Pore Structure Petrophysical Characterization of the Upper Cretaceous Oil Shale from the Songliao Basin (NE China) Using Low-Field NMR

Exploration and Development Research Institute of Daqing Oilfield Co Ltd., Daqing 163712, China Heilongjiang Key Laboratory of Tight Oil and Shale Oil Accumulating Research, Daqing 163712, China Institute of New Energy and Low-Carbon Technology, Sichuan University, Chengdu 610065, China Beijing Key Laboratory of Unconventional Natural Gas Geological Evaluation and Development Engineering, China University of Geosciences, Beijing 100083, China


Introduction
Against this background of growing energy demand to support the betterment of mankind, shale oil is treated as a remarkable supplement and plays an increasingly important role in the world's energy portfolio [1][2][3][4]. Following the successful oil extraction from shale in North America, China recently started to explore shale oil resources. Chinese oil shale resources are widely distributed and occur in 47 basins, among which the Songliao basin (NE China) is one of the most important and largest oil fields [5,6]. Furthermore, the shales from Upper Cretaceous Qingshankou (UCQ) and Nenjiang (UCN) formations in the Songliao basin are organic-rich and thus have an excellent hydrocarbon potential for shale oil exploration [7,8].
According to the U.S. Geological Survey Estimation, the average technically recoverable oil shale resources from UCQ and UCN formations in the Songliao basin stands at 3.3 billion barrels [9], allowing these two formations to be hotspots for shale oil development in China and to attract extensive attention in the recent years. Geochemical analyses suggested that UCQ and UCN formations were deposited in a eutrophic and alkaline palaeolake with high salinity and anoxic bottom water conditions during organic matter accumulation [6,10]. Jia et al. [10] also stated that clay minerals, microbial activity, and detrital matter input have notable influences on the enrichment of organic matter in the UCQ and UCN formations. Sequence stratigraphy and geochemistry indicated the UCQ formation has a better hydrocarbon source rock potential and slightly higher oil yield during pyrolysis than UCN formation [11]. For UCQ formation, its clay content averages 55% [12], raising concerns that it might not be amenable to effective hydraulic fracturing. Although UCQ and UCN shales are mentioned a lot, the characterization of their pore structure is rarely presented in recent research.
Pore space determines the storage capacity for shale oil in the free phase, while the connectivity between pores controls the fluid flow in shale [13][14][15].
ese make the advanced characterization on the pore structure be necessary for better development of shale oil with regards to UCQ and UCN formations. Many approaches are capable of characterizing the shale pore structure, including scanning electron magnetic (SEM), helium porosimetry (HP), and transmission electron microscopy (TEM) [16]. Due to the ultrafine grained microfabric and microlevel heterogeneity of shale, the conventional methods above are defective and unable to examine the entire pore network [16]. In order to remedy this defect, low-field NMR technique, as a time-consuming, convenient, and nondestructive method, gradually enters the mainstream for measuring the pore structure by quantifying the interactions of protons and the porous media [17,18]. In this study, low-field NMR methodology is introduced to describe the pore structure of the UCQ and UCN shales, which is beneficial for the shale oil industry in the Songliao basin, especially for the resource estimation and reservoir evaluation. In addition, the approaches of mercury injection porosimetry (MIP) and low-temperature N 2 adsorption/desorption (LTNA) are also employed to verify the accuracy of the NMR-based strategy.

