Nonvolatile nuclear spin memory enables sensor-unlimited nanoscale spectroscopy of small spin clusters

In nanoscale metrology, dissipation of the sensor limits its performance. Strong dissipation has a negative impact on sensitivity, and sensor–target interaction even causes relaxation or dephasing of the latter. The weak dissipation of nitrogen-vacancy (NV) sensors in room temperature diamond enables detection of individual target nuclear spins, yet limits the spectral resolution of nuclear magnetic resonance (NMR) spectroscopy to several hundred Hertz, which typically prevents molecular recognition. Here, we use the NV intrinsic nuclear spin as a nonvolatile classical memory to store NMR information, while suppressing sensor back-action on the target using controlled decoupling of sensor, memory, and target. We demonstrate memory lifetimes up to 4 min and apply measurement and decoupling protocols, which exploit such memories efficiently. Our universal NV-based sensor device records single-spin NMR spectra with 13 Hz resolution at room temperature.

In this paper the authors describe and experimentally demonstrate a protocol for achieving a tenfold improvement in the nuclear spin sensing sensitivity using single NV center in diamond. This is achieved by using an ancillary nitrogen nuclear spin as a memory to partially store phase information during two separate interrogation steps. Further, the authors show that this memory is robust enough such that the electronic sensor spin and the target spin can be decoupled during the waiting time by optical pumping or even ionization techniques. This suppresses the sensor induced dephasing of the target spin. Both techniques combined allow very narrow linewidths, which are no longer limited by the inverse of the T1 time of the electron spins.
The experimental demonstration of a tenfold improvement of the resolution of NV-based nanoscale NMR spectroscopy is certainly a nontrivial achievement, given the already quite advanced stage of this field, and also an important step towards real chemistry applications. The authors combine a several experimental techniques and identify an optimal combination to achieved a real spectroscopic benefit and I think that many of the insights presented here will be very important for a further advancement of the field. The results and conclusions of this work are well supported by the presented data and the extensive analysis given in the supplementary material. Overall this is a very professional paper and I recommend it for publication in Nature Communications.
My only minor complaint about this work is that it is very technical and hard to follow for readers who are not immediately working at the forefront of the field. In particular, already in the introduction one gets easily lost with all the definition of T_1 and T_2 times for three different types of spins (sensor, target, memory), which are not properly introduced at this stage. Also the concepts of a classical memory or that of a "in-situ correlation spectroscopy measurement" are not commonly known and maybe these concepts could be further illustrated by explicit formulas or illustrative graphs. However, I leave this up to the authors.
Reviewer #2: Remarks to the Author: The manuscript by Matthias Pfender et al. investigates the NV electron spin -host nuclear spin system for high spectral resolution sensing and demonstrates techniques to decouple the sensor spin from both the host nuclear spin and the target spin. Using these techniques they achieve spectral linewidths of 13 Hz, which exceed previous reports by an order of magnitude, bringing them closer to being able to achieve linewidths of a few Hz required for NV-based NMR.
The dissipative decoupling techniques shown in this paper are novel in this particular realisation and I believe the results to be of interest to researchers in the community and in the wider field. Their exploration and analysis of different decoupling regimes sheds light on the mechanisms at play and indicates which technique is the most beneficial for a particular target spin. However, I think that the title and the main text don't reflect their achievement, but instead place too much emphasis on the N14 as a quantum memory, which is not a novel idea and has already been demonstrated multiple times.
In a reviewed manuscript I would like to see the following criticisms addressed: 1) The title doesn't reflect what is new it this work. Especially the word 'nonvolatile' and 'sensor unlimited' do not describe the achievements of this paper.
• Nonvolatile: The N14 spin is still very much a volatile memory, as demonstrated by its finite T1.
The word 'nonvolatile' suggests a new kind of memory, but the 14N is already being used as a memory in multiple groups. • Sensor unlimited: the linewidth is still very much limited by sensor T1 as can be seen in Fig 6 b (T1 sens limits T2* target), 'sensor unlimited' is therefore misleading. 2) In the abstract the authors claim to use the N14 spin as a quantum and classical storage, but no data is shown for quantum storage during the 'storage and target manipulation' part. If the authors mean the encoding part of the scheme to involve quantum storage, this should be specified.
