A novel multidisciplinary approach for crude oil classification based on structural characteristics of asphaltenes Authors: Hashem Sarafdokht, Ahmad Reza Rabbani, Morteza Asemani Published: 2025 ABSTRACT AbstractCrude oil is a complex mixture of various chemical components, and each crude oil may behave differently during production, transportation, storage, and refining. The typical approach to treating crude oils produced from different wells or oil fields is to classify them into distinct groups and treat each group similarly. However, traditional methods of crude oil classification focus on particular characteristics and ignore others. Hence, a multidisciplinary approach is necessary to benefit both upstream and downstream industries. Employing the structural characteristics of asphaltenes to classify crude oils is the most effective solution. This study introduces an innovative chemometric method for the first time, integrating the Modified Moving Window Correlation Coefficient (MMWCC), Multidimensional Scaling (MDS), and Hierarchical Cluster Analysis (HCA) techniques. This approach enabled a robust and comprehensive classification of 93 crude oil samples from the Asmari reservoir, spanning 21 different oil fields in the Dezful Embayment, Southwest Iran. By analyzing the FTIR spectra of asphaltenes, a dissimilarity matrix was constructed using the MMWCC method. Subsequently, MDS was employed to visualize these dissimilarities in a 2D and 3D space. Finally, HCA was utilized to cluster the samples into distinct groups based on Euclidean distance in the 3D space. This approach optimizes the clustering threshold by considering the spatial proximity of samples. The proposed method successfully classified the Asmari reservoir crude oils into nine distinct groups, providing insights into their asphaltene origins. Complementary analyses, including XRD and ICP-OES, further validated the defined oil groups. This novel approach offers an efficient, straightforward, cost-effective, and multidisciplinary solution for crude oil classification, providing valuable information for both upstream and downstream industries.Keywords: FTIR spectroscopy, Modified moving window correlation coefficient (MMWCC), Crude oil classification, Asmari reservoir, Dezful embayment, IranSubject terms: Chemistry, Energy science and technology, Engineering, Mathematics and computing Introduction Crude oils are naturally composed of a complex mixture of many organic compounds, and their physicochemical properties influence their production, transportation, storage, and refinery1,2. To simplify the complexity of different crude oils, they are often grouped into distinct categories based on similar characteristics. This classification approach allows for similar treatment of oils within each group. Given the numerous benefits of crude oil classification, various methods have been developed and applied in the literature3–5. Traditionally, organic geochemistry methods are standard techniques for classifying crude oils based on their origin and source rock characteristics5,6. The organic geochemical methods mainly rely on source-specific markers, including saturate biomarkers and aromatic hydrocarbons, extracted from the light-end fraction of crude oils to provide information about the source rock characteristics, depositional environment, geological history, and organic matter input7,8. Despite the intrinsic benefits of these conventional classification methods, they have certain limitations. For example, secondary alteration processes (e.g., biodegradation, water washing, and evaporative fractionation) can obscure molecular signatures, while mixed or altered oils are difficult to classify reliably. In addition, the complex sample preparation and potential contamination during extraction may further compromise analytical accuracy. Consequently, there has been a growing demand for rapid, low-cost, and less preparation-dependent techniques, prompting the application of spectroscopic methods such as Fourier-Transform Infrared (FTIR) and Attenuated Total Reflection FTIR (ATR-FTIR)9,10. These approaches have proven powerful for the characterization and classification of crude oils11. It has been demonstrated that these techniques can yield reliable and precise data on asphaltene structure and its functional groups12,13. Asphaltenes are the heaviest, most complex fraction of crude oil, operationally defined as being soluble in