Skip to content

Microwave Near Field Imaging of Externally Injected Signals in an Encapsulated Electronic Device

FieldValue
JournalMicromachines
Published2026-06-10
AuthorsQiang Zhu, Yangfan Zhang, Xin Li, Huan Fei Wen, Jun Tang
DOI10.3390/mi17060711

Diamond nitrogen-vacancy centers resolved the microwave field over a 1 × 1 mm² region of a sealed 128 MB black-epoxy USB flash drive. The maps linked four external pins to internal signal lines and showed channel-dependent crosstalk without opening the package. The authors swept 2.82 GHz to 2.97 GHz at 5 dBm and reconstructed each field map by optically detected magnetic resonance.

The tested sample was a commercially available 128 MB black-epoxy USB flash drive with four exposed interface pins. The pins were, in order, the power supply positive terminal, the data line negative terminal, the data line positive terminal, and the ground terminal. Six signal lines appeared in the selected camera region, but their links to those pins were unknown before testing.

Exposed circuitry was photographed only after acquisition for verification. The drive was new and fully functional during imaging. Each injected channel produced a distinct normalized field pattern. These patterns allowed the authors to infer line-to-pin relationships and crosstalk.

Time-domain reflectometry, continuity testing, and impedance analysis were described as lumped-parameter methods. They use external interfaces but cannot provide spatial distributions of internal signal lines or map a line to a specific interface. The paper addresses that gap by recording the near-field microwave distribution above the encapsulated conductors.

The bulk diamond sensing layer was fabricated by the high pressure high temperature method. It was irradiated with 10 MeV electrons at a dose of 9.8 × 10⁻¹⁸ cm⁻² and vacuum annealed at 850 °C for 2.5 h. The authors applied an NV concentration calibration from their previous work. They state that the compensation removed systematic imaging error caused by nonuniform NV concentration.

Optical readout used 532 nm green excitation and fluorescence collected near 680 nm. The metastable ±1 state has a lifetime of about 300 ns and reduces resonance fluorescence by about 30%. Without an external magnetic field, the zero-field splitting is 2.87 GHz. Applying a magnetic field splits the ±1 levels and shifts the resonant frequency around that center.

This optical readout allowed the diamond to remain a passive sensing layer. The USB drive supplied the microwave excitation, so no external radiation antenna was required.

A beam expander and homogenizer converted the laser output into a flat-top beam. The beam filled a 20× objective with NA 0.45 and illuminated the USB drive. The diamond was held in close contact with the drive surface to collect the local microwave field.

SMA connectors were soldered to the four exposed pins. The Agilent N5183B signal generator covered 9 kHz to 40 GHz, with adjustable power from -144 dBm to 30 dBm. The swept signal was coupled into each interface in sequence.

ParameterSettingRole in the experiment
Measured frequency2.82 GHz to 2.97 GHzStraddle the 2.87 GHz NV resonance
Frequency step400 kHzDefine the spectral sampling interval
Microwave power5 dBmMaintain weak-drive contrast conditions
Optical excitation532 nmPrepare and read NV fluorescence
Objective20×, NA 0.45Focus illumination and collect fluorescence
Translation step40 µmAssemble the complete field map
Imaging region1 × 1 mm²Cover the six observed signal lines

The signal generator and CMOS camera shared a clock source. This synchronization tied each camera frame to a specific swept frequency. The USB drive sat on a motorized translation stage and moved in 40 µm steps during acquisition.

Each camera pixel was treated as an independent photodetector. The authors fit the ODMR spectrum at every pixel with a Lorentzian and extracted contrast and linewidth. The fit used resonance area, off-resonant fluorescence, and full width at half maximum.

Microwave field strength is encoded in both ODMR contrast and linewidth, but under different conditions. At high microwave power, power broadening makes the linewidth proportional to the perpendicular field component. At low power, the linewidth stays constant while contrast is proportional to the square of that field component.

The experiment used 5 dBm to remain in the weak-drive regime. Laser power was held constant and at optical saturation. The reconstruction used saturated optical contrast as its reference, supported by the linear microwave-power and contrast relationship verified in the authors’ previous work.

The concentration correction was then applied to each pixel. This step removed the measured systematic variation caused by NV concentration inhomogeneity in the diamond. The resulting field images were normalized, so they describe relative spatial intensity rather than absolute field amplitude.

The four channel maps showed distinct relationships between the injected interface and the visible signal lines.

Interface and terminalObserved field patternAuthors’ connection inference
Interface 1, power supply positiveStrongest radiation between lines 3 and 4. Line 2 produced a weaker, radiating response.Lines 3 and 4 were the main connection. The line 2 response was attributed to electromagnetic diffusion.
Interface 2, data line negativeStrong radiation between lines 3 and 4, stronger than for interface 1. A clear boundary formed at the upper edge of line 2.Lines 2, 3, and 4 were connected, with line 4 producing the strongest response.
Interface 3, data line positiveEqual intensity appeared between lines 2 and 3 and between lines 3 and 4. The field directly over line 3 was weak.Line 3 was driven. Its near field was distributed symmetrically on both sides.
Interface 4, groundMaximum intensity occurred at line 2. A clear signal appeared at line 4, with weak radiation at line 5.Line 2 was the primary connection. Lines 4 and 5 received crosstalk.

