The Effect of Aging on the Molecular Distribution of Crumb Rubber Modified Asphalt Based on the Gel Permeation Chromatography Test

Round 1

Reviewer 1 Report

In this contribution, the authors used GPC to analyze the components in crumb rubber modified asphalt before and after aging. The GPC results showed an increase in the macromolecular content after aging. This topic is inspiring to the readership of Buildings. However, this research overlooked critical discussions. The following questions and comments need to be addressed before making a further decision.

1. Figure 1 and 2 show the normalized RI response. What component/molecular weight is the normalization based on? Does the fraction of this internal reference change before and after aging? If the normalization is based on a changing peak, the normalization will vertically shift the entire GPC curve, so the heights of different curves cannot represent their actual concentrations for comparison. Can the authors show the GPC curves normalized using an external reference, or all samples of the same concentration without normalization?

2. The crumb rubber modified asphalt shows an increase in the macromolecular content after PAV aging (Line 274). If the macromolecular content is from the degradation of crumb rubber (Line 286), what reactions dissociate the crosslinked rubber into a product soluble in THF? Can SARA analysis isolate this product from the components of asphalt? If the neat crumb rubber undergoes PAV aging, is the aged crumb rubber soluble or partially soluble in THF?

3. Following Question 2, how to confirm whether the increasing macromolecular content is from the degrading crumb rubber, or due to the aging of asphalt, e.g., small asphaltic molecules forming macromolecules? Why does the SK70-PAV in Figure 1 also show the high-molecular-weight product with a molecular weight of 10000 even without any crumb rubber? Do the other two types of asphalt also show an increase in macromolecular contents? Does elongating the PAV time affect the macromolecular contents?

4. What is the difference between the GPC samples discussed in 3.2 and 3.3? Most of the crumb rubber is not soluble but swollen in THF due to its crosslinked structure, therefore, the crumb rubber is filtered and removed to prepare the GPC samples. If filtering is necessary, what is the difference between GPC samples and filtered samples? The parameters in Equation 2.1 and 2.2 needs clear definitions to elaborate the method of filtering test.

5. Other amendments:

Table 3.1 in Line 236, but Table 3 in Line 261.

The MSD (Line 105) and CRE (Line 375) need definition.

Line 201, both macromolecules and intermediate molecules are denoted as LMS.

Line 244, 10.9%.

The x and y axes are not shown in Figure 1 inset. What is the 104 in Line 226?

In Figure 8, the vertical axes are labeled as M_n instead of M_z.

Comments for author File: Comments.pdf

Author Response

In this contribution, the authors used GPC to analyze the components in crumb rubber modified asphalt before and after aging. The GPC results showed an increase in the macromolecular content after aging. This topic is inspiring to the readership of Buildings. However, this research overlooked critical discussions. The following questions and comments need to be addressed before making a further decision.

1. Figure 1 and 2 show the normalized RI response. What component/molecular weight is the normalization based on? Does the fraction of this internal reference change before and after aging? If the normalization is based on a changing peak, the normalization will vertically shift the entire GPC curve, so the heights of different curves cannot represent their actual concentrations for comparison. Can the authors show the GPC curves normalized using an external reference, or all samples of the same concentration without normalization?

Response:  

Thank your comments. 

Before testing asphalt samples, it is necessary to calibrate the molecular weight leaching volume standard curve to ensure the accuracy of measuring the molecular weight of the sample. The calibration curve for this test is made from polystyrene standards. Five different molecular weight standard samples are selected at certain molecular weight intervals for testing to obtain the molecular weight of the standard sample and its corresponding retention time. The obtained data are processed and fitted using Excel software, and a universal calibration curve is established using a cubic polynomial. The obtained relative molecular weight standard curve is shown in Figure 1.

Figure 1 GPC correction curve cubic polynomial fitting curve graph

The regression formula is:

LogM = -0.003086361 V³ + 0.203362277 V ² - 4.628988063 V + 40.162712410

R² = 0.998599623 （1）

Where: LogM represents the logarithmic value of the molecular weight of the sample; V represents the elution volume.

In order to verify the accuracy of the obtained formula, the residual percentage is calculated using formula (2), and the calculation result is shown in Figure 2. The percentage of residual error is between -8.05% and 7.16%, and the prediction accuracy of the obtained formula is good, meeting the test requirements.

           （2）

Figure 2 Percentage of fitted residuals

Using the fitting formula obtained above, the normalized data obtained from the GPC test are processed to obtain the relative molecular weight of the test sample. At the same time, calculate the weight average molecular weight Mw, number average molecular weight Mn, Z average molecular weight, and dispersion degree D corresponding to each component according to formula (3) (4) (5) (6).

