Formation of two-dimensional multichannel vertical optical waveguides in a nitrogen-vacancy center diamond using a femtosecond Bessel beam laser for local quantum sensing Open Access

In diamond crystals, nitrogen vacancy (NV) centers function as quantum sensors that can measure magnetic fields, temperature, and stress with high sensitivity without making contact. NV centers have garnered increasing attention for their unique properties, particularly in biological applications, where local information is essential. For instance, NV centers in nanoparticles have been utilized to measure intracellular temperature and magnetic fields.¹ Moreover, NV centers have been employed to assess the electrical and magnetic characteristics at the tissue level, including measurements of nerve action potentials.^2,3 To understand the functions of biological tissues, it is crucial to acquire associated information, such as local magnetic and electrical fields, signal propagation from tissues, and local interaction dynamics.

Ota et al.⁴ recently attempted to elucidate the details of brain functions across entire micrometer-scale spatial regions by optically measuring the functional networks of cerebral cortical neurons. The process entailed designing specialized optical systems to detect the two-dimensional spatial distributions of local networks and performing invasive observations by making incisions in the skull. While their proposed optical method has vital implications in the field of medical research, its application in clinical settings remains impractical owing to the need for room-temperature measurements, high sensitivity, in situ observation capabilities, and noninvasiveness.

As discussed above, in order to apply diamond NV centers to the medical field, it is extremely important to develop techniques that enable these devices to obtain localized spatial information on a cellular scale.

For that purpose, a sensing array composed of cell-sized high-resolution elements is required for such applications. If an observation system that uses an NV center array in diamond could noninvasively map neural activity from the magnetic fields generated by brain nerve cells, it could pave the way for elucidating neural interactions and activity with unprecedented ease.

A critical challenge in achieving high-sensitivity magnetic field measurements using NV centers is the efficient detection of the red luminescence emitted by NV centers, whose intensity changes in response to magnetic field variations. However, the isotropic nature of the NV center luminescence results in significant light dispersion, reducing the observable intensity and, consequently, the magnetic field sensitivity.

To address this problem, recently, femtosecond lasers (hereinafter referred to as fs lasers) have been employed to create lateral NV center waveguides^5–13 parallel to the diamond substrate surface. These waveguides are type-II horizontal waveguides that provide a method to partially confine and enhance luminescence. However, such type-II horizontal waveguides cannot completely confine luminescence within the structure. If microscale vertical optical waveguides can be realized, an NV center-sensing array composed of cell-sized high-resolution elements can be realized.

In this study, we have realized the fabrication of a high-spatial-resolution NV center micro-scale type I vertical waveguide (hereinafter referred to as waveguide-element) on the surface of diamond substrates using a Bessel beam femtosecond laser¹⁴ (hereinafter referred to as Bessel beam laser) that enables complete core confinement of light in the waveguide due to the total reflection of the light at waveguide wall and enhancement of collection efficiency of the NV center emissions. The size of the waveguide was controlled from 5 to 33 μm.

Furthermore, we developed an innovative sensor, micro-NV center array element of diamond (MAED), for the purpose of sensing the spatial distribution of magnetic field, temperature, and strain over a wide range and with high resolution. Since there are no examples of research on waveguide-elements of NV center with spatial resolution equivalent to the size of cells, we demonstrated the observation of magnetic field distribution with cell size accuracy using MAED.

We used single-crystal (100)-plane CVD diamond substrates with a nitrogen content of approximately 1 ppm, manufactured by Element Six (UK). The dimensions of the substrate were 3 × 3 mm² with a thickness of 250 μm. Surface cleaning was performed using a mixture of nitric acid and sulfuric acid at 200 °C. Once the waveguides were fabricated, they were annealed at 1000 °C for 3 h¹⁶ via rapid thermal annealing in a furnace, and the surfaces were cleaned via acid treatment.

Unlike conventional fs lasers, which focus the laser energy at a single point and process materials only within a confined focal area, the Bessel-beam laser sustains a uniform beam diameter along its axial direction over tens of micrometers to several millimeters. The extended axial uniformity enables the laser to process the diamond substrates uniformly from the front surface to the back. Additionally, the distinct intensity distribution and heat dissipation characteristics of the Bessel-beam laser offer unique modification conditions compared to traditional fs lasers. Uniform processing was achieved throughout the diamond depth by designing the main beam length such that it exceeded the thickness of the diamond substrate. Therefore, the Bessel beam laser is not only suitable for forming vertical waveguides in diamond but can also be used to create NV centers anywhere in the diamond, making it a powerful tool in the fabrication of highly sensitive, vertical waveguide-type diamond NV center sensors. Based on the size of the measurement target, the spatial resolution of the generated vertical waveguide can be designed arbitrary. In our experiments, it was between 5 and 33 μm.