3) The focus on using N14 in the first part of the main text is misleading as it isn't a new technique. It distracts from the novelty of the the 3-part sensing scheme and the decoupling using the NV0 state or using continuous optical yellow illumination. 4) As the authors focus on molecular NMR in the abstract and introduction, I would like to see a discussion of more 'real life' measurements, not detecting spins internal to the diamond lattice but trying to resolve chemical shifts within a molecule. What limits T2 of target spins in that case?
Minor comments: Fig. 1d) (in the main text) b) Check use of 'further' and 'furthermore' in lines 32 and 50, 'deliberately' in line 70. c) Last paragraph of Introduction presents the concept of using a nuclear spin as a memory for quantum and classical information as the novel result in this work. This has been reported several times already (including by the same group in Nat. Common 7:12279 (2016)). This memory hasn't changed and only seems to have gained the new name of 'non-volatile memory'. Emphasis in the last paragraph of the introduction should be on the new dissipative decoupling techniques. d) Figure 1 (d) caption: define 'proper phases' e) Line 88: 'As a main point of this work we utilise the expectation value of the N nuclear spin…' This is not the main novelty of the work. f) Line 131: 'less volatile' T1 storage, becomes non-volatile? Not a new idea. g) Suggestion: Give names to the two 13C atoms, e.g. target A, target B, to avoid referring to spins as in line 171 'previous one' or having to repeat '-1.8 kHz coupled spin' and '2.8 kHz coupled spin' multiple times. h) Caption Fig. 3: '…, using only the conversion from applied laser power…' Please specify what these parameters are used as, presumably fitting parameters. i) Change Fig. 4 d) Axis label to 'Target spin T2' from 'Nuclear spin T2' for consistency. j) Line 198: '… capable of quantum and classical metrology data storage'. True but misleading as authors don't perform quantum storage. k) Line 199: 'We have developed…' again, misleading. Focus on new aspects of correlation spectroscopy.

a) No mention of what open and filled circles mean in
Supplementary: l) B1: How is the memory spin initialised? m) The authors speak about their three storage steps in the supplementary. It would benefit the clarity of the main text to adopt this description in the main text as well. n) C1: In line 368 'green' should be 'orange' o) Figure 6 could benefit from labels for each measurement (such as in Fig. 1b) Reviewer #3: Remarks to the Author: This is an excellent manuscript. I recommend it for publication in Nature Communications. The presented work is very interesting both for the quantum sensing and solid-state quantum information communities and the manuscript is well written. However, before a decision can be made on publication, the authors should address the following: 1) The manuscript does a good job of making clear what problems are actually being solved/investigated: 1) protecting target spin from sensor spin relaxation and 2) precise spectroscopy by in-situ storage of correlations in the memory spin. However, these topics are presented as very general challenges for sensing/metrology (see e.g. the abstract and the introduction). The actual situation appears to be more subtle in that these challenges appear only under certain conditions. In particular, NV sensor relaxation is only an issue at room temperature and correlations can alternatively be stored classically by reading out the system. In those cases similar or better spectroscopy results than reported here have already been reported. It is clear that the situation the authors consider is intrinsically interesting, but many of the statements made are too general and need to be refined.
First, regarding sensor relaxation, this is not an issue at cryogenic temperatures where T1 times are minutes or hours. Therefore many of the statements throughout the manuscript are only true with the addition of "at room temperature". The authors should add a sentence to the introduction that makes clear that they are limiting the discussion to room temperature (and therefore sensor T1 = ~6 ms).
Second, regarding storing correlations, it does not become entirely clear if there is a fundamental advantage of storing the classical correlation in an in-situ memory over reading out the result and storing it on a computer. Given that there is a (single-shot) readout mechanism of the sensor state, high-resolution spectroscopy appears to be available directly by simply sensing + readout, waiting or manipulating the target, then again sensing + readout.