toluene but insoluble in light n-alkanes14,15. They are often depicted as polycyclic aromatic structures with alkyl side chains containing heteroatoms (i.e., N, S, and O) and trace elements (e.g., Ni and V). In geochemistry, it is believed that these noxious components have a close structural similarity to the precursor kerogen of reservoir oil4,16–18. Additionally, they are a primary cause of various problems in refinery processes, and their structural characteristics are crucial for optimizing refinery operations. FTIR spectroscopy offers a straightforward, rapid, cost-effective, and reliable method for investigating structural characteristics of asphaltenes. Although the FTIR spectra provide valuable structural characteristics, advanced and efficient chemometric tools are required to extract subtle structural differences among samples. Lamontagne et al.19 defined several structural indices (e.g., aromaticity index, aliphatic index, carbonyl index, and sulphoxide index) from FTIR spectra to compare road bitumens. These indices were also used in several papers for various studies, including reservoir compartmentalization and oil-oil correlation4,20–24. Routinely, bivariate cross plots are used to compare different samples based on these FTIR indices. Overlapping bands and a complex mixture of asphaltenes make functional group index analysis challenging to reveal subtle differences between highly similar samples, thereby hindering the efficient classification and correlation of crude oils. The moving window correlation coefficient has been established as an efficient method for constructing a similarity matrix, particularly for highly similar FTIR spectroscopy data25. Despite the many inherent benefits of the moving window correlation coefficient technique, it suffers from several disadvantages (e.g., baseline variation, band shifting, and sensitivity to noise), which were addressed later by Asemani et al.1and called as modified moving window correlation coefficient. Asemani et al.1,7, also combined the modified moving window correlation coefficient (MMWCC) with hierarchical cluster analysis (HCA). This method revealed subtle differences between the various samples and efficiently classified them. The efficiency of this method compared to the traditional method was calculated and found to be significantly increased. Nevertheless, a persistent limitation of the conventional HCA approach lies in defining the cut-off threshold for determining the number of clusters. This step often requires complementary analyses or subjective judgment, reducing reproducibility and consistency across datasets. In this study, we address this long-standing challenge by introducing an integrated algorithm that efficiently identifies the optimal cut-off point within the HCA dendrogram based on the spatial proximity of samples. This innovation eliminates the ambiguity of threshold selection and enhances the robustness and objectivity of clustering, without the need for other supplementary data. The primary objective of this study is to develop a comprehensive and robust chemometric classification method based on the structural characteristics of asphaltenes. To achieve this, the structural characteristics of asphaltenes from 93 different crude oils collected from 21 oil fields in the Asmari reservoir throughout the Dezful Embayment were investigated using FTIR spectroscopy. The modified moving window correlation coefficient technique (MMWCC) was employed to identify subtle differences between pairs of samples and generate a dissimilarity matrix. Subsequently, multidimensional scaling (MDS) was utilized to visualize these dissimilarities as points in 2D and 3D Euclidean space. Subsequently, HCA was employed for sample classification, while the proposed algorithm for cut-off detection was implemented to define clusters objectively. Finally, the classification results were validated by complementary XRD and ICP-OES analyses performed on representative samples from each group. Geological setting The oblique collision of Arabia with the Eurasian Plate has significantly influenced the tectonostratigraphic evolution and hydrocarbon enrichment within the Zagros Fold and Thrust Belt (ZFTB). The convergence started at the beginning of the Late Cretaceous26 and intensified during the Late Miocene and Pliocene27. The ZFTB has been structurally divided into three regions: Lurestan, Khuzestan (including Izeh and Dezful Embayment), and