Photoluminescence dari NV− Centers di 5 nm Detonation Nanodiamonds:Identifikasi dan Sensitivitas Tinggi terhadap Medan Magnet

Hasil dan Diskusi

Menghitung NV ⁻ Pusat di DND dengan Metode Half-Field EPR

Karena radiasi yang dipancarkan oleh pusat warna dapat diserap oleh sp ² -mengandung spesies yang ada di dalam agregat DND dan bahkan mungkin terhalang oleh fitur morfologi tertentu di ruang antara partikel, sangat penting untuk mengontrol dan memantau NV ⁻ berpusat di berlian dengan metode non-optik. NV ⁻ pusat dapat dideteksi dengan spektroskopi EPR bahkan ketika menanamkan pembawanya ke dalam media padat yang tidak transparan yang menghalangi radiasi optik yang menarik dan bercahaya atau dalam kasus tidak adanya NV ⁻ radiasi sekunder terkait dengan mekanisme pendinginan tertentu. Konsentrasi total mereka selanjutnya dapat dengan mudah dievaluasi dengan analisis spektrum EPR [16, 26, 27]. Pada saat yang sama, konsentrasi NV yang ditentukan ⁻ pusat memberikan informasi berharga tentang potensi (setelah pemurnian ekstensif) kecerahan NV ⁻ PL Spektrum EPR dari bubuk DND yang dimurnikan dari pengotor magnetik logam transisi 3d (kebanyakan Fe, Ni) di wilayah yang disebut setengah medan magnet ditunjukkan pada Gambar 2 (kurva hitam). Spektrum terdiri dari dua garis dekat (1 dan 2) dengan g -faktor g ₁ = 4,27 dan g ₂ = 4.00, dan lebar ΔH _pp1 2,0 mT dan ΔH _pp2 1,4 mT. Baris pertama, bidang rendah, sesuai dengan transisi terlarang m _s = 2 antara tingkat energi Zeeman dari keadaan triplet NV ⁻ pusat dalam medan magnet. Baris kedua, bidang tinggi, sesuai dengan transisi terlarang yang sama m _s = 2, yang terjadi ketika radiasi gelombang mikro diserap oleh pusat lain, yaitu multivacancies yang memiliki S = 1 Dalam Ref. [16], garis-garis ini pertama kali ditugaskan ke pusat triplet berbeda yang hidup berdampingan di DND, dan khususnya, g ₁ = 4.27 baris ditugaskan ke NV ⁻ Tengah. Posisi g ₁ = 4.27 baris berhubungan dengan transisi terlarang m _s = 2 seperti yang ditunjukkan pada diagram energi keadaan dasar NV singlet-triplet ⁻ pusat di medan magnet hingga 300 mT (Gbr. 3). Parameter pemisahan medan-nol besar dari spin-Hamilton (D = 850 × 10 ⁻⁴ cm ⁻¹ ), yang disebabkan oleh interaksi pertukaran yang kuat dari putaran konstituen dalam triplet NV ⁻ tengah, menyebabkan sensitivitas tinggi dari posisi g ₁ = 4.27 baris (bergeser 0,5% pada g -skala) ke frekuensi gelombang mikro tertentu dalam kisaran dari 9,0 hingga 9,9 GHz (band-X) [26]. Intensitas integral dari garis g ₁ = 4.27 dapat digunakan untuk memperkirakan secara tepat konsentrasi NV ⁻ pusat bahkan ketika luminescence dari nanodiamonds tidak terdeteksi karena penyerapan atau pendinginan karena spesies optik aktif tetangga, atau bahkan masking oleh luminescence latar belakang parasit yang kuat dari spesies co-ada ketiga. Berlian submikron sintetik yang diiradiasi elektron berenergi tinggi (ukuran rata-rata 100 nm) dengan konsentrasi NV yang diketahui ⁻ center digunakan sebagai sampel referensi (sampel Ib HPHT FND). Spektrum EPR sampel Ib HPHT FND yang diambil pada daya gelombang mikro rendah P _MW = 3 μW ditunjukkan pada Gambar. 2 (kurva merah) untuk perbandingan. Dalam rentang di bawah 320 mT, ini terdiri dari tiga sinyal EPR yang terdefinisi dengan baik sesuai dengan "terlarang" m _s = 2 (pada 157,85 mT), dan dua diizinkan z , m _s = 1 dan x ,y , m _s = 1 (pada 234,39 mT dan 281,27 mT) transisi antara tingkat energi Zeeman dari keadaan triplet dasar NV ⁻ dalam medan magnet. m _s = 2 dan z , m _s = 1 transisi juga ditampilkan pada skema pada Gambar. 