Examples where equal or better spectroscopic resolution was already demonstrated are reference 27 ( Figure 3C, spectroscopy of a C13 spin at room temperature, decoupling from sensor relaxation by laser illumination, correlations between consecutive repeated readouts) and Nature Communications 7:11526, 2016 ( Figure 1C, spectroscopy of C13 spins, no sensor relaxation due to cryogenic temperature, direct correlations between single-shot readouts of sensor spin). The authors should revise the manuscript to reflect that equal or better spectroscopic resolution has already been performed in the mentioned papers, but using alternative methods under different conditions. Again, these comments do not compromise the presented results in any way. But they do seem to challenge the validity of some of the very broad general statements made throughout the manuscript. Here are some examples (from the abstract and beginning of the manuscript): "The dissipation of nitrogen-vacancy centers in diamond is sufficiently small for detection of individual or small ensembles of nuclear spins, yet limiting spectral resolution of nuclear magnetic resonance (NMR) spectroscopy to a few hundred Hz" ¬-only true at room temperature and if no decoupling of sensor (for example, this statement disregards reference 27).
"NV centers enable target spin identification with spectral linewidths of several hundred Hz"much better linewidths were already demonstrated in the above-mentioned works.
"In previous NV-based NMR experiments the achievable linewidth was limited by the finite relaxation time of the sensor…" -not true for experiments at cryogenic temperatures or previous experiments with decoupling of sensor/target (for example this statement disregards reference 27).
"The longitudinal relaxation time of the NV center sensor spin's expectation value … of … 6ms is usually one to three orders of magnitude longer than its transverse relaxation time (see Fig. 1b)." -this value is only accurate at room temperature.
2) I am a little confused about the title, which seems inaccurate.
-"Nonvolatile quantum memory" the main idea in this paper is to use the Nitrogen nuclear spin as a classical memory, not as a quantum memory. The word quantum should be removed here.
-"of finite quantum systems". It is not clear to me what is meant with finite quantum systems, and this does not become clear from the text. Please explain.

General remarks
We thank the referees for carefully reviewing our manuscript.
First we will give a few general comments before we give point by point answers to the raised issues and questions. Typically, our answers are shown in blue, whereas the reviewer comments are displayed in black. Finally, we will attach a highlighted version of the manuscript showing in red all changes to the previous version. In this document you will nd a separate reference list too follow our arguments more easily.
Although our work was appreciated, the reviewers had three main concerns.
First, our manuscript was lacking a general motivation for applying correlation spectroscopy with NV centers at room temperature. We have added a required description.
Second, there were doubts about the novelty of using the 14 N nuclear spin as classical memory.
We have now emphasized that this is indeed the rst work, where the robustness of the 14 N spin under free evolution, during readout, during sensor reinitialization and during decoupling was demonstrated. We have analyzed the scaling of this robustness for a huge magnetic eld range.
Consequently, it is also the rst work, where classical information was stored beyond the T sens Reviewer #1 (Remarks to the Author): In this paper the authors describe and experimentally demonstrate a protocol for achieving a tenfold improvement in the nuclear spin sensing sensitivity using single NV center in diamond. This is achieved by using an ancillary nitrogen nuclear spin as a memory to partially store phase information during two separate interrogation steps. Further, the authors show that this memory is robust enough such that the electronic sensor spin and the target spin can be decoupled during the waiting time by optical pumping or even ionization techniques. This suppresses the sensor induced dephasing of the target spin. Both techniques combined allow very narrow linewidths, which are no longer limited by the inverse of the T 1 time of the electron spins.
The experimental demonstration of a tenfold improvement of the resolution of NV-based nanoscale NMR spectroscopy is certainly a nontrivial achievement, given the already quite advanced stage of this eld, and also an important step towards real chemistry applications. The authors combine a several experimental techniques and identify an optimal combination to achieved a real spectroscopic benet and I think that many of the insights presented here will be very important for a further advancement of the eld. The results and conclusions of this work are well supported by the presented data and the extensive analysis given in the supplementary material.
