Van den haute 1992). The latest study reported that ZFT partial annealing zone was at geological timescales as ～190–380,～180–350 and ～160–330°C for heating durations of 1,10 and 100 Ma,respectively(Tagami 2005),AFT annealing is a thermally activated process occurring over a range of temperatures typically up to ～100–120°C over geological time scales,and lower temperature portion of thermal history(＜ ～60–70°C)is poorly constrained by AFT data(Kohn et al.2005).In this study,we assume that the closure temperature of ZFT of the studied samples is 210°C and the partial annealing zone of AFT is between 60 and 110°C.

The apatite(U–Th–Sm)/He(AHe)dating system has a helium partial retention zone(HePRZ)of between ～40 and 80°C(Wolf et al.1998;House et al.1999;Stockli et al.2000),and has a closure temperature of ～75°C(Farley 2000). The zircon (U–Th–Sm)/He (ZHe) thermochronometer has a HePRZ of 140–200°C(Wolfe and Stockli 2010),and a closure temperature that ranges from 175 to 193°C given a cooling rate of 10°C/Ma for typical grain sizes(Reiners 2005).

Robust thermal histories were obtained by jointly inverting the analytical data using a‘Frequentist’approach described by Ketcham(2005).We report,in this study,new low-temperature thermochronology of apatite and zircon data from the granitoid samples in the Yanji area,including 16 AFT ages and 7 track length data,16 ZFT ages,8 AHe ages and 3 ZHe ages,and integrate with 6 AFT ages,6 track length data and 6 ZFT ages from Li et al.(2010).These data can help us to obtain a clearer insight into the late-stage cooling and inferred exhumation history of granitoids in the Yanji area,NE China,and to place this within the context of the regional tectonic history of the western margin of the Pacific plate.This can also be usedto recognize the Late Cretaceous-Cenozoic tectonic relief and evolution of the Yanji area,which can be an indication of Mesozoic-Ceozoic tectonic evolution in the NE China and neighboring area influenced by the effects arising from the Pacific Plate subduction(Sun et al.2007;Yang et al.2018).

Fig.2 a Geological sketch showing the sampling locations(Modified after JBGMR 1988).b Topographic map of the Yanji area,showing zircon and apatite fission track ages and(U-Th–Sm)/He ages

2 Geological and geomorphological setting

The Yanji area,located at the junction of the major tectonic units in the area(NCB and CAOB),is in close proximity to the microcontinental blocks of the Khanka and Jiamusi Massifs(Fig.1b)(JBGMR 1988;Jia et al.2004;Zhang et al.2004).Generally,the basement is composed of the Archaean Jiapigou Group and the Late Proterozoic Zhangguangcailing Group.The Paleozoic strata are widely distributed and have undergone various degrees of metamorphism and deformation(JBGMR 1988),intruded by immense volumes of the Phanerozoic granitic rocks(JBGMR 1988;Zhang et al.2004).Available zircon U–Pb data indicated that emplacement of these Phanerozoic granitoids extended from the Late Paleozoic(285 Ma)to the Early Cretaceous(116 Ma)(Zhang et al.2004;Wu et al.2011).The Mesozoic geology is characterized by the eruption of voluminous lavas and deposition of terrestrial sediments, and the Cenozoic intermontane basins are developed in the Yanji area(Fig.2a).

The topographic relief is higher in the west than in the east for the Yanji area.They are ＞1000 m a.s.l.in the southwest,in the northwest and in the northeast,decreasing gradually to ＜100 m a.s.l.in the southeast for the Yanji area,with the lowest in Hunchun area(Fig.2b).The whole landscape of the Yanji area presents three gradients of the mountains,hilly regions and basins.Most of the mountains are located at the surrounding areas,most of the hilly regions are situated on the edges of the mountains,and the basins are mainly distributed between the banks of rivers and mountains in the Yanji area.

