Estimation of re-refined engine oil bottom content in bitumen and its chemo-rheological effects as an additive and rejuvenator

Abstract

This study investigates the performance of three types of re-refined engine oil bottom (REOB) in bitumen modification, focusing on their roles as additives, anti-aging agents, and rejuvenators. Neat and aged 60/70 penetration-grade bitumen samples were modified with 5–11 wt% REOB and evaluated using penetration, softening point, ductility, viscosity, FTIR, and inductively coupled plasma mass spectrometry (ICP-MS). The results indicate that adding 9 wt% REOB increased penetration by up to 45% and reduced the aging index by 14%, demonstrating its effectiveness as a softening and anti-aging additive. However, ductility and FTIR analyses revealed that REOB lacks functional rejuvenating capabilities, acting primarily as a physical softener rather than chemically restoring aged binders. A multiple linear regression model was developed using ICP-MS results to estimate REOB content in bitumen with high accuracy (R² = 0.98), providing a reliable quality control tool. Additionally, a human health risk assessment showed that REOB-modified bitumen does not increase carcinogenic or non-carcinogenic risks compared to neat bitumen. These findings contribute to a deeper understanding of REOB’s limitations and potential in sustainable bitumen applications.

Introduction

The growing global demand for sustainable infrastructure has intensified efforts to improve the environmental performance of road construction materials^1,2. In response these challenges, researchers and industry practitioners have investigated the use of alternative materials, particularly waste-derived products, as additives or partial substitutes in bituminous materials^3,4. One such material is re-refined engine oil bottom (REOB), a residue from the vacuum distillation of used engine oils^5,6.

Used engine oils are commonly refined using acid-clay treatment and thin-film distillation⁵. Acid-clay treatment, an older method, removes impurities with sulfuric acid and clay but produces hazardous acid sludge^7,8. Thin-film distillation, a more advanced method, separates oil components through staged distillation, leaving behind re-refined engine oil bottom (REOB), a non-acidic by-product containing heavy contaminants⁵. Both acid sludge and REOB pose environmental risks, prompting interest in their reuse⁹. Due to their physical similarity to bitumen, researchers have explored their potential in asphalt applications as additives, rejuvenators, or bitumen substitutes.

Aliakbari et al. (2023, 2024) proposed a sustainable solution for managing acid sludge by neutralizing it with quicklime and incorporating the resulting powdery sludge (PS) as a fine aggregate in asphalt mixtures^7,8. Their investigations demonstrated that PS significantly improved bitumen performance by enhancing rutting resistance by up to 34.7%, reducing aging and moisture susceptibility by up to 36.4%, and increasing both resilient modulus and indirect tensile strength by 17.3%. These studies further confirmed the environmental safety of this approach through heavy metal leaching tests and life cycle assessment, showing that PS can serve as a low-cost, eco-friendly additive that safely immobilizes contaminants within the asphalt matrix.

REOB was initially used in cold climates to improve the low-temperature performance of bitumen¹⁰. However, its effectiveness remains debated. Some researchers attribute increased aging to metal residues in REOB that catalyze oxidation^11,12, while others argue these effects are unrelated to REOB itself^13,14,15. Bennert et al. (2016) reported that long-term aging potential doubled with REOB, although high-temperature rutting resistance improved at 20 wt%¹⁰. In contrast, Wielinski et al. (2015) found no significant changes at 9 wt%¹⁵. Similarly, Li et al. (2017) observed negligible effects at 2.5 wt%, but noted performance degradation at higher dosages, including reduced rutting and thermal cracking resistance¹⁶.

Despite extensive research on REOB-modified bitumen, key challenges persist, particularly regarding its variable composition, uncertain dosage effects, and traceability. Mortier (2024) conducted a study evaluating REOB-blended binders using XRF, ATR-FTIR, GPC, and SARA analyses¹⁷. The study confirmed that REOB increases aging susceptibility and alters the chemical structure of bitumen by raising saturate levels and shifting molecular weight distributions, ultimately reducing low-temperature flexibility. Moreover, Mortier and others note the absence of regulatory standards, making it possible for REOB to be added without disclosure, particularly in Europe, where REOB lacks a unique CAS number¹⁷.

Li et al. (2023) investigated REOB/SBS-modified bitumen and reported that low REOB dosages improved SBS swelling and high-temperature rheology by introducing light aromatic fractions that enhance dispersion¹⁸. However, high REOB contents were found to compromise durability and low-temperature cracking resistance, with increased phase separation and weaker cohesion between SBS domains and the bitumen matrix¹⁸.

From an environmental perspective, recent studies highlight both promise and caution. Aliakbari et al. (2024) showed that while heavy metal concentrations in REOB exceed regulatory limits, asphalt mixtures containing 14 wt% REOB did not exhibit leachate levels above safety thresholds⁵. Nonetheless, mechanical performance, particularly tensile strength and moisture resistance, deteriorated at higher dosages, emphasizing the need for a balanced approach to maximize environmental benefits without sacrificing pavement durability.