Fars, separated by north-south and east-west trending fault zones (Fig. 1)28. The Balarud Flexure bounds the Dezful Embayment to the northwest, the Izeh Fault Zone to the southeast, the mountain front fault to the north, and the Zagros Front Fault to the south (Fig. 1)29. Due to its enormous petroleum significance, this region has been extensively studied from various perspectives, including structural geology30–32, stratigraphy33–35, and petroleum systems28,36–39. The Dezful Embayment, situated in the Zagros Fold-Thrust Belt (ZFTB) of southwest Iran, is a globally significant petroliferous region that hosts approximately 8% of the world’s oil reserves in more than 50 different oil fields36,39. The Asmari carbonate reservoir is the primary producing reservoir in the Dezful Embayment, accounting for over 75% of the region’s crude oil production. It is primarily charged by the Aptian- Albian Kazhdumi source rock, with minor contributions from the Palaeocene-Oligocene Pabdeh Formation36,37,40,41. The complex charge history, involving multiple source rocks and migration pathways, presents significant challenges in understanding the origin and properties of the crude oils in the Dezful Embayment24. Structural map of the Zagros region showing the distribution of the studied oil fields in the Dezful Embayment. The map is generated by authors using CorelDRAW Graphics Suite software 2025 (version 26.2.0.170; Corel Corporation; https://www.CorelDRAW.com). The Kazhdumi Formation is a mature, organic-rich unit that exhibits shaley facies in the central parts of the Dezful Embayment, transitioning to carbonate facies towards the basin margins42,43. The Pabdeh Formation, deposited throughout the Dezful Embayment, is characterized by relatively organic-rich, immature, and marly facies (Fig. 2). While the formation has entered the oil window in the northeastern parts of the embayment, it remains immature in other areas36,41,44. Accordingly, the Kazhdumi source rock has made a significant contribution to the charging of most oil fields in the Dezful Embayment. While the Pabdeh source rock has also contributed, its impact is more limited41. Other formations have made minor, localized contributions to the charge of the Asmari and Bangestan reservoirs. Different source rocks led to the emergence of complex oil plays in the Dezful Embayment, resulting in the development of diverse oil families43–45. The stratigraphic column of different formations in the Dezful Embayment, Zagros basin46. Materials and methods Samples Ninety-three crude oil samples were collected from the Asmari reservoir in 21 different fields across the Dezful Embayment (Table 1). Asphaltene extraction and FTIR spectroscopy were performed on all samples, followed by complementary analyses, including XRD and ICP-OES on selected samples. Table 1:The list of samples and Asphaltene content.NOFieldsSamplesAsphaltene (wt.%)API (°)NOFieldsSamplesAsphaltene(wt.%)API (°)NOFieldsSamplesAsphaltene(wt.%)API (°)1AhwazAZ-952.9829.0632Qale NarQR-76.0328.4563MarunMN-2270.6331.092AZ-1190.531.4833QR-94.4328.9264MN-3872.1928.773AZ-1424.6329.6734QR-104.4228.8565ShadeganSG-24.3228.954AZ-1470.5729.8535QR-145.8128.6066SG-262.98-5AZ-1487.9830.1336**QR-175.3528.5367**SG-276.527.786*AZ-1500.9729.3737Balarud*BR-41.5233.9868AghajariAJ-841.3830.687AZ-1683.1929.0838BR-51.3234.0569AJ-971.35-8*AZ-2552.1428.7839HaftkelHK-321.2137.7270AJ-1041.3930.479AZ-3491.3830.3940HK-481.337.0571AJ-1081.431.5010GachsaranGS-33.3831.0141HK-502.0736.7872AJ-1101.4630.2211GS-2293.4830.9642HK-641.2737.5273AJ-1231.0932.0712GS-2323.3431.8743HK-671.4836.9574AJ-1851.4730.1513GS-2443.1939.9544HK-711.2537.6575Rag-e SefidRS-136.6125.3714Gulkhari*GL-47.1428.4845HK-741.3337.1676**RS-516.0927.0615GL-127.4227.9546Lab-e SefidLS-11.635.1077RS-626.9126.5716GL-1543.2129.3447LS-123.0434.6778PazananPZ-1211.4435.0017KoupalKL-100.729.1448KarunKN-22.1332.2779PZ-1292.0734.7018KL-170.8827.3549KN-32.7232.1680RamshirRR-93.5525.7219KL-291.0125.4550KN-72.1632.2581RR-112.3825.1820KL-360.6129.4851PalanganPL-11.2530.9382RR-175.6323.9321MansuriMI-126.4926.4552MasjidSoleymanMIS-280.8238.0883RR-211.7626.1822MI-484.09-53MIS-1621.8437.3384MansurabadMB-16.3227.1023**MI-585.6125.0254MIS-2620.6138.2285MB-25.8328.5824MI-869.1725.3255MIS-2960.7738.2086MB-36.2327.1625MI-1135.2927.3656Naft-e