This manuscript describes measurements of light-absorption properties and chemical composition of ambient water-soluble HULIS samples collected during haze episodes and clean periods. The measurements involve comprehensive chemical analyses including carbon analysis (OC/EC and TOC), ion chromatography for inorganic ions, and ESI FTICR MS for organics. The light absorption properties were quantified using UV-vis spectrometry.

General comments:

1. The data set produced in this study, especially the ESI FTICR MS data, is extensive and informative. However, there are various instances where assertions are made that are not supported by the data, and are at some points contradictory with other assertions in different parts of the manuscript. Please see specific examples under ‘Specific comments’ below.
2. The term HULIS as used in this manuscript needs to be better defined. HULIS is a vague term – much like brown carbon – that has been used to refer to different things in different studies. Here, HULIS is obtained based on an extraction procedure that isolates the less polar fraction (~50%) of WSOC. It would be helpful for the reader to explicitly indicate in the methods section that this definition is operational, and also contrast the definition of HULIS in this study with other studies. This is important because the results are compared to multiple previous studies on HULIS, and it should be noted that not all HULIS are defined the same way.

Specific comments:

1. Section 2.4 and 2.5 should be combined: ESI-MS is also chemical analysis.
2. Line 165: The manuscript presents results of PM2.5 concentrations, but there is not description of how the PM2.5 concentrations are measured in the methods section.
3. Figure 1: There are inconsistencies in the x-axis values: the distance between the major ticks changes between 1 day (e.g. 1/24 – 1/25) and 2 days (e.g. 1/10 – 1/12).
4. Figure 1f: how come the Lev/OC value are larger than 1? Lev is one of many OC species.
5. Line 218-219: The statement that Lev/OC increased in haze-II is not accurate. There are 2 data point for Lev/OC in haze-II (Figure 1f): one is higher than haze-I and one is lower than haze-I.
6. Line 232-233: This is not valid. AAE is a measure of the wavelength dependence of light absorption, not the magnitude of light absorption.
7. Line 248-260: The statement on line 251 that MAE of HULIS is generally higher than WSOC is not valid. The values for HULIS (1.1 +/- 0.27) and WSOC (1 +/- 0.21) are virtually the same. In Lines 254-256, MAE of HULIS (1.1 +/- 0.27) is said to be “comparable” to other values ranging between 0.91 and 1.84. Then in line 257-259, MAE of 0.91 is said to be “much lower” than MAE of 1.3. These statements are subjective and inconsistent.
8. Line 263-266: The argument that stagnant conditions lead to prolonged oxidation thus lower MAE for haze versus clean days is not convincing. It is not clear that the PM sampled during the haze days had longer atmospheric lifetime / OH exposure. What if the PM in the clean days had more contribution from long-range transported PM?
9. Line 285-286: It is not clear how the presence of these 3 molecules suggests contribution from biomass burning and vehicular emissions.
10. Line 308-318: This paragraph mentions that HULIS in haze days had higher MW than in clean days, and makes the point that high MW HULIS is more resistant to chemical transformation. This is in contrast with the assertions in section 3.2 that MAE on haze days were smaller than on clean days because HULIS on haze days underwent more chemical transformation.
11. Line 319-330: This paragraph makes the point that lower AI_mod for haze days can be due to photooxidation during haze days and explain that lower MAE for haze days. How is this assertion reconciled with the larger MW and resistance to oxidation mentioned in the previous paragraph?
12. Line 374-378: It does not look like the statement “relatively low BBOA content” is supported by the data in Figure 3. Most of the molecules are clustered in the region identified as BBOA. In any case, previous parts of the manuscript mention BBOA as being an important contributor to HULIS measured in this study, but this paragraph mentions that traffic sources are more important.

Minor comments:

1. Line 80-83: The statement talks about ‘recent years’ but is supported by a reference from 2014. A newer reference is needed.
2. Line 91: What is meant by ‘exact’?
3. Line 99-102: This sentence is not comprehensible.
4. Line 180-182: This statement is not valid. Wind speed alone does not dictate stability (See stability classifications by Turner 1970). In fact, for an unstable atmosphere, increasing wind speed makes the atmosphere less unstable.
5. Line 277-279: Vague statement. In what sense are the peaks “comparable” with peaks from other studies?
6. Line 337-339: I assume you mean biomass burning aerosol (not biomass burning mixture). In any case, what does “comparable” mean here?

​​This manuscript reports the measurements of the water-soluble organic carbon (WSOC) and WS-HULIS fraction of PM_2.5 samples collected in a developed region with dense populations during a haze event; the authors conducted a comprehensive analysis of the chemical composition and light absorption with their samples. They investigated the evolution of light absorption and molecular properties of these samples during one haze cycle (clean-haze bloom-haze decay-clean). While I think the subject is very interesting, there are several issues with this manuscript, including the analysis and conclusions, which are detailed below.