Geological Background
As the largest continental sedimentary basin in northeastern China, the Songliao basin is a typical Mesozoic basin superimposed on the Palaeozoic basement [11]. e basin evolution can be subdivided into prerift doming, synrift subsidence, postrift thermal subsidence, and structural inversion stages. Referring to the caprock features, the Songliao basin includes six first-order structural units ( Figure 1): Northeastern Uplift Zone, Western Slope Zone, Central Depression Zone, Northern Plunge Zone, Southeastern Uplift Zone, and Southwestern Uplift Zone [6,10]. e thickness of the Cretaceous strata in this continental retroarc basin reaches up to 7000 m, where the UCQ and UCN formations are mainly distributed at the Central Depression Zone and were deposited during the postrift phase characterized by strong subsidence [11,19,20]. e sedimentary of UCQ formation occurred from 92.0 to 86.2 Ma. e organicrich member (lowest of UCQ formation) is 60-135 m thick, covering an area of 87 km 2 . e UCN formation was deposited from 84.5 to 79.1 Ma [19,20]. e oil shale (first and second members) of UCN formation has a composite thickness of 200-400 m and covers an area smaller than that of the UCQ formation ( Figure 1) [19]. e UCQ formation has a maximum depth of about 2500 m, while the UCN formation is 0-1900 m deep [21,22]. e UCQ and UCN formations both contain Type I kerogen and are mainly within the oil window with vitrinite reflectance (R o ) values in the basin center exceeding 1.1%, decreasing below 0.5% near the basin margins [12,19]. As a result, in the Songliao basin, UCQ and UCN formations are the vital source rocks for conventional petroleum development, as well as the primary target for shale oil exploration.

Samples.
A total of fourteen oil shale core samples were collected for this study, among which seven samples were from UCQ formation and the rest came from UCN formation (Table 1). e sampling wells are located on the Central Depression Zone of the middle-north Songliao basin ( Figure 1). By X-ray diffraction analysis, minerals in both UCQ and UCN shale samples were found to be mainly quartz and clay minerals, with a supplement of feldspar, carbonate minerals, and pyrite. Compared with the UCQ shale, the UCN shale sample had higher quartz content and lower content of clay minerals, on average ( Figure 2). For the UCQ shale, the R o varies from 0.75% to 1.30% and averages 0.99%, and the total organic carbon (TOC) is in the range 0.71%∼5.43% with a mean of 2.33% ( Table 1). As far as the UCN shale, the R o values have a distribution of 0.60%∼0.75% (average of 0.67%), and the TOC averages 1.36% with a range of 0.79%∼3.78% (Table 1).

Low-Field NMR Measurement
3.2.1. Basic Principle of Low-Field NMR. Generally, NMR signals are usually motivated by the activity of magnetic nuclei (e.g., hydrogen proton) in the magnetic field [23,24]. In this study, the NMR measurements were performed using an analyzer of the MicroMR12-025V type (Shanghai Niumag Corporation, PRC) with a magnetic strength of 0.28 T using a 25.4 mm diameter magnet coil which generates a homogeneous and stable field gradient. e low magnetic field with a frequency of 11.792 MHz makes the NMR signal come from the 1 H-fluid in the pore rather than from the solid skeleton of the oil shale [25,26].
By using low-field NMR, the number of hydrogen atoms present in the CH 4 molecule can be detected through the measurement of transverse relaxation times (T 2 ) [27]. In general, a typical T 2 expression keeps connection with the surface relaxation related to the pore structure, the bulk relaxation of fluid precession, and the diffusion relaxation resulted from the gradient field [28,29]. Accordingly, in a porous media, the compete T 2 relaxation submits to the following equation: 1 where the subscripts B, S, and D represent the bulk, surface, and diffuse relaxations, respectively. Among the above parameters, the diffuse relaxation is generated in a magnetic field gradient. erefore, the homogeneous and stable field gradient enables the ratio 1/T 2D to be small enough to be ignored. Bulk relaxation of a pure liquid fluid is an intrinsic property and relaxes slowly with a usual time of 2-3 s, making T 2B ≫ T 2 and thus the ratio 1/T 2B will also be negligibly small in equation (2) [30]. Consequently, the subsistent T 2 relaxation in this study yields [31,32]   1 where ρ 2 is the surface relaxivity with a constant value of about 0.05 μm/ms for shale based on the work of Sondergeld et al. [33] and S/V is the surface-to-volume ratio (namely, specific surface area). For simplification purpose, the pore geometry of shale is regarded as a cylinder, so S/V equals 2/r [34]. In this situation, equation (2) can be transformed into where r is the shale pore radius (nm) and the conversion parameter (100) has a unit of nm/ms. Equations (2) and (3) indicating a faster T 2 relaxation correspond with a smaller pore size, and vice versa. During the NMR experiments, the main setting parameters, including waiting time of 3500 ms, echo interval of 0.12 ms, echo number of 10000, and scan times of 64, were adopted to enable the measurements to capture the maximum recovery of the polarized NMR T 2 signal and the fast relaxation components.

