This work presented high-precision measurement results of CO2 triple oxygen isotopes from upper troposphere lower stratosphere air samples up to 21 km collected during past aircraft campaigns. The results are interesting as it showed distinct relationship between triple oxygen isotopic compositions and N2O for air in the upper troposphere vs. lower stratosphere. Such observation is critical to enable CO2 triple oxygen isotopes as a tool to understand stratosphere-troposphere exchange, as well as global carbon cycle. This work highlighted the importance of high-precision triple oxygen isotopes measurements during quantification of the downward net isoflux of O-MIF signal. While the presentation of the results is clear, interpretation of the data is mostly adequate, I have a few minor general comments:

1. It would be great if there is more discussion about the de-coupling of chemical mechanisms of CO2-O17 generation and N2O loss in the stratosphere due to stratospheric dynamics (Lines 325 - 344). The CO2-O17 signal is originated from ozone chemistry, therefore the path history (O1D abundance vary greatly in the stratosphere) and age of the air parcel are both important; while N2O is more sensitive to altitude as the photochemical lifetime of N2O decrease exponentially in the stratosphere. Therefore, air parcels that are relatively “young” but have been to mid-stratosphere (~30 km) could have significant N2O loss but low O17, and vise versa. Clarifying some of these mechanisms could be useful.
2. More discussion may be needed to support the argument that the slope from CARIBIC samples can represent a “global average” N2O-O17 slope. Because of the observed potential “de-coupling” of N2O-O17 slope, it could be useful to discuss what are the potential factors that could result in different slopes. If the well-mixed upper trop air from CARIBIC represents global average slope, StratoClim gives you “below average” slope, where can you anticipate “higher than average” slopes? How will such variations impact the uncertainty of the global average slope?
3. If the uncertainty in “global average” slope changed because of 1) and 2), how does it impact the uncertainties in global estimation of O17 isoflux?

More detailed comments:

Sections 2.2 & 2.3: since these were not mentioned until section 5, maybe considering moving these down (after 2.5) a little bit?

Line 174: maybe briefly mention how age of air is calculated?

Line 249: CO2 not CH2.

Figure 5: subpanel titles (a, b, c, d) are not lined up.

Figures 3-8: ∆^{′ 17}O is used in your text but in figures you used “∆¹⁷O”, please consider using consistent notations.

Line 368: uncertainty inconsistent with figure.

Lines 388 & 394: repetitive sentences.

This study presents triple oxygen isotope (Δ′17O) data of CO2 of tropospheric and stratospheric samples derived from two aircraft campaigns. Δ′17O is a novel isotope tracer that can provide an additional constraint on troposphere-stratosphere exchange and help to more accurately quantify exchange fluxes in the global carbon cycle. The authors show that Δ′17O-N2O relationship differs for upper tropospheric and lower stratospheric air. Using a mass balance model, they use their data to quantify global carbon fluxes. Their results demonstrate the potential of high-precision Δ′17O(CO2) measurements for quantifying and refining grow carbon fluxes and understanding mixing, transport and production processes at the troposphere-stratosphere boundary.

The manuscript is well-written, methods are described in detail, results are well-illustrated and key findings are highlighted in the discussion. I have only some very minor comments, as outlined below.

Minor comments:

Line 119: “the box model used in this study” you refer to the study of Koren et al (2019) or to the present study? Please clarify. Also, can you briefly describe the model of Hoag et al (2005 and Liang et al (2017b) on which the model you used in this study is based?

Line 149: which type of distribution did you use for the input parameters? Is it uniform or normal?

Line 156: I found this sentence confusing. Consider reformulating to something like: To ensure comparability of previously published data and the CARIBIC and StratoClim samples measured in this study, …

Line 168/169: Can you explain briefly how this correlation allows differentiation between tropospheric and stratospheric samples?

Section 2.5-2.7: I suggest following the chronological order and presenting the sampling and isotope analysis of the samples used in this study before presenting the data-processing, calculations and models applied, where these are used.

Figure 3: I was confused by the illustration as histograms. First, I was thinking the height reflects the number of samples. You may think about illustrating the data just a points with SE?

Line 269 ff: I suggest presenting data related to Figure 6 before Figure 5. Like this, you first present correlations found in your study before comparing both the triple oxygen isotope plot as well as correlations between N2O-Δ′17O and CH4-Δ′17O with previously published data.

Conclusions: It would be great if the authors could highlight some key findings on the application of Δ′17O(CO2) here. I refer here mainly to the ability to identify mixing, transport and production processes at the troposphere-stratosphere boundary.

Technical comments:

Line 114: doubling “depends on the uncertainty”. Remove one.

Line 118: change point to comma after “(see Figure 1)”.

Line 210: the third and “fourth” traps.

Line 320: Remove “are”. “The samples they analyzed are mostly represent upper troposphere air”

Line 137: Repetition “however” two times in a row. Consider changing to “in contrast”. Also in this line: “In the middle and stratosphere”. Here seems to be missing a word.