                  （3）

                 （4）

                 （5）

                            （6）

In the formula, Ni is the number of molecules with a molecular weight of Mi;

Wi is the weight of the component with a molecular weight of Mi

Since the intensity of the detector signal (the ordinate yi of the GPC chromatogram) is proportional to the weight concentration of the molecule corresponding to the current retention time ti, it is possible to convert the retention time ti to the molecular weight Mi and directly apply the formula (2.33) to calculate the weight average molecular weight of the specified component. Calculate the ratio of detector signal strength yi to its corresponding molecular weight Mi to obtain the relative molar number (proportional to ni), and then follow formula (2.34) to obtain the number average molecular weight. The corresponding dispersity D and average molecular weight Z can be calculated from the above formula. Among them, the weight average molecular weight (Mw) is obtained by statistical averaging based on molecular weight, and the number average molecular weight (Mn) is obtained by statistical averaging based on molecular number. Both of these molecular weight parameters have clear physical meanings, while the physical meaning of the Z average molecular weight is less clear.

The weight average molecular weight is sensitive to changes in the number of large molecules, while the number average molecular weight is more sensitive to the number of small molecules. The dispersion degree D is the ratio of the weight average molecular weight to the number average molecular weight, indicating the degree of dispersion of the sample molecular weight. The specific situation of the dispersion of the molecular weight of polymers is most accurately characterized by the molecular weight distribution. The molecular weight and molecular weight distribution of asphalt are the intrinsic reasons for characterizing the performance of asphalt pavement. Reasonable analysis of relevant information has important reference value for understanding the intrinsic mechanism of asphalt performance.

2. The crumb rubber modified asphalt shows an increase in the macromolecular content after PAV aging (Line 274). If the macromolecular content is from the degradation of crumb rubber (Line 286), what reactions dissociate the crosslinked rubber into a product soluble in THF? Can SARA analysis isolate this product from the components of asphalt? If the neat crumb rubber undergoes PAV aging, is the aged crumb rubber soluble or partially soluble in THF?

Response:  

Thank your comments. 

In this paper, only GPC method was used to analyze the molecular weight change characteristics of rubber powder modified asphalt. From other studies, it can be seen that the four components of rubber powder modified asphalt contain rubber powder substances, including functional groups such as 1100cm^-1, 965cm^-1, and 670cm^-1, which directly proves that rubber powder and asphalt undergo complex physical and chemical reactions. These material exchanges play a positive role in improving the low-temperature performance of asphalt. At the same time, rubber powder can effectively inhibit the increase of polar functional groups such as aromatics, resins, and asphaltene carbonyls. From the perspective of component composition, the addition of rubber powder can effectively inhibit the evolution of asphalt aging process and improve its low-temperature performance.

The addition of rubber powder can effectively increase the content of light components (saturated and aromatic components) in asphalt itself, while reducing heavy components (colloid and asphaltene), indicating that rubber powder can significantly delay and inhibit the transfer of components within asphalt. Rubber powder can be used as a "sponge" in asphalt to absorb light components into the interior of rubber powder particles for storage. During the aging process, the absorbed light components will be continuously released to supplement the light components lost by the asphalt itself due to thermal oxygen aging. At the same time, the rubber powder itself also undergoes a degradation process, resulting in a reduction in its own quality, which to some extent delays the aging process of asphalt.

In other studies, the author has analyzed the four components of asphalt, proving that the modification effect of rubber powder on asphalt can be understood through four component analysis.

Due to the complex composition and structure of asphalt, complex physical and chemical changes have taken place under the aging action after the addition of rubber powder, resulting in many changes in the microstructure and chemical behavior of asphalt. In order to analyze the specific structural changes, GPC testing was conducted on the four components of asphalt obtained by using the component separation method, summarizing the molecular weight and distribution changes of asphalt components before and after aging, and adding rubber powder, pointing out the molecular weight and distribution changes of asphalt components in the rubber powder The combined effects of aging and other factors provide a theoretical basis for further exploring the aging mechanism of asphalt and clarifying the impact of component changes on low-temperature performance of asphalt.

The relevant experiments of neat rubber crumbs undergoing PAV aging have not yet been conducted, and further research can be conducted on this issue to more fully demonstrate the modification mechanism of rubber powder on asphalt.

3. Following Question 2, how to confirm whether the increasing macromolecular content is from the degrading crumb rubber, or due to the aging of asphalt, e.g., small asphaltic molecules forming macromolecules? Why does the SK70-PAV in Figure 1 also show the high-molecular-weight product with a molecular weight of 10000 even without any crumb rubber? Do the other two types of asphalt also show an increase in macromolecular contents? Does elongating the PAV time affect the macromolecular contents?

Response:  

Thank your comments. 