To reduce the high cost of conventional rejuvenators, researchers have explored various low-cost, waste-derived alternatives such as used cooking oil, used engine oil, and REOB^19,20,21,22. Cooper et al. (2017) observed that adding 5–15 wt% REOB to asphalt mixtures decreased cracking resistance at low and intermediate temperatures, with little effect on rutting or moisture susceptibility²³. Kilger et al. (2019) found that REOB, unlike bio-oil, retained anti-oxidation additives from engine oils and performed best in binders modified with Elvaloy and PPA, although SBS-modified binders became more aging-sensitive²⁴. Xu et al. (2021) reported similar initial performance between REOB and bio-oil but noted that bio-oil offered superior fatigue life under extended PAV aging²⁵.

Cui et al. (2022) developed a composite SBS/REOB-modified binder that showed significantly improved high- and low-temperature properties compared to REOB alone, suggesting that SBS enhances REOB’s mechanical performance and microstructural integrity²⁶. In a full-scale field simulation, Li et al. (2022) found that asphalt mixtures with 7 wt% REOB maintained similar long-term performance to conventional pavements after 700,000 loading cycles, despite minor declines in fatigue and skid resistance²⁷. These results underscore REOB’s practical potential, while also reinforcing the importance of dosage control and binder system compatibility.

Recent research shows that effective binder recovery is not solely a function of softening but also depends on chemical compatibility and microstructural restoration^28,29. These findings highlight the need for rejuvenators to be assessed not only by their effect on penetration and viscosity but also through their chemical affinity with the binder matrix, an aspect explored in this study through both physical testing and chemical composition analysis of REOB.

Scope and objective

The incorporation of re-refined engine oil bottom (REOB) into bituminous binders has gained traction due to its potential environmental and economic benefits. Some suppliers and researchers claim that REOB enhances binder quality while reducing carbon emissions by up to 85%³⁰. However, contrasting studies have reported that REOB may accelerate binder aging, reduce pavement durability, and introduce performance inconsistencies. This lack of consensus has led to confusion among engineers, material specifiers, and regulators, particularly in the absence of standard quality control methods or regulatory guidelines. Addressing these knowledge gaps is essential for enabling informed, evidence-based decisions about REOB’s use in asphalt infrastructure.

This study responds to these uncertainties by evaluating the chemo-rheological performance of three different REOB sources when used as both additives and rejuvenators in neat and aged 60/70 bitumen. A suite of laboratory tests, including penetration, ductility, viscosity, FTIR, and ICP-MS analysis, is employed to differentiate whether REOB restores aged binder functionality or merely acts as a softener. Given the challenges of identifying undisclosed REOB in commercial bitumen, this research also introduces a metal-based regression model to quantify REOB content, offering a novel tool for forensic analysis and quality assurance.

Beyond technical performance, this research addresses a significant oversight in current literature: the potential environmental and human health risks associated with REOB use. By assessing heavy metal exposure and carcinogenicity indices, this study provides critical insight into whether REOB-modified binders pose safety concerns. Taken together, this integrated approach seeks to clarify REOB’s role in sustainable pavement engineering while supporting the development of more transparent, safe, and performance-driven asphalt specifications.

Materials and test methods

Materials

REOB is the residual by-product of used engine oil refining, typically obtained via vacuum distillation. It contains a mix of heavy hydrocarbons, metal particles, degraded additives, and residual waste engine oil, giving it a dark appearance and high viscosity similar to bitumen. Due to variations in used oil sources and refining processes, REOB properties, such as viscosity and elemental composition, can differ significantly. In this study, three REOB samples (R1, R2, and R3) were sourced from different re-refining facilities using distinct processing technologies. These were selected to represent the range of physical and chemical variability found in commercially available REOB. Fig. 1 presents the rotational viscosity of the REOBs.

A 60/70 penetration grade bitumen (PG 64 − 22) was used in this study, and the neat bitumen’s detailed specifications are shown in Table 1.

Fig. 1

Rotational viscosity of Neat, R1, R2, and R3 versus temperature.

Table 1 Specifications of the neat bitumen.

Sample Preparation

According to the usage of REOB, the samples are divided into the following categories:

• Group sample No. 1 (GS1): Neat bitumen blended with REOB to assess additive performance.

• Group sample No. 2 (GS2): Aged bitumen blended with REOB to evaluate rejuvenating potential.

In the GS1 group, 60/70 penetration-grade bitumen was mixed with three REOB types (R1, R2, R3) at 5, 7, and 9 wt%. Blending was performed at 125 °C (corresponding to 600 mPa·s viscosity) using a mechanical stirrer at 5000 rpm for 30 min. The selected mixing conditions were determined based on the literature and trial-and-error assessments (based on the rotational viscosity of the REOBs). These samples were also subjected to short- and long-term aging simulations to assess the impact of REOB on aging behavior. Aging was performed following ASTM D1754³⁷ and ASTM D6521³⁸ by placing a 2.3 mm-thick bitumen layer in an oven at 163 °C for 5 h (short-term aging) and then at 95 °C for 7 days (long-term aging), with plate positions rotated daily for uniform exposure^39,40,41. The GS1 sample codes are summarized in Table 2.

