Robust all-optical single-shot readout of nitrogen-vacancy centers in diamond

High-fidelity projective readout of a qubit’s state in a single experimental repetition is a prerequisite for various quantum protocols of sensing and computing. Achieving single-shot readout is challenging for solid-state qubits. For Nitrogen-Vacancy (NV) centers in diamond, it has been realized using nuclear memories or resonant excitation at cryogenic temperature. All of these existing approaches have stringent experimental demands. In particular, they require a high efficiency of photon collection, such as immersion optics or all-diamond micro-optics. For some of the most relevant applications, such as shallow implanted NV centers in a cryogenic environment, these tools are unavailable. Here we demonstrate an all-optical spin readout scheme that achieves single-shot fidelity even if photon collection is poor (delivering less than 103 clicks/second). The scheme is based on spin-dependent resonant excitation at cryogenic temperature combined with spin-to-charge conversion, mapping the fragile electron spin states to the stable charge states. We prove this technique to work on shallow implanted NV centers, as they are required for sensing and scalable NV-based quantum registers.

In this paper, the authors present the single shot readout method based on spin to charge conversion for NV center. Single shot readout is a powerful tool for quantum information technology. For sensing, it greatly boosts up the sensitivity. For quantum computing and communication, it gives the deterministic readouts instead of the probabilistic ones. Before this work, there are two single shot readout protocols in NV center system. Both lead to successful results, but with certain limitations. One utilizes nuclear spin for repetitive readouts, it can work in any temperature and almost with any NV centers. But it requires to use strong magnetic field. The other method uses the spin selective resonant transitions. It only works for the NV centers with good optical spectrum, low strain, and at low temperature. Spin to charge conversion was proposed as an alternative method for the single shot readout. It was believed that this approach can give broader working ranges. This readout has been demonstrated before, but the fidelity was not high enough for reaching the single shot readout level. In this work, the authors cleverly combine the weak resonant laser and a high power below-band laser to realize the high fidelity single shot readout based on spin to charge conversion. This is a huge step in the field. The authors also made systematic analysis and detailed explanations. It is amazing to see that without special optical structure, this single shot readout works well. The paper is well written and deserves the publication in Nature Communications. But still, I have several questions/comments. 1. Usually, by exciting NV centers with the green laser, it can induce spectrum jumps due to its disturbance to the local charge surroundings. This can lead to the shifts of the resonant transitions. Can the authors comment on this effect, since the green laser is used in the beginning of all sequences, as shown in Fig.1 and measurements? After the green pulse, the resonant pulses may become non-resonant. Also, for Fig.2a, have the authors considered this effect to the result? Furthermore, this kind of spectrum jump is quite common in shallow NV centers, even with just resonant red excitation. Does it influence the results? 2. Can the authors comment on the working temperature range of this method? The spin selective transitions will be mixed up when the temperature rises up. 3. Similarly, can the authors also comment on the working range of the strain of NV centers? 4. For the working range of the off-axis magnetic field, could the authors give some estimations? This is one of the bright spots of this new protocol. Nevertheless, it is worth to analyze this in more details. For example, the off-axis magnetic field will mix the spin states. This will influence in every part of the protocol, from the initialization to the readout. Also, in high field, the optical transitions are much more crowded. If the sub-GHz linewidth is taking into account, how large can the magnetic field be? 5. About the Pi pulse for spin initialization step. Since the excited state life time for spin 0 is around 12ns, to reach the spin flips in the ground state, the Pi pulse should be shorter than this time. Was this the case? This fast pulse needs a lot of power. Also, why not use A1 sub-level resonant transition for the spin initialization? 6. In Fig. 1a, would it be better to mark what MW1, MW2 and 637nm in the level structure? For people working in NV center field, these are obvious. But for people from other fields, seeing labels in Fig.1c may be confusing.