3 Sampling and experimental methods

Twenty-two representative samples,in this study,were collected for FT and(U–Th–Sm)/He analyses(Fig.2a,Table 1)from the Phanerozoic granitoids in the Yanji area.Especially eight samples were collected from the Tianfozhishan at ～100–200 m elevation intervals.The Tianfozhishan is a hill in the southwestern Yanji area(Fig.2b),over 1200 m a.s.l.,composed of granite with zircon U–Pb age of 196±7 Ma by LA-ICP-MS method(Wu et al.2011).Most samples are of the Jurassic in crystallization age,with a few samples of the Permian,the Triassic or the Early Cretaceous(Table 1).These samples are fresh fineto coarse-grained granite or diorite,without any significant tectonic or magmatic overprinting.Sampling was conducted by avoiding possible thermal effect related to active faults and/or Cenozoic magmatisms.

We measured the coordinates of each sampling location using a portable GPS with precision in the order of 10 m,along with its elevation with precision in the order of 5 m.

Apatite and zircon grains were obtained by the conventional crushing,sieving,magnetic and heavy liquids separation techniques.ZFT and AFT ages were analyzed at the Guangzhou Institute of Geochemistry,Chinese Academy of Sciences,by using the external detector method(Hurford and Green 1983).Apatite and zircon grains for FT dating were mounted by epoxy resin and FEP(fluroethylenepropylene)Teflon sheets,respectively.Next,the prepared samples were carefully ground and polished to an optical finish to expose internal grain surfaces.Then,apatite spontaneous fission tracks were revealed by 5.5 M HNO3at 21°C for 20 s(Carlson et al.1999),and zircon by a molten 8 g NaOH+11.5 g KOH eutectic at 220°C for 20–60 h(Gleadow et al.1976),as necessary to obtain a high-quality etch(Gleadow 1981).

The mounts were covered with a flake of low-U muscovite,packed for irradiation,along with standard uranium dosimeter glass SRM612 and standard sample FCT(27.8±0.7 Ma)(Hurford and Green 1983),and irradiated in a well-thermalized(Cd for Au ＞100)neutron flux 492 Swim reactor,Chinese Institute of Atomic Energy.Afterwards,muscovite was etched in 40%HF at 25°C for 20 min to reveal induced fission tracks.The mean etch pit diameter(Dpar)was measured for each grain as a representative of their chlorine content to be used in annealing models.Only those crystals with prismatic sections parallel to the c-crystallographic axis were accepted for Dpar analysis,because of their high etching efficiency.The FT ages were calculated following the method recommended by Hurford (1990), with the zeta calibration method(Hurford and Green 1983).Ages were calculated using the Trackkey software®(Dunkl 2002),and ages quoted are pooled fission track ages with one standard error.Zeiss Axioplan microscope at 1250× magnification with AUTOSCAN system was used to observe and measure the spontaneous and induced densities of zircon and apatite FT populations,as well as confined lengths and their related Dpar values of apatite fission tracks(Table 2).

AHe and ZHe thermochronology was performed at the University of Melbourne,Australia.Euhedral zircon and apatite crystals without fracture or U/Th-rich inclusion were handpicked for analysis. Grains with a diameter greater than 70 μm were preferred where possible in order to ensure helium gas values were optimal for measurement and to minimize the alpha ejection correction.

