• General comments

The paper presents a new procedure for single grain IR-RF dating using a CCD camera fitted on a Lexsyg Research device. The manuscript presents a successful spatially resolved (SR) IR-RF dose recovery test on an artificially irradiated sample. A second dose recovery test on a naturally sample however gives more scattered dose results with several components. While the lowest component agrees with the expected dose, the average dose overestimates the expected dose by 50%. These experiments also reveal an unstable signal background due to a degenerated cooling system, which is important to check before conducting any SR IR-RF measurements. Besides these issues, the SR procedure seems robust and simple to use. I appreciate the description of the camera settings, and the statement on the scientific dimension. This paper is a great contribution for IR-RF users and I therefore recommend its publication in Geochronology.

• Specific comments

I only have minor comments.

Page 6, Figure 2: how was the exposure time determined at 4.2s? Why does the channel time (5s) higher than the exposure time?

Page 7: The solar simulator settings here are not the same as Frouin et al. 2015. Please check the numbers.

Moreover, the UV intensity has been doubled. An increase in UV induces a higher temperature during bleaching, which can activate a shallow TL peak (at ~120C at 10C/s, Huot et al., 2015). I am therefore not convinced that:

1. 1hour pause is sufficient for the thermocouple to completely cool down (see Huot et al., 2015 fig. 3) and for the phosphorescence signal to completely disappear,
2. and that the RFreg signal truly compares with the RFnat signal (if the shallow TL peak has been indeed activated during bleaching).

I  wonder if the large scatter on the natural sample can be due to this variation in temperature during the IR-RF measurement procedure (due to high UV contribution) and/or insufficient pause.

Page 14, line 281: what signal did you use for the Feldspar paleodose?

Page 14, line 289: same question

Page 17, figure 6. It would have been nice to see a picture of each aliquot in daylight for the readers to have an idea where the signal is coming from and see how close to each other the grains actually are. It looks like the light is coming from a much smaller area than the grain. Could you comment on that?

• Technical corrections

Page 2, line 28: there are three “and” in one sentence. Please remove one at least.

Page 2, line 42: (e.g. Duller and Roberts (2018)) replaced by  (e.g. Duller and Roberts, 2018) 

Page 6, Figure 2, caption: change the quote marks

Page 6, line 141: “measurements: one….”

Page 15, figure 5, B, the graph shows a IRreg signal measured for 8000s but line 298, it is written that the measurements were done until 10,000s.

In caption, please add “grain diameter: 7 px”

Page 15, line 295: did you use silicon to mount the grains?

Page 18, line 357: “IR-RF age of ca 31 ….Gy” and “quartz age of 26.1… Gy”, you mean “ka” I guess?

General comments

The paper from Dirk Mittelstraß and Sebastian Kreutzer entitled "Spatially Resolved Infrared Radiofluorescence: Single-grain K-feldspar Dating using CCD Imaging" appears as an experimental or technical report on the instrumental prerequisites for spatially resolved Infra-Red Radiofluorescence (IR-RF). It is a pure factual description of many technical aspects. The authors promise to give full access to experimental details which should be of great interest for further investigations in this field.

It should be emphasised that the paper does not deal with all the issues necessary for age determination, it is focused on the reliable determination of D_e values by IR-RF on a set of single K-feldspar grains.

The paper is written in a clear and understandable way. Many technical details are given; theoretical principles are summarised in the appendix. The list of references includes the most important investigations of the investigated topic. I was particularly pleasured by the fact that the authors digged for the first publication on spatially resolved IR-RF by M.Krbetschek (although it was only a conference contribution).

Specific comments

There are just a few remarks and/or questions which should be considered in the final version of this paper:

• the bright spots were associated by the authors with bremsstrahlung photons from the beta source. To my knowledge, M.Krbetschek correlated this effect with energy deposition from cosmic radiation (muon component). Was this idea checked? Are there arguments against this idea?
• Figure 3: the before/after picture for step 2 (image alignment) requires some explanation. A blurry picture is presented as an improvement, this is not understandable at first glance.
• Concerning the discussion about the D_e distribution of BDX16651: Dosimetric aspects should be more discussed. The existence of a broad D_e distribution does not necessarily result in a broad age distribution, if the K content varies from grain to grain. Here spectrometric methods for the K determination of single grains could help, e.g. far-red radiophosphorescence (Dütsch, C., Krbetschek, M.R., 1997. New methods for a better K-40 internal dose rate determination. Radiation Measurements 27, 377-381. and Krbetschek, M.R., Goetze. J, Irmer, G., Rieser, U. and Trautmann, T., 2002. The red luminescence emission of feldspar and its wavelength dependence on K, Na, Ca-composition. Mineralogy and Petrology 76, 167-177.)
• In Sec. 4.1 the proposed mirror is called in the text "parabolic" but in the picture "spherical". What is correct? And - would it be beneficial to use an elliptical mirror with camera and sample in the focuses?
• Although the main goal of the paper are technical aspects, in Sec. 4.2 the authors should give a list of problems to be solved before the method can be used for "real" dating purposes.