Table 2 Sample codes for GS1.

In GS2, REOB was added to previously aged bitumen to simulate field rejuvenation. REOB dosages of 7, 9, and 11 wt% were evaluated, with no REOB present during the aging stage. This approach enabled the assessment of REOB’s rejuvenating effectiveness when blended after aging. The GS2 sample codes are provided in Table 3.

Table 3 Sample codes for GS2.

Test methods

This study consists of four main components, as outlined in Fig. 2, evaluation of REOB’s performance as a bitumen additive, evaluation of its rejuvenating potential in aged binders, quantification of REOB content using elemental analysis, and human health risk assessment based on heavy metal exposure.

A comprehensive set of physical and chemical tests was conducted for compositional and elemental analysis. Statistical analyses were applied to validate observed trends, and a regression model was developed to estimate REOB dosage based on metal concentrations.

Fig. 2

Flowchart of experimental program and analysis of the results.

Bitumen’s characterization tests

The penetration, softening point, and ductility tests were conducted on both GS1 and GS2 samples in accordance with ASTM D 5³², ASTM D 36³³, and ASTM D 113³⁴ standards, respectively.

The rotational viscosity test was conducted to assess the impact of REOB on bitumen viscosity and to determine appropriate mixing and compaction temperatures. This test followed the ASTM D 4402⁴² standard and was carried out at temperatures of 115, 135, 160, and 180 °C.

Fourier transform infrared spectroscopy (FTIR) test

FTIR analysis was employed to assess the chemical aging of binders based on functional group identification. Carbonyl (C = O, ~ 1700 cm⁻¹) and sulfoxide (S = O, ~ 1030 cm⁻¹) absorption peaks were used to quantify oxidation levels, as they are sensitive indicators of bitumen aging^43,44,45,46. Among the different indicators, carbonyl and sulfoxide are known as practical parameters in estimating bitumen aging. Therefore, the aging index (AI) is determined using Eq. (1):

$$AI=\frac{{{A_{1030}}+{A_{1700}}}}{{{A_{1376}}+{A_{1460}}}}$$

(1)

Where AI is the aging index, A_i is the area under the absorbance-wavenumber graph around peak i, peak numbers 1030 and 1700 correspond to the sulfoxide and carbonyl, and peak numbers 1376 and 1700 correspond to the aliphatic.

FTIR tests were performed on GS1 and GS2 samples using a Thermo Smart iTX spectrometer, following ASTM D5477⁴⁷. Spectra were collected over a range of 600–4000 cm⁻¹ with a resolution of 1 cm⁻¹ and 0.48 cm⁻¹ scanning intervals.

Inductively coupled plasma mass spectrometry (ICP-MS) test

ICP-MS effectively measures metal concentration within a material up to parts per billion (ppb). Based on US EPA SW846 Method 3050⁴⁸ and Method 6022B⁴⁹, an acid-dissolving process has been utilized to determine the extent of heavy metals in the samples. The ICP-MS test was performed on samples containing different R1, R2, and R3 dosages using an Agilent ICP-MS 7900. These results have provided an appropriate solution for determining the REOB percentage in bitumen.

X-ray fluorescence (XRF) tests

To complement ICP-MS, X-ray fluorescence (XRF) techniques were used to detect and compare mineral elements in REOB and binder samples. XRF is primarily suited for elemental identification, offering semi-quantitative insights. Three methods were employed:

• XRF-Ash: Following ASTM E1621⁵⁰, samples were incinerated at 600 °C for 90 min to remove organics, then ground to < 75 μm powder for analysis. Testing was conducted using a Philips PW1480 spectrometer.

• XRF-Direct: This non-destructive technique was applied directly to samples without prior treatment, using an ARL PERFORM’X analyzer (Thermo Fisher Scientific), based on NIST Technical Note 1816⁵¹.

• XRF-Handheld: A portable Niton XL3t Ultra Analyzer was used for rapid on-site scanning. This device allowed non-invasive surface testing, supporting quick screening and repeatability.

While XRF offered a convenient screening tool, ICP-MS was ultimately selected for its superior sensitivity and precision in supporting REOB quantification modeling.

Data analysis

Data collection and preparation

Elemental data were collected using ICP-MS on bitumen samples prepared with 0–25 wt% REOB using three REOB sources (R1, R2, R3). The experimental design aimed to reflect realistic compositional variation across dosage and source. Values below the instrument’s detection limit were replaced with the corresponding detection threshold to maintain a consistent matrix for modeling. Outlier detection was conducted via box plots, and no extreme values were identified. Consistent sample preparation procedures minimized internal variability across the dataset.