Apatite and zircon aliquots were loaded into platinum capsules and degassed under vacuum at ～900°C for 5 min and ～1300°C for 15 min,respectively,using an 820 nm fiber-optically coupled diode laser(Li et al.2016).4He abundances were determined as an isotope ratio,using a pure3He spike which has been calibrated against an independent4He standard.The uncertainty in the sample4He measurement is estimated at less than 1%.Apatite U–Th–Sm data were obtained using an Agilent 7700 quadrupole ICP-MS after complete dissolution of the degassed apatite aliquots in HNO3.For zircon U–Th data,aliquots were transferred to Parr bombs where they were spiked with235U and230Th and digested at 240°C for 40 h in HF.Standard solutions containing the same spike amounts as samples were treated equally,as were a series of unspiked reagent blanks.A second bombing in HCl for 24 h at200°C insured dissolution of fluoride salts and final solutions were diluted to 10%acidity for analysis on an Agilent 7700 quadrupole ICP-MS.Analytical uncertainties are estimated at about 6.2%(±1σ),incorporating α-correction related constituent and taking into account an estimated 5 μm uncertainty in grain size measurements,gas analysis and ICP-MS uncertainties.Durango apatite and Fish Canyon Tuff zircon were carried out as internal standards with each batch of samples and served as an additional check internal monitor for analytical accuracy.Further,weighted mean ages of AHe and ZHe are used to discuss individual sample cooling age and the number of single grains analyzed per sample is shown in Tables 3 and 4,respectively.

Table 1 Sample details

4 Results

We analyzed 22 samples,from which we obtained 22 AFT ages and 22 ZFT.

Ages(16 new FT ages and 6 FT ages from Li et al.(2010),therein 13 AFT with track length data),9 AHe ages(41 single-grain ages, therein 34 accepted single-grain ages),and 3 ZHe ages(10 accepted single-grain ages).Detailed results are given in Tables 2,3 and 4.

The measured data,including the pooled ages of both zircon and apatite FT,the mean track length(MTL)and Dpar value of apatite FT are listed in Table 2,and apatite FT length distributions of seven new samples as well as Sample No.YJ3 are illustrated in Fig.3.The results show that zircon and apatite FT ages for these samples range from 91.7±7.6 to 101.0±9.2 Ma and 70.2±4.0 to 85.4±3.8 Ma,respectively(Table 2).All of apatite FT length distributions are unimodal with a mean length of 12.0–13.9 μm and a standard deviation of 1.3–2.4 μm,and the mean Dpar values of apatite FT vary from(1.6±0.2)μm to(2.0±0.3)μm(Table 2).

All AHe analytical data from 9 samples are summarized in Table 3,with weighted average ages presented(at the 95%confidence level)for multiple analyses of each sample calculated using Isoplot(Ludwig 2012).The data can be divided into three groups.

1. Five samples(YJ3,YJ8,YJ10-1,YJ10-8 and YJ10-13)yield weighted mean ages in the range of 30.4±4.8 to 60.9±13 Ma with analyses from aliquots of the same sample being replicated within analytical error at the±2σ uncertainty.Dispersion(standard deviation of age/mean age)is varied largely for samples in our data set,ranging from 0.6 to 31%.

?

2 elbaT deunitnoc 2 2-5 2-52-5 μσ μσ χρ ρ ρ)m(1±rapD naeM skcart fo.oN )m(1±LTM )%()(P )aM(ega delooP )N(mc01×/)N(mc01×/)N(mc01×/N lareniM .on elpmaS d diiss a–688.1±2.31538.3±4.58)1125(57.5)2681(38.42)3361(77.1202etitapA4JY 797.7±6.99)0923(13.1)052(50.12)6511(53.797nocriZ a–256.1±0.31361.4±3.18)5185(18.5)0721(23.02)9401(87.6103etitapA5JY 999.8±3.89)5214(83.1)371(54.81)947(98.976nocriZ a–737.1±1.31790.5±1.28)5185(18.5)447(71.41)126(38.1103etitapA6JY 7.5±1.69)0923(13.1)415(07.52)2922(06.411 01nocriZ a–343.1±9.21890.5±6.18)1125(57.5)957(96.12)636(71.8122etitapA8JY 693.9±6.49)5214(83.1)541(00.92)406(08.021 4nocriZ ylevitcepser,nocriz dna etitapa rof 6.9±5.133 dna 5.8±8.043 fo seulav atez gnisu dohtem rotceted lanretxe yb denimreted segA ρρρlanretxe etivocsum a ni derusaem elpmas a fo ytisned kcart decudni,;enimreted ot detnuoc skcart fo rebmun,N;elpmas a fo ytisned kcart suoenatnops,;detnuoc slatsyrc fo rebmun,n is s s ρρ ot detnuoc skcart fo rebmun,N;rotceted lanretxe etivocsum a ni derusaem 216MRS retemisod ssalg fo ytisned kcart decudni,;enimreted ot detnuoc skcart fo rebmun,N;rotceted d di i 2 2 χ χρmodeerf fo seerged)1-n(gnideecxe ro gnilauqe fo ytilibaborp,)(P;enimreted d a)0102(.la te iL retfA