The increase in macromolecular content is mainly due to aging. The addition of rubber powder resulted in an increase in the overall molecular weight of GPC. The specific analysis of the impact of rubber powder on the molecular weight change of asphalt can be determined by referring to the third part, the relationship between PE and IE.

After aging, the macromolecular content of the other two types of asphalt also increased. The increase range of macromolecular content of three different types of asphalt is different.

According to relevant regulations, prolonging the aging time of PAV will, to a certain extent, lead to an increase in macromolecular content. As no relevant experiments have been conducted in this article, it is necessary to conduct relevant experiments to judge this issue in the future.

4. What is the difference between the GPC samples discussed in 3.2 and 3.3? Most of the crumb rubber is not soluble but swollen in THF due to its crosslinked structure, therefore, the crumb rubber is filtered and removed to prepare the GPC samples. If filtering is necessary, what is the difference between GPC samples and filtered samples? The parameters in Equation 2.1 and 2.2 needs clear definitions to elaborate the method of filtering test.

Response:  

Thank your comments. 

The GPC samples discussed in 3.2 is rubber powder modified asphalt, and the sample contains rubber particles. The GPC samples discussed in 3.3 is a filtered asphalt modified with rubber powder, and the sample does not contain rubber particles.

Existing experiments have shown that rubber particles not only undergo physical modification, but also undergo chemical degradation in asphalt. Part of the rubber material is incorporated into the asphalt. The purpose of this article is to explore the effect of the chemical reaction of rubber particles on asphalt by using filtration methods. From the actual situation of filtration, compared to the molecular weight distribution diagram of matrix asphalt and rubber powder modified asphalt, filtered asphalt exhibits different characteristics of change, which fully demonstrates the necessity of filtration.

The modification of asphalt by rubber powder involves both the interaction effect (IE) between rubber powder and asphalt, and the particle effect (PE) of rubber powder particles in asphalt, as shown in Figure 1. In order to verify the impact of rubber powder on the composition and structure of asphalt, extraction and filtration were used to filter rubber powder from modified asphalt, and then relevant tests were conducted on the filtered rubber powder and extracted asphalt to explore the practical role of rubber powder in asphalt. The specific process is shown in Figure 2. First, extract about 40 g of rubber powder modified asphalt and put it into a conical flask, dissolve it with 200 ml of trichloroethylene, and then place it in an ultrasonic instrument for 30 minutes to quickly dissolve the rubber powder modified asphalt. Due to the high quality of asphalt filtered in this test, the filtration efficiency of using traditional filter paper is relatively low. This step can be improved by using a vacuum filter to greatly accelerate the filtration efficiency, requiring only replacement of the filter paper each time. The extracted asphalt solution is evaporated using a rotary evaporator. In order to improve the distillation efficiency, multiple experiments were conducted for the solution of asphalt trichloroethylene. At the beginning, the oil bath temperature was set to 55 ℃, the rotation rate was 50 rpm/min, and the rotary distillation was conducted for 30 minutes to quickly recover most of the trichloroethylene. When it is observed that there is no trichloroethylene in the conical flask, increase the oil bath temperature to 155 ℃ and heat for 15 minutes to completely drain the residual trichloroethylene in the asphalt. After the completion of this step, place the conical flask in a vacuum drying box, set the temperature to 105 ℃, and dry it for 1h under a vacuum of 93 kPa. After that, place the asphalt sample in an aluminum box for storage for relevant test purposes.

Figure 1  Effect of IE and PE on asphalt properties in rubber powder modified asphalt

Figure 2 Extraction process of rubber powder modified asphalt

5. Other amendments:

Table 3.1 in Line 236, but Table 3 in Line 261.

Response:  

Thank your comments. Table 3.1 has been modified to read Table 3.

The MSD (Line 105) and CRE (Line 375) need definition.

Response:  

Thank your comments. The MSD is an abbreviation for molecular size distribution. CRE is an abbreviation for crumb rubber effect.

Line 201, both macromolecules and intermediate molecules are denoted as LMS.

Response:  

Thank your comments. MMS is an abbreviation for middle molecules size.

Line 244, 10.9%.

Response:  

Thank your comments. This error has been corrected.

The x and y axes are not shown in Figure 1 inset. What is the 104 in Line 226?

Response:  

Thank your comments. The coordinates shown in Figure 1 are only molecular size distributions, so it is not necessary to display the x and y coordinate axes.

104 is wrong, it should be 10⁴. Related errors have been corrected.

In Figure 8, the vertical axes are labeled as M_n instead of M_z.

Response:  

Thank your comments. The related error has been modified.

Author Response File: Author Response.docx

Reviewer 2 Report

The Manuscript "Study on molecular distribution of crumb rubber modified asphalt under aging effect based on the Gel Permeation Chromatography test" is very interesting.