Feature importance of variables

ICP-MS identified 26 elemental components, but many were either invariant or irrelevant for REOB quantification. A Random Forest algorithm was applied to identify the most influential predictors. Although a relatively simple regression model was ultimately developed, Random Forest was chosen due to its robustness in handling multicollinearity and identifying non-obvious variable relationships⁵².

Multiple linear regression model

The Shapiro-Wilk test is employed to assess the normality of the data. Correlations between variables are examined using either Pearson’s or Spearman’s correlation tests, depending on the data’s normality. In order to determine the REOB percentage in bitumen, a multiple linear regression model is utilized, with the independent variables being the concentration of metals in parts per million (ppm) and the dependent variable being the REOB percentage. This model was developed using IBM-SPSS 26.0 and the forward method. After the model’s development, residual diagnostics were controlled, and both autocorrelation and heteroscedasticity were evaluated using the Durbin-Watson test and the White test, respectively.

Additional samples were prepared using REOB and bitumen from different sources at known dosages (0, 5, 10, 15, and 20 wt%) to evaluate the model’s generalizability. These samples were used exclusively for cross-validation to assess the model’s ability to estimate REOB content in materials beyond the training dataset.

Human risk assessment

Humans can be exposed to heavy metals in three ways: ingestion, inhalation, and dermal contact^53,54. This assessment focused specifically on the presence of heavy metals commonly found in REOB, which may pose health risks under certain exposure scenarios. This risk is divided into non-carcinogenic and carcinogenic classifications for children and adults. Chronic daily intake (CDI) of heavy metals through ingestion (CDI_ing), inhalation (CDI_inh), dermal contact (CDI_derm), and total (CDI_total) can be calculated from Eq. (2) to (5), respectively. The variable description and values have been presented in Table 4^55,56,57.

$$CD{I_{ing}}=\frac{{C \times I{R_{ing}} \times CF \times EF \times ED}}{{BW \times AT}}$$

(2)

$$CD{I_{inh}}=\frac{{C \times I{R_{inh}} \times EF \times ED}}{{PEF \times BW \times AT}}$$

(3)

$$CD{I_{derm}}=\frac{{C \times SA \times AF \times ABS \times CF \times EF \times ED}}{{BW \times AT}}$$

(4)

$$CD{I_{total}}=CD{E_{ing}}+CD{E_{inh}}+CD{E_{derm}}$$

(5)

Table 4 Descriptions and values of variables used in health risk assessment equations.

The hazard index (HI) presented in Eq. (6) is used for non-carcinogenic health risks^55,56,57, in which HQ is the risk quotient for each element obtained from Eq. (7), and RFD is the reference dosage of each element obtained from Table 4. exceeds one, there is a potential risk of non-carcinogenic effects from heavy metal exposure in individuals. Conversely, if the HI is below one, adverse health effects are unlikely to occur in exposed individuals^55,56,57.

$$HI=H{Q_{ing}}+H{Q_{inh}}+H{Q_{derm}}$$

(6)

$$HQ=\frac{{CD{I_{N - C}}}}{{RDF}}$$

(7)

There is a possibility of developing any cancer in people who are exposed to high-risk substances. To evaluate the carcinogenic risk of substances, the total carcinogenic risk index (TCR) is used, which is calculated from the sum of the carcinogenic risk (CR) of ingestion, inhalation, and dermal contact (Eqs. (8) and (9))^55,56. SF is the slope factor for each element, which is presented in Table 4. The acceptable threshold value for TCR is 0.0001. TCR value exceeding 0.0001 indicates a potential lifetime carcinogenic risk^55,56.

$$TCR=C{R_{ing}}+C{R_{inh}}+C{R_{derm}}$$

(8)

$$CR=CD{I_{Ca}} \times SF$$

(9)

Results and discussion

REOB as an additive

Penetration and softening point

Based on the observations of R1, R2, and R3, it has been found that REOB is entirely permeable at an ambient temperature, and it is not feasible to measure its degree of penetration and softening point. Among the three, R1 exhibited the lowest viscosity and highest softening effect, while R2 showed the greatest stiffness, consistent with its higher resistance to deformation. As shown in Fig. 3, the addition of 5 wt% REOB increased the penetration of neat bitumen to over 70 dmm, indicating a noticeable softening effect. At a dosage of 9 wt%, the penetration values increased by 45%, 32%, and 42% for R1, R2, and R3, respectively, relative to the unmodified (Neat) binder.

As shown in Fig. 4, REOB addition consistently reduced the softening point across all REOB types and dosages. This softening behavior is influenced by the proportion of unrecovered used engine oil in the REOB, higher oil content results in lower viscosity and a more pronounced softening effect on the binder. Therefore, the degree of softening varies with REOB source and composition, emphasizing the importance of material characterization prior to application.