General comments:

1. It will be necessary for the author to provide more details in the results and discussion, especially when making deductions and conclusions. This manuscript provides a very comprehensive dataset of the chemical and optical analysis of their samples. However, In the data interpretation, some of their conclusions/statements are given too simply and vaguely, which needs to be supported with more details and be more specific. For example, in explaining the variance in properties (MW, MAE, etc) of HULIS and WSOC obtained during different stages of the haze event (e.g., haze day vs clean days), the authors mostly attribute these differences to statements such as “effects of aging/oxidation/degradation” without further explanations or details. Since the aging process involves many different pathways and mechanisms, the authors will need to be more specific in the results when explaining the data other than just simply stating “aging”.
2. The “HULIS fraction” used in this manuscript needs to be better described and defined since this is the major substance studied here. Adding some brief descriptions in the introduction would be necessary. HULIS was first reported as macromolecular organic substances in atmospheric aerosol particles; and when used to refer to the light absorbing properties of the atmospheric aerosols, “HULIS” – humic-like substance, is more describing the similarities in the light absorption of the light absorbing organic carbons (e.g., brown carbon) with humic substances, which is a light absorption that sharply decreases from UV to visible wavelength. Also, many different methods for the extraction/isolation of HULIS have been reported, and the potential effects of the extraction/isolation procedure on the chemical/physical nature of HULIS have aroused concerns as well. For instance, the HULIS part could be large molecules formed by intermolecular force (aggregates) and the procedure (extraction solvent, PH adjustment, SPE, etc.) would largely change it. I think a discussion of the nature of HULIS and a brief explanation of your choice for the isolation method, at least under the circumstances of this manuscript, is important for the data interpretation later in the result section.
3. In order to highlight the novelty of this manuscript, it’s better to have a brief summary of the work about HULIS fractions/WSOC of PM_2.5 in the Pearl River Delta with emphasis on the major contributions of this manuscript to the current state of knowledge regarding this topic. I do notice there is one sentence mentioning that there are previous works regarding this topic(line 83-86), however, I think brief descriptions of the referred studies are necessary. Also, is this the only work studying the chemical/optical evolution of WSOC/HULIS/Brown carbon collected in PRD? If not, what have the other studies done? 

Specific comments:

Line 37: comment for writing — “stronger” than what?

Line 44: “natural environment” It is better to be more specific here about what environment (e.g., natural aquatic / soil environment).

Line 45-46: “> 70% of light absorption in water-soluble brown carbon (BrC)” This need to be more specific about the wavelength/wavelength range or what parameters they used to compare (e.g., mass absorption coefficients, etc), if possible.

Line 114: “Field blank samples were collected without power on.” Does this mean that the blank filter is “conditioned” in the air sample holder rather than conditioned in the sampling environment (e.g., passing particle-free air through the filter)? If so, can you explain why you chose this way as “blank control”?

Line 121-122: “Briefly, portions of the PM_2.5 samples (100 cm²) were ultrasonically extracted with 50 mL of ultrapure water for 30 min. The extracts were filtered through a 0.22-μm PTFE syringe filter and then adjusted to pH of 2 with HCl…”. I didn’t find the description of the WSOC fraction here. Or is the extraction of WSOC the same as described for HULIS except for the PH adjustment and following procedure? I think this needs to be clarified here.

Line 232-233 “This difference may be related to the higher enrichment of light-absorbing organic species in HULIS.” The statement is oversimplified and needs more specific explanation and evidence to describe why the enrichment of light-absorbing OC doesn’t simply enhance the light absorption at certain wavelengths but the wavelength dependence (AAE). 

Line 263-265: it might need to be more specific about the “secondary oxidation reaction” and “photolytic aging” here. It is "prolonged" or "enhanced" oxidation during haze? for example, the increased ozone levels associated with high levels of PM_2.5 could be a drive for oxidation during haze days.

Line 315-318: The author needs to mention that the samples used in Dasari and Wong’s work are from different sources (mostly biomass burning aerosols) and are different from the samples used in this manuscript. In addition, the method for molecular weight estimation used in Wong’s paper is also very different from this method (size-exclusion chromatography), in which the MW is estimated by the SEC column retention time, and it’s also highly dependent on the column, the mobile phase, and the sample itself (e.g., polarity, aggregation, etc). Also, line 317-318 is a bit self-contradictory: HULIS in haze undergo stronger oxidation, which usually leads to fast degradation. However, "a longer aging process" means higher stability (longer lifetime), which means the HULIS is more stable during haze days. Or, is the “longer aging” simply implying that because they have a higher MW, they will have a longer lifetime in the atmosphere? There needs some clarification. 

Even if the author wanted to use these two references to support their inference about the higher stability of HULIS in this manuscript,  they didn’t try to explain the variance in HULIS MW from different stages of the haze event (haze days vs clean days). 

Line 324-325: “These differences can be attributed to bleaching or degradation of aromatic compounds.” As I understand it, “degradation” is one pathway that leads to bleaching, i.e., the degradation/oxidation of aromatic compounds could result in the bleaching of HULIS. There might need some clarification or rephrasing.

Line 573-574: It is unclear if the “longer aging process” here means “a longer lifetime." Is this an inference or an observation?