Below are some comments to the authors:

1. Line 169: It should be RTFOT - Rolling Thin Film Oven Test

2. Formula 2.1 and 2.2: Please explain I_CRM, I_BB, I_EB.

3. Line 201 and 202: ...between 3000 and 19000 for intermediate molecules (LMS), and less than 3000 for small molecules (SMS). It should be "...between 3000 and 19000 for intermediate molecules (MMS), and less than 3000 for small molecules (SMS)"?

4. Line 207: Mn, the Z average...Please explain "Z"

5. Figure 4: There are references to Figure 4a, 4b, 4c and 4d in the text. Figures 4a and 4c are missing. I don't understand these charts. Identical data, equations, R², but different symbols.

6. Line 336: Mn and LMS, SMS and SMS is studied in Figure 7. There are two "SMS". It should be MMS

7. Figure 8: The vertical axis is incorrectly described. It should be Mz

8. Line 375: Please explain symbol "CRE". No formula for calculating CRE. it is the sum of IE and PE? 

9. Line 392: Figure 11 shows the CRE of Mw....But in Figure 11b it is "CRE on D". Is this correct?

Author Response

The Manuscript "Study on molecular distribution of crumb rubber modified asphalt under aging effect based on the Gel Permeation Chromatography test" is very interesting.

Below are some comments to the authors:

1. Line 169: It should be RTFOT - Rolling Thin Film Oven Test

Response:  

Thank your comments. The error has been corrected.

2. Formula 2.1 and 2.2: Please explain I_CRM, I_BB, I_EB.

Response:  

Thank your comments. Existing experiments have shown that rubber particles not only undergo physical modification, but also undergo chemical degradation in asphalt. Part of the rubber material is incorporated into the asphalt. The purpose of this article is to explore the effect of the chemical reaction of rubber particles on asphalt by using filtration methods. From the actual situation of filtration, compared to the molecular weight distribution diagram of matrix asphalt and rubber powder modified asphalt, filtered asphalt exhibits different characteristics of change, which fully demonstrates the necessity of filtration.

The modification of asphalt by rubber powder involves both the interaction effect (IE) between rubber powder and asphalt, and the particle effect (PE) of rubber powder particles in asphalt, as shown in Figure 1. In order to verify the impact of rubber powder on the composition and structure of asphalt, extraction and filtration were used to filter rubber powder from modified asphalt, and then relevant tests were conducted on the filtered rubber powder and extracted asphalt to explore the practical role of rubber powder in asphalt. The specific process is shown in Figure 2. First, extract about 40 g of rubber powder modified asphalt and put it into a conical flask, dissolve it with 200 ml of trichloroethylene, and then place it in an ultrasonic instrument for 30 minutes to quickly dissolve the rubber powder modified asphalt. Due to the high quality of asphalt filtered in this test, the filtration efficiency of using traditional filter paper is relatively low. This step can be improved by using a vacuum filter to greatly accelerate the filtration efficiency, requiring only replacement of the filter paper each time. The extracted asphalt solution is evaporated using a rotary evaporator. In order to improve the distillation efficiency, multiple experiments were conducted for the solution of asphalt trichloroethylene. At the beginning, the oil bath temperature was set to 55 ℃, the rotation rate was 50 rpm/min, and the rotary distillation was conducted for 30 minutes to quickly recover most of the trichloroethylene. When it is observed that there is no trichloroethylene in the conical flask, increase the oil bath temperature to 155 ℃ and heat for 15 minutes to completely drain the residual trichloroethylene in the asphalt. After the completion of this step, place the conical flask in a vacuum drying box, set the temperature to 105 ℃, and dry it for 1h under a vacuum of 93 kPa. After that, place the asphalt sample in an aluminum box for storage for relevant test purposes.

Figure 1  Effect of IE and PE on asphalt properties in rubber powder modified asphalt

3. Line 201 and 202: ...between 3000 and 19000 for intermediate molecules (LMS), and less than 3000 for small molecules (SMS). It should be "...between 3000 and 19000 for intermediate molecules (MMS), and less than 3000 for small molecules (SMS)"?

Response:  

Thank your comments. This error has been corrected.

4. Line 207: Mn, the Z average...Please explain "Z"

Response:  

Thank your comments. Mz is a parameter in the GPC test. Z represents a higher average molecular weight. Dependent on components with high molecular weight, it is more difficult to accurately determine. It is often necessary to use techniques such as diffusion and sedimentation to measure the behavior of polymer molecules for measurement.

5. Figure 4: There are references to Figure 4a, 4b, 4c and 4d in the text. Figures 4a and 4c are missing. I don't understand these charts. Identical data, equations, R2, but different symbols.