SefidNS-101.9634.1287MB-46.2227.1326NargesiNI-10.7435.7957NS-170.3536.6388MB-65.1628.6727NI-21.234.6558NS-180.5936.5489MB-85.3928.3328**NI-30.9635.0859NS-451.1835.3790MB-126.0227.5629NI-51.3533.3460MarunMN-241.3830.3391MB-135.9528.5430NI-71.4734.0761MN-1902.7330.2392MB-144.4429.0331NI-111.6533.2762MN-2000.4931.6493MB-155.7928.04Selected sample for replication analysis, ** Selected samples for complementary analysis. The list of samples and Asphaltene content. Masjid Soleyman Selected sample for replication analysis, ** Selected samples for complementary analysis. Asphaltene extraction The extraction of asphaltenes from the studied crude oils was conducted according to the D6560-17 standard47. The details of this analysis can be found elsewhere48. Briefly, the n-hexane at the ratio of 25:1 (ml: g) was added to the crude oil and refluxed for 1h. The mixture was stored in a dark place overnight before being filtered. The filtrate was washed with n-hexane to remove impurities, and asphaltenes were subsequently recovered using toluene. The solvent was evaporated, and the asphaltenes were dried at 100 °C for 30 minutes. To assess repeatability and establish classification thresholds, asphaltene extraction was performed multiple times on four distinct samples. FTIR spectroscopy The extracted asphaltenes were prepared as KBr pellets by mixing asphaltenes and KBr powder at a 1:100 weight ratio and pressing under 10 tons of force. The prepared KBr pellets were dried in an oven at 80 °C for 8 hours before FTIR spectroscopy analysis. FTIR spectra were recorded using a Bruker-Tensor 27, in the range of 4000 to 400 cm−1. Each spectrum was obtained by averaging 32 scans at a resolution of 4 cm⁻1. Three spectra were recorded for each sample, and the average spectrum was used for subsequent analysis. Complementary analyses An Inel EQUINOX 3000 diffractometer was used to record the X-ray diffraction (XRD) patterns of the extracted asphaltenes using Cu Kα radiation. The asphaltenes powder was scanned over a 2θ range of 10° to 80° with a step size of 0.02° and a scan rate of 2°/min. The XRD patterns were curve resolved using the Gaussian method before calculating the structural parameters. For measuring the concentrations of trace metals, the crude oil samples were prepared as aqueous solutions through acid digestion. The prepared samples were analyzed using a PerkinElmer Sciex Elan 6100 Inductively Coupled Plasma-Mass Spectrometry (ICP-MS) system to determine the concentrations of nickel (Ni) and vanadium (V). Chemometrics Chemometrics is a set of mathematical tools and statistical approaches for optimally analyzing large sets of data. Multivariate chemometric methods can be broadly categorized into two types: Exploratory or data mining (unsupervised) methods: These methods are used to identify trends or clusters within data without any prior knowledge or labels. They treat the data objectively, allowing patterns to emerge from the data itself. Supervised methods: These methods utilize labeled data to train models that can classify or predict outcomes for new, unseen data. Specific objectives guide them and aim to make informed decisions based on the learned patterns. A key characteristic of chemometrics is its ability to discriminate and classify large sets of analytical data into distinct groups. This capability is crucial for various applications such as identifying different types of samples, monitoring process variables, and predicting product quality49,50. The FTIR spectra provide valuable insights into the molecular characteristics and chemical structure of the asphaltene fractions of crude oils1. Asemani et al.7demonstrated the effectiveness of the modified moving window correlation coefficient (MMWCC) for pairwise comparison of crude oil fingerprints. This technique has higher efficiency than other methods to reveal subtle changes between samples. Briefly, the selected characteristic peaks in the second derivative FTIR spectra are chosen, and peak ratios are calculated by dividing the intensity of each peak by the other peaks. Then, the MMWCC method is applied to the calculated peak ratios for the pairwise comparison of samples. The MMWCC calculates the Pearson correlation coefficient in successive windows, rather than across the entire dataset. After each calculation, the window advances one point, and a new correlation coefficient is calculated. Hence, a series of correlation coefficients is calculated, and detailed information about the similarity or dissimilarity of pair samples is provided. Multidimensional scaling (MDS) is a