3. Cukup mengejutkan bahwa dibandingkan dengan sampel referensi submikron dengan NV ⁻ pusat, partikel DND hanya menunjukkan g = 4.27 baris EPR terkait dengan m _s = 2 transisi "terlarang". Khususnya, z . yang diizinkan , m _s = 1 dan x ,y , m _s = 1 transisi gelombang mikro tidak diamati. Penjelasan yang mungkin adalah pelebaran abnormal dari garis EPR terkait dengan m _s = 1 transisi karena variasi besar dalam parameter utama spin-Hamilton (D dan E ) untuk S = 1 NV ⁻ atas ansambel partikel DND. Dengan demikian, sama sekali tidak ada garis yang terkait dengan z . yang diizinkan , m _s = 1 dan x,y, m _s = 1 transisi adalah fitur kunci yang tidak biasa dari NV ⁻ di DND. Menurut pemasok komersial, konsentrasi yang diharapkan dari NV ⁻ dalam referensi Ib HPHT FND dievaluasi berdasarkan dosis iradiasi yang diserap adalah ~ 5.3 ppm. Intensitas integral dari g ₁ Garis = 4,27 dalam spektrum EPR DND ditemukan ~ 4,8 kali lebih kecil dari pada spektrum Ib HPHT FND EPR, dan oleh karena itu, konsentrasi NV ⁻ pusat di DND ditentukan menjadi 1.1 ± 0.3 ppm. Prosedur integrasi yang sesuai dari kedua sinyal EPR dalam rentang setengah medan magnet (130–180 mT) bersama dengan koreksi latar belakang secara khusus ditunjukkan pada Gambar. 4. Gambar 4 a berisi spektrum EPR asli yang diukur dari turunan pertama penyerapan gelombang mikro, sebelum perlakuan matematis (kurva 1); spektrum EPR yang sama setelah integrasi (kurva atas 2); dan spektrum EPR asli yang sama seperti yang diukur, tetapi setelah mengurangi garis tipe Lorentz yang luas terkait dengan kompleks yang mengandung besi yang tersisa (kurva bawah 3). Gambar 4 b menunjukkan spektrum EPR dari sampel referensi (kurva atas) dan DND (kurva bawah) setelah integrasi spektrum EPR yang dikoreksi latar belakang dan diukur. Spektrum terintegrasi yang sebenarnya untuk DND itu sendiri (Gbr. 4b, kurva bawah) cukup berbeda dari spektrum kedua pada Gbr.4a, di mana hanya spektrum EPR yang tidak dikoreksi dan diukur dari DND yang terintegrasi. Spektrum yang diperoleh (Gbr. 4b, kurva bawah) menunjukkan lembah yang sangat dalam antara sinyal EPR tetangga (g ₁ = 4.00 dan g ₂ = 4.27) tipe Lorentz dipisahkan oleh jarak ~ 10 mT. Area yang diarsir di bawah spektrum EPR yang ditunjukkan pada panel (b) dari Gambar 4 memberikan intensitas integral dari garis g ₁ = 4,27 untuk kedua sampel yang dibandingkan. Rasio area ini ketika dinormalisasi dengan bobot sampel adalah sekitar ~ 4,8. Menariknya, untuk DND, bentuk garisnya g ₁ = 4.27 simetris dan luas. Garis ini sebagian ditumpangkan dengan bidang tinggi g ₂ = 4.00 EPR line disebabkan oleh multivacancy. Beberapa latar belakang yang tidak dikoreksi yang tersisa dalam bentuk dasar yang kuat (terutama dari kompleks terkait besi yang tidak dapat dilepas) terlihat jelas pada Gambar 4a (spektrum 2). Dengan demikian, dekomposisi spektrum EPR DND yang benar untuk setidaknya tiga komponen yang tumpang tindih dalam kisaran 130–180 mT adalah kebutuhan utama untuk evaluasi yang benar dari intensitas terintegrasi g = 4.27 garis dan konsentrasi NV ⁻ . Kami juga menemukan bahwa berbagai perlakuan pada suhu di bawah 500 °C (dalam vakum dan di udara) tidak memiliki efek praktis pada konsentrasi NV ⁻ pusat di DND. Intensitas sinyal terintegrasi ganda g ₁ = 4.27 (area bayangan di bawah kontur Lorentzian berpusat pada ~ 151 mT untuk ν = 9.0785 GHz) sebagian besar tetap sama setelah perawatan seperti sebelumnya. Ini menunjukkan bahwa NV ⁻ pusat terkubur cukup dalam di kisi, pada kedalaman dari permukaan nanopartikel setidaknya konstanta kisi (~ 0,36 nm), sehingga mereka tidak dapat bereaksi dengan agen kimia eksternal yang tidak mengetsa atau menembus fase berlian.