Technical comments

Besides these notes, some additional technical comments can be found directly in the attached copy of the paper.

This paper presents a new approach to date feldspar samples using infrared radiofluorescence (IR-RF) signal by targeting individual grains. Authors named this approach as spatially resolved (SR) IR-RF. This approach exposes all the grains with radiation at a time and captures the luminescence emission from all of them in a single frame. The image (at a certain amount of dose) is then processed to extract luminescence/counts from each grain.

Such a development is not easy and, therefore, authors must be appreciated. They have succeeded in presenting their work quite well. As a first paper discussing single-grain IR-RF measurements, it has enough information and results too. Therefore, I recommend this paper to be published in Gchron. Though I have a few queries and suggestions; I want authors to address them.

The paper is written clearly and well-presented but the authors are suggested to check a few grammar mistakes (such as the use of articles) and typos or misplaced text within a sentence including the type of English (American, British or others). I have pointed out some (find them below) but still, authors need to look at such errors throughout the text. My comments and general queries are as follows (P stands for page number):

• Figure 5: do authors see an initial rise in the signal in these samples? The signals used as the examples in Figure 3 have it. Does this mean it is not there in the samples investigated here (Figure 5)? Have they checked this between the grains? How many initial channels (time) were removed in De determination?
• Authors say they see at least four-dose components. Have they compared their dose response curves (DRCs)? Could they share the data? The authors have not presented any data comparing DRCs within a sample. This is crucial since one would want to know the extent of variation in the DRCs (i.e. if any saturates later or not) in order to evaluate grain-dependent saturation (or dynamic range).
• Do authors think that the grains which were not found bright, in reality, photons were produced but not detected due to the overall noise? In that case, noise can still be suppressed and therefore EMCCD is required since below a threshold detection point EMCCD is still better than sCMOS cameras. Do authors think that it is not important to look for those grains and only the bright grains (and resulting DRCs) will do?

Specific comments are:

P2, line 39-40: I really think this sentence needs a word ‘from’ before single grains.

P2, line 43: It is ‘charged coupled device (CCD)’.

P3, line 67:  Seems ‘a’ is in the wrong place. Further, I do not think it is important to expand SR again since it is already done earlier. One time explanation is enough.

P7, line 148-150: Actually, most of the solar simulator settings differ a bit when compared to Frouin et al. (2015). Therefore, it is good to mention about all of them, not just UV.

P8, line 178: ‘a’ setting?

P8, line 183: ‘(2) it reduces the dynamic range and the linearity of the signal acquisition’…! I think this is partially true. There is a limit to which EM gain can still be applied maintaining full well capacity. Under this limit, the EM gain can increase the dynamic range by reducing the read noise.

P10, line 210: ‘images are combined to one image by taking the median pixel value for each pixel location….! Does this mean the total no. of images are reduced by the group size during median filtering and the measurement time of each image is summed?

P11, line 222: Authors should provide more information on interpolation methods or provide references, therefore, must expand this paragraph.

P14, line 290, The sigma here represents error?

P15, Figure 5: Good to use unit i.e. px for diameter.

P16, line 329: define sigma_m.

P17, Figure 6: Is it possible to re-number the grains 1-10? As the text says only 10 grains were selected. Further, It would be good to keep the horizontal label within the image’s boundaries.

P18, line 356-357: The unit will be in years, not Gy.

P19, line 371: It would be good to cite any work relevant to the potential degradation of the filter under ionising radiation.

P20, line 408: Not sure what effect?

P21, lin3 433: Close the bracket around SR IR-RF.

P23, line 481: the image ‘is’.

P25, line 514: dpixel = 16 μm – Is this value from the manufacturer?

P25, line 520: Luminescence signal (phi_grain) per grain is basically the sum of counts from each pixel in an ROI (that has a fixed number of pixels), isn't it? The difference between eq A1 and A6 lies between how phi_grain is chosen: per pixel or in an ROI. am I correct? I think the sum needs to be mentioned.