well-established exploratory data analysis technique that visually represents the similarity or dissimilarity between pairs of samples as distances in a multidimensional space. This method effectively transforms correlation coefficients into a distance-based representation, allowing for the visualization of sample proximities. The MDS is beneficial when dealing with a large number of samples, where traditional correlation analysis may become cumbersome. By projecting samples onto a multidimensional space, MDS enables visual inspection and exploration of relationships between samples. Samples with a high correlation coefficient are placed at a closer Euclidean distance, while those with a low correlation are placed farther apart. Hierarchical cluster analysis (HCA) is a known technique for clustering samples based on their similarities or dissimilarities. This technique employs a non-iterative hierarchical clustering strategy, where each sample is considered a class. So, the distance between different classes is calculated, and the sample pairs with the smaller distance are merged into a new class. Different similarity coefficient measurements, such as Euclidean distance and correlation coefficient, are commonly used to measure the distance between classes. The distance between this new class and other classes is calculated, and classes with small distances are merged into a new class. This procedure is continued until all samples are clustered. The clustering results are presented in a tree-like representation known as a dendrogram7. The proposed chemometrics technique for classification of samples This paper presents a novel algorithm that combines MMWCC, MDS, and HCA methods to classify the FTIR spectra of asphaltenes separated from crude oil samples. The flowchart of the proposed algorithm is illustrated in Fig. 3. Firstly, the peak ratios are calculated based on the intensity of the characteristic peaks from the second derivative spectra of asphaltenes. The outliers of each peak were identified in the studied batch using the Interquartile Range (IQR) technique. If the outlier samples are associated with analytical errors, they will be removed from subsequent statistical processing. Outliers not related to analytical errors are replaced by the upper or lower extreme values52. The MMWCC technique is applied to peak ratios, and the minimum correlation coefficient in all moving windows (MMWR) is calculated for pairs of samples. The width of the moving window is calculated based on the repeatability of the samples, as mentioned by Asemani et al.1. Then, the dissimilarity matrix is constructed based on the value of 1-MMWR. Subsequently, the dissimilarity between pairs of samples is converted to a 3D Euclidean space for a visual representation of the samples’ relationship. Finally, the distances between samples are used to classify samples into different groups using the HCA technique. The proposed algorithm for classifying crude oils based on structural characteristics of asphaltenes. Results and discussion Bulk characteristics of the studied oil samples API gravity of studied oil samples ranges from 23.93° to 39.95°, with an average of 31.54°. The API gravity value has a direct relationship with the asphaltene content. The asphaltene content of the 93 oil samples from the Asmari reservoir ranges from 0.35% (i.e., NS-17) to 9.17% (i.e., MI-86), with an average of 3.02% (Table 1). The lowest average asphaltene content belongs to the Kupal field (i.e., 0.8%, 4 samples), and the highest average asphaltene content relates to the Reg-e Sefid field (i.e., 6.54 %, 3 samples). The relative amount of different fractions in crude oil can be influenced by various processes, including the type of precursor kerogen, thermal maturity, migration distance, secondary processes (e.g., biodegradation, water washing, and evaporative fractionation), and oil-rock interaction8. For example, high-maturity crude oils usually have relatively lower asphaltene contents. On the other hand, the relative concentration of the polar fraction of crude oils (i.e., resins and asphaltenes) usually decreases with increasing migration distances because these compounds form hydrogen bonds with the minerals along the migration pathways. Furthermore, physical factors also influence the asphaltene content of crude oils. In the reservoir that has reached the thermodynamic equilibrium, gravity segregation leads to the enrichment of