Overall this is a very professional paper and I recommend it for publication in Nature Communications.
My only minor complaint about this work is that it is very technical and hard to follow for readers who are not immediately working at the forefront of the eld. In particular, already in the introduction one gets easily lost with all the denition of T 1 and T 2 times for three dierent types of spins (sensor, target, memory), which are not properly introduced at this stage. Also the concepts of a classical memory or that of a "in-situ correlation spectroscopy measurement" are not commonly known and maybe these concepts could be further illustrated by explicit formulas or illustrative graphs. However, I leave this up to the authors.
We have tried to improve the introduction regarding general accessibility, furthermore we have added a paragraph motivating in-situ correlation spectroscopy. In the remaining text we tried to nd a balance between technical relevant details and clarity.
For proper denition and overview of timescales we have added a table.
Reviewer #2 (Remarks to the Author): The manuscript by Matthias Pfender et al. investigates the NV electron spin -host nuclear spin system for high spectral resolution sensing and demonstrates techniques to decouple the sensor spin from both the host nuclear spin and the target spin. Using these techniques they achieve spectral linewidths of 13 Hz, which exceed previous reports by an order of magnitude, bringing them closer to being able to achieve linewidths of a few Hz required for NV-based NMR.
The dissipative decoupling techniques shown in this paper are novel in this particular realisation and I believe the results to be of interest to researchers in the community and in the wider eld.
Their exploration and analysis of dierent decoupling regimes sheds light on the mechanisms at play and indicates which technique is the most benecial for a particular target spin. However, I think that the title and the main text don't reect their achievement, but instead place too much emphasis on the 14 N as a quantum memory, which is not a novel idea and has already been demonstrated multiple times.
In a reviewed manuscript I would like to see the following criticisms addressed: 1) The title doesn't reect what is new it this work. Especially the word`nonvolatile ' and sensor unlimited' do not describe the achievements of this paper.
Nonvolatile: The 14 N spin is still very much a volatile memory, as demonstrated by its nite T 1 . The word`nonvolatile ' suggests a new kind of memory, but the 14 N is already being used as a memory in multiple groups.
Below we answer the referee's criticism of using the word nonvolative. First, we motivate the use of the word nonvolatile by the fact that it survives much longer than any time scale relevant in our measurements. Furthermore, nonvolatile also includes a certain robustness to various external inuences. As pointed out in Figure   1c and partially anticipated from previous work (e.g. [Neumann et al., 2010]) we make the memory robust by going to higher magnetic elds, here > 1 T. Nonvolatile refers to several aspects: (i) Classical information is stored well beyond all other relevant time scales in the experiment, rst demonstrated here.
(iv) Stored information survives dissipative decoupling of target from the sensor, rst demonstrated here.
Second, regarding novelty, we characterize and use the 14 N spin as robust/nonvolatile memory for the rst time. For comparison, in previous experiments, it has been used as quantum and therefore explicitly volatile memory [Zaiser et al., 2016]. The 14 N spin was also previously used as classical memory [Laraoui et al., 2013a] but then it has neither been proven to be robust nor operated in robust mode, as we know today. In fact, the latter paper does speculate about their achievable spectral resolution of 200 Hz, which has been achieved recently even with a volatile quantum (not classical) memory in [Zaiser et al., 2016]. A similar work, also using the 14 N spin as classical memory, was presented on the arXiv one week before ours (compare [Pfender et al., 2016, Rosskopf et al., 2016). Therefore, we still think that investigating and demonstrating the NV intrinsic nitrogen spin as robust classical memory is indeed an appropriate achievement of our work, which deserves to be highlighted.
Sensor unlimited: the linewidth is still very much limited by sensor T 1 as can be seen in Fig 6 b (T sens 1 limits T * ,target 2 ),`sensor unlimited' is therefore misleading.
