Characterization of Autofluorescence in Optical Fibers for NV-based Sensing Applications
Paper details
Section titled “Paper details”| Field | Value |
|---|---|
| Journal | Applied Optics |
| Published | 2026-02-07 |
| Authors | Sigfrid Lindén Johansson, Alexander Bukschat, Dennis Lönard, Alena Erlenbach, Jonas Gutsche |
| DOI | 10.1364/ao.611880 |
Under the paper’s comparison conditions, a 2 m optical fiber produced background equal to fluorescence from a diamond with a 740 nm diameter. The estimate assumed an NV concentration of 3.5 ppm and used a 15 µm diamond as its reference. It also assumed that NV fluorescence scales linearly with diamond volume at constant doping. This estimate marks a practical limit for small fiber-coupled sensing diamonds.
Fiber background in NV readout
Section titled “Fiber background in NV readout”The background contained two components with different spectral behavior. Raman scattering dominated below 600 nm, while defect fluorescence dominated above 600 nm. The broad defect band overlapped the 600 to 800 nm NV emission region. Typical green excitation near 520 nm therefore produces background within the same spectral region used for magnetic-field readout. Reflected and scattered excitation light can be removed with filters. Fiber fluorescence at fixed wavelengths can remain inside the accepted detection band. This overlap lowers signal contrast and the signal-to-noise ratio.
The longer excitation settings represent wavelengths near those used for NV spin-to-charge conversion. These measurements help determine whether fiber background remains manageable when excitation moves into the NV emission region.
Measurement method
Section titled “Measurement method”The excitation path included a fiber coupler, half-wave plate, polarizing beam cube, bandpass filter, optional neutral density filter, and dichroic mirror. The spectrometer recorded light generated inside each test fiber in the reverse direction. The output path used a blazed grating, lens, and camera-based spectrometer. Excitation power measured at position p2 was held constant for comparative spectra.
A 650 nm calibration laser measured coupling losses through each test fiber. Known laser wavelengths calibrated the pixels, and a second-degree polynomial mapped pixel position to wavelength. Long-pass filters removed wavelengths below 550 nm for 515.8, 520.2, and 532.0 nm excitation. They removed wavelengths below 650 nm for 593.8 and 636.0 nm excitation. Some configurations also used notch filters to suppress excitation leakage. Changing the beam-dump material caused no measurable spectral change.
Raman and defect contributions
Section titled “Raman and defect contributions”Raman peak intensity increased linearly with excitation power. The peaks below 600 nm shifted with excitation. The authors associate this shift with an approximately 2 nm change in the diode laser wavelength during the power series.
For 520 nm excitation, the expected largest fused-silica Raman displacement was about 34.5 nm. Measurements also showed Raman shifts near 48 nm for every fiber. Replotting spectra from 515.8, 520.2, and 532.0 nm against Raman shift aligned the Raman maxima. Five peaks occurred at 613, 809, 1056, 1186, and 1612 cm⁻¹. The 1612 cm⁻¹ peak is not usually reported for fused-silica glass. Peaks below 600 cm⁻¹ could not be detected with the selected filters.
Above 600 nm, the broad fluorescence maximum stayed at 644.9 ± 1.3 nm. This position matched the reported emission of non-bridging oxygen hole centers (NBOHCs). The NBOHC-region signal was much stronger under 593.8 and 636.0 nm excitation. The plotted spectra were divided by 40 and 160, respectively, to match the 532.0 nm curve. The authors attribute this increase to an NBOHC absorption band near 630 nm.
Raman peaks under these longer excitation wavelengths were two orders of magnitude weaker. They remained hidden beneath the stronger fluorescence band.
Assembly and operating conditions
Section titled “Assembly and operating conditions”Epo-Tek 353ND gave the lowest autofluorescence among the tested epoxy glues, although the epoxy effect was weak. A self-assembled fiber using 353ND had a spectrum comparable to the commercial assembly. An illuminated empty ceramic connector produced a distinct maximum near 695 nm. The same peak was absent when excitation was properly coupled through a connectorized fiber.
Protective tubing did not worsen fluorescence above 600 nm. Removing it increased intensity in that region, while the trend reversed below 600 nm. The authors associated the difference with an unpolished end, lower coupling efficiency, and external coupling along the fiber. Scratching a polished facet with a coarse 6 µm-grid pad increased intensity across the spectrum. A local 695 nm maximum suggested connector material on the scratched facet.
Shortening the fiber reduced the total signal. This trend indicated that background light was generated throughout the fiber. Small deviations were attributed to imperfect cutting and remaining dirt. Bending had no measurable effect above the minimum short-term radius of 16 mm.