polar compounds, including asphaltenes and resins, at the bottom of the oil column. Consequently, wells producing from the lower parts of the oil column are expected to yield crude oil with higher asphaltene contents compared to those producing from the upper parts51. Since different factors affect the concentration of various crude oil fractions, the asphaltene content of wells from the same oilfield or those from different oil fields should be compared cautiously. In other words, differences in the asphaltene content alone may not be a reliable indicator of variations in crude oil type or source rock characteristics. FTIR assignments The raw FTIR spectra of the studied asphaltene samples are depicted in Fig. 4. These patterns are more or less similar, meaning that all asphaltene samples are characterized by identical functional groups in their chemical structures. However, minor differences in samples are associated with the relative concentration of functional groups that cause peak height differences in FTIR bands. The detailed assignment of FTIR spectra of asphaltenes can be found elsewhere53. Generally, a broad band greater than 3100 cm−1 is related to the O–H and N–H stretching vibrations. This band can occur due to moisture absorbed by the KBr pellets and, therefore, is not a characteristic peak for comparing different asphaltene samples. The aromatic C-H stretching becomes evident as a small shoulder peak, detectable at approximately 3050-3060 cm−154. This peak is resolved as two characteristic peaks at 3052 and 3014 cm−1 in the second derivative spectrum55. The 3000-2750 cm−1 region is assigned to the C–H stretching of aliphatic functional groups (i.e., methyl, methylene, and methine)55. The ring breathing mode of aromatics (C=C stretching) gives rise to a band at approximately 1600 cm−1. The characteristic band at 1460 cm−1 can be assigned to either CH3 asymmetric bending, CH2 scissors, or both, and the 1376 cm−1 band is attributed to the CH3 symmetric bending vibration. The sulfoxide functional group appears at approximately 1030 cm−1; however, this band is not a characteristic peak for comparing different asphaltenes. The number of substitutions in benzene is recognized by the C-H out-of-plane bending vibration of aromatic rings that appears in the 900-700 cm−1 region. The small shoulder band at about 723 cm−1 is attributed to methylene rocking vibration in straight-chain alkanes containing more than four carbons53–56. This peak shows variations between replicate samples and is not considered characteristic for comparing different asphaltene samples in this study. The FTIR spectra of all studied asphaltene samples from the Asmari reservoir in the Dezful Embayment. Classification of crude oils The genetic classification of crude oils based on the structural characteristics of asphaltenes can provide valuable insights into their habitat and migration pathways. This information can aid in optimizing treatment processes for crude oils with diverse physicochemical properties. In this study, a novel algorithm was applied to classify the 93 studied crude oils from the Asmari reservoir of the Dezful Embayment fields. For crude oil classification, eleven characteristic peaks (3052, 3014, 2958, 2923, 2850, 1600, 1460, 1375, 862, 808, and 746 cm⁻1) were used to calculate 110 peak ratios. Pairwise comparisons of samples were made based on these peak ratios. Asphaltene separation procedures were performed on multiple samples (AZ-255, AZ-150, GL-4, and BD-4) at different times to assess repeatability (Table 1). The optimal moving window width was determined to be 11 points based on replicate sample analysis, as described elsewhere1. A dissimilarity matrix was constructed using the minimum moving window correlation (MMWR) for replicate samples. The resulting 2D and 3D scatter plots from MDS analysis (Fig. 5) demonstrate the proximity of replicate samples, validating the results. While visual differentiation of oil types is feasible in 3D plots for smaller datasets, it becomes challenging as the number of samples increases. Using the HCA method, the Euclidean distance between different samples in the 2D and 3D plots was used to draw the dendrogram. One of the primary challenges associated with the HCA method is determining the threshold value. By combining MDS and HCA, a suitable threshold can be identified. The optimum threshold is that the samples in each cluster should have an appropriate