Fig. 3

Results of penetration test for GS1 samples.

Fig. 4

Results of softening point test for GS1 samples.

The penetration ratio (aged/unaged) was used to assess aging resistance, with higher values indicating improved resistance to oxidative hardening⁵⁸. The calculated ratios for Neat, N-9R1, N-9R2, and N-9R3 were 50%, 57%, 56%, and 58%, respectively, indicating that REOB-modified samples had 12–16% higher resistance to aging compared to unmodified binder. This enhancement may result from two main factors: (1) REOB’s ability to physically soften the aged binder, thus reducing the apparent aging effect, and (2) the lower susceptibility of REOB itself to oxidation due to its residual oil content, which dilutes the aged bitumen fraction, as also noted by^17,23.

Ductility

Bitumen ductility can provide an estimation of bitumen brittleness and adhesion. Unaged samples of GS1 have achieved a minimum ductility of 100 cm. These results indicate that using REOB up to 9% seems flawless. Due to the device’s limitation, more than 100 cm ductility is not measurable, and it is impossible to comment on the effect of REOB on the ductility.

In comparison of various samples, the ductility measurements for A(Neat), A(N-9R1), A(N-9R2), and A(N-9R3) were found to be 26 cm, 39 cm, 46 cm, and 48 cm, respectively. This indicates that samples A(N-9R1), A(N-9R2), and A(N-9R3) exhibit 50%, 77%, and 85% better ductility than the A(Neat) sample, which is a noteworthy outcome. Interestingly, despite R1 being softer than R2, the ductility of A(N-9R1) is lower than that of A(N-9R2). Consequently, the ductility test results demonstrate that using a softer REOB, which yields a softer sample, does not necessarily guarantee better performance in the ductility test.

Rotational viscosity

As shown in Fig. 1, R2 exhibits higher viscosity than R1 and R3 across all temperatures, suggesting it contains fewer light oil fractions due to more efficient oil separation during re-refining. This makes R2 heavier and more viscous. Viscosity variation among REOBs reflects differences in re-refining efficiency, making viscosity a reliable classification index. Additionally, REOB samples show 37–40% lower viscosity sensitivity to temperature compared to neat bitumen, indicating reduced temperature susceptibility.

Figure 5 illustrates the rotational viscosity of the samples at a test temperature of 135℃ (rotational viscosity at 115, 160, and 180℃ can be found in the data source file). The viscosity of A(N-9R) is lower than that of A(Neat), which may be attributed to a decreased aging rate or a lower initial viscosity of N-9R compared to Neat before aging.

The ratio of the aged to the unaged viscosity indicates an aging index. The average of this index across all four tested temperatures was utilized to compare the samples. The aging rate index based on viscosity for Neat, N-9R1, N-9R2, and N-9R3 samples is 5.64, 4.24, 4.10, and 4.25, respectively. Therefore, the addition of 9 wt% REOB has reduced the aging rate of bitumen by 25–27%, making the sample more resistant to aging. Despite viscosity differences, R1, R2, and R3 show similar aging performance (variation < 4%) when blended with bitumen at 9 wt%. The reduction in aging could stem from either:

• Improving bitumen resistance to aging due to chemical reaction with REOB and strengthening bonds against oxidation.

• Bitumen bonds have not been reinforced, and less REOB aging has caused less drop in the viscosity; therefore, the effect of bitumen aging on the index value has become less significant.

Fig. 5

Rotational viscosity of GS1 samples at the temperature of 135℃.

Chemical analysis

FTIR-based aging index (AI) is a reliable method to evaluate oxidation in bitumen. As shown in Fig. 6, all three REOB types (R1, R2, R3) exhibited similar chemical profiles, while sulfoxide (1030 cm⁻¹) and aromatic (1600 cm⁻¹) peaks were observed only in the Neat sample, confirming REOB’s distinct composition.

As shown in Fig. 7, the AI values of unaged N-9R1, N-9R2, and N-9R3 were 14% higher than Neat, indicating REOB’s initial chemical impact. Upon aging, the AI ratios for Neat, N-9R1, N-9R2, and N-9R3 were 2.94, 1.79, 1.51, and 1.50, respectively. These results confirm that REOB significantly improves aging resistance, with R2 and R3 outperforming R1.

Although R1 was the softest REOB physically, it offered the least chemical resistance to aging. This reinforces that aging behavior is influenced by both chemical composition and softening capacity, highlighting REOB’s role as more than just a viscosity modifier.

Fig. 6

Absorption spectrum of Neat, R1, R2, and R3 in FTIR test.

Fig. 7

The AI aging index of Neat, N-9R, and A(N-9R) samples.

REOB as a rejuvenator

This section evaluates REOB’s rejuvenation performance. As mentioned in the “Sample Preparation” section, R1, R2, and R3 have been utilized at concentrations of 7%, 9%, and 11% by weight in the Neat bitumen.