2. A further three samples(YJ1,YJ5 and YJ10-15)yield single aliquot ages within a relatively restricted range and some replicate well at±2σ,but some do not(a clear old age outlier is excluded for YJ1,YJ5 and YJ10-15, respectively), and the weighted average calculation for the moderate dispersion shown yields ages with greater uncertainties ranging between 61±17 and 72.2±4.2 Ma.Dispersions range from 0.6 to 16%in our data set.

3. One sample(YJ10-14)shows wide age dispersion,and AHe grain ages are evidently older than their coexisting AFT age, so the weighted mean age is not analyzed,and the AHe data were excluded in our interpretation.

Possible explanations for He age dispersion and apparent excess He ages have been outlined by Fitzgerald et al.(2006)and Shuster et al.(2006).Namely,in theory,there is a positive correlation between single grain AHe ages and the effective U concentration eU,expressed as U ppm+0.235 Th ppm.However,in nature,plots of eU versus single-grain age are more complex,and different samples show correlations that are positive,negative,or entirely absent,because the effect of radiation damage accumulation and annealing is not linear,and it is dependent on the thermal history(Flowers et al.2009;Gautheron et al.2013;Mbongo-Djimbi et al.2015).Other factors that can cause single-grain age dispersion must also be considered,such as the possible presence of U and Th zonation,unidentified U-and Th-bearing inclusions,implantation,the contribution of Sm,and the dominant influence of the thermal history of the sample in controlling He diffusion(Brown et al.2013;Wildman et al.2016;and therein references).

In conclusion,the accepted AHe grain ages range from 26.0±1.6 to 87.9±5.5 Ma,ZHe grain ages range from 74.8±4.6 to 145.6±9.0 Ma.Although it is not a positive correlation between single grain AHe ages and eU for the eight samples(Fig.4),the AHe and ZHe ages can be grouped in a similar way to the FT ages,as a product of a thermal history that is initially unknown.

5 Thermal modeling

5.1 Apatite fission track and(U-Th-Sm)/He modeling

To better interpret the significance of our data,we modeled the data by HeFTy software(Ketcham 2005)to quantify the timing and amount of cooling at specific sites with eight apatite FT data,in combination with AHe data sets as far as possible.Time–temperature histories were calculated using the inverse modeling approach of HeFTy(Ketcham 2005)with a multi-compositional annealing model (Ketcham et al.2007b)and the Dapr value as the kinetic parameter.Models were run using c-axis projected lengths(Ketcham et al.2007a),and initial track lengths were calculated according to the mean Dpar value of each sample using the formula of Carlson et al.(1999).

?

?

?

The thermal modeling focuses mainly on samples with AFT data,but for samples YJ3,YJ10-8 and YJ10-13,where AHe results showed reasonable analytical reproducibility,we have used both AFT and AHe data of grains with one or two terminations.AHe data were modeled using the radiation damage accumulation and annealing model(RDAAM)(Flowers et al.2009).

