the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Integration of electrically detected magnetic resonance on a chip (EDMRoC) with charge pumping for low-cost and sensitive defect characterization in SiC MOSFETs
Abstract. Integration of microwave sources and detection circuits has led to the design of very compact electron paramagnetic resonance (EPR) instruments, so-called EPR on-a-chip (EPRoC). As recently demonstrated, this approach also offers opportunities for electrical detection of magnetic resonance (EDMR), a variant of EPR in which the magnetic resonance effect is detected via changes in the electrical properties of materials or devices. Here, we report the demonstration of EDMRoC on lateral SiC MOSFETs under charge pumping (CP) conditions. The detected CP current gives direct access to microscopic information about the recombination centers within the transistor gate inversion region under the gate dielectric. Efficient and selective microwave excitation of the region of interest of the device can be obtained by only modest modifications to both the MOSFET and the EPRoC electronic board. A comparative study between EDMRoC and a traditional resonant cavity configuration reveals comparable signal-to-noise ratios for CP-detected EDMR spectra. In addition to space- and cost-efficiency, EDMRoC offers alternative detection modes with scanning and modulation of the microwave frequency, as well as potentially easier sample mounting and exchange. We end with a discussion of the advantages, limitations, and perspectives of the EDMRoC set-up compared to EDMR in a conventional EPR spectrometer.
- Preprint
(1551 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on mr-2024-11', Anonymous Referee #1, 10 Aug 2024
The paper describes the use of a miniature EPR-on-a-chip device to measure the electrically detected spin resonance signal from a custom-made test sample of a SiC MOSFET device. Details are provided regarding the measurement setup, and test results are presented in comparison to the use of Bruker's conventional cylindrical cavity. The EPR chip is tested first using a conventional large electromagnet and then also with a custom-made compact permanent magnet.
Overall, it is a fair paper that describes a valuable use of the EPR chip for electrical spin detection, and I recommend accepting it subject to the following corrections:
General Comments:
-
The EPR chip is small, but the entire experimental setup includes many components (synthesizer, current amplifier, etc.). It would be helpful to show all these devices in a schematic drawing and highlight the role of the EPR chip and what it saves over a conventional setup. Specifically, it would be useful to understand if the EPR chip in this context can simply be replaced by a small broadband antenna.
-
It would be beneficial to add a figure that summarizes and explains the CP mechanism in the MOSFET device.
-
Regarding the text "MOSFET device," does it resemble a real device, or is this just a test sample? What does a real sample look like, and what are the prospects of measuring a real sample?
-
Figure 2c - Where are the n, p, and gate parts in the structure shown in 2c?
-
Figure 2e - Please add a scale bar to this figure to appreciate the distances.
-
Line 157 - An illustrative figure for the spin resonance effect in the measured material would be helpful to include.
-
Line 205 - It would be useful to see the B1 field superimposed on the measured sample in Figure 2c.
-
Line 257 - This section is repetitive of the previous subsection. Consider rearranging it to avoid redundancy.
-
What is the (calculated?) static magnetic field profile/homogeneity for the permanent magnet?
-
Line 360 - What about the use of methods like those in 10.1016/j.jmr.2015.02.010?
Typos:
- Line 182 - "2(c)"
- Line 145 - "2(e)"
Citation: https://doi.org/10.5194/mr-2024-11-RC1 - AC1: 'Reply on RC1', Sofie Cambré, 20 Sep 2024
- AC2: 'Reply on RC2', Sofie Cambré, 20 Sep 2024
-
-
RC2: 'Comment on mr-2024-11', Anonymous Referee #2, 12 Sep 2024
The paper describes the application of EDMR on a chip (EPRoC) to detect spin-dependent transitions in SiC-based MOSFETs that may determine the functionality of the device. The paper compares the signal to noise of EPRoC with and without a permanent magnet to conventional EDMR performed on a commercial EPR setup with electromagnet. The scientific content of the paper is interesting enough to be considered for publication, because the EDMRoC opens up a huge opportunity for the scientific community for novel applications and so far. The paper deals more with the measurement setup and less with the physics of the device, so that it is rather difficult to read for a community that is not an expert in the topic.
Nevertheless, I recommend the paper for publication, providing that the following suggestions have been considered by the authors.
General comments:
- There is a large number of publications using EDMR applied to solid state device, among others transistors, Schottky and pn diodes and solar cells that date back to the last century. I recommend to also cite that work, specifically the work of Martin Brandt and collaborators (e.g. Genshiro Kawachi et al 1997 J. Appl. Phys. 36 121), who have also performed fundamental work developing the EDMR technique.