The Bessel beam laser system used in this study is shown in Fig. 1(a). A fs laser (280 fs pulse width, 200 kHz repetition rate, and 532 nm wavelength) with a Gaussian beam profile was converted into a ring beam profile using an axicon lens. Subsequently, the ring beam was focused and interfered with additional lenses and objectives to create a long Bessel beam with a nearly uniform intensity profile along its axis, as described in Ref. 14. The length of the uniform beam was controlled by adjusting the ring width, while the beam diameter was controlled using the objective lens magnification. For photoluminescence (hereinafter referred to as PL) measurements, we employed a 532 nm CW-pumped Raman micro-spectrometer and a custom-built 532 nm CW-pumped transmission micro-PL spectrometer shown in Fig. 1(b). The transmission spectrometer enabled the simultaneous measurement of PL spectra and spatial profiles. Magnetic field observations were conducted using optical detection of magnetic resonance¹⁵ (hereinafter referred to as ODMR) as shown in Fig. l(c) in which wide excitation green laser (532 nm) is used for the measurement of spatial distribution of fine magnetic field generated by a current through Cu wire embedded on cell^-size vertical waveguides of MAED.

Square vertical optical waveguides were fabricated by irradiating a diamond substrate using a computer controlled Bessel beam laser, which maintained a uniform beam diameter along its axis from the front surface. This enabled the entire substrate thickness to be graphitized in a single step. The states of the front and back surfaces of the diamond were confirmed using optical microscopy, and internal graphitization was examined by shifting the focus of the microscope through the depth of the substrate. The micrographs at the front and the back surfaces of the fabricated waveguide array are shown in Figs. 2(a) and 2(b), respectively. The micrographs contained nine vertical square waveguides with sizes ranging from 5 to 20 μm. Even the back surface was graphitized by the laser irradiated from the front surface, and the sides of the waveguide were surrounded by graphite, creating a wall that reflected the red luminescence from the NV center inside the waveguides.

Considering that diamond with a stable structure has a high refractive index and that the occurrence of defects reduces the refractive index of a material, irradiating diamond with a high-intensity fs laser may cause defects, thereby reducing the refractive index of the diamond. In other words, a low-refractive-index diamond layer is formed between the graphitized waveguide wall and the core diamond, which facilitates the formation of a waveguide that resembles a typical conventional optical fiber. We believe that if this low-refractive-index diamond layer can be formed, a waveguide-element can be created in the diamond, which can be used as an element to perform micro-magnetic field measurements. All the red light emitted from NV centers in the waveguide can be confined in the waveguide, and additionally, a large number of NV centers can be created in the core of the waveguides under proper irradiation condition of Bessel beam laser, thereby increasing the sensitivity of the micro-magnetic field measurement.

The formation of the waveguide walls is highly dependent on the irradiation intensity and scanning speed of the Bessel beam laser. Once the laser intensity exceeds a certain level, the wall becomes thicker, and the defect area created in the surrounding diamond widens, increasing the optical confinement effect. However, the NV center light-emitting waveguide area is significantly thinner. In this case, the red luminescence of the NV center in the waveguide can be confined efficiently; however, the amount of red luminescence may also reduce because the NV center is reduced by the many defects in the waveguide. Thus, for waveguides fabricated with Bessel-beam lasers, it is crucial to find a balance between the optical confinement effect and defect formation.

Figure 3(a) shows a crosshatched waveguide array written using a Bessel beam laser. Figure 3(b) illustrates a charge-coupled device (CCD) image of the red luminescence that propagates through the waveguide captured using the PL measurement setup illustrated in Fig. 1(b). A 532-nm continuous wave laser was used to excite the central region of the crosshatched waveguide from the back surface of the diamond substrate, and the 532-nm excitation light was filtered out and only the red light was observed by the CCD. The red light emitted from the excited NV centers propagated to the surface of the diamond while being reflected within the waveguide. Considering the relationship with other waveguides, the red light that propagated through the waveguide appeared to couple with adjacent waveguides, with less coupling with the diagonal waveguide. This indicates that the crosshatched waveguide did not confine the emitted light well. A multi-walled waveguide design was developed to minimize coupling and light leakage.