According to the aforementioned conditions as well as the generally average surface temperature of 20°C,the thermal history modeling results are illustrated in Fig.3,from which it indicates that the measured and modeled apatite FT ages and lengths are consistent within the scope of standard errors.Based on the Kolmogorov-Smimov test(probability values ＞0.5),the modeled and the measuredtrack lengths vary from 11.9 to 13.3 μm,and 12.0 to 13.2 μm,respectively,and the modeling and the measured apatite FT ages range from 74.8 to 85.5 Ma,and 76.5 to 85.4 Ma, respectively. In particular, the eight cooling curves exhibit roughly similar patterns but exhibit some different time sequence of multi-stages of cooling(Fig.3).

Fig.3 Modeling sketch of time–temperature paths and AFT lengths in the Yanji area.Thermal modeling of apatite FT length and age data modeled time-temperature paths for eight apatite samples,computed with HeFty program by Ketcham(2005).Several conditions are as follows:(1)the present day temperature is set to a constant value of 20°C,AHe age,AFT age and ZHe age constraints(see text for further details)are drawn as a thick blue box;(2)kinetic variable of Dpar was input;(3)models for samples YJ3,YJ10-8 and YJ10-13 also incorporate representative AHe data;and(4)modeling scheme was Monte Carlo.We modeled 100,000 paths for each plot.Thick lines show best-fit solutions obtained for these model run,and red and green colors show good-fit solutions and accept-fit solutions obtained for same model run,respectively

Fig.3 continued

5.2 ZHe,AFT and AHe modeling

Fig.4 Dependence of AHe ages on the eU-factor(see text for further detail)

ZHe ages were basically similar to ZFT ages within two standard errors.Thus we use ZFT age at ～100 Ma as ZHe age to modify the thermal history of samples.Thereinto,we modeled the thermal history of Sample No.YJ3,combining the measured ZHe with AFT and AHe data,and those of other samples,integrating the measured AFT with assumed ZHe data at ～100 Ma.ZHe data were modeled using the helium diffusion model of Reiners(2005).Where applicable,broad temperature constraints 140–200°C were set around the time range covering the ZHe ages with 2σ to set an initial box.These pre-modeling settings were always included with large uncertainties so as to give the inversion algorithm sufficient freedom to search for a wide range of possible thermal histories.Inverse thermal history modeling was run until 100 good paths were obtained,which in most cases resulted in ＞1000 acceptable paths.The results were shown in Fig.5.All modeling results predict a rapid to slow cooling,then an accelerated to slow cooling with varied styles.The results are different from those of Li et al.(2010)in the latest cooling.

Fig.5 Cooling histories of eight samples based on thermal history modeling of ZHe,AHe and AFT data.Envelopes encompass all of‘‘Good Fit’’models obtained for the eight samples using a merit value of 0.5 in HeFTy,while thick lines represent best-fit models to analytical data

6 Discussion

6.1 Exhumation history of the Tianfozhishan and paleogeothermal gradient

Figure 6 shows the AFT and ZFT ages versus elevations in the Tianfozhishan. The exhumation rates are ～0.049 mm/year and ～0.073 mm/year (1 mm/year=1000 m/Ma),linear fitting by the elevations and AFT and ZFT ages with the least square,respectively(Fig.6).These estimates are much lower than the exhumation rate of ～0.19 mm/year during ～100–80 Ma from Li et al.(2010),assuming the steady-state geothermal gradient of 35°C/km(Li et al.2010).

Thermal history analysis results indicate that the heat flux of Yanji Basin even reached 94 mW/m2during early Longjing stage(～90 Ma),and it decreased during late Longjing stage for denudation(Chen et al.1997).In the neighboring area,such as the Songliao Basin,the average geothermal gradient is 37°C/km in the present-day,and decreases gradually from the basin center to the margin at ～30°C/km,which is less than its paleogeothermal gradient(Ren 1999).The above facts show that the paleogeothermal gradient might be decreased gradually since the Late Cretaceous.Thus,the paleogeothermal gradient of the Tianfozhishan,even extending to the Yanji area,was possibly greater than 35°C/km in the Late Cretaceous.