Response:  

Thank your comments. A and c have been omitted and relevant information has been added in Figure 4. The samples of a and b diagrams are the same, representing different parameters. Figure a shows the relevant parameters under different aging conditions. Figure b shows the molecular weight distribution of different asphalt samples. The c and d diagrams are similar.

6. Line 336: Mn and LMS, SMS and SMS is studied in Figure 7. There are two "SMS". It should be MMS.

Response:  

Thank your comments. This error has been corrected.

7. Figure 8: The vertical axis is incorrectly described. It should be Mz.

Response:  

Thank your comments. The related error has been modified.

7. Line 375: Please explain symbol "CRE". No formula for calculating CRE. it is the sum of IE and PE? 

Response:  

Thank your comments. The relevant formula has been added to the article, and the relationship is shown in Figure 2.

Figure 2  Effect of IE and PE on asphalt properties in rubber powder modified asphalt

9. Line 392: Figure 11 shows the CRE of Mw....But in Figure 11b it is "CRE on D". Is this correct?

Response:  

Thank your comments. This error has been corrected.

Author Response File: Author Response.docx

Reviewer 3 Report

This is a well written manuscript and I would like to take this opportunity to congratulate the authors for writing this good manuscript. However, I just had several minor concerns, to be addressed by the authors before this manuscript can be accepted for publication:

1.      The words “study on” can be omitted from the title.

2.      The term or symbol of Mw and D should be expressed in full words first.

3.      Figure 4….b and d? Where are a and c?

4.      Results and Discussion section. The authors only reported what had been done in this study, They should compare and discuss their findings with previous studies done by other researchers as well. Please revise this section accordingly.

Author Response

This is a well written manuscript and I would like to take this opportunity to congratulate the authors for writing this good manuscript. However, I just had several minor concerns, to be addressed by the authors before this manuscript can be accepted for publication:

1. The words “study on” can be omitted from the title.

Response:  

Thank your comments. The words “study on” has been omitted form the title.

2. The term or symbol of Mw and D should be expressed in full words first.

Response:  

Thank your comments. Mw is an abbreviation for Weight average molecular weight. D is an abbreviation for dispersion.

3. Figure 4….b and d? Where are a and c?

Response:  

Thank your comments. A and c have been omitted and relevant information has been added in Figure 4.

4. Results and Discussion section. The authors only reported what had been done in this study, They should compare and discuss their findings with previous studies done by other researchers as well. Please revise this section accordingly.

Response:  

Thank your comments. 

The previous studies are supplemented and compared. GPC analysis confirmed that the content of maltene fraction in crumb rubber modified asphalt decreased from around 75% to 63% after long-term aging effects, the main reason accounting for this phenomenon was that the maltene fraction was potentially absorbed by rubber particles during swelling process or associated in asphaltenes [1]. With the aid of GPC technique, the degradation phenomenon of crumb rubber modified asphalt during aging process was proved by the decreased average molecular weight of the asphalt phase, which was resulted from the low molecular weight processing oil released from rubber particles [2]. GPC analysis also showed that the percent increase of LMS in crumb rubber modified asphalt during aging process was smaller than that of base asphalt, indicating an improved aging resistance caused by released component from rubber particles [3]. As expected, the GPC test for the crumb rubber-modified binders, in which the rubber in asphalt binders was removed using a syringe filter, was effective in evaluating the aging effect of rubber-modified binders for different RTFOT aging times[4].

Reference:

[1] W.H. Daly, S.S. Balamurugan, I. Negulescu, M. Akentuna, L. Mohammad, S.B. Cooper, G.L. Baumgardner, Characterization of Crumb Rubber Modififiers after Dispersion in Asphalt Binders, Energy and Fuels. 33 (2019) 2665–2679.

[2] S. Wang, X. Zhao, Q. Wang, Rheological and structural evolution of rubberized asphalts under weathering, J. Mater. Civ. Eng. 29 (2017) 1–7.

[3] S.J. Lee, J. Hu, H. Kim, S.N. Amirkhanian, K.D. Jeong, Aging analysis of rubberized asphalt binders and mixes using gel permeation chromatography, Constr. Build. Mater. 25 (2011) 1485–1490.

[4]Soon, Lee J , Serji N, et al. Short-term aging characterization of asphalt binders using gel permeation chromatography and selected Superpave binder tests[J]. Construction and Building Materials, 2008, 22(11):2220-2227.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

1. To my first question, the authors explained how to generate the GPC calibration curve. However, my question is focused on the normalized RI response on the vertical axes in Figure 1 and 2 instead of the calibration curve. In Figures 1 and 2, all the curves are normalized to the maxima of 100. What component is this normalization based on? Besides, The authors emphasized, "the intensity of the detector signal is proportional to the weight concentration of molecule corresponding to the current retention time." The normalization changes the height, aka intensity, of the RI response, so those curves cannot represent the actual concentrations of different components when comparing between curves. For example, the 20%-PAV curve in Figure 2 shows a higher RI response than other curves at molecular weights of above 20000. Is this higher RI response due to the actual higher concentration of high-molecular-weight species, or due to the curve shifting upward because of the normalization?