Reviewer #2 (Remarks to the Author): The authors have performed a measurement of the NV spin at cryogenic temperatures using a spin to charge conversion sequence based on resonant excitation of one spin sublevel of NV-. The resonant spin-selective excitation enables a highly selective conversion of the spin state to a charge state distribution between NV-and NV0 that can then be measured with high fidelity. This work expands the toolbox of available techniques for NV spin readout, and is of significant interest to the NV community.
The key question of how this method compares to alternative fluorescence based readout protocols remains largely unaddressed in the manuscript. The authors compare their method to readout via detection of fluorescence under off-resonant pumping with green laser light, which pumps all spin states and consequently results in relatively low fluorescence contrast (this is also what limits the contrast of non-resonant spin to charge conversion). However, the state of the art method for spin readout in cryogenic conditions, where resonant excitation of the NV-ZPL is accessible, is rather to selectively drive one spin sublevel (as the authors do here for their resonant spin to charge conversion) while detecting fluorescence. The resonant driving results in a very high contrast (and is also what enables the high contrast for the authors' measurement). So it would be much more meaningful for the authors to directly compare their resonant spin to charge conversion method with resonant fluorescence readout in their setup. This is especially the case given that the resonant fluorescence method is limited by imperfect cycling of the ms=0 spin state, a limit that could potentially be overcome by employing spin to charge conversion, if the ionization step is sufficiently selective. It would greatly expand the impact of this method if it is possible to overcome the cycling limit, but it is not yet clear from the manuscript whether that could be the case. The authors' demonstrated end-to-end fidelity is already quite impressive given the absence of any optical collection enhancements, so a measurement of the resonant fluorescence readout fidelity in their setup would provide a very useful comparison.
A few more minor/technical points: 1. A level of 50kcps is a fair assumption as a "standard" collection efficiency, and seems to be more than adequate for proving the efficacy of the method. I don't see the relevancy of comparing to subkcps count rates, as the authors do in their abstract and again in lines 16-21 of p.4. section S.5 contains first a high-power pulse of 642nm light. In principle, this shouldn't excite any transitions in the NV center. Why is it necessary, or what improvement does it bring over other initialization sequences that were attempted? What sequences were attempted to optimize this initialization? 5. I don't see a specific mention of the 637nm power used for the ionization pulse. Did the authors vary or optimize this power? In any case, it would be good to specify the power used.
6. Lines in some plots are too thick and obscure the relevant details (particularly fig. 1d and insets of figs 2d & 3b). Also the authors could consider plotting the insets to Fig. 2d and 3b on a vertical log scale so that the relevant information is visible.

Brendan Shields
Reviewer #3 (Remarks to the Author): The ability to read out a quantum state with high fidelity, and ultimately in a single shot is important for implementing quantum protocols. For optically active solid-state qubits, single-shot readout remains challenging due the limited photon flux during the transient spin-dependent fluorescence.
While the NV center in diamond has been extensively studied for many years, the high refractive index of diamond, low photon collection efficiency and limited contrast for spin-dependent fluorescence result in low single-shot SNR for most of the experiments. The conventional spin state readout scheme at room temperatures is based on spin-dependent inter-system crossing, and this scheme requires many repetitions (~10^5) to improve the averaged SNR. Previous work has tackled this problem in a number of ways: by using spin-dependent shelving and converting the spin state to charge state, the readout fidelity can be improved. Alternatively, the spin state can also be stored in a nearby nuclear spin and readout repetitively. So far, single-shot readout of NV centers has only been realized using highly cyclic optical transitions at cryogenic temperatures, using a solid immersion lens to improve the optical collection efficiency.
In this work, the authors demonstrate single-shot readout of NV centers by combining two wellestablished techniques: spin selective resonant excitations and spin-to-charge conversion at cryogenic temperatures. The spin-dependent ionization is achieved by exciting to the excited state with a resonant excitation and ionizing the center from the excited state with a high-power reddetuned laser. Therefore, only the spin-state being addressed by the resonant excitation will be ionized. This work presents the first demonstration of single shot spin readout of NV centers without directly reading out highly cycling optical transitions. This relaxes the conditions for single shot readout significantly.