The authors described interface 4 as connected to line 2 with an auxiliary ground. Their mappings are field-based inferences rather than direct electrical continuity measurements. The field patterns nevertheless separated the driven lines from weaker responses caused by coupling.

The USB encapsulation also prevented the illumination problem reported in earlier work. No metal wires were exposed, so reflections from those wires did not disturb the excitation light.

Comparison with electrical and prior NV methods

Section titled “Comparison with electrical and prior NV methods”

The paper distinguishes its measurement from conventional interface testing. Time-domain reflectometry, continuity testing, and impedance analysis return lumped electrical information but do not show where fields occur within the package. Sequential injection and field imaging add that spatial information for each external channel.

The cited commercial NV approaches measure stray magnetic fields generated by currents in devices. Their main targets are wafer-level or chip-level magnetic imaging, failure analysis, and defect localization. This work instead maps the microwave field radiated by signal lines inside a packaged product.

The drive served as the microwave source during imaging. This setup avoided the nonuniform excitation field that an external horn antenna could introduce. The earlier concentration-compensation work also supplied the correction needed for reliable reconstruction across the diamond surface.

Limits and device engineering implications

Section titled “Limits and device engineering implications”

The experiment covered one new, fully functional USB drive. It examined four interface pins and a 1 × 1 mm² region containing six signal lines. The reported maps were normalized and did not provide absolute field values. The study therefore did not establish sensitivity to defects in a failed product or quantitative agreement with a calibrated probe.

The injected waveform was a continuous swept microwave signal, not an actual USB data communication signal. The authors identify this distinction as a limitation of the demonstration. Their results support near-field characterization, electromagnetic compatibility diagnosis, and internal functional verification, but broader data operation remains untested.

For device engineers, the demonstrated workflow can associate an external pin with internal routing and reveal crosstalk without exposing the conductors. It is most useful when the channel produces a distinct field fingerprint that can be compared across injections.

Diamond selection also affects measurement validity in this implementation. The demonstrated sensor was an irradiated and annealed bulk crystal whose NV concentration distribution had been calibrated in prior work. The method therefore depends on controlling and compensating the sensing layer’s response.

The demonstrated result remains a four-pin, 1 × 1 mm² field map of one sealed USB drive under a continuous 2.82 GHz to 2.97 GHz sweep. Operation with real USB data and quantitative absolute-field readout remain untested.

View Original Abstract

Addressing the challenges of electromagnetic compatibility testing and non-destructive inspection of internal structures in miniaturized electronic devices. This paper reports a non-destructive testing method based on wide-field imaging using diamond nitrogen-vacancy (NV) centers, and systematically demonstrates its application on a black-epoxy-encapsulated Universal Serial Bus (USB) flash drive. In the experiment, a swept microwave signal from 2.82 GHz to 2.97 GHz was sequentially injected into the four external interface pins of the USB drive. A bulk diamond served as the quantum sensing layer, and optically detected magnetic resonance (ODMR) was employed to perform wide-field imaging of the microwave field distribution on the surface of the signal lines within a 1 × 1 mm2 region of interest. The experimental results show that the microwave field distributions corresponding to different interface channels are significantly different. Based on these differences, the connection relationship between each signal line and its corresponding interface pin can be clearly identified, and the differences in field distribution as well as crosstalk characteristics among channels can be revealed. The method established in this work provides an effective technical pathway for non-destructive electromagnetic testing and functional verification of electronic products.

  1. 2022 - Highly Plasticized Lanthanide Luminescence for Information Storage and Encryption Applications [Crossref]
  2. 2008 - Integrated Microwave and Optical Scanning Probe for Magnetic Resonance Imaging [Crossref]
  3. 2021 - Real-Time Removal of Topographic Artifacts in Scanning Microwave Microscopy [Crossref]
  4. 1995 - Imaging Radio-Frequency Fields Using a Scanning SQUID Microscope [Crossref]
  5. 2017 - On-Chip Quantum Tomography of Mechanical Nanoscale Oscillators with Guided Rydberg Atoms [Crossref]
  6. 2017 - Characterizing the Local Vectorial Electric Field near an Atom Chip Using Rydberg-State Spectroscopy [Crossref]
  7. 2022 - Measurement and Simulation of the Magnetic Fields from a 555 Timer Integrated Circuit Using a Quantum Diamond Microscope and Finite-Element Analysis [Crossref]
  8. 2011 - Electric-Field Sensing Using Single Diamond Spins [Crossref]
  9. 2023 - Radio-Frequency Electric Field Sensing Based on a Single Solid-State Spin [Crossref]
  10. 2019 - Robust and Accurate Electric Field Sensing with Solid State Spin Ensembles [Crossref]