Penetration and softening point

Figure 8 and Fig. 9 show that REOB addition effectively recovers the penetration and softening point of aged bitumen toward the levels of the original (Neat) binder. Among the three REOB types, R1 achieved the highest recovery, aligning with its higher oil content and softer characteristics.

Comparing AN-9R and A(N-9R) samples (both with 9 wt% REOB) reveals that post-aging addition (AN-9R) leads to higher penetration and lower softening point than pre-aged REOB samples. This may reflect either (1) effective rejuvenation of the aged binder or (2) degradation of REOB properties during the aging process.

Fig. 8

Results of penetration test for GS2 samples.

Fig. 9

Results of softening point test for GS2 samples.

The per-percent improvement in penetration for AN-9R samples was 5.93%, 4.97%, and 5.45% for R1, R2, and R3, respectively, higher than in unaged systems. This confirms REOB is more effective in restoring aged binder flexibility than in simply softening fresh bitumen.

The optimum REOB dosage was determined based on its ability to restore the penetration and softening point of aged bitumen to values close to those of the unaged control sample, as in the literature^59,60,61,62. This approach reflects the practical goal of regaining original binder consistency for mixing and compaction purposes. While more advanced performance grading (PG) methods account for binder behavior over a broader temperature and loading spectrum, conventional penetration-based specifications remain widely used in practice. Thus, these physical indicators were deemed sufficient for defining a baseline dosage range, although PG-based evaluation is recommended for high-performance applications in future studies. As shown in Fig. 10, the optimal percentages of R1, R2, and R3 are 8, 10, and 9 wt%, respectively.

Fig. 10

The optimum percentage of rejuvenators based on the penetration and softening point.

Ductility

As illustrated in Fig. 11, the ductility of the aged control sample (A(Neat)) was 26 cm. Adding REOB significantly improved flexibility, with AN-7R samples more than doubling the ductility. However, even at higher dosages, none of the samples reached the unaged binder’s minimum ductility of 100 cm.

Notably, while initial REOB additions improve ductility, the marginal benefit decreases with higher dosages, suggesting a limit to REOB’s rejuvenating capacity. This trend supports the view that REOB acts more as a softening agent than a true chemical rejuvenator.

Fig. 11

Results of ductility test for GS2 samples.

Rotational viscosity

One of the factors influencing the determination of the optimal rejuvenator percentage is viscosity. The rotational viscosity of GS2 samples at a test temperature of 135℃ is shown in Fig. 12 (rotational viscosity at 115, 160, and 180℃ are provided in the data source file). In this study, viscosity has been used to control the rejuvenating performance of REOB.

The rotational viscosity of samples treated with optimal percentages of REOB (determined from penetration and softening point tests) at temperatures of 115℃ and 135℃ closely matches the viscosity of the Neat sample. This demonstrates that the selected REOB percentages effectively restore the viscosity of aged bitumen to that of unaged bitumen. However, at temperatures of 160℃ and 180℃, samples treated with optimal percentages of REOB exhibit higher rotational viscosity than the Neat sample. This observation may be attributed to the lower temperature sensitivity of REOB compared to the Neat sample, resulting in a greater viscosity for REOB at higher temperatures.

Fig. 12

Rotational viscosity of GS2 samples at the temperature of 135℃.

Chemical analysis

The AI index is an appropriate metric for investigating the rejuvenation performance of REOB. This test was conducted on rejuvenated samples using the optimal REOB percentage, and the results are displayed in Fig. 13. Ideally, the optimal percentages of REOB for rejuvenation should recover the AI index from 0.20 (AI of A(Neat)) to its initial value of 0.07 (AI of Neat). The rejuvenated samples AN-8R1, AN-10R2, and AN-9R3 achieved AI values of 0.18, 0.17, and 0.16, respectively. These modest improvements, corresponding to reductions of only 10%, 15%, and 20%, indicate that REOB lacks sufficient chemical rejuvenating capacity.

This finding contrasts with results from penetration, softening point, and viscosity tests, which suggested that REOB restored binder properties. However, those improvements are primarily due to REOB’s low viscosity and light oil content, which physically soften the aged binder rather than chemically reversing oxidation.

Fig. 13

AI aging index of rejuvenated samples.

FTIR spectra showed that key oxidation markers (carbonyl and sulfoxide peaks) remained significantly elevated after REOB addition. This confirms that REOB does not effectively reverse the chemical aging process and instead acts as a physical softener. Unlike functional rejuvenators that contain reactive polar compounds capable of rebalancing aged binder colloidal structure, REOB merely dilutes the asphaltene content without restoring chemical integrity.

Additionally, when comparing results from A(N-9R) samples (where REOB was present during aging) in Fig. 7 to those from rejuvenated GS2 samples (Fig. 13), the former exhibited lower AI values. This suggests that REOB is more effective at slowing down oxidation when blended before aging than reversing it afterward. Therefore, REOB’s role is more consistent with that of an anti-aging additive than a functional rejuvenator.