We see that gure 6b and the accompanying caption are misleading. Although, there is the comment that the presented decaying oscillation is for the non-decoupling case, in the end we give the hint that T * 2 is limited by T sens 1 , omitting the necessary statement that this is only true without decoupling. We have improved the caption.
Nevertheless, gure 4d illustrates the limit of our decoupling techniques. Indeed, we are not completely sensor-unlimited, however, much less limited. Our experiments yield an increase of the sensor limit by a factor of four for coupling strengths of about 2kHz, at a distance of up to 3nm. When we extrapolate our technique to single nuclear spins at the coherent interaction limit the improvement is already 20-fold. Current, nanoscale NMR experiments with target spin distances of about 10 nm would benet from two orders of magnitude improvement. The latter limit is then beyond one second, i.e. beyond the free evolution times in conventional NMR experiments. Therefore, we think it is fair to say that our technique enables sensor-unlimited spectroscopy. We now make clear in the text what we mean by the statement sensor-unlimited.
2) In the abstract the authors claim to use the 14 N spin as a quantum and classical storage, but no data is shown for quantum storage during the`storage and target manipulation' part. If the authors mean the encoding part of the scheme to involve quantum storage, this should be specied.
The current measurement scheme exploits both, the quantum and the classical memory aspect of the 14 N nuclear spin. Indeed, the referee is right that the gain in spectral resolution is mainly due to the classical storage aspect during the storage and target manipulation part. We have added a more detailed explanation of the measurement sequence, saying that quantum storage only occurs during short intervals during encoding and decoding and that during storage and manipulation we use the classical memory aspect of our nitrogen spin. Note that quantum storage during the encoding and decoding parts is indeed a benet because (i) it increases sensitivity [Zaiser et al., 2016] and (ii) it is faster due to lack of additional RF pulses for quantum ↔ classical memory switching. To avoid confusion we have also changed the title.
3) The focus on using 14 N in the rst part of the main text is misleading as it isn't a new technique. It distracts from the novelty of the the 3-part sensing scheme and the decoupling using the NV 0 state or using continuous optical yellow illumination.
As pointed out above (points about robustness), we think that our demonstrated utilization of the 14 N nuclear spin is indeed novel. Actually, the presented robustness enables the 3-part sensing scheme including decoupling in the rst place. Therefore, we think it deserves to be mentioned. Furthermore, we have added more details about the decoupling techniques to the introduction.
However, we agree that more distinction to previous experiments is necessary. We have improved the text accordingly.

4)
As the authors focus on molecular NMR in the abstract and introduction, I would like to see a discussion of more`real life' measurements, not detecting spins internal to the diamond lattice but trying to resolve chemical shifts within a molecule. What limits T 2 of target spins in that case?
Indeed, the current measurements are very much dierent from NMR on external nuclear spins. The intrinsic 13 C spins of the diamond lattice are pretty isolated stationary spins.
There are existing proposals to use them as a resource for qubits or as a quantum simulation testbed [Cai et al., 2013]. The latter is another application of the demonstrated method.
Hence, we do not exclusively want to concentrate on NMR of external spins. Typically, in previous NV-based NMR experiments with external spins the limiting factor for NMR linewidth has been for example diusion of target spins or dipolar broadening but also insucient sensor coherence time [Staudacher et al., 2013, Mamin et al., 2013, DeVience et al., 2015, Staudacher et al., 2015, Häberle et al., 2015, Kong et al., 2015, Lovchinsky et al., 2016, 2017. However, it is very likely that all mentioned limitations can be overcome and high-resolution NMR on external nuclear spins can be demonstrated. That's where our novel technique is going to be applied.
We have added a more thorough discussion about potential applications of our method and additional current limits to nano-scale NMR spectroscopy on`real-life' samples. We have made clear that no demonstration of nitrogen as robust classical memory was presented before and we have added more detailes about our dissipative decoupling techniques. d) Figure 1 (d) caption: dene`proper phases' We have added a comment that the phases are given as subscripts x and y in the wire diagram.
e) Line 88:`As a main point of this work we utilise the expectation value of the N nuclear spin . . . ' This is not the main novelty of the work.