Experimental Operations.
All oil shale samples were cut into core plugs with a diameter of 2.5 cm and a length of 5 cm. Each sample was detected twice by the NMR instrument under two different pretreatments-hydrocarbon/ water removal (Step I) and hydrocarbon saturated (Step II). e difference between two measurements is adopted to expose the hydrocarbon information in shale pores, enabling the investigation on the pore morphology, porosity, pore size distribution (PSD), and specific surface area.
Step I. Hydrocarbon/water removal: the NMR T 2 signal from residual 1 H-fluid in completely closed pores of the shale.
Firstly, collected samples were placed in an oil-cleaned instrument after being wrapped with the filter paper, under a constant temperature of 90°C (363.15 K). Secondly, hydrocarbon and water in connected pores were extracted by circulating the dichloromethane and acetone vapor for 72 h. en, the treated samples were placed in an NMR detector to record their T 2 spectra at 35°C (308.15 K).
e T 2 measurements were used to explore the information of closed pores in the shale with the assumption that the completely closed pores are 100% filled by the same 1 Hfluid for all samples.
Step II. Hydrocarbon saturated: the NMR T 2 signal from the saturated solution in an effective pore volume of shale.
Firstly, after hydrocarbon/water removal, all shale samples were vacuumed in a vacuum oven for 24 h and then fully saturated with n-dodecane (n-C 12 ) for 48 h under a pressure of 10 MPa. Secondly, these n-C 12 -saturated shale plugs were wrapped with a nonmagnetic film to prevent the evaporation of n-C 12 . en, n-C 12 -saturated samples were conducted by NMR measurements at 35°C (308.15 K).
Subsequently, the T 2 signals from n-C 12 -saturated plugs (Step II) minus those from hydrocarbon/water removed plugs (Step I) are supposed to be the T 2 signals of n-C 12 in the effective pore volume of shale. Accordingly, by employing n-C 12 as a probe, these NMR measurements are the basis in this study to enhance the knowledge of connected pores in the shale.

Accuracy Validation of NMR Measurement.
Currently, there is no universal operation standard for the NMR methodology, in spite that it is described as a sophisticated approach.
us, the accuracy of NMR measurements needs to be validated to enhance its reliability. In this study, the pore structure derived from the NMR approach was verified by LTNA and MIP operations which have sophisticated and universal operation standard. e LTNA and MIP methods have different strengths and weaknesses [35], thus their combination is more reliable in the accuracy validation of the NMR measurement.
LTNA experiments were performed for all collected samples by using a BSD-PS Type surface area and porosity analyzer, following the standard of SY/T 6154-1995 (that is, characterization on specific surface and pore size distribution of rock from N 2 adsorption). For sample preparation, the oil shale samples were grinded and sieved to 60-80 mesh. en, the powders were dried at 150°C (423.15 K) for 3 h in a vacuum oven. During LTNA measurements, the N 2 adsorption isotherms were obtained from recording adsorption and desorption volumes at − 196°C (77.15 K) with the relative pressure (P/P 0 ) ranging from 0.01 to 0.995.
Complying with the standard of GB/T21650.1-2008 (namely, pore size distribution and porosity of solid materials by mercury porosimetry and gas adsorption-Part 1: mercury porosimetry), MIP measurements for all collected samples were conducted by a GT60 Type instrument which has a capacity to recognize pore sizes of 0.0036∼950 μm. Before MIP tests, oil shale samples were crushed and sieved into fine fragments with a diameter of 2∼3 mm. e fragments were transferred into a dilatometer (volume of 1 cm 3 ) after being dried at 110°C (383.15 K). Followed by vacuuming the dilatometer, the intrusion and extrusion curves were obtained from recording the injected and ejected volumes of mercury. In this study, the maximum pressure of mercury injection reaches to as high as 200 MPa (i.e., ∼29000 psi). e relationship between pore size and MIP pressure is reported by Fangwen et al. [35] and Seiphoori et al. [36].