Assuming a ～0.049–0.073 mm/year exhumation rate of the Tianfozhishan in the Late Cretaceous,our thermal history models predict the paleogeothermal gradient at ～90–135°C/km.However,the exhumation rate was probably more than ～0.049–0.073 mm/year,because both the little difference of elevations and the errors of AFT and ZFT ages may lead to larger errors of the exhumation rate.

Fig.6 AFT and ZFT ages versus elevations in the Tianfozhishan,as well as linear fitting by the elevations and AFT and ZFT ages with the least square,respectively

Consequently,it should be reasonable for the exhumation rate of ＞0.049–0.073 mm/year and the paleogeothermal gradient of ＞35°C/km in the Yanji area during the Late Cretaceous.

6.2 Cooling and exhumation history of each sample

All samples were collected from the granites formed between 241 and 129 Ma as shown by LA-ICP-MS in zircon U–Pb ages(Wu et al.2011)(Table 1).The measured apatite and zircon FT ages and(U–Th–Sm)/He ages are markedly younger than the formation ages of the hostrocks,representing the cooling ages of the samples during the uplift and denudation processes because of their negligible effects from magmatic or tectonic activities since the Late Cretaceous.

Thermal modeling history reveals a cooling record from ～210 to ＜40°C between ～100 Ma and the present day (Fig.5). The models show that all samples experienced rapid cooling (～6.7°C/Ma) during ～210 Ma to the Late Cretaceous,then slowed dramatically to ～0.8°C/Ma between the Late Cretaceous and the Oligocene-Early Miocene,followed by increased cooling(～2–3°C/Ma)from the Oligocene-Early Miocene to the Middle Miocene(Fig.5).Assuming exhumation-induced cooling and a paleogeothermal gradient of ＞35°C/km(see above),the above cooling rates suggest that exhumation rates slowed down to ＜0.02 mm/year during the Late Cretaceous-Early Miocene,following an earlier phase of relatively rapid Late Cretaceous exhumation (＜0.19 mm/year), then an accelerated exhumation rate of ＜0.06–0.08 mm/year during the Oligocene-Early Miocene to the Middle Miocene,and finally a slow exhumation (＜0.01 m/year) occurred since the Middle Miocene(Fig.5).

6.3 Differential episodic exhumation among various locations

Table 2 shows approximately similar pooled ages of zircon and apatite FT in different sampling locations,suggesting a uniform exhumation for the Yanji area during the Late Cretaceous.The uniform pattern is probably the result of both similar onset time and rate of exhumation at the different locations,namely,the entire exhumation process occurred in the Yanji area.The pooled ages of zircon and apatite FT suggest that these samples once cooled to the closure temperatures of zircon and apatite FT, and exhumed to the equivalent burying depths during(91.7±7.6)–(101.0±9.2)Ma and (70.2±4.0)–(85.4±3.8)Ma,respectively.Using the exhumation rate of ＞0.049–0.073 mm/year)(see above),the exhumation of ＞0.9 km was suggested during (91.7±7.6)–(101.0±9.2)Ma and (70.2±4.0)–(85.4±3.8)Ma(Fig.5).

However,Table 3 shows varied AHe ages in different sampling locations,suggesting a differential cooling and exhumation for the Yanji area during the Cenozoic.

Figure 5 also indicates that a cooling rate of ～0.8°C/Ma and a correspondingly slow exhumation rate of ＜0.02 mm/year,assuming the average paleogeothermal gradient of ＞35°C/km(see above)in the Yanji area between the Late Cretaceous and the Oligocene-Early Miocene.However, the cooling and inferred exhumation rates have become evidently accelerated during the Oligocene-Early Miocene to the Middle Miocene,whose values are ～2–3°C/Ma and ＜0.06–0.08 mm/year, assuming the average paleogeothermal gradient of ＞35°C/km (see above), respectively. The rapid cooling and inferred exhumation process started from the apatite FT partial annealing zone with the maximum exhumation of ～2.2 km between the Late Cretaceous and the Middle Miocene.

