THz Metal-Mesh Bandpass Filters with Diamond-shaped Apertures for Sensitive THz Receivers

Boris S. Karasik; Changyun Yoo; Jeffrey L. Hesler

REVIEW 3 major objections 4 minor 14 references

THz Metal-Mesh Bandpass Filters with Diamond-shaped Apertures for Sensitive THz Receivers

T0 review · 3 major / 4 minor · reviewed 2026-08-10 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A single fitted equation sizes diamond-shaped copper meshes for >90% THz passbands from 1.5 to 5.5 THz, confirmed by three fabricated filters.

desk verdict Useful, honest design paper: the diamond aperture and fitted scaling law are new, and the three measured filters back them up; the main weakness is the unmeasured robustness claim. read the letter →

arxiv 2501.10843 v1 pith:DNFSQNIO submitted 2025-01-18 physics.ins-det physics.app-ph

Changyun Yoo , Jeffrey L. Hesler , Boris S. Karasik This is my paper · ORCID

classification physics.ins-detphysics.app-ph

keywords THzbandpassfiltermetal-meshdiamond-shapedaperturefrequencyselectivesurfacehot-electronbolometerdirectdetectioneffectelectroformingreceiver

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper claims that a freestanding 5-$\mu$m-thick copper mesh with diamond-shaped apertures can serve as a high-transmission THz bandpass filter, passing more than 90% of normally incident power at center frequencies from 1.5 to 5.5 THz with a 5-15% fractional bandwidth. The authors derive a simple design equation, $f_0[\mathrm{THz}] = 300/(aJ[\mu\mathrm{m}])$ with $a = -1.861(J/G)+2.881$, so the aperture diagonal $J$ and the period $G$ follow directly from the target frequency. They report that three fabricated filters, at 1.9, 2.5, and 4.7 THz, match finite-element simulations in center frequency, bandwidth, and peak transmission, and that the diamond shape tolerates the corner rounding typical of electroformed meshes better than cross-shaped apertures. Because such filters narrow the RF bandwidth reaching a hot-electron bolometer mixer, they reduce the direct-detection error in THz receiver calibration, for example from about 3 nW to about 0.4 nW of absorbed-power difference at 1.9 THz. If the claims hold, these filters fill a gap where few filters above 2 THz combine >90% transmission with <15% bandwidth.

What carries the argument

The central object is the diamond-shaped aperture, a periodic opening in a freestanding copper sheet defined by its diagonal $J$ and the mesh period $G$. The aperture acts as a resonant frequency-selective surface: transmission peaks when the aperture dimensions are comparable to the free-space wavelength, and the fitted equation $f_0[\mathrm{THz}] = 300/(aJ[\mu\mathrm{m}])$ with $a = -1.861(J/G)+2.881$ converts a chosen center frequency and $J/G$ ratio into physical dimensions. The absence of re-entrant corners, unlike cross-shaped apertures, is the load-bearing feature: the passband resonance is set by overall aperture size rather than by corner geometry, so the rounding typical of electroforming shifts the center frequency much less. Finite-element simulations of transmission versus rounding radius carry the robustness claim, while the measured spectra of the three filters carry the design-equation claim.

What would settle it

Electroform two diamond-aperture meshes with the same nominal dimensions but controlled corner-rounding radii differing by about 4-5 $\mu$m, measure both with an FTIR spectrometer, and compare the center-frequency shift with the finite-element prediction; a shift clearly larger than the predicted few percent would show the simulated rounding tolerance is optimistic.

Watch

Extended reading notes

Core claim

The central discovery is that changing the aperture shape from a cross to a diamond preserves the high transmission and narrow bandwidth of resonant metal-mesh filters while removing the sharp corners that make the center frequency vulnerable to fabrication rounding. The paper establishes, by finite-element simulation and by three measured filters, that a 5-$\mu$m copper mesh with diamond apertures obeys a linear scaling law $f_0[\mathrm{THz}] = 300/(aJ[\mu\mathrm{m}])$, with the fitted coefficient $a = -1.861(J/G)+2.881$; keeping $J/G$ between 0.6 and 0.7 keeps the -3-dB bandwidth near 5-15% and peak transmission above 90% across the 1.5-5.5 THz range. Measured filters at 1.91, 2.50, and 4.67 THz show center frequencies within roughly 0.01-0.03 THz of design and transmission within a few percent of simulation, and the 2.5 THz filter was independently checked at ~95% transmission using a molecular-gas laser line. The practical payoff is that THz heterodyne receivers can be fitted with a narrowband front-end filter that is inexpensive to electroform, stable at cryogenic temperatures, and able to reduce direct-detection calibration errors by limiting the RF bandwidth reaching the mixer.