Results and Discussion
According to the series of experiments, this section will discuss how NMR relaxation (T 2 spectra) can be used as an independent tool for classifying the shale pore types and morphology, calculating porosity, and evaluating pore size distribution (PSD) and specific surface area. Considering that the petrophysical properties are of significance for shale oil exploration and exploitation, this section is supposed to be a great theoretical implication for the shale oil industry in UCQ and UCN shale formations of Songliao basin, NE China. Note that this section works with the precondition that the in-site oil (in fact, multifluid mixtures) is simulated by pure n-C 12 , and with the overlook of the extremely small pores cannot be filled by n-C 12 .

NMR T 2 Spectra.
e NMR T 2 spectra from a sample with different pretreatments have obviously variable characteristics. Under the circumstance of hydrocarbon/water removal, NMR T 2 spectra exhibit small amplitude and thus indicate that the residual 1 H-fluid in completely closed pores of shale is scarce (Figure 3). Comparatively, distinct amplitude emerges to the samples saturated with n-C 12 , suggesting that considerable amount of n-C 12 is injected into effective pores in the oil shale to enable the T 2 amplitude to be raised significantly ( Figure 3). By subtraction processing, the T 2 spectra motivated by the n-C 12 in effective pores are plotted (Figures 3 and 4), which is actually adopted for the investigation on the pore structure of the collected oil shale samples in this work.
It is intuitively plausible that three kinds of T 2 spectra are obtained by NMR experiments: unimodal distribution (e.g., sample D05), bimodal distribution (e.g., sample Y40), and trimodal distribution (e.g., sample F34) (Figures 3 and 4). For all samples, the NMR spectra have a dominant peak at a T 2 value of about 1∼10 ms. Basically, the measured NMR spectra exhibit a diversity morphology and a wide T 2 value of the dominant peak, which notes that the pore structure of the Upper Cretaceous oil shale is complicated. is is because that the T 2 spectra with diverse characteristics usually represent different pore structures.
Furthermore, the characteristic of NMR T 2 spectra for the UCQ shale differs from that for the UCN shale. For example, more unimodal T 2 spectra emerge to the UCN shale rather than the UCQ shale, according to the NMR measurements (Figures 3 and 4). is phenomenon intuitively reveals the difference of pore structure in UCQ and UCN shale samples, further enabling the significance of petrophysical characterization in this study.

Pore Type and Morphology.
Because of the sedimentary, diagenesis, and tectonism, shale is usually characterized as strong heterogeneity and thus contains various pore types [37,38]. As verified by Yao et al. [25], NMR T 2 amplitude is able to indicate the pore type and morphology. In this study, e NMR T 2 distributions are commonly unimodal and bimodal and sometimes trimodal (Figures 3 and 4). Referring to the criterion created by Yao et al. [25], a narrow unimodal T 2 distribution represents a single pore type (e.g., samples G80 and F63), whereas a wide one suggests multiple pore types (e.g., sample G57) (Figure 4). For multiple NMR T 2 peaks (bimodal and trimodal), the connection among these peaks can be used to identify the connectivity among pores [25]. For example, the wellconnected bimodal/trimodal T 2 distributions (e.g., samples G24 and G82) suggest that well-connective multipores exist in these oil shale samples (Figure 4). By contrast, the less well-connected bimodal/trimodal T 2 distributions (e.g., samples Y39 and T49) may indicate that pores in different size are disconnected in these samples. In addition, as per the achievements by Li et al. [34], adsorption pores commonly exist in the samples with unimodal, bimodal, or trimodal T 2 distributions, while seepage pores are mainly developed in the samples with bimodal or trimodal T 2 distributions.
Both UCQ and UCN shale samples had pores poorly connected because well-connected bimodal/trimodal T 2 spectra are less frequent in study samples (Figure 4). Nevertheless, UCQ and UCN shale samples, by contrast, have some different pore types, on the basis of the NMR T 2 spectra. It seems that the sample with a single pore type and adsorption pore is more common for the UCN shale since there are more unimodal T 2 spectra with narrow distribution (Figure 4). Furthermore, LTNA is employed to analyze the correctness of NMR T 2 distribution in describing the pore type and morphology. According to the International Union of Pure and Applied Chemistry (IUPAC) classification, the N 2 adsorption/desorption curve would be described as the Type IV isotherm when a noticeable hysteresis loop exists, which is associated with the limiting uptake over a range of high P/P o [39]. Accordingly, the N 2 adsorption/desorption curves of the two representative samples are of Type IV ( Figure 5). Furthermore, Sing et al. [39] noted that hysteresis appearing in the multilayer range of physisorption isotherms is usually associated with capillary condensation in mesopore structures, where four specific pore structures are identified according to the shape of the hysteresis loops. By this classification, the pore type of sample Y39 belongs to H2 (inkbottle-shaped pore)-a poorly connective pore type, while that of sample G82 primarily is H3 (plate-like pore)-a well-connective pore type ( Figure 5). e LTNA results illustrate that the NMR T 2 measurement is reliable in characterizing the pore type of oil shale.