The results show that the maximum exhumation is likely ＞3.0 km since the Late Cretaceous,and the current elevation is commonly a few 100 m in the Yanji area,which has been characterized by continental deposits since the Early Mesozoic(JBGMR 1988),thus suggesting the occurrence of negligible surface uplift and total unroofing of at least ＞2 km since the Late Cretaceous.This also implies this exhumation may be dominated by the uplift of rocks concurrent with the erosion of rocks based on the relationship among surface uplift, uplift of rocks,and exhumation of rocks(England and Molnar 1990),and suggests that the topographic frame of the Yanji area might be formed in the Middle Miocene.

The modeling time–temperature results show varied cooling for different samples since the Oligocene-Early Miocene(Fig.5).This was caused by the differential uplift of land mass and low surface erosion in the Yanji area,because the main uplift patterns of the Yanji area were considered to be the thrusting uplift and differential uplift of land mass during the Late Cretaceous-Cenozoic(Ge and Ma 2007).

This study revealed the more details about the cooling and inferred exhumation,for example,differential uplift and denudation between the Oligocene-Early Miocene and the Middle Miocene,and proposed higher paleogeothermal gradient(＞35°C/km),reflecting exhumation of ＞3 km,probably less than of ＞5 km from Li et al.(2010).

In a word,it has probably experienced a sequential cooling,and inferred exhumation in the Yanji area since the Late Cretaceous,involving one rapid(in the Late Cretaceous),one slow(during the Late Cretaceous to the Oligocene-Early Miocene),and one accelerated with varied styles cooling(between the Oligocene-Early Miocene and the Middle Miocene), and a latest extremely slow exhumation processes(Fig.5).

6.4 Implications for the(paleo-)Pacific Plate subduction

The rate of Pacific-Eurasia convergence varied episodically since the Late Cretaceous(Northrup et al.1995).The eastward narrowing of the region of magmatism from the late Early Cretaceous to Paleogene suggests the eastward drift of Eurasian continent and rollback of the Paleo-Pacific subducted slab(Tang et al.2018).Both of them support the episodic cooling rate of the phanerozoic granitoids in the Yanji area since the Late Cretaceous.

The Yanji area and Hida massif in southwestern Japan,by comparative studies on regional geology,were inferred to have experienced similar tectonic evolution history and belong to the same tectonic unit before the opening of the Japan Sea back-arc basin,and they were believed to separate from the East Asian continent during the Late Oligocene-Early Miocene to the Middle Miocene(Shao et al.1991;Jolivet et al.1994;Tang et al.2004;Oh 2006).Regional evolution since the Late Cretaceous was probably related to the subduction of the Farallon-Izanagi and Kula-Pacific ridges toward NE Asian continental margin(Engebratson et al.1985;Northrup et al.1995;Kinoshita 1995;Taria 2001),and the time of Izanagi-Farallon ridge subduction into NE China may be between 106 and 55 Ma(Guo et al.2007;Li et al.2007).

Rifting of the northeastern margin of the Asian continent started at the Late Oligocene-Early Miocene and was followed by development of the Japan Sea basin,separated from the Asian continent between the Late Oligocene-Early Miocene and the Middle Miocene(Nakamura et al.1990;Yamaji 1990;Jolivet et al.1994;Taria 2001;Yang et al.2018).Back-arc extension continued into the Quaternary,associated with the eruption of Cenozoic-Quaternary basaltic lavas in NE China(e.g.Basu et al.1991;Liu et al.2001).The intraplate Changbai volcano is not a hotspot like Hawaii but a kind of back-arc volcano related to the deep subduction and stagnancy of the Pacific slab under Northeast Asia(Lei and Zhao 2005).