Load-bearing premise

The load-bearing premise is that the simulated tolerance of the diamond aperture to fabrication rounding, tested only in finite-element modeling, holds in real electroformed meshes; if rounding shifts a real filter's center frequency by more than the few percent modeled, the advantage over cross-shaped apertures above 2 THz would shrink.

Editorial extensions

If this is right

At 1.9 THz, the filter narrows the effective RF bandwidth of a hot-electron bolometer receiver from about 2 THz to about 270 GHz, reducing the absorbed-power difference between 295 K and 77 K calibration loads from about 3 nW to about 0.4 nW and thereby shrinking direct-detection calibration errors.
With $J/G$ between 0.6 and 0.7, the design equation yields >90% peak transmission and 5-15% fractional bandwidth at any center frequency from 1.5 to 5.5 THz, so a target spectral line fixes $J$ and $G$ directly.
The three fabricated filters (1.91, 2.50, and 4.67 THz) match finite-element simulation closely, with center frequencies within roughly 0.01-0.03 THz of design and transmission within a few percent.
The freestanding copper filters are stable at cryogenic temperatures because thermal contraction is negligible relative to $J$ and $G$, so they can be mounted at the 4.2 K stage in front of the mixer.
The diamond aperture's tolerance to rounding errors removes the main obstacle that previously kept cross-aperture metal-mesh filters below about 2 THz, extending high-transmission narrowband filtering to 5.5 THz.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Beyond the paper, the same scaling idea should extend to other corner-free apertures, such as circles, which the authors note can also give >90% transmission with 5-15% bandwidth; fitting an analogous equation for circular holes is a testable next step.
A practical consequence of the reported ~2 µm electroforming overshoot is that batch-to-batch dimension control, not the design equation, sets the floor on center-frequency accuracy; a ±1 µm variation in $J$ corresponds to roughly a 1-3% frequency shift, which matters when a filter must sit on a narrow molecular line.
Because all transmission measurements were at normal incidence, the behavior under the oblique illumination found in real quasi-optical receivers remains open; if the passband shifts with angle, the receiver optics would have to be included when choosing $f_0$.
If the simulated rounding robustness holds, diamond apertures could extend electroformed metal-mesh filters to even higher frequencies, where shrinking feature sizes make cross-aperture corners a proportionally larger source of error.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Read this one. It is a practical THz filter paper with real measured results and a transparent design rule. The diamond aperture is a genuine variation on the cross-aperture mesh: for the first time someone shows systematic FEM scaling of J and G, fits f0 = 300/(a J) with a = -1.861(J/G)+2.881, and then demonstrates three freestanding electroformed filters at 1.91, 2.50, and 4.70 THz whose measured center frequencies, bandwidths, and peak transmission match HFSS. That is a complete design-and-verify loop. The paper also gives a useful comparison table against commercial filters, which argues convincingly that above 2 THz there was a real gap.

The design equation is explicitly a retrofitted curve, not a derivation, so don't expect physics from it. That is fine: the measured validation is what matters, and the agreement is good (0.01–0.03 THz in f0, 1–3% in Tp). The paper is honest about the electroforming overshoot and pre-compensates it. No red flags in the citation pattern; the Porterfield cross-aperture work and Melo review are the relevant benchmarks and they are cited.

The soft spots are real but not damning. First, the headline advantage—diamond apertures tolerate rounding errors better than crosses—rests entirely on HFSS corner-fillet simulations plus a supplementary comparison. The three fabricated filters had sharp apertures; the ~2 µm overshoot was systematic and compensated, so it never exercises the tolerance claim. If the simulation overstates the diamond's advantage, the motivation for diamond over cross weakens at >2 THz. That is a missing-experiment concern, not an internal inconsistency, and the three filters would remain valid. Second, the FTIR spectra have no error bars; the authors attribute <3% Tp error to dynamic range, which is plausible but not shown. Third, cryogenic compatibility is asserted from copper's thermal expansion rather than measured; minor.

The paper deserves a serious referee. It is a design contribution with supporting measurements, and the robustness question is a legitimate thing for a reviewer to ask for. I would engage with it.