We have altered the corresponding paragraph.
f ) Line 131:`less volatile' T 1 storage, becomes non-volatile? Not a new idea.
We have changed the sentence. g) Suggestion: Give names to the two 13 C atoms, e.g. target A, target B, to avoid referring to spins as in line 171`previous one' or having to repeat`−1.8 kHz coupled spin' and`2.8 kHz coupled spin' multiple times.
We followed the suggestion and gave names to the target spins, A 1 and A 2 for target spins around NV A and B 1 for the observable target spin around NV B.
h) Caption Fig. 3:`. . . , using only the conversion from applied laser power . . . ' Please specify what these parameters are used as, presumably tting parameters.
We have improved caption 3.
i) Change Fig. 4 d) Axis label to`Target spin T 2 ' from 'Nuclear spin T 2 ' for consistency.
We have changed the axis label.
j) Line 198:`. . . capable of quantum and classical metrology data storage'. True but misleading as authors don't perform quantum storage.
We have changed the summary accordingly.
k) Line 199:`We have developed . . . ' again, misleading. Focus on new aspects of correlation spectroscopy.
We make clear that we have adapted correlation spectroscopy to support quantum and classical storage on a separate memory spin along with suitable decoupling of target and memory spins from sensor dissipation.
Supplementary: l) B1: How is the memory spin initialised?
We have added a comment to the main text and the appendix saying that a readout result reports the initial memory state for the next run.
m) The authors speak about their three storage steps in the supplementary. It would benet the clarity of the main text to adopt this description in the main text as well.
We have now mentioned all three storage steps in the main text. n) C1: In line 368`green ' should be`orange ' We have corrected the color reference. o) Figure 6 could benet from labels for each measurement (such as in Fig. 1b) We have added the same labels to gure 6.

Reviewer #3 (Remarks to the Author):
This is an excellent manuscript. I recommend it for publication in Nature Communications.
The presented work is very interesting both for the quantum sensing and solid-state quantum information communities and the manuscript is well written. However, before a decision can be made on publication, the authors should address the following: 1) The manuscript does a good job of making clear what problems are actually being solved/ investigated: 1) protecting target spin from sensor spin relaxation and 2) precise spectroscopy by in-situ storage of correlations in the memory spin.
However, these topics are presented as very general challenges for sensing/metrology (see e.g. the abstract and the introduction). The actual situation appears to be more subtle in that these challenges appear only under certain conditions. In particular, NV sensor relaxation is only an issue at room temperature and correlations can alternatively be stored classically by reading out the system. In those cases similar or better spectroscopy results than reported here have already been reported. It is clear that the situation the authors consider is intrinsically interesting, but many of the statements made are too general and need to be rened.
First, regarding sensor relaxation, this is not an issue at cryogenic temperatures where T 1 times are minutes or hours. Therefore many of the statements throughout the manuscript are only true with the addition of at room temperature. The authors should add a sentence to the introduction that makes clear that they are limiting the discussion to room temperature (and therefore sensor T 1 =∼ 6 ms).
We have now claried the room temperature aspect of our work.
Second, regarding storing correlations, it does not become entirely clear if there is a fundamental advantage of storing the classical correlation in an in-situ memory over reading out the result and storing it on a computer. Given that there is a (single-shot) readout mechanism of the sensor state, high-resolution spectroscopy appears to be available directly by simply sensing + readout, waiting or manipulating the target, then again sensing + readout.
We have added a discussion/comparison about non-in-situ or conventional correlation spectroscopy as described by the reviewer. In particular we have discussed why this is feasible for NV centers at cryogenic temperatures and not at room temperature. We have also mentioned that single shot readout of the sensor (e.g. via the memory, without storage) has been demonstrated to aect target spins. In addition, single shot readout of the NV center is likely to ionize the NV leading to reduced correlation between subsequent measurement results [Waldherr et al., 2011, Aslam et al., 2013. In general, a non-ideal readout visibility V further reduces the visibility of the conventional correlation signal as V 2 , whereas insitu correlation spectroscopy exhibits visibility scaling as V .