Spektrum EPR dari DND yang dimurnikan asam di wilayah setengah medan magnet (kurva hitam) dan nanopartikel berlian Ib HPHT fluoresen iradiasi elektron referensi dengan ukuran rata-rata ~100 nm (kurva merah) dalam kisaran hingga 320 mT. Kedua garis spektrum DND EPR ditandai dengan angka 1 dan 2. Garis medan rendah 1 dengan g = 4.272 sesuai dengan NV ⁻ pusat. Garis yang terkait dengan bidang rendah (LF) diizinkan z m _s = 1 dan x ,y m _s = 1 transisi dalam FND 100 nm ditandai dengan panah untuk spektrum atas. Garis terlemah, hampir tidak dapat dibedakan terkait dengan z . yang diizinkan m _s = 1 transisi juga ditandai dengan lingkaran. Frekuensi gelombang mikro adalah 9,4393 GHz

Energy diagram of ground singlet-triplet levels ³ A₂ of NV ⁻ in magnetic field up to 300 mT. “Forbidden” Δm _s  = 2 and LF allowed Δm _s  = 1 transitions caused by absorption of microwave radiation (ν  ≈ 9.44 GHz) are marked by vertical red arrows. The position of the ground state spin level anti-crossing (|0 〉_GS and |−1〉_GS ) is marked by a circle

a Background subtraction in EPR signal of DND and b estimations of the double-integrated intensities of the g = 4.27 line for a DND sample and reference fluorescent Ib HPHT synthetic diamond. Panel a :as-registered first derivative EPR signal of the DND in the ± 15 mT half magnetic field range (curve 1, blue); the same, but integrated EPR signal in the same ± 15 mT range of magnetic field (curve 2); the same first derivative EPR signal of DND, but after subtraction of the broad parasitic EPR signal from remaining non-removable iron-containing paramagnetic complexes shown by red contour of Lorentzian shape in the upper curve (curve 3, blue). Panel b :Estimation of the double integrated intensity of the g  = 4.27 line for a fluorescent Ib HPHT diamond having NV ⁻ (upper curve, shaded area) and the DND sample (bottom curve, shadow area). The bottom spectrum in panel b consists of two contours of Lorentzian shape, one of which centred at lower magnetic field (≈ 150.932 mT), is assigned to the NV ⁻ centres of DND (the area below this contour in red is highlighted). Microwave frequency:ν = 9.0785 GHz