Moreover,the upper Cretaceous stratigraphic units such as Dalazi (～100 Ma) and Longjing formations were developed in the Yanji Basin(JBGMR 1988),which was an imaginable depocenter for the rapid exhumation process during the Late Cretaceous,and intense sedimentation in the Cenozoic intermontane basins was potentially in response to the accelerated with varied style exhumation process between the Late Ologocene-Early Miocene and the Middle Miocene. This also implies the rapid and accelerated exhumation was predominantly constrained by the uplift of rocks as well as strong erosion of rocks.Further,these apatite FT track ages are similar to those of the Sulu terrane(centralized in 55–90 Ma)(Grimmer et al.2002).

To summarize the above discussion,in combination with our low-temperature thermochronological data and thermal modeling results, the cooling and exhumation processes of granitoids in the Yanji area can be depicted in the following four stages(Fig.7).

1. During the Late Cretaceous,the first rapid exhumation process and contemporary magmatic activities,as well as thrusting and uplifting, were mainly driven by tectonic uplift and denudation,for the Paleo-Pacific Plate subduction (the subduction of the Farallon-Izanagi and Kula-Pacific ridges)toward the NE Asian continental margin(Fig.7a).

2. During the Late Cretaceous to the Late Ologocene-Early Miocene, the weak exhumation possibly occurred owing to the extensional process in the Yanji area,driven by low regional erosion rates(＜0.02 mm/year)(Fig.7b).

3. During the Late Oligocene-Early Miocene to the Middle Miocene,the Japan Arc was separated from the East Asian continent due to the opening,spreading and closure of the back-arc Japan Sea basin.Simultaneously,the Kula,Farallon and Izanagi plates were diminished because the Pacific Plate was extended and subducted into the Eurasian Plate(Fig.7c).Therefore,it shows a pronounced exhumation process with differential uplift of land mass,driven by both,the regional mantle-driven uplift and erosional denudation.

4. During the Middle Miocene to the present,a nearly quiescent exhumation associated with very low erosion(＜0.01 mm/year)occurred in the Yanji area,driven by the compression mainly by the subduction of the Pacific Plate beneath the Eurasian Plate(Fig.7d).

This episodic exhumation is roughly similar to those of Li et al.(2010).However,this study used more robust evidence to propose higher paleogeothermal gradient(＞35°C/km), reflecting exhumation of ＞3 km in the Yanji area since the Late Cretaceous,probably less than of ＞5 km from Li et al.(2010).

7 Conclusions

The zircon and apatite FT and He ages and thermal history modeling results from this study show that the Yanji area has been experienced a four-stage cooling history spanning from the Late Cretaceous to Cenozoic.The rapid cooling,and inferred exhumation history,during the Late Cretaceous,can be identified from zircon and apatite FT and ZHe ages.The accelerated cooling with varied styles wasfrom modeling time–temperature paths,mainly led by the differential exhumation during the Oligocene-Early Miocene to the Middle Miocene.During the interval between the two pronounced exhumation processes, there was insignificant cooling and weak exhumation driven by low regional erosion rates.Finally,the latest cooling shows a

nearly quiescent exhumation associated with low erosion.We conclude that the two pronounced exhumation processes occurred in response to tectonic uplift and denudation in the Yanji area.Accordingly,this exhumation is possibly linked to the subduction of the Pacific Plate beneath the Eurasian Plate since the Late Cretaceous.Further,this study used more robust evidence to propose higher paleogeothermal gradient(＞35°C/km),reflecting exhumation of ＞3 km in the Yanji area since the Late Cretaceous.

Fig.7 Cartoon for the geotectonic framework and exhumation history of the Yanji area since the Late Cretaceous(see text for further details)

Acknowledgements This work is supported by the DREAM project of MOST China(2016YFC0600406),and the National Natural Science Foundation of China(Grant Nos.41072158,41372227).The authors thank Prof.Kohn BP for help in performing He analyses and discussing on the data at the University of Melbourne,and Dr.Ruohong Jiao for his constructive comments and revision to improve the manuscript greatly.