2. To my second question, the authors elaborated on how rubber powders improve the low-temperature performance and aging resistance of asphalt, including the physical and chemical reactions between rubber powder and asphalt in other studies. However, my question on whether the macromolecules are from rubber degradation is still not answered. What molecular weights are the macromolecules produced from the rubber degradation in Figure 2? The shoulder at the molecular weight of 10000 is due to the formation of asphaltenes (Line 235), and the subsequent tailing is suspected to be due to the curve shifting because of the normalization, as discussed in the previous comment.

Besides, FTIR detects the chemical reactions between rubber powders and asphalt, but forming soluble species requires the dissolution of crosslinked rubber into low-cross-degree products. Reference 29 illustrates how the partial swelling and degradation of crumb rubber into linear molecules when mixing in asphalt. Can PAV dissolute rubber powder into soluble species as the high-temperature mixing does?

3. In Response 4, if the filtered components cannot pass through the Buchner funnel, can they be eluted from the GPC column?

Author Response

1. To my first question, the authors explained how to generate the GPC calibration curve. However, my question is focused on the normalized RI response on the vertical axes in Figure 1 and 2 instead of the calibration curve. In Figures 1 and 2, all the curves are normalized to the maxima of 100. What component is this normalization based on? Besides, The authors emphasized, "the intensity of the detector signal is proportional to the weight concentration of molecule corresponding to the current retention time." The normalization changes the height, aka intensity, of the RI response, so those curves cannot represent the actual concentrations of different components when comparing between curves. For example, the 20%-PAV curve in Figure 2 shows a higher RI response than other curves at molecular weights of above 20000. Is this higher RI response due to the actual higher concentration of high-molecular-weight species, or due to the curve shifting upward because of the normalization?

Response:  

Thank your comments. 

The raw data obtained through GPC may experience baseline drift in some samples due to instrument and solvent reasons, and it is necessary to use Origin software to correct the baseline of the data. Due to the impossibility of identical sample configuration concentrations, in order to horizontally compare the chromatograms of samples with different concentrations, the corrected baseline data is normalized based on the peak signal of the chromatogram peak. This allows for a visual comparison of the differences between different test samples without affecting the average molecular weight and molecular distribution results.

The GPC test data processed through normalization can only be used for qualitative analysis of samples, but cannot establish quantitative relationships with other test methods through quantitative analysis. Since the area under the elution curve is directly proportional to the relative concentration of each component, it is feasible to divide the area for quantitative analysis. In the field of road engineering, GPC data is usually divided into three parts based on size, namely large molecules (LMS), medium molecules (MMS), and small molecules (SMS). The area of each part is calculated using different partitioning methods based on the area enclosed by the chromatogram curve and the x-axis. According to different research needs, there are slight differences in the methods used by researchers in the field of roads when dividing chromatograms. The mainstream partitioning methods include dividing the chromatogram into 13 equal parts and into 3 parts: large particle size molecules (LMS, 1-5 equal parts), medium particle size molecules (MMS, 6-9 equal parts), and small particle size molecules (SMS, 10-13 equal parts). Alternatively, according to the molecular weight range, the test sample molecules can be divided into three parts: polymer (molecular weight greater than 19000), asphaltene (molecular weight 3000-19000), and soft asphalt (molecular weight less than 3000). Alternatively, through statistical methods, it can be directly divided into three parts based on the outflow time. After a unified comparison of the three methods mentioned above, it was found that there was little difference in the three processing methods for asphalt samples. However, when processing data on saturation, aromatics, resins, and asphaltenes, there was a significant error in using the 13 equal division method. In this regard, this article converts the measured sample data into molecular weight, using values greater than 19000 to represent large molecules (LMS), values between 3000-19000 to belong to medium molecules (LMS), and values less than 3000 to belong to small molecules (SMS), in order to quantitatively analyze the molecular weight of asphalt and its components. Calculate and process all data through Excel.

Meanwhile, the 20%-PAV curve in Figure 2 shows a higher RI response than other curves at molecular weights of above 20000. The higher RI response due to the actual higher concentration of high-molecular-weight species.

2. To my second question, the authors elaborated on how rubber powders improve the low-temperature performance and aging resistance of asphalt, including the physical and chemical reactions between rubber powder and asphalt in other studies. However, my question on whether the macromolecules are from rubber degradation is still not answered. What molecular weights are the macromolecules produced from the rubber degradation in Figure 2? The shoulder at the molecular weight of 10000 is due to the formation of asphaltenes (Line 235), and the subsequent tailing is suspected to be due to the curve shifting because of the normalization, as discussed in the previous comment.