The main high-intensity EPR signal of the DND lies above 320 mT at ν  = 9.4393 GHz and has a g -factor g  ≈ 2.0027. It has a Lorentzian curve profile of linewidth ΔH _pp  = 0.84 mT [13]. This linewidth is greater than that of fluorescent nanodiamonds (FNDs, 100 nm) milled from microcrystalline diamond synthesized by a high-pressure and high-temperature method. The broad signal can be explained by greater exchange and dipole-dipole interactions between the S  = 1/2 spins in the spin ensemble within an individual DND nanoparticle than those within a FND. The intensity of the main EPR signal collected from all S  = 1/2 paramagnetic centres, both of nitrogen (P1 centres [28]) and non-nitrogen origin, indicates a spin concentration of ~ 1300 ppm, corresponding ^{Footnote 4} to around 15 S  = 1/2 spins in each DND [13]. However, it can be concluded from the earlier obtained data that approximately 40–50% of the total number of all paramagnetic centres in the DNDs are due to half-populated antibonding orbitals of isolated P1 centres. Thus, the huge total of spin-half, point-like agents located in DNDs (N _PC ) can be represented as the sum of at least two contributions, from P1 centres and from elemental point defects having dangling bond spins ½, for example based on vacancies like H1 centres (VH ^o ):\( \left({N}_{\mathrm{PC}}=\left[{\mathrm{N}}_{\mathrm{s}}^{\mathrm{o}}\right]+\left[\mathrm{et}\ \mathrm{al}\ \right]\ \right) \). Here we assume that specific EPR signatures from the hyperfine structures of P1 and H1 centres are absent or greatly “smeared” through the dense arrangement of localized spins within each particle. This sum gives us a clue about the approximate content of both isolated nitrogen and monovacancies inside the DND, although isolated nitrogen and monovacancies can also be present in the diamond lattice in nonparamagnetic forms such as \( {\mathrm{N}}_{\mathrm{s}}^{-},{\mathrm{N}}_{\mathrm{s}}^{+} \), V ^o . The presence of H1 centres together with neutral monovacancies in DND at some minor level (< 700 ppm) is reasonable in principle because rapid assembling of diamond lattice during the detonation takes place from hydrogen-containing products of TNT-hexogen pyrolysis such as CH₃ ^* or CH₂ ^* radicals. Although the mutual charge transfers between the main groups of centres present in the lattice in various charge and spin states (\( {\mathrm{N}}_{\mathrm{s}}^{\mathrm{o}},{\mathrm{N}}_{\mathrm{s}}^{-},{\mathrm{N}}_{\mathrm{s}}^{+};{\mathrm{V}}^{-},{\mathrm{V}}^{\mathrm{o}} \)) are possible in principle, the foremost contribution to paramagnetism comes from only \( {\mathrm{N}}_{\mathrm{s}}^{\mathrm{o}} \). Let us therefore estimate the maximal possible concentration of NV in a DND on the basis of assumptions about the known total amounts of substitutional nitrogen and monovacancies inside the particles. The actual charge state of NV is not essential for such an estimation. A statistical approach gives the following simple formula for the probability (p _NV ) of finding at least one NV in one 5-nm DND particle consisting of N nodes of covalent diamond lattice:\( {p}_{\mathrm{NV}}=\frac{2 nv}{N} \) , where n and v are the mean numbers of isolated nitrogen atoms and monovacancies inside the DND particle, respectively. Because N _PC  ≈ 1300 ppm, we can approximately assume that n  + v  ~ 15 for N  = 11,500. The maximal value of p _NV is achieved when \( n\approx v\approx \frac{15}{2} \). This gives p _NV  ≈ 0.0098. This value corresponds to around ~ 1 ppm of NV in DND, as was obtained previously from comparison of the g  = 4.27 EPR signals of the DND and the reference sample. Excluding the surface nodes linked with surface functional groups and interior nodes occupied by A-centres (up to 2 at.%) from the formal integration procedure, using N  = 9950–10,000 gives a slightly greater concentration of NV, up to 1.1 ppm. The estimated experimental concentration of NV ⁻ centres in the DNDs is in close agreement with the theoretical estimation made above, and about 1000 times smaller than the concentration of all S  = 1/2 paramagnetic species in the system.