Figure 4(a) shows a fourfold walled waveguide (15 μm square core) with every 3 μm wall distance fabricated to reduce the leakage and the coupling with the neighboring waveguides. Figures 4(b) and 4(c) show the PL intensity distributions at the diamond surface with and without the waveguide structure under a broad-area excitation of a 532 nm laser from the back surface of the substrate, respectively. In the absence of a waveguide, the PL intensity spreads uniformly across the substrate. By contrast, for the waveguide, the PL from the central core was strong, indicating that the emission from the waveguide was confined. The luminescence intensity in the central region was approximately 1.7 times higher than the intensity in the absence of the waveguide. In this case, the waveguide length was almost 200 μm: therefore, a similar enhancement of almost eight times would be expected if a substrate with a thickness of 1 mm were used for the vertical waveguide. In addition, when the NV center was formed in the core by a fs laser, the luminescence intensity increased by a factor of approximately 7 in our experiment, suggesting an overall PL intensity enhancement of over 50 times, that is, a sensitivity enhancement of approximately seven times.

The minimum observable magnetic field is given by Eq. (1). According to this equation, the higher the number of NV centers in the waveguide, the higher the sensitivity; however, if the density of NV centers is too high, the interaction between NV centers increases and T₂^* decreases. Thus, there exists a trade-off between the density of NV centers and T₂^* to increase the sensitivity. However, in a measurement system that uses the MAED, the sensitivity can be increased without decreasing T₂^* because the absolute number of NV centers can be increased by increasing the volume of the core region of the waveguide, that is, increasing the number of NV centers inside the waveguide by increasing the width and length of the waveguide without increasing the density of the NV centers. In addition, the collection efficiency can be increased because of the confinement effect of the waveguide, even though light can be generated anywhere in the waveguide, as shown in Fig. 4(b).

A copper wire with a diameter of 25 μm was placed between the MAED and the microwave antenna to investigate the sensing availability of magnetic field distribution using the MAED, as shown in Fig. 1(c). The magnetic fields were generated by controlling the current through a wire. A 5 × 5 waveguide array (MAED) was constructed inside the diamond for the magnetic field distribution measurements, as shown in Fig. 5(a). Each of the waveguide-element contains a quadruple wall structure, the distance from wall to wall is 3 μm, and the distance of each element is 20 μm. Each element contains a core section with a 15 μm square that detects the red emission from the NV center to sense the magnetic field. The four-layer wall blocks light leakage. Each element is a square with dimensions of 33 μm.

The brown lines in Fig. 5(a) represent copper wires placed across the MAED to generate a small magnetic field. In this study, we controlled the magnetic field with the current in the copper wire to confirm that each waveguide-element of the MAED detects the magnetic field individually. This is a proof-of-concept experiment to demonstrate that we can measure the spatial distribution and propagation of magnetic fields generated by neurons with cellular resolution, which is the goal of this study. Experiments measuring actual nerves will be described in a future paper.

Figure 5(b) presents the changes in the ODMR spectra of representative waveguide-elements as the magnetic field strength increases with current (0–20 mA). It is evident that the graph shifts to the right as the magnetic field strength increases. A graphical representation of the relationship between the magnitude of this shift and current shows a clear linear dependence [Fig. 5(c)], which can be observed near the microwave frequency of 2.764 GHz. The microwave frequency shift is proportional to the magnetic field strength (which is proportional to the current flow). The magnetic field strength is calculated to be about 2 μT at a position 100 μm away from the copper wire when a current of 1 mA is applied directly to the copper wire. In this experiment, the distance from the observation position of the PL of the NV center during ODMR measurement to the copper wire was more than 100 μm, and from Fig. 5(c), it was possible to measure the 1 mA case, and the magnetic field strength when a current of 1 mA was applied would be about 2 μT. Therefore, the magnetic field sensitivity of this study is considered to be about 2 μT.