Besides, FTIR detects the chemical reactions between rubber powders and asphalt, but forming soluble species requires the dissolution of crosslinked rubber into low-cross-degree products. Reference 29 illustrates how the partial swelling and degradation of crumb rubber into linear molecules when mixing in asphalt. Can PAV dissolute rubber powder into soluble species as the high-temperature mixing does?

Response:  

Thank your comments. 

Through infrared spectroscopy and GPC testing, it can be shown that rubber undergoes a degradation reaction in asphalt, with some rubber substances entering the asphalt, resulting in a change in the molecular weight of the asphalt. Regarding the issue you mentioned, regarding whether the macromolecular substances in asphalt come from rubber, I believe the proportion is relatively large. Because the more rubber powder is added, the higher the proportion of macromolecular substances in asphalt. This to some extent proves that some of the macromolecular substances in asphalt are degraded from rubber. Currently, regarding the high molecular weight of rubber in asphalt

By observing the infrared spectrum (Figure 1) of the modified asphalt with 20% rubber powder, it can be observed that, taking JB70 rubber powder modified asphalt as an example, compared to the base asphalt, some new characteristic peaks appear at 670 and 1100cm-1, corresponding to S-C bonds and Si-O-Si, respectively. The appearance of S-C bonds is due to the presence of cross-linking bonds in the rubber powder, while Si-O-Si is caused by the entry of silica (the main component of SiO2) into the asphalt, which is the filler inside the rubber powder. The emergence of these new peaks indicates that rubber powder undergoes complex material exchange and reactions in asphalt. At the same time, intuitively, the addition of rubber powder leads to changes in the peak strength of some functional groups. Taking the carbonyl group at 1700cm-1 as an example, it can be seen from Figure 1 that compared to the base asphalt, its peak strength is significantly reduced. The carbonyl group can qualitatively indicate the oxygen-containing polar functional groups such as aldehydes, ketones, esters, carboxylic acids generated during the asphalt aging process, which is an important indicator of asphalt aging. The reduction indicates that rubber powder can significantly delay asphalt aging and extend the service life of asphalt. The other two types of asphalt also have similar conclusions.

Figure 1 Infrared spectrum of rubber powder modified asphalt

The infrared functional group spectrum of rubber powder modified asphalt can only qualitatively analyze the changing characteristics of each functional group. Due to the different peak heights and absorption strengths, it is impossible to analyze the specific changes in chemical functional groups before and after aging. Quantitatively analyze the changes in chemical functional groups of rubber powder modified asphalt using the above determined analysis method.

The carbonyl index and sulfoxide index are two important indicators to characterize the degree of asphalt aging. In order to analyze the impact of different rubber powder dosages on the degree of asphalt aging, the carbonyl index and sulfoxide index of modified asphalt with different rubber powder dosages under long-term aging were first analyzed. The specific results are shown in Figure 2. From it, it can be seen that compared to the control group, the carbonyl index of asphalt after adding rubber powder has significantly decreased. Taking the 20% rubber powder dosage as an example, the carbonyl index of SK70 asphalt has decreased by 85.28% compared to the matrix asphalt, JB70 asphalt has decreased by 47.89%, and JB90 asphalt has decreased by 81.69%. This indicates that rubber powder can effectively delay the production of polar functional groups such as carbonyl, thereby delaying the aging process of asphalt. The effect of delaying aging varies slightly depending on the type of asphalt. From the trend of sulfoxide based index changes, for different types of asphalt, the sulfoxide based index generally shows a decreasing trend with the increase of rubber powder content. Taking 20% rubber powder as an example, compared with the base asphalt, the sulfoxide base index of SK70 asphalt decreased by 48.43%, JB70 asphalt decreased by 15.97%, and JB90 asphalt decreased by 59.68%. This shows that the addition of rubber powder can effectively inhibit the increase of sulfoxide S=O content, indicating that rubber powder can inhibit the addition reaction between sulfur elements and asphalt components in asphalt.