The concentration ratio for interior S  = 1/2\( {\mathrm{N}}_{\mathrm{s}}^{\mathrm{o}} \) and S  = 1 NV ⁻ centres is therefore qualitatively consistent with the main idea of rapid NV centre formation, that is, from the “random inclusion” of both substitutional nitrogen atoms and vacancies to the growing diamond nano-crystallites during the overall ~ 13–20 μs duration of the detonation wave propagation. Here, we intuitively assume that each NV ⁻ is formed as a result of the random embedding and occurrence of impurity nitrogen atoms and vacancies in the nearest neighbour lattice sites during the period of time prior to the subsequent rapid cooling of the products to temperatures of the order of ~ 500–650 °C at which the diffusion of vacancies in the lattice is practically stopped.

Nitrogen Impurity Concentrations

XPS is a powerful tool for studying the DNDs’ composition and the chemical state of the main alien elements present on the surface of the DNDs (and also within 2 nm under the surface) [29]. Our main interest was focused on the XPS signal from nitrogen and the evaluation of the interior nitrogen content, since nitrogen is the predominant inner impurity. The XPS signals of carbon (C1s), nitrogen (N1s) and other elements have been recorded after etching of the surface with Ar ions. An overview of the XPS spectrum plotted in the wide range of binding energies from 250 to 600 eV is shown in Fig. 5a. Although the data indicates the presence of a large amount of oxygen-containing atomic groups at the DND’s surface (5.5 at.%), the analysis of the O1s signal is not of particular relevance to this work. The concentration of both inner and exterior nitrogen was preliminarily evaluated to be between 1.7 and 2.4 at.% [30]. The concentration of all the residual elements found (a small number of metals) did not exceed ~ 0.32 at.% in the as-received DND, and it could be reduced by 20–30 times by etching the DND powder in boiling aqua regia and hydrochloric acid [30]. The photoemission peak of nitrogen (N1s) is shown in Fig. 5b. Etching with Ar ions results in the removal of weakly bound adsorbed species from the surface, while the covalent diamond lattice remains unaffected. The characteristic high-energy peaks (~ 404.6 eV and 407.3 eV) in the N1s photoemission signal have very low intensities when compared with an untreated pristine sample. These peaks demonstrate the presence of any remaining nitrate ions (peak at ~ 407.3 eV) and nitrite groups (peak at ~ 404.6 eV) on the Ar ion-treated surface. Further complete removal of nitrate ions and nitrite groups from the surface can be achieved only by annealing the DND powder in air at temperature> 350  ^o C. The main peak of the N1s signal at ~ 401 eV, which is not influenced by Ar ion treatment, corresponds to chemical bonds of the type N–sp ³ –C. This peak is a characteristic of interior elemental nitrogen covalently bound within the diamond lattice. It appears due to various forms of nitrogen present in the diamond matrix, including NN dimers, next nearest neighbouring N ⁺ ...N ⁻ pairs, and more complex nitrogen clusters. Similar data were obtained for the photoemission peak of carbon C1s (sharp, intense signal seen in Fig. 5a, see also Ref. [30] for details). The C1s XPS signal consists of two main peaks:one centred at 284.9 eV corresponding to C–C bonds in the diamond matrix and another peak centred at 287.3 eV corresponding to C–N bonds. Only diamond carbon and sp ³ /sp ² carbon bound to nitrogen, even in the form of atomic-scale disordered nitride phase species (where neighbouring carbon and nitrogen atoms can have up to three C–N bonds), are represented in the C1s signal of the Ar ion-treated surface. Our careful analysis of the integrated intensity of only the N1s 400.9 eV peak and C1s signals (together with O1s and residual elements’ signals) suggests that nitrogen is contained within the diamond lattice of selected DND sample in the amount of 1.65 ± 0.05 at.% and mainly in the form of complex clusters which are not paramagnetic. This value was obtained after special correction by a reducing factor taking into account that the actual size of DND particles is larger than the whole depth of crystal lattice from which the excited photoelectrons are readily emitted. It seems that only a small part of nitrogen is present in the diamond lattice in the form of isolated paramagnetic nitrogen atoms with spin S  = 1/2 (no more than 8–9% of all nitrogen). The isolated paramagnetic nitrogen atoms are only accessible for observation by EPR as was shown in the previous section.