Referee Report

3 major / 4 minor

Summary. The paper reports free-standing 5-µm-thick copper-mesh bandpass filters with diamond-shaped apertures for THz receivers. The authors use HFSS simulations to map center frequency, -3-dB bandwidth, and peak transmission as functions of aperture diagonal J and period G, and they fit the simulated design curves to the simple equation f0[THz] = 300/(aJ[µm]) with a = -1.861(J/G) + 2.881. Three filters with center frequencies of 1.91, 2.50, and 4.70 THz are fabricated by electroforming and measured with FTIR; the measured transmission profiles agree well with simulations. An independent measurement using a 2.522 THz molecular-gas FIR laser confirms the high peak transmission of the 2.5 THz filter. The paper motivates the diamond aperture by its robustness to fabrication rounding errors compared with cross apertures, and argues these filters can reduce direct-detection effects in HEB mixers by limiting the RF bandwidth.

Significance. If the claims hold, the filters fill a practical gap for >2 THz narrowband, high-transmission bandpass filters that are compatible with cryogenic HEB receivers. The work has clear strengths: three fabricated filters with measured performance, FTIR and independent FIR-laser cross-checks, a direct comparison table with literature and commercial filters, and a simple design equation that is quick to use. The main limitations are that the robustness advantage over cross apertures rests on simulation rather than on fabricated samples with controlled rounding, and that the design equation is an empirical retrofit validated only near J/G ≈ 0.7, so its claimed range of applicability is broader than what the measurements actually probe.

major comments (3)

[§II, Fig. 1(b)] The central motivation for the diamond aperture is its robustness to fabrication rounding errors, but this claim is supported only by HFSS corner-fillet simulations. No fabricated filter with intentionally varied rounding is measured in the main text, and the systematic ~2 µm electroforming overshoot is pre-compensated in the nominal design, so it does not exercise the tolerance claim. Because the advantage over cross apertures at f0 > 2 THz is the main reason for the new geometry, this load-bearing point needs either experimental verification (e.g., filters or test structures with deliberately varied corner radii, with measured f0 shifts) or an explicit statement that the robustness is a simulation-based expectation whose experimental confirmation is pending.
[§II, Eq. (1) and Table II] The design equation is a retrofitted fit to FEM design curves, and the three measured filters all have J/G ratios near 0.7 (0.697, 0.704, and 0.690). The stated usable range J/G = 0.6–0.7 and f0 = 1.5–5.5 THz is therefore not actually validated across that range; the equation may perform well, but its accuracy at, say, J/G = 0.6 is unverified. The authors should either add measured filters at another J/G ratio or restrict the claim of experimental validation to the tested region and show residuals for the full simulated range.
[§III, Figs. 3(b–d)] The measured and simulated spectra are said to agree excellently, but no error bars, repeated measurements, or quantitative residuals are shown. The only uncertainty statement is a <3% peak-transmission error attributed to the spectrometer dynamic range. Without an uncertainty estimate for f0 and Tp, the claimed design accuracy and the quantitative agreement with simulation cannot be fully assessed; the authors should report measurement uncertainties for each filter.

minor comments (4)

[§IV] The conclusion states the frequency range as '1.5 – 5.5 GHz'; this should be THz.
[Abstract] There is a typographical error: 'The se metal-mesh filters' should read 'These metal-mesh filters.'
[§II, Figs. 2(a)–2(c)] The statement that 'the errors observed in each simulation result are less than 1 %' would be clearer if the convergence criterion and mesh settings were specified, since the reader cannot otherwise assess what this error represents.
[References and Supplementary Material] The manuscript repeatedly refers to Supplementary Material sections (e.g., 'Section I of the Supplementary Material,' 'Section II of the Supplementary Material') without including them in the submitted text; please ensure the supplementary document is available to reviewers and readers.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the design equation is an openly acknowledged fit to FEM design curves and the measured filters provide independent validation.

full rationale

The paper's derivation chain is self-contained. The design equation f0[THz] = 300/(a J[um]) with a = -1.861(J/G) + 2.881 is explicitly presented as a retrofit to the FEM design curves ('acquired by retrofitting the design curves'), so it is an openly empirical parameterization of the simulations, not a hidden first-principles prediction. The measured 1.9, 2.5, and 4.7 THz filters were designed from these curves and then independently characterized by FTIR and by a 2.522 THz molecular-gas-laser check; their agreement with HFSS simulations therefore provides external validation rather than circular confirmation. The robustness-to-rounding claim in Fig. 1(b) is supported only by simulation and a supplementary comparison, which is a gap in experimental demonstration but not circularity, since the simulations are not fitted to the measured transmission data. The one self-citation (Porterfield et al. 1994, ref. [6], with coauthor Hesler) is used as background on prior cross-aperture filters and is not load-bearing for the paper's central design or validation. No fitted parameter is renamed as a prediction, and no equation reduces by construction to its input. The main limitations, such as the lack of measured filters with intentionally varied rounding errors, are concerns about experimental support rather than circular reasoning.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central design equation is an empirical fit to simulation, and the fabrication process includes a calibrated overshoot correction. No new physical entities are introduced. The main unstated inputs are the accuracy of the HFSS model and the assumption that three measurements span the design range.