spatial distribution in the Euclidean space plot. The optimum threshold was determined from replicate samples, as these samples should be placed close in Euclidean space. The illustration of the similarity or dissimilarity of different replicate samples as points in 3D (a) and 2D (b) space by the MDS method. The dendrogram of replicate samples is shown in Fig. 6. The threshold of the HCA method should be determined so that the replicate samples are classified in the same cluster. Hence, the best threshold for the studied samples would be between 0.32 and 0.61. Therefore, these thresholds were applied to classify the studied crude oils. Different thresholds were used, and the proximities of samples were also checked in 2D and 3D scatter plots. The MDS and dendrogram results should be compatible to determine the optimum threshold. Different thresholds, such as 0.31, 0.49, and 0.61, were checked. The optimum threshold for the studied samples was 0.61 based on the appropriate spatial distribution of different samples for each cluster in the 2D and 3D scatter plots (Figs. 7 and 8). According to these results, the Asmari samples from the Dezful Embayment were classified into nine distinct families (AS-1 to AS-9) (Fig. 8). The geographic distribution of these oil families is also logical from the regional geology perspective. Differences in the chemical structure of the asphaltenes can be used to identify genetically distinct oil families in the Dezful Embayment. These oil families are derived from various source rocks, which also indicates that different crude oils have experienced varying degrees of source rock contributions and migration pathways during reservoir charging. The distinct migration pathways can induce subtle alterations in the primary geochemical signatures of the precursor oils. The HCA dendrogram of replicate samples shows that a threshold between 0.32 and 0.61 should be employed to classify similar samples into a group. The illustration of the similarity or dissimilarity of different oil groups as points in 3D (a) and 2D (b) space by the MDS method. The HCA dendrogram of all Asmari reservoir crude oils by a 0.61 threshold. The Asmari crude oils are classified into 9 distinct groups based on the asphaltenes structure throughout the Dezful Embayment. Detailed characteristics of different asphaltenes groups Due to insufficient sample amount, detailed chemical analyses were performed only on the 5 selected samples from different groups. X-ray diffraction (XRD) is a non-destructive and viable analytical technique to obtain valuable information about the structural features and aggregation characteristics of asphaltene clusters. This technique determines microstructural information and crystalline parameters of asphaltenes including the interlayer distance between aromatic sheets (dm), the interchain distance between saturated parts (dr), the diameter of aromatic clusters perpendicular to the plane of the sheet (Lc), the diameter of the aromatic planar sheet (La), the average number of carbon atoms per aromatic structure (Cau), the average number of aromatic rings per sheet (Ra), the average number of aromatic sheets in an asphaltene cluster (Me), and aromaticity index (fa)57–59. These parameters are calculated from the three peaks in the XRD pattern of asphaltenes by the equations mentioned elsewhere57,58. Fig. 9 shows an XRD pattern for the MI-58 sample from the AS-1 group, which is representative. The XRD pattern was curve-fitted to obtain three peaks of γ, Graphene (002), and 10 additional peaks using the Gaussian method. Curve fitting was performed on 5 selected samples from different groups, and their crystalline parameters were calculated (Table 2). The Structural parameters obtained from the XRD analysis are presented in Fig. 10 and Table 2. XRD pattern of a representative sample (i.e., MI-58) from the AS-1 group and its deconvolution by the Gaussian method. Table 2:The structural parameters of asphaltenes extracted from the XRD for different oil types.NO.SampleGroupReservoirStructural parametersdrdmLaLcCauRaMefa1MI-58AS-1Asmari2.093.5815.0228.7425690.562NI-3AS-22.123.5414.1638.47245120.463RS-51AS-32.053.5314.2730.25245100.514SG-27AS-42.113.5513.2627.9722590.515QR-17AS-82.083.5414.9729.0525690.6Min2.053.5313.2627.9722590.46Max2.123.5815.0238.47256120.6Average2.093.5514.3430.90245100.53 The structural parameters of asphaltenes extracted from the XRD for different oil