Examples where equal or better spectroscopic resolution was already demonstrated are reference 27 [here: [Maurer et al., 2012]] ( Figure 3C, spectroscopy of a 13 C spin at room temperature, decoupling from sensor relaxation by laser illumination, correlations between consecutive repeated readouts) and Nature Communications 7:11526, 2016 [here: [Cramer et al., 2016]] ( Figure 1C, spectroscopy of 13 C spins, no sensor relaxation due to cryogenic temperature, direct correlations between single-shot readouts of sensor spin). The authors should revise the manuscript to reect that equal or better spectroscopic resolution has already been performed in the mentioned papers, but using alternative methods under dierent conditions. Now, we have clearly emphasized the results of [Maurer et al., 2012] in the text. However, we have also mentioned, that this demonstration is not universal, as it requires a particular 13 C nuclear spin. Furthermore, it inhibits storage on the nitrogen spin and readout of the demonstrated memory takes about 100 times longer than our readout. We have estimated the time of a basic single shot readout step of ≈ 170 µs vs. ≈ 1.5 µs in the nitrogen case. A complete readout takes multiples of such a basic step.
We have now added the given example of a cryogenic experiment, where the demonstrated spectral resolution is ≈ (π · 18.2 ms) −1 = 17.5 Hz ( Figure 1C, qubit 3). That is slightly broader than the best value we have presented, however, does not seem to be limited by the applied technique but by the coherence time of the target spin itself.
Again, these comments do not compromise the presented results in any way. But they do seem to challenge the validity of some of the very broad general statements made throughout the manuscript. Here are some examples (from the abstract and beginning of the manuscript): The dissipation of nitrogen-vacancy centers in diamond is suciently small for detection of individual or small ensembles of nuclear spins, yet limiting spectral resolution of nuclear magnetic resonance (NMR) spectroscopy to a few hundred Hz only true at room temperature and if no decoupling of sensor (for example, this statement disregards reference 27 [here: [Maurer et al., 2012]]). Now, we have restricted our claim to room temperature operation of NV sensors. Apart from that, dissipation of the NV electron spin at room temperature indeed limits resolution in general (i.e. if no decoupling is applied). So we keep this information. As before, however, we explicitly say, just a few sentences later, that we need to employ decoupling. As the abstract is not referenced in nature communications we have not given any reference here. Now, however, we have more clearly mentioned the achievements of [Maurer et al., 2012].
NV centers enable target spin identication with spectral linewidths of several hundred Hz much better linewidths were already demonstrated in the above-mentioned works.
We have restricted this statement to room temperature applications. Furthermore, We have considerably revised this part of the manuscript. However, keep in mind that this statement in general is correct for room temperature. Only if you apply a suitable decoupling method you break the sensor limitation, which is what we clearly say.
In previous NV-based NMR experiments the achievable linewidth was limited by the nite relaxation time of the sensor . . . not true for experiments at cryogenic temperatures or previous experiments with decoupling of sensor/target (for example this statement disregards reference 27 [here: [Maurer et al., 2012]]).
We have now incorporated also cryogenic measurement results and we have discussed the achievements of [Maurer et al., 2012] The longitudinal relaxation time of the NV center sensor spin's expectation value . . . of . . . 6ms is usually one to three orders of magnitude longer than its transverse relaxation time (see Fig. 1b). this value is only accurate at room temperature.
We have now restricted the insitu correlations spectroscopy discussion to room temperature.
2) I am a little confused about the title, which seems inaccurate.
"Nonvolatile quantum memory" the main idea in this paper is to use the Nitrogen nuclear spin as a classical memory, not as a quantum memory. The word quantum should be removed here.
We have removed the word quantum from the title. "of nite quantum systems". It is not clear to me what is meant with nite quantum systems, and this does not become clear from the text. Please explain.
We have changed the title, which now refers to small spin clusters.