Figure 2: Change trend of chemical functional group index of CRMB under PAV state: (a) carbonyl index; (b) Sulfoxide based index

Figure 3 shows the changes in the Si-O-Si functional group index of the 1100cm^-1 peak. By analyzing the infrared spectrum of SK70 asphalt, no characteristic peak was found at 1100cm^-1. Only the changes in the characteristic peak of the aging state of PAV modified asphalt with JB70 and JB90 rubber powder were analyzed at this location. From Figure 3 (a), it can be seen that the I_-1100 characteristic peak index of the two types of rubber powder modified asphalt gradually increases with the increase of rubber powder content, indicating that the SiO₂ content in the asphalt continues to increase, indicating that the white carbon black filler used in the rubber powder is gradually released into the asphalt. This result is caused by the desulfurization degradation reaction of the rubber powder. The more the amount of rubber powder added, the more obvious the effect is. This is related to the low-temperature performance of two types of rubber powder modified asphalt. As the amount of rubber powder added increases, the low-temperature performance gradually improves, indicating that under the aging state of PAV, the release of white carbon black in the rubber powder into the asphalt plays a positive role in improving the low-temperature performance of asphalt. Figure 3 (b) shows the changes of two types of asphalt under different aging conditions. As the aging degree gradually deepens, the index in the asphalt gradually increases, which means that the deepening of aging degree promotes the release of more white carbon black from the rubber powder into the asphalt. It indicates that under the combined action of heat and oxygen, the rubber powder will continuously undergo desulfurization and degradation reactions, releasing substances such as white carbon black from the rubber powder into the asphalt phase, and engaging in complex material exchange activities with the asphalt.

Figure 3: Change trend of 1100cm^-1 peak functional group index: (a) Different rubber powder dosages; (b) Different aging states

3. In Response 4, if the filtered components cannot pass through the Buchner funnel, can they be eluted from the GPC column?

Response:  

Thank your comments. 

If the filtered components cannot pass through the Buchner funnel, they can not be eluted from the GPC column.

Author Response File: Author Response.docx

Round 3

Reviewer 1 Report

1.       Thanks to the authors’ detailed explanation, now the GPC results in Figure 1 and 2 are better understood. I have one concern left: “All the corrected baseline data are normalized based on the peak signal of a chromatogram peak”, does the intensity of this chromatogram peak change between tests or samples? Consider a scenario of RI response as follows.

After correcting baseline, but before normalization:

Sample 1: the chromatogram peak of 50, component X of 10.

Sample 2: the chromatogram peak of 60, component X of 10.

After normalization:

Sample 1: the chromatogram peak of 100, component X of 20,

Sample 2: the chromatogram peak of 100, component X of 17.

Comparing the normalized RI response associated with component X will mislead to an incorrect conclusion that Sample 1 contains more component X than Sample 2, though they actually have the sample amount of component X.

2.       In the first revision, the authors clarified that the GPC samples discussed in Session 3.2 are rubber powder modified asphalt, and the samples contain rubber particles. Whereas the GPC samples discussed in Session 3.3 are filtered asphalt modified with rubber powders, and the samples do not contain rubber particles. Here the authors confirmed that the filtered components that cannot pass through the Buchner funnel will not be eluted from the GPC column. If so, either filtered using a Buchner funnel or not, the components that pass through the GPC column should not contain rubber particles. So why are the binders tested again after filtration/extraction?

Author Response

Comments and Suggestions for Authors

1.Thanks to the authors’ detailed explanation, now the GPC results in Figure 1 and 2 are better understood. I have one concern left: “All the corrected baseline data are normalized based on the peak signal of a chromatogram peak”, does the intensity of this chromatogram peak change between tests or samples? Consider a scenario of RI response as follows.

After correcting baseline, but before normalization:

Sample 1: the chromatogram peak of 50, component X of 10.

Sample 2: the chromatogram peak of 60, component X of 10.

After normalization:

Sample 1: the chromatogram peak of 100, component X of 20,

Sample 2: the chromatogram peak of 100, component X of 17.

Comparing the normalized RI response associated with component X will mislead to an incorrect conclusion that Sample 1 contains more component X than Sample 2, though they actually have the sample amount of component X.

Response:  

Thank your comments.

The question you raised is very good. In order to avoid errors caused by testing strength, when preparing GPC samples, it is required that the mass of GPC samples must be within 2mg, and the error cannot exceed 5%. Due to the fact that the sample mass and solvent volume used are the same, the final test concentration and strength of the obtained sample are considered to be the same.

2. In the first revision, the authors clarified that the GPC samples discussed in Session 3.2 are rubber powder modified asphalt, and the samples contain rubber particles. Whereas the GPC samples discussed in Session 3.3 are filtered asphalt modified with rubber powders, and the samples do not contain rubber particles. Here the authors confirmed that the filtered components that cannot pass through the Buchner funnel will not be eluted from the GPC column. If so, either filtered using a Buchner funnel or not, the components that pass through the GPC column should not contain rubber particles. So why are the binders tested again after filtration/extraction?

Response:  

Thank your comments.

In the second revision, the opinion proves that the rubber particles undergo degradation reaction, and some rubber components are integrated into the asphalt. The rubber particles are filtered to obtain the filtered asphalt. Testing the molecular weight of filtered asphalt further proves that rubber undergoes degradation reactions in asphalt due to aging, thereby proving the mechanism of rubber's effect on asphalt performance.

Author Response File: Author Response.docx