Estimating the REOB dosage

This section explores elemental analysis techniques for determining the REOB percentage in bitumen. Using ICP-MS and various XRF methods, the study identifies key metal markers, selects the most effective testing method, and develops a regression model. A Random Forest algorithm is employed for feature selection, followed by model validation.

Comparison of elemental analysis methods

Previous studies have primarily used the XRF-Ash method to detect metals in REOB-modified binders. However, XRF is a semi-quantitative technique, reporting metal concentrations based on relative weight and is susceptible to error, especially in samples with organic content. In contrast, ICP-MS provides highly precise elemental concentrations in the ppb range and has not yet been widely used for REOB detection.

In this study, four techniques (ICP-MS, XRF-Ash, XRF-Direct, and XRF-Handheld) were applied to Neat, R1, R2, and R3 samples. As shown in Table 5, ICP-MS outperformed all XRF methods, detecting metals that were undetectable or inaccurately quantified by XRF. For instance, aluminum in the Neat sample was measured at 731 ppm by ICP-MS, compared to 7 ppm by XRF-Ash, and was undetected by XRF-Direct and XRF-Handheld.

XRF tests, especially XRF-Direct and XRF-Handheld, suffer from reduced accuracy due to interference from organic compounds. While XRF-Ash provides better accuracy than its portable counterparts, it still lacks precision in determining true metal concentrations.

Given the comparable local costs of XRF-Ash and ICP-MS, the latter is recommended for accurate REOB quantification. Nevertheless, the XRF-Handheld device offers value for on-site screening, delivering immediate results and operational convenience. It can be used as a preliminary tool to identify potential REOB presence before more rigorous analysis.

Table 5 Comparison of various elemental analysis test results on neat, R1, R2, and R3.

Data analysis

Following the evaluation of elemental analysis methods, ICP-MS was selected to develop a numerical model for estimating the REOB percentage in bitumen. Table 6 presents descriptive statistics for each detected metal, including minimum, maximum, standard deviation, and the Neat/Average (R1, R2, R3) ratio. This ratio highlights the relative enrichment or depletion of metals between the Neat binder and the REOB sources.

Among the findings, zinc (Zn) showed the greatest increase in REOB samples compared to Neat, making it a strong indicator of REOB presence. In contrast, vanadium (V) had a Neat/REOB ratio of 242, indicating it is significantly more concentrated in the original binder than in REOB, thus, less useful for detecting REOB.

Essential metals

A Random Forest algorithm was applied to the ICP-MS dataset to identify the most relevant predictors. As shown in Fig. 14, the five most influential metals for REOB detection were zinc (Zn), phosphorus (P), calcium (Ca), aluminum (Al), and sulfur (S).

These metals consistently increased in concentration with higher REOB content across R1, R2, and R3, making them reliable markers. Their concentration trends in relation to REOB dosage are presented in Fig. 15, confirming their strong correlation with REOB percentage and validating their selection as independent variables in the predictive model.

Fig. 14

The feature importance of the sample’s metals in determining the REOB dosage.

Table 6 Metal’s concentrations of samples (ppm).

Fig. 15

The concentration of Al, Ca, P, S, and Zn metals versus REOB dosage.

The presence of specific metals in bitumen can be used to identify additives introduced during modification. For instance, zinc is commonly associated with crumb rubber, while calcium, iron, and silicon are prevalent in iron furnace slag^63,64,65. Similarly, silicon is a key component of recycled glass, and both silicon and calcium can originate from aggregates or lime⁶⁶. This overlapping elemental presence can cause misidentification when using univariate models to estimate additive content.

The mutual presence of specific elements, such as zinc and phosphorus or zinc and calcium, can serve as more robust indicators of REOB. Additionally, aluminum and sulfur, when identified in combination with these key metals, help distinguish REOB from other common additives. This multi-element approach reduces the risk of false positives in REOB detection.

Given that zinc, phosphorus, calcium, aluminum, and sulfur were identified as the most influential metals, the XRF-Handheld test can serve as a practical pre-screening tool for REOB detection. Although it lacks the precision of ICP-MS, it offers notable advantages for field applications, including immediate results, operational simplicity, elimination of sample preparation, and the ability to conduct testing without the need for laboratory facilities. Its relatively low cost further enhances its suitability for on-site evaluations. Therefore, XRF-Handheld devices can be effectively used for initial confirmation of REOB presence, while ICP-MS remains the preferred method for precise quantification. To standardize the field application of XRF-Handheld analysis, future studies should focus on defining threshold control limits for the key metals identified in this study.