XPS spectra of acid-purified DNDs after Ar ion etching:a overview of the spectrum in the range 250-600 eV and b N1s photoemission peaks

XPS spectroscopy was also applied to estimate the nitrogen content in other DND samples provided by our suppliers. We found that the nitrogen content varies in these samples from ~ 1.6 at.% (minimal value) to ~ 2.1 at.% (maximal value). We simultaneously noticed that higher concentrations of nitrogen correspond to the samples with a smaller X-ray CSR size. Such values of nitrogen content were also roughly confirmed by elemental analysis with a Micro Corder JMC10 device in the course of sample combustion in oxygen flow (30 ml/min) at 1000  ^o C. As a reference source of nitrogen, carbon and hydrogen for this method, we used the antipyrine C₁₁ H₁₂ N₂ O, having nitrogen in its structure in the form of N–N groups.

DND Fluorescence Intensity and Its Dependence upon the Nitrogen Content

The PL spectrum of one selected powder DND sample is shown in Fig. 6a. It has the maximum intensity at 680 nm. Following [31], a characteristic spectrum with peak position above 650 nm indicates the presence of both NV ⁻ and NV ⁰ defects, although the contribution from some light-emitting defective sites of remaining sp ² -coordinated carbon species around DND particles cannot be excluded completely. However, the characteristic zero-phonon lines (ZPL) at 638 nm (NV ⁻ ) and 575 nm (NV ⁰ ), respectively, are not detected. This probably occurs due to the changing positions of the NV ⁰ and NV ⁻ excited/ground states inside the bandgap for centres lying in the vicinity of the particle edge and the subsequent broadening of the 638-nm and 575-nm ZPL spectral components for the ensembles of such NV ⁰ and NV ⁻ centres with slightly different electronic parameters. Let us mark that the absence of featured NV ⁻ ZPL peak at 638 nm was confirmed in many studies related with photoluminescence of DND aggregates lying free on the substrate or embedded inside polymers [31]. Sometimes it (suppressed or very poorly recognized ZPL feature at 638 nm) even happens for larger isolated fluorescent Ib HPHT nanodiamonds of a size about 40 nm [4].

a Photoluminescence spectrum of DND powder pressed flush in a shallow hole with a diameter of 2 mm made in a copper plate and b the interior nitrogen content measured by XPS and c the intensity of NV ⁻ PL at 680 nm as a function of X-ray CSR size for a series of selected DNDs synthesized in the presence of some intentionally added inorganic additives in the detonation zone (charge and water shell) as provided by the manufacturer. Excitation laser wavelength λ  = 532 nm, power ~ 0.5 mW. The diameter of the focused laser spot on the sample surface was 2 μm. Conditions of recording were temperature T  = 293 K, and an air environment

We also studied the PL intensity for a series of DND samples with different CSR sizes. The CSR size characterizes the average size of perfect diamond domains or the mean size of diamond nano-crystallites in the powder even if they are randomly arranged in large-scale polycrystalline aggregates with size exceeding 30–50 nm. The CSR size varied from 4.3 to 5.6 nm in the series of DND samples selected for our studies (similar to the results explained in Ref. [32] although we used another more traditional approach for analysing X-ray diffractograms). Figure 6b shows the dependence of interior nitrogen content evaluated by means of the XPS method versus the CSR size of the DNDs. The larger the CSR size, the smaller the nitrogen content. This seems reasonable as powders with larger DND particles were synthesized as a result of the addition of some inorganic substances having the elements playing the role of nitrogen-getter inside the detonation zone (charge or charge enclosure). Such elementary additives probably promote the reduction of the overall nitrogen content in the growing diamond lattice during the explosion process or change the conditions of the diamond lattice assembly to slightly prolong the synthesis (on the order of microseconds). In addition to the overall nitrogen content, we also recorded the reduced amount of nitrogen-related paramagnetic centres in these DNDs, as confirmed by EPR spectroscopy. PL spectra of all DNDs having different CSR sizes are practically the same in shape in the range 550–900 nm, but this is not the case for the absolute intensities of the PL at the maxima of the PL spectra at 650–680 nm. The intensity at the maximum of PL spectrum is plotted in Fig. 6c as a function of the X-ray CSR size of the DNDs. Comparing both panels (Fig. 6b, c) it is clearly seen that the smaller the nitrogen content in DND, the higher the NV ⁻ PL intensity. Again, this seems reasonable as nitrogen-related centres and especially A-centres can act as effective quenchers of PL if the NV ⁻ light-emitting centres in some DND particles (one per a hundred of 5 nm DND particles at least) are surrounded by a “gas” of A-centres and other lattice imperfections, similar to the approach proposed in Ref. [33]. This trend gives us a hint at possible ways to enhance the intensity of luminescence from ensembles of DND particles by manipulating their size and nitrogen concentration. Among them, there is the technological enlargement of the mean size of DND particles in the course of their treatment at high pressure and high temperature at appropriate conditions, promoting their recrystallization and further crystal growth [34]. The probable reduction of NV ⁻ content through treatment leading to the reduction of A-centres and other interior defects in the diamond lattice may be compensated in principle by the opposite trend, promoting the brightness of NV ⁻ emission, and reducing the amount of all types of PL quenchers in the system, and hence, substantially improving the crystal quality of particles with sizes exceeding a dozen nanometres [34, 35]. Further works on these topics are now in progress.