Figure 6 shows the magnetic field distribution measured by the MAED. Figure 6(b) shows a CCD image of the 4 × 4 MAED inside the red enclosure in Fig. 6(a). The MAED was excited over a wide area by widening the spot size of the 532 nm wavelength excitation laser, as shown by the bright areas in Fig. 6(b). The numbers on the right and bottom of the image in Fig. 6(b) indicate the addresses of each waveguide-element and correspond to the matrix of the 3D plot shown in Fig. 6(c). Figure 6(c) shows a graph of the microwave frequency shift observed for each waveguide-element due to the generated magnetic field. Each waveguide-element corresponds to the magnetic field strength distribution when a current of 20 mA flows through it. The microwave shift shown in the figure is calculated as the average of the shifts measured at two points on each of the steep slopes on both sides of the ODMR spectrum peak, for a total of four points. The shift is commonly calculated from the derivative and change in the intensity of the ODMR spectrum. In this experiment, a microwave frequency of 2.7704, 2.7700, 2.7646, and 2.7648 GHz was used to calculate the shifts of the microwave frequency induced by 20 mA current flow in the wire. The ODMR spectra were measured from 163 CCD images of PL as a function of the microwave frequency. The microwave frequency shift (proportional to the magnetic field strength) is shown as the height of the vertical bar in the figure. Partially missing elements were ignored owing to their limited measurement ranges. The arrangement of the elements was determined using the sheet address (x, y). The typical position of the vertical waveguide-elements (2,3), (3,3), and (4,3) and magnetic field induced by the current in the Cu wire are shown in Fig. 6(d). As shown in this figure, the diamond NV center devices measure the magnetic field within a diamond waveguide of 250 μm thickness located 100 μm above the Cu wire. As depicted in the figure, each device observes the magnetic field distribution over a range of approximately 150 μm in the lateral direction. For example, at positions 2-3 and 4-3, the magnetic field has components in both the x-direction and the perpendicular (z-) direction, while at position 3-3, which is directly above the Cu wire, the magnetic field has only an x-direction component. (In this experiment, the x-component was observed via the Zeeman effect.)

Figure 6(c) shows the average shift of the ODMR spectra across all microwave frequencies for each device. These results reveal that the magnetic field generally becomes stronger near the wire and decreases rapidly with distance at the devices located farther from the wire, providing a magnetic field strength distribution with resolution at the cell size level. However, at specific microwave frequencies in devices 4-3 and 4-2, we observed a reversal in the magnetic field direction. That is, at waveguide device positions 2-3 and 4-3, the magnetic field has both x- and z-direction components, as shown in Fig. 6(d). When a device is located away from the wire in the x-direction, the z-direction magnetic field becomes stronger relative to the x-direction, potentially resulting in a negative observed magnetic field direction. The z-component of the magnetic field may cause mixing of Zeeman levels, leading to the appearance of related peaks or changes in linewidth at specific microwave frequencies. In the future, a more detailed analysis may yield intriguing insight into the local variation of the magnetic field direction.

When observing neurotransmission on a macroscopic scale, it is crucial to obtain the associated information, such as the local magnetic and electrical information of the nerve, as well as the propagation of information from there and local interaction information. Accordingly, the MAED may be suitable for macroscopic observation of neurotransmission. At present, the sensitivity of magnetic field observation is approximately 2 μT, which is still low for measuring neurotransmission, but the sensitivity can be greatly increased by optimizing the NV center fabricated in the core using the fs laser, increasing the waveguide length, optimizing the waveguide performance, using highly doped diamond, improving crystal quality, and further optimizing crystal annealing and measurement methods (including differential measurement).

MAED enables the precise measurement of local fine magnetic field spatial distributions, highlighting its potential for ultra-sensitive spatial magnetic field sensing, and MAED shows promise for the functional mapping of action potential dynamics in axons and other biological systems, offering insight into neural interactions and dynamics.

In this experiment, we found the multi-walled vertical waveguide structure significantly improves the light collection efficiency from the NV centers, paving the way for highly sensitive NV center sensor arrays. Although this multi-wall waveguide is effective for optical confinement, it is unsuitable for large-scale array fabrication. However, depending on the fabrication conditions, optical confinement can still be achieved with a design that uses fewer walls, and thus, it is necessary to explore such designs in the future.

This innovative approach demonstrates the feasibility of using MAED sensors in a wide range of applications ranging from quantum sensing to biological research. These findings lay the foundation for future developments in high-resolution noninvasive sensing technologies.

We express our gratitude to Professor Tsutomu Araki of Ritsumeikan University and Professor/M.D. Manabu Honda of the National Center of Neurology and Psychiatry for their collaboration during the early stages of this study. We also extend our gratitude to Professor Itaru Kamiya and Professor Naotaka Iwata of the Toyota Technological Institute for their valuable discussions and advice. This research was supported by the Amada Foundation, Iketani Science and Technology Foundation, and the Uehara Memorial Foundation.