Cooling approximately half of the fiber in liquid nitrogen changed the 600 to 750 nm region clearly. Fluorescence first increased, then decreased slightly while remaining above the pre-cooled level. The region below 600 nm stayed roughly constant. The authors suggest a stress contribution from the coating and silica core. They also cite known NBOHC temperature dependence.
Coupling angle had little effect while light remained guided. Beyond about 200 degrees of mirror rotation, no guided light reached the fiber and the ferrule spectrum dominated. A slight intensity increase before coupling failed could indicate a cladding contribution.
Comparing available fibers
Section titled “Comparing available fibers”The comparison covered 14 fiber configurations. Core diameters ranged from 3.45 µm to 400 µm, and numerical apertures ranged from 0.12 to 0.50. Most fibers were 2 m long, while FT400EMT and OM2 were 5 m long.
| Fiber | Core diameter | Numerical aperture | Reported result |
|---|---|---|---|
| AFM200L | 200 µm | 0.22 | Lowest fluorescence above 650 nm; aluminium-coated |
| UM22-100 | 100 µm | 0.22(2) | Second-lowest above 650 nm; sharp feature near 645 nm |
| FP200ERT | 200 µm | 0.50 | Low total intensity; polymer cladding |
| FG050LGA | 50 µm | 0.22(2) | Broad 600 to 750 nm band; used in external-factor tests |
| OM2 | 50 µm | 0.2 | Highest Raman intensity in the 50 to 105 µm group |
AFM200L produced the lowest fluorescence above 650 nm. UM22-100 followed it and showed a sharp feature near 645 nm. FP200ERT also had low total intensity and a higher numerical aperture. Its polymer cladding may contribute more background below 650 nm than doped fused silica.
OM2 had the strongest Raman intensity in the 50 to 105 µm group. The authors propose its graded-index construction, germanium doping, and manufacturing process as possible causes. Small-core fibers also showed a 695 nm peak found in the empty-connector spectrum. No monotonic dependence appeared between fluorescence and core diameter or numerical aperture. The authors recommend measuring candidate spectra before selecting a fiber.
Limits and device design
Section titled “Limits and device design”The 740 nm crossover estimate applies only to the stated fiber and assumed volume scaling. It does not establish a universal diamond size limit. Other fluorescence sensors in comparable spectral ranges may face similar background, but the direct calculation applies to fiber-based NV sensors.
Measured intensities were mainly relative because some optical transmission curves were unknown. The fiber set also included different lengths, which affect total background. Long-term bleaching, aging, and drift were not quantified. The authors leave these effects and repeated fibers with varied core and cladding diameters for future work.
Hollow-core fibers can have lower background, but the study excluded them because of cost and integration difficulty. They are difficult to integrate into setups requiring a closed, solid fiber facet. Excitation modulation can also complicate background correction. NV charge-state populations change with excitation intensity, so the NV spectrum changes during modulation.
Fiber selection must balance background against collected light. Numerical aperture and core diameter determine coupling from the source. The following measures follow from the measured results:
- Minimize fiber length.
- Compare candidate spectra instead of selecting only by core diameter or numerical aperture.
- Avoid illuminating the ceramic ferrule.
- Polish and clean the fiber facet.
- Control temperature when operating conditions vary.
- Match the core diameter to the source and sensing diamond, or add suitable optics.
Materials and support
Section titled “Materials and support”The fiber-coupled NV sensor requires an NV-containing diamond; the paper’s reference used a 15 µm diamond with 3.5 ppm NV centers. The 6CCVD diamond materials catalog lists single-crystal and polycrystalline diamond grown by MPCVD, but its boron-doped option does not supply NV centers. An NV-specific material specification would therefore need confirmation through engineering support.
View Original Abstract
Optical fibers are crucial for guiding light in various sensing applications. Especially for quantum sensors such as the nitrogen-vacancy (NV) center in diamond, they enable light control and device miniaturization. However, fluorescence and scattering within the fiber, often referred to as fiber background, autofluorescence, or autoluminescence, can overlap spectrally with the NV centers’ fluorescence, degrading the signal-to-noise ratio and thus limiting sensor sensitivity. Here, we investigate the optical spectra of standard optical fibers, considering material dependencies, physical influences, and their fluorescence scaling with excitation power and wavelength. Our results identify spectral components and fiber types with minimal unwanted background signals, guiding the selection of optimal fibers for NV-based quantum sensing.
Boron Doped Single Crystal Diamond Plate
View Specs →
Boron Doped Diamond Plate
View Specs →