free parameters (3)

Coefficient a in design equation = a = -1.861(J/G) + 2.881
The center frequency equation f0 = 300/(aJ) is obtained by retrofitting HFSS design curves (Section II, Eq. (1)). The linear dependence of a on J/G is an empirical fit, not a derived relation.
Electroforming overshoot compensation = 2 um
Fabricated patterns were consistently about 2 um larger than the stencil; the authors reduce the design dimensions by this amount to hit target J and G. This process calibration is folded into the measured geometry and therefore into the frequency agreement (Section III).
J/G ratio = 0.6-0.7
The paper selects this ratio range to keep Tp>90% and bandwidth 5-15%; it is a design choice drawn from simulation sweeps, not a derived optimum.

assumptions (4)

domain assumption HFSS FEM simulations accurately model power transmission of thin perforated copper meshes from 1.5 to 5.5 THz at normal incidence.
The design curves, design equation, and rounding robustness all come from HFSS (Section II, Fig. 2). Material parameters (conductivity, surface roughness) are not given in the main text and no independent mesh-convergence study beyond <1% error claims is shown.
domain assumption Three measured filters at 1.9, 2.5, and 4.7 THz are representative of the full 1.5-5.5 THz design range.
The design equation is claimed valid over 1.5-5.5 THz, but only three center frequencies are experimentally verified (Table 2). Interpolation between these points and extrapolation to the range edges rests on the assumption that simulation error stays small.
domain assumption Cryogenic operation does not change filter performance.
Section III states performance remains stable at cryogenic temperatures because copper thermal contraction is negligible, but no cryogenic measurement is reported.
domain assumption Direct detection power scales linearly with receiver RF bandwidth and the assumed 3 dB optical coupling loss is typical.
The estimate that absorbed power falls from ~3 nW to ~0.4 nW uses a bandwidth reduction factor of 7.4 and a 3 dB coupling loss; this is an estimate, not a measurement (Section III).

how reviews work

0 comments

Cite this review

Pith. "Pith review of THz Metal-Mesh Bandpass Filters with Diamond-shaped Apertures for Sensitive THz Receivers." pith.science (2026). https://pith.science/paper/DNFSQNIO

@misc{pith2026250110843,
  author       = {Pith},
  title        = {Pith review of: THz Metal-Mesh Bandpass Filters with Diamond-shaped Apertures for Sensitive THz Receivers},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DNFSQNIO}},
  note         = {Machine review of arXiv:2501.10843}
}

read the original abstract

We present high-performance terahertz (THz) metal-mesh bandpass filters developed to mitigate direct detection effects in sensitive THz receivers based on superconducting hot-electron bolometer (HEB) mixers. These metal-mesh filters are free-standing 5-micron thick sheets of copper perforated with a periodic array of diamond-shaped apertures. The simple aperture design minimizes the effect of fabrication (rounding) errors on filter performance, allowing precise engineering of the center frequency (1.5-5.5 THz) while maintaining high (>90 %) peak power transmission at normal incidence and a relatively narrow bandwidth of 5-15 %. Based on finite-element method (FEM) simulation results, we provide simple design equations that can be used for rapid design with high accuracy. The measured transmission profiles of 1.9-THz, 2.5-THz, and 4.7-THz filters show excellent agreement with the simulation results. These filters are compatible with cryogenic operation and can substantially reduce the direct detection effects in HEB mixers.

Figures

Figures reproduced from arXiv: 2501.10843 by the authors.

**Figure 2.** Fig.2 [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

14 extracted references · 14 canonical work pages

[1]

Direct detection effect in small volume hot electron bolometer mixers,

J. J. A. Baselmans, A. Baryshev, S. F. Reker et al. , "Direct detection effect in small volume hot electron bolometer mixers," Applied Physics Letters, vol. 86, 2005

work page 2005
[2]

Influence of the direct response on the heterodyne sensitivity of hot electron bolometer mixers,

J. J. A. Baselmans, A. Baryshev, S. F. Reker et al. , "Influence of the direct response on the heterodyne sensitivity of hot electron bolometer mixers," Journal of Applied Physics, vol. 100, 2006

work page 2006
[3]

The Direct Detection Effect in the Hot-Electron Bolometer Mixer Sensitivity Calibration,