- The readability of the paper would be dramatically enhanced if a scheme of the electronic processes of the MOSFET are included. The authors should show and discuss the specific spin-dependent transitions and their influence on the recombination current and the functionality of the device.
- The paper of McCrory (ref. McCrory 2019) demonstrates how the limitation for the EDMR application is indeed the resonator and not the price of the instruments. McCrory et al. also show many new applications and experiments that can be performed using a different detection scheme. The authors should comment on this.
- A reference should be given that describes in a bit more detail how the MOSFETs have been prepared.
- A definition of the EDMR signal intensity should be given. Usually the MW-induced current change is normalized to the total current. This, however, seems not to be the case here. The authors should explain, why they choose to use ΔI instead of ΔI/I and how they can compare signal intensities in this way. Moreover, signal amplitudes are compared although different frequencies are being used. EDMR shows a weak MW frequency dependence and this should then be considered (see e.g. Fukui et al. Volume 149, Issue 1, March 2001, Pages 13-21).
- The origin of the EDMR signal is not discussed. What type of a defect is observed and how does this defect inter act with the CP condition. The authors have to describe details about the mechanism (is it a spin-pair mechanism (KSM)?) or cite accordingly. If this is not a KSM process, the magnetic field dependence will be strong and must be considered.
- Since this paper deals with spectroscopy, the spectroscopically determined line parameters should be stated in this paper and compared to literature values.
- A large portion of the discussion is about the low-cost approach. How does the EDMRoC compare to EDMR performed with a strip-line approach as is frequently used? This setup is extremely easy to established and does not require a by far more complex device like the EPRoC. The authors should discuss this to some extend.
Line 28: The statement “…but also more complex devices…“ is too bold. A solar cell, for instance, can easily exceed the complexity of a transistor, because this device has to be considered under illumination (large quasi Fermi splitting) and under high charge injection. Leave out the word “more complex”
Line 45: The authors should report on previous works where the two suggested ways to improve the sensitivity are described.
Line 53: How compact and expensive is the setup? The word “extreme” demands an explanation. And what other setup do the authors compare their’s to? For example, the antenna used by McCrory for the EDMR applications is rather small and not to costly. Please comment.
Line 60: Which Si technology was used for the fabrication of the chip?
Line 74: Omit the word quantitative. In this report the argumentation is mainly qualitative.
Line 82: What type of TIA was used?
Line 83: What is an application specific ASIC?
Line 86: Is there any reference to the custom made PLL? Can the authors give details?
Line 108: I suppose you mean a sweepable magnet?
Line 145: Should read Fig. 2e
Line 173: The sentence is incomprehensible. Please rephrase.
Line 177: Please provide the settings and details of the TIA (see comment line 82).
Line 208-216: What is the distance between EPRoC and the MOSFET? Has this been considered for the calculation? The MOSFET only covers a part of the coils of the resonator array. How does this affect the results?
Fig. 3b,c,d: The y-axis should read DI/I or DI.
Line 241: What was the modulation frequency? Where any phase shifts observed? In particular, the EDMR signal often also has a non-resonant signal which would only appear in the BM signal. Is such a signal observed?
Line 260-265: As stated above, this discussion is critical with respect to the definition of the signal intensity. Please comment.
Line 278: I assume that the authors refer to g-strain, that leads to the line broadening. This effect can be easily simulated. With that, further details of the line parameters should be given and compared. What do the authors mean by “leading to reduced resolution of sidebands”? From Fig. 4c it becomes obvious that no field dependent process determines the line shape. How does this correlate with “sidebands”?
Line 312: The authors should fit the lines of all three measurements and then give quantitative numbers on the field inhomogeneity. This adds to the quality of the paper and makes it more quantitative.
Line 327: In the discussion the mechanism of the observed EDMR lines should be mentioned or cited, if well accepted.
Line 343: How would these parallel components of B1 change the saturation behavior of EDMR and why are they not considered?
Line 423: Please label the y-axis and the two resonance curves. Is the solid line a fit?
Fig. F1: The distance of Chip to sample is very critical. How do the authors take this experimental uncertainty into account? Please comment.
Fig. H1: What leads to the shift and narrowing of the line? Please discuss in the text.
Citation: https://doi.org/10.5194/mr-2024-11-RC2 - AC2: 'Reply on RC2', Sofie Cambré, 20 Sep 2024
- AC1: 'Reply on RC1', Sofie Cambré, 20 Sep 2024
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
302 | 80 | 27 | 409 | 14 | 13 |
- HTML: 302
- PDF: 80
- XML: 27
- Total: 409
- BibTeX: 14
- EndNote: 13
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1