types. The visual difference of structural parameters extracted from the XRD analysis for different oil types. Some parameters are relatively similar for different samples. Notably, the interlayer distance between aromatic sheets remains nearly constant, averaging 3.55 Å. This suggests that the spacing between aromatic sheets in asphaltene clusters is consistent across various oil types, with differences less than 0.05 Å. This issue is also true for the interchain distance between the saturated parts of the asphaltenes cluster (dr). The dr has an average of 2.09 Å and a difference of less than 0.7 Å between samples. These results suggest that the distance between clusters of aromatic sheets or aliphatic chains is subtle and does not serve as a determinative parameter in the asphaltene structure. There are 5 or 6 aromatic rings in each sheet of the cluster (Ra), and each cluster contains 9 to 12 aromatic sheets (Me) for different oil types. The La, Lc, Cau, and fa values vary between different oil types, indicating that different groups have distinct structures, particularly in the number of Aromatic rings, which may be influenced by the thermal maturity of the oil samples. Nickel (Ni) and vanadium (V) are well-known trace elements in crude oil, often used to determine geochemical characteristics and classify crude oils3,60,61. These trace elements, typically found as metalloporphyrin complexes, are directly incorporated into the sedimentary organic matter during sedimentation62. Consequently, these elements can differentiate between various oil families and classify crude oils62,63. Each oil family exhibits a similar value of the Ni/V ratio. The trace element content for different oil types is summarized in Table 3. The Ni/V ratio for the analyzed samples ranges from 0.9 (AS-3 group) to 6.43 (AS-2 group), with an average of 2.33. Table 3:Trace element content of different oil types.NoSampleGroupReservoirNi (ppm)V (ppm)Ni/V1MI-58AS-1Asmari172.591850.932NI-3AS-2117.7818.316.433RS-51AS-3165.935184.870.94SG-27AS-4164.81167.120.995QR-17AS-8163.9268.682.39Min117.7818.310.9Max172.591856.43Average157.01124.802.33 Trace element content of different oil types. Conclusions A multidisciplinary approach, integrating MWCC, MDS, and HCA techniques, was employed to classify crude oils based on their asphaltene structural characteristics. This novel approach effectively reveals subtle differences in oil fingerprints and optimally organizes crude oils into distinct groups. By addressing the challenge of determining the optimal threshold in HCA, this integrated procedure provides a robust solution. Samples with proximity in both 2D and 3D space are assigned to the same cluster. This technique was applied to the FTIR spectra of 93 crude oils from the Asmari reservoir in the 21 different oil fields throughout the Dezful Embayment, SW Iran. The Asmari reservoir crude oils were classified into 9 distinct groups based on their asphaltene structures. Geochemical characteristics primarily influence the structural signature of asphaltenes, while differences along migration pathways exert a subtle impact. Consequently, the defined groups provide valuable insights into genetic types and oil plays in the Dezful Embayment. Most fields are classified into different groups, which illustrates that reservoir compartmentalization is a primary phenomenon in most oil fields of the Dezful Embayment. This phenomenon may be attributed to the active tectonics of the Zagros region during the deposition of the Asmari reservoir. Acknowledgements The authors wish to thank National Iranian South Oil Company (NISOC), especially the Research and Technology office, for the provision of oil samples. The authors also acknowledge Dr. Majid Alipour for his time and comments Author contributions A.R.R. conceived the idea, supervised the project, directed the research, and revised the manuscript. M.A. conceptualized the research, performed the experiments, and interpreted the spectroscopy results. The first draft of the manuscript was written by H.S., who also helped provide the test samples, and interpreted the spectroscopy results. Data availability The datasets used and/or analyzed during the current study available from the corresponding author on reasonable request. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement The datasets used and/or analyzed during the current study available from the corresponding author on reasonable request.