S. Cherednichenko, V. Drakinskiy, T. Berg et al., "The Direct Detection Effect in the Hot-Electron Bolometer Mixer Sensitivity Calibration," IEEE Transactions on Microwave Theory and Techniques, vol. 55, pp. 504 - 510, 2007

work page 2007
[4]

Analytical correction for direct detection in the retrieval of gas spectral lines measured with superconducting hot -electron bolometer mixer,

A. D. Semenov, M. Wienold, M. Sidorova et al. , "Analytical correction for direct detection in the retrieval of gas spectral lines measured with superconducting hot -electron bolometer mixer," Journal of Applied Physics, vol. 135, 2024

work page 2024
[5]

Stabilized hot electron bolometer heterodyne receiver at 2.5 THz,

D. J. Hayton, J. R. Gao, J. W. Kooi et al., "Stabilized hot electron bolometer heterodyne receiver at 2.5 THz," Applied Physics Letters, vol. 100, 2012

work page 2012
[6]

Resonant metal -mesh bandpass filters for the far infrared,

D. W. Porterfield, J. L. Hesler, R. Densing et al. , "Resonant metal -mesh bandpass filters for the far infrared," Applied Optics, vol. 33, pp. 6046 -6052, 1994

work page 1994
[7]

Development of metal mesh based quasi -optical selective components and their application in high - power experiments at novosibirsk terahertz FEL,

S. A. Kuznetsov, V. V. Kubarev, R. V. Kalinin et al., "Development of metal mesh based quasi -optical selective components and their application in high - power experiments at novosibirsk terahertz FEL," 29th International Free Electron Laser Conference, FEL 2007, pp. 89-92, 2007

work page 2007
[8]

Direct fabrication of terahertz optical devices on low - absorption polymer substrates,

Y. Ma, A. Khalid, T. D. Drysdale et al. , "Direct fabrication of terahertz optical devices on low - absorption polymer substrates," Optics Letters, vol. 34, pp. 1555-1557, 2009

work page 2009

Show all 14 references

[9]

Verification of Metal -Mesh Filter Response via ANSYS HFSS Simulation,

H. R. J. Stevenson, J. E. Payne, J. Eddy et al. , "Verification of Metal -Mesh Filter Response via ANSYS HFSS Simulation," in 2023 Intermountain Engineering, Technology and Computing (IETC) , 2023, pp. 288-292

2023
[10]

B. A. Munk, Frequency selective surfaces: theory and design. John Wiley & Sons, 2005

2005
[11]

Cross-Shaped Terahertz Metal Mesh Filters: Historical Review and Results,

A. M. Melo, A. L. Gobbi, M. H. O. Piazzetta et al., "Cross-Shaped Terahertz Metal Mesh Filters: Historical Review and Results," Advances in Optical Technologies, vol. 2012, p. 530512, 2012

2012
[12]

THz Band Pass Filters

TYDEX. "THz Band Pass Filters." TYDEX. https://www.tydexoptics.com/products/thz_assemblie s/thz_band_pass_filter/ (accessed January 2nd, 2025)

2025
[13]

Multimesh filters

QMC. "Multimesh filters." QMC Instruments. https://www.qmcinstruments.co.uk/multi-mesh-filters (accessed January 2nd, 2025). 5 > REPLACE THIS LINE WITH YOUR MANUSCRIPT ID NUMBER (DOUBLE-CLICK HERE TO EDIT) <

2025
[14]

THz Bandpass Filters: 10 micron - 590 micron center wavelength

Thorlabs. "THz Bandpass Filters: 10 micron - 590 micron center wavelength." Thorlabs. https://www.thorlabs.com/catalogpages/obsolete/201 8/FB19M10.pdf (accessed January 2nd, 2025)

2025

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.

[1] [1]

Direct detection effect in small volume hot electron bolometer mixers,

J. J. A. Baselmans, A. Baryshev, S. F. Reker et al. , "Direct detection effect in small volume hot electron bolometer mixers," Applied Physics Letters, vol. 86, 2005

work page 2005

[2] [2]

Influence of the direct response on the heterodyne sensitivity of hot electron bolometer mixers,

J. J. A. Baselmans, A. Baryshev, S. F. Reker et al. , "Influence of the direct response on the heterodyne sensitivity of hot electron bolometer mixers," Journal of Applied Physics, vol. 100, 2006

work page 2006

[3] [3]

The Direct Detection Effect in the Hot-Electron Bolometer Mixer Sensitivity Calibration,

S. Cherednichenko, V. Drakinskiy, T. Berg et al., "The Direct Detection Effect in the Hot-Electron Bolometer Mixer Sensitivity Calibration," IEEE Transactions on Microwave