Modeling and validation

A multiple linear regression model was developed to quantify the REOB percentage in bitumen, using REOB% as the dependent variable and the concentrations of zinc (Zn), phosphorus (P), calcium (Ca), aluminum (Al), and sulfur (S) as independent variables. A Shapiro-Wilk test confirmed the normal distribution of all variables, enabling the use of Pearson’s correlation, which showed significant correlations between REOB% and all five metals⁶⁷. The final model is expressed as:

$$REOB\% = 0.030Zn + 0.024P - 5.612$$

(10)

As shown in Table 7, the model yields an R² of 0.98 and an adjusted R² of 0.98, indicating excellent fit. The Durbin-Watson statistic was 2.22, suggesting no autocorrelation in the residuals. White’s test results confirmed the absence of heteroscedasticity (p-values > 0.47 for all predictors), validating model assumptions. The high Adjusted R² value may be attributed to the number of observations. In this regard, random data was used to evaluate model validation, which was not used in developing the model.

Table 7 Summary of model’s outputs and ANOVA table.

In order to test external validity, the model was applied to five new samples containing 0, 5, 10, 15, and 20 wt% REOB (Table 8). As shown in Fig. 16, the predicted REOB values closely match observed values, with a regression slope of 1.03 and a maximum deviation of 1.3%, confirming the model’s high predictive accuracy. This model, based on Zn and P concentrations, offers a reliable, validated method for estimating REOB content in bitumen and represents a practical tool for quality control in field and lab settings.

Table 8 Metals’ concentration of samples used for model cross-validation.

Fig. 16

Prediction versus observation values of REOB dosages.

Human risk assessment

As shown in Table 9, the HI values for all samples are below the critical threshold of 1, indicating no significant non-carcinogenic risk. In adults, HI values were more than ten times lower than the limit, while children’s values ranged from 0.68 to 0.91, approaching but not exceeding the threshold. Interestingly, the Neat sample exhibited the highest HI, while REOB-modified samples demonstrated lower values, suggesting that REOB incorporation does not increase the non-carcinogenic risk and may even reduce it.

The TCR values indicate a more nuanced risk pattern. For adults, all TCR values ranged between 5.4E-05 and 6.6E-05, staying below the accepted carcinogenic threshold of 1.0E-04. However, for children, the TCR values exceeded this limit, ranging from 1.3E-04 to 1.5E-04, highlighting a potential carcinogenic risk. Among the tested samples, the Neat binder posed the highest TCR, while REOB-modified samples (R1, R2, R3) showed 25–37% reductions compared to the critical limit.

These findings suggest that REOB-modified bitumen does not increase human health risks and is potentially safer than conventional bitumen. While no significant concern is raised for adults, the elevated TCR in children warrants caution. To mitigate exposure risks for children, especially in areas near asphalt or waterproofing facilities, it is highly recommended to:

• strictly refrain from the frequent presence or accommodation close to bitumen and suppliers, waterproofing shingle factories, and other sites that use bitumen on a daily basis.

• Implement and enforce an appropriate protocol to regulate and control the exposure of children to bitumen-related substances.

Table 9 The hazard index and total carcinogenic risk for adults and children.

Summary and conclusions

This study evaluated the effectiveness of re-refined engine oil bottom (REOB) as both an additive in neat bitumen and a rejuvenator in aged binders. Three REOB samples (R1, R2, R3) from different refining processes were analyzed to assess their chemical and physical behavior. Although REOB consistently softened the binder and reduced viscosity, FTIR results indicated that its rejuvenating capabilities are limited. REOB primarily acts as a softener, not a chemical rejuvenator.

A practical model was developed using ICP-MS data to quantify REOB content in bitumen based on Zn and P concentrations. Among several elemental analysis methods tested, ICP-MS proved the most reliable and accurate. This model enables binder producers and regulators to verify REOB dosage with minimal testing requirements. Additionally, a human health risk assessment showed that REOB-modified binders do not pose elevated risks compared to neat bitumen under standard exposure conditions.

From an economic perspective, REOB offers a low-cost, waste-based alternative that can improve the sustainability of asphalt production. In regions producing over two million tonnes of bitumen annually, utilizing REOB at 9 wt% in just 15% of total binder use would demand over 27,000 tonnes of REOB, exceeding the typical yearly supply (~ 25,000 tonnes). Still, this strategy can significantly reduce disposal costs and improve the softness of the binder.

Transport costs are negligible: delivering 1000 tonnes of REOB over 100 km costs the equivalent of just 0.43% of the price of neat PG 64 − 22 bitumen. Furthermore, REOB’s low viscosity allows for easy handling without the need for additional heating.

In summary, incorporating REOB into bitumen offers both economic and environmental benefits; however, its use should be limited to applications where softening, rather than complete rejuvenation, is the primary objective.

Future studies

Future research should focus on field validation of REOB-modified asphalt mixtures to assess long-term performance under real conditions and using advanced bitumen tests. Investigating storage stability and potential phase separation is also recommended. Environmental studies on emissions during asphalt production would further support safe application. Finally, the REOB detection model could be refined and expanded using larger datasets or machine learning methods.