DND Aggregate-Specific Fluorescence and Discrimination of Diffraction-Limited Emitters

Characterizing the fluorescence from isolated DND particles or submicron aggregates is crucial both for understanding their potential use as fluorescent markers and to help to mitigate disadvantages related with a relatively low concentration of NV ⁻ in them as determined by EPR. Photoluminescence was recorded for DND aggregates spin-coated on a glass microscope coverslip from an aqueous suspension with an average size of DND aggregates about 30 nm (as measured by dynamic light scattering).

Figure 7a shows two PL 2D maps obtained by confocal microscopy, using a 532-nm wavelength excitation laser with 100-μW output optical power. Bright spots corresponding to DND aggregates up to 500–700 nm in lateral size are observed. Dimmer spots of size around the optical diffraction limit are also observed after selection of the appropriate isolated spots. Figure 7c shows the intensity distribution along the specially selected straight line aa plotted in Fig. 7a. This line crosses about six dim spots of smallest diameter—five spots (1–5) are crossed by the straight line fairly precisely along their centres and one spot (6) is with a small displacement from this line. The corresponding five peaks in intensity distribution are clearly seen in Fig. 7c. Thus, each dim spot 1–5 laying on the line aa corresponds to the DND aggregate of smallest lateral size (in the range below 70 nm). It is possible that all of them are fixed on one V-shaped straight groove existing on the glass coverslip. We successfully fitted each peak in the intensity distribution with a 2D Gaussian \( {A}_i{e}^{-\left[{\left(x-{x}_{oi}\right)}^2+{\left(y-{y}_{oi}\right)}^2\right]/2{s}_i^2} \), where x _oi , y _oi are the Cartesian coordinates of the centres of the dim round spots, A _i -maximum PL intensity of each isolated spot, s _i is a parameter close to s _o  ≈ r _o /3, where s _o is a 1/3 part of the Airy disk diameter r _o . In our case, for λ  = 532 nm radiation (in vacuum) and numerical aperture of microscopic objective NA = 1.40, we have the following values for r _o dan s _o :r _o =1.22λ/2NA ≈ 232 nm and s _o  ≈  77 nm. The s _o value in the 2D Gaussian determines the point spread function (PSF) of and ideal point-like emitter, i.e. the diffraction limitation related with the smallest possible interference ring. For the five peaks mentioned above (i  = 1–5) we found the following s _i values, respectively:85, 77, 77, 84 and 77 nm. Peak 5 has both the highest intensity and s -value equal to the theoretical value s _o  ≈  77 nm. This means that the lateral size of the corresponding emitter for peak 5 is significantly smaller than the PSF size. The same is also true for peaks 1–4. We can conclude that the lateral sizes of DND aggregates laying along line aa are in the range below 70 nm. Each spherically shaped DND aggregate of ~ 30 nm in size ^{Footnote 5} consisting of individual 5-nm particles (percentage of voids is 50%) has about 1.3 NV ⁻ centres in accordance with the 1.1 ppm content of NV ⁻ determined previously. This value is great enough in principle to distinguish aggregates of this size by optical methods. Each of the five aggregates lying along the line aa probably has between 2 and 10 colour centres. The larger the height of the DND aggregate lying on the substrate, the higher the density of luminescence centres (per unit square) for the light emitted more or less perpendicular to the surface. For aggregates with height up to 300–350 nm, the brightness of PL intensity can be at least one order of magnitude greater than that for aggregates of smaller 25–30 nm size. The PL spectrum of one selected aggregate of high brightness (marked by a circle in the 2D map presented in Fig. 7a) was studied in detail. It has roughly the same shape (especially on the right wing above 680 nm) as that for the spectrum shown in Fig. 6a and indicates the presence of both NV ⁻ and NV ⁰ defects. However, the characteristic zero-phonon lines (ZPL) at 638 nm (NV ⁻ ) and 575 nm (NV ⁰ ), respectively, were again not detected. ZPLs are usually well resolved only for diamond crystals of micron size or larger, while for nanometre-scale crystals, they are typically not well-observed due to some experimental or other physical reasons ^{Footnote 6} . Let us mark that for DND aggregates lying on the cover glass, the final treatment with UV/O-Cleaner removed the sp ² -coordinated carbon species around the particles inside the aggregates and as a result the PL spectrum is free from the contribution of light-emitting centres of disorganized sp ² -carbon phase.