Effective Porosity.
Pores and microfractures provide storage, migration, and seepage channels for oil in shale and thus are of significance for shale oil extraction. Porosity is an extensively used index to evaluate the pore volume of inner pores/fractures in oil shale. Previous studies noted that the signal intensity of the hydrogen-containing fluid (n-C 12 in this study) in samples is able to investigate the NMR porosity [40][41][42]. Accordingly, a series of NMR measurements were, respectively, conducted on five standard n-C 12 samples (0.2, 0.4, 0.6, 0.8, and 1.0 mL). As shown in Figure 6, in this study, there is a clear linear relationship between amplitude and n-C 12 volume: where V n− C 12 is the n-C 12 volume (unit: mL) and T 2 is the NMR amplitude motivated by n-C 12 in pores (10 3 a.u.). According to the NMR T 2 spectra motivated from n-C 12 in effective pores (Figures 3 and 4), the effective porosities of collected samples are obtained and plotted in Figure 7, referring to equation (4). In order to clarify the accuracy of NMR-based porosity, gravimetric measurements were employed by weighing the n-C 12 quality in effective pores of oil shale. e porosity values from NMR and gravimetric approaches present a strong resemblance (Figure 7),  signaling the NMR strategy is reliable in porosity measurement of oil shale. Results exhibit the porosities of collected samples have a range of 2.3%∼12.5% with an average of 7.3%, according to the NMR measurements (Figure 7). e porosity is 7.57% in average for the UCQ shale with a wide range of 2.6%∼12.2%, while that averages 6.98% for the UCN shale and varies from 4.5% to 11.6%. e wide range of measured porosity suggests the strong heterogeneity of collected samples, suggesting the nonuniform horizontal distribution of shale oil in UCQ formation or UCN formation.
us, more attentions should be paid to the selection of area with greater porosity in the future, aiming to guide the shale oil extraction from the UCQ and UCN shale reservoirs.