This work introduces a new protocol for reading out solid-state qubits. While the result is sound and convincing, the presentation of the material is sometimes confusing and difficult to understand, probably especially for people outside the field. I would recommend that the authors address the following points before publication.
A few scientific questions that should be addressed: 1. In the PLE spectra ( Fig. 1d and Fig. 4a), there are extra peaks for |+1> spectra that are not present in the simulated spectra. The extra peaks align with peaks in |0> spectra. Is this from imperfect spin initialization with the green laser? If so, I would expect some extra peaks in the |0> spectra that align with peaks in |+1> spectra as well. 2. The authors claim that their use of the resonant laser to read out the charge state is novel. I do not think this claim is correct-see for example, the supplementary information to Andersen et al, Science 364, 154-157 (2019). 3. In Fig. 3a, why is there a decay for the |+1> trace? Is this due to some residual 'off-resonant' excitation of the other transitions with the resonant laser? 4. The authors mentioned that the spectral diffusion is compensated for the shallow NV with higher excitation power (Line 30 -34 of page 4). However, the postselection on the charge state is accomplished with resonant excitation which in principle should also postselect the transition frequency. The authors should comment on this. 5. Under Section S.3, the authors were not able to saturate the shallow implanted NV with a resonant laser. They attribute this to the spectral diffusion of shallow NV centers. How do they exclude other possibilities such as charge state instability? 6. Why is the spin polarization from green laser quoted differently in the caption of Fig. 2 (85%) and line 18 of page 3 (70%)? 7. In Fig. 4b, the authors attribute the background after 20 us to finite excitation of close by transitions. There is another |0> transition that is further away from other transitions (around -7 GHz shown in Fig. 4a). Why is that transition not used here?
Below are points of clarification and presentation issues to be addressed: 1. The color scheme used in Fig. 1d, Fig. 2a, Fig. 4a and Fig. 4b is confusing. The contrast between |0> and |+1> is not significant enough. Similarly, the inset pulse sequences throughout the paper are too small to be legible. 2. In the inset of Fig. 1d, 'df' is not defined. 3. The inset for Fig. 2a is confusing. The pulse sequence for |+1> initialization should be placed next to the actual data (lower panel). The presentation could also be made clearer by having separate panels for the pulse sequences instead of the indiscriminate uses of insets. 4. The presentation of Fig. 2d and Fig. 3b is confusing. For both figures, the inset for the upper panel has no 'real data' and the y-axis range for the lower panel is cut-off well below the data range. I understand that the authors are trying to show the contrast in the low photon number regime. But it's not informative for the readers to look at an almost blank figure. In this case, it might be better to use log-scale for the y-axis. 5. Line 6 of page 3: what do the authors mean by spin-state stability? 6. Line 12 of page 3: the citation for Robledo et al is missing. 7. In Fig. 4e, the color scheme is confusing, the highlight of |0> transitions is hard to see. 8. Typo: under Section S.1, it should be "… measurements were not able to resolve a more complex hyperfine structure". 9. Typo: the caption of Fig. S1 should be "Hyperfine ODMR transitions …" instead of "Hyperfine ODRM transitions".    Fig.1c a be c f i g. A fe cie ific e i ha h d be add e ed: 1. I he PLE ec a (Fig. 1d a d Fig. 4a  Be a e i f c a ifica i a d e e a i i e be add e ed: 1. The c che e ed i Fig. 1d, Fig. 2a, Fig. 4a a d Fig. 4b i c f i g. The c a be ee 0> a d +1> i ig ifica e gh. Si i a , he i e e e e ce h gh he a e a e a be egib e. I Fig. 1d, 2a, 4a a d 4b