a 2D-colour mapping of the PL signal intensity of DND spin-coated on a glass microscope coverslip together with b schematics of the experimental setup. c The PL intensity profile along the “aa” cross-section. d The photoluminescence intensity versus time for the selected DND aggregate marked with a circle in the upper side of the 2D map shown in panel a , in the presence or absence of an external magnetic field. Laser excitation at 532 nm wavelength. Square, 201 × 201 pixels. Integration time is 3 ms/pixel. Step—100 nm. The excitation radiation is focused on the upper surface of the glass coverslip with deposited DND. In zero field (B  = 0), only small changes of intensity, due to the blinking of some of the NV ⁻ colour centres occur. When magnetic field is temporally applied a large decrease of the PL intensity occurs. Upper left panel a :the left scale for PL intensity corresponds to the 2D mapping of another DND aggregate shown in the left bottom corner of the large 2D map. Upper right panel b :schematics of the experimental setup explaining the displacement of the permanent magnet above the coverslip relative to the deposited DND

To confirm the presence of NV ⁻ centres in the DND, we studied the influence of an external magnetic field on the PL intensity. No PL modification is expected from NV ⁰ , which have no magneto-optical properties unlike NV ⁻ . Figure 7d shows the meander-like time variation of the PL intensity from an isolated DND aggregate in the presence of an external magnetic field switched “ON” by quickly bringing a compact permanent magnet close to the DND or by removing it (“OFF”). This occurs as a result of the mixing of the |0〉_GS and |−1〉_GS states of NV ⁻ centres at the ground-state spin level anti-crossing (GSLAC), marked with a circle at magnetic field ~ 100 mT in Fig. 3. Such mixing leads to a change in the populations of these states, accompanied by a PL intensity decrease. The optical transitions between the ground ³ A₂ and excited ² E triplet states preserve the spin quantum number (Δm _S  = 0). However, from the m _S  = ± 1 excited state, the optical excitation of NV ^– also decays with no radiation in the visible domain, through a system of two metastable singlet states before coming back to the ground state [5]. This process is accompanied by radiation at a longer wavelength of 1042 nm defined by the gap between these two singlet levels with S  = 0. This additional decay pathway results in a lower fluorescence intensity from the main radiation transition within the m _S  = ± 1 subsystem of ground and excited triplet states. The experimentally detected decrease of NV ^– centre PL intensity is quite reasonable in the presence of a weak (≤ 100 mT) magnetic field [36], as observed in Fig. 7d. Surprisingly, in our case (for DNDs), this drop is essentially larger than those reported in the literature for Ib HPHT fluorescent microdiamonds and even evaluated theoretically in the framework of the standard NV model [37].