4.4.
Pore Size Distribution. n-C 12 in different pore sizes shows different relaxation velocities due to the difference of relaxation mechanism and relaxation velocity during NMR measurements. erefore, the distribution characteristics of the NMR T 2 spectrum have the ability to indirectly reflect the pore size distribution and fluid distribution, according to equation (3): larger pores correspond to longer relaxation time and smaller pores to shorter relaxation time. Taking sample Y40 as a typical example, the PSD derivations from different methods take an attitude that the NMR measurement covers a broader PSD than the LTNA or MIP method ( Figure 8). In addition, the NMR-based PSD is regarded as reliable as it is very close to the combined result of LTNA and MIP approaches (Figure 8). Regarding to sample Y40, the pore size varies considerably with a span from 1 nm to 10000 nm, while the pores with diameters of ∼100 nm act as the major contributors for total porosity ( Figure 8).
Furthermore, the relationships of dV/dD pore volume vs. pore width for all samples are plotted in Figure 9, on the basis of NMR experiments. For both UCQ and UCN formations, there exist two types of PSD curves in Figure 9-peaked at ∼10 nm (Type I) and peaked at ∼100 nm (Type II). Around 30% of the collected samples hold the Type I PSD curve (Figure 9), where pores with diameters of 1 ∼100 nm occupy a dominant proportion among all inner pores of oil shale. PSD curves of major samples (∼70%) belong to Type II (Figure 9), with the vast majority of pores being 10∼1000 nm in size. Considering the polarization of Type I and Type II, more attentions are expected to be drawn to the mechanism of this phenomenon.   with strong interaction between pore surface and 1 H-fluid (e.g., H 2 O and n-C 12 ). Hence, the interaction between n-C 12 molecules and pore surface can be characterized by T 2 relaxation in this study. Combining equation (2) with equation (3), the S/V of oil shale yields As per equation (5), the S/V distribution of samples can be obtained. Setting sample Y38 and J100 as representative samples, the S/V distributions are plotted in Figure 10. Basically, shorter T 2 indicates smaller pore size and greater S/V value. On the contrary, the longer the T 2 is, the larger the pores in oil shale and the lower the value of S/V is ( Figure 10). Regarding to sample Y38, the S/V ranges from 0.1 nm − 1 to 200 nm − 1 and mainly centered at ∼2 nm − 1 . Comparatively, sample J100 has a broader S/V range of 0.08 nm − 1 ∼ 1000 nm − 1 and mainly peaked at ∼20 nm − 1 . Similar with samples Y38 and J100, the S/V distributions of the rest of the samples are obtained by coordinate transformation of Figures 3 and 4, according to equation (5). is phenomenon indicates that the S/V distribution is able to be clarified as two patterns, that is, peaked at ∼2 nm − 1 (Pattern A) and peaked at ∼20 nm − 1 (Pattern B), based on Figure 9 and equation (5). As for the sample with a Pattern A S/V distribution, the S/V is mainly contributed by the pores with a diameter of 10-200 nm. Comparatively, if the S/V distribution is described as Pattern B, the S/V is primarily donated by bigger pores (100-2000 nm).
Basically, in shale samples, pores with smaller S/V benefit to adsorb fluid, while pores with bigger S/V are helpful for the storage-free fluid [38]. us, the NMR-based approach for the S/V measurement is conducive to filter the favorable target for shale oil exploration because free oil (rather than adsorbed oil) is potentially the most producible component of tight shale reservoirs [43].

Outlook of NMR-Based Approach for the Shale Oil Development in Songliao
Basin. In this study, the NMR-based methodology exposed a preliminary understanding on the pore structure of oil shale from the Songliao basin, in spite that only fourteen core samples were collected. Compared with the petrophysical characterization using the conventional technique like LTNA, the NMR-based method with the n-C 12 probe is probably more suitable and reliable to the study on the shale oil reservoir. is is because the N 2 molecule has a smaller diameter than the n-C 12 molecule and thus tends to probe more pores with extremely small size which cannot store oil. at is to say, the NMR measurement is a more targeted approach to focus on the effective pores for oil storage in shale and thus provide more real scientific investigations.

Conclusions
(1) e NMR spectra of all samples have a dominant peak at a T 2 value of about 1∼10 ms. e narrow and wide unimodal T 2 distribution represents a single and multiple pore type, respectively. e well-connected bimodal/trimodal T 2 distributions suggest that pores in different size have good connectivity, and vice versa. By contrast, the UCN shale has more single pore type and adsorption pores than the UCQ shale. (2) As verified by gravimetric measurements, the NMR strategy is reliable in determining the porosity of oil shale. According to the NMR measurements, the porosities of collected samples vary from 2.3% to 12.5% with an average of 7.3%. e wide range of measured porosity suggests that the strong heterogeneity emerges to both UCN and UCQ shale reservoirs from Songliao basin. (3) As for both UCN and UCQ samples, PSD curves are intuitively clarified as two types-peaked at ∼10 nm and peaked at ∼100 nm, while S/V distributions are also categorized into two patterns-peaked at ∼2 nm − 1 and peaked at ∼20 nm − 1 . However, more attempts are needed to explain the polarization of PSD and S/V distribution.

Data Availability
e data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
e authors declare that they have no known competing financial interests or personal relationships that could influence the work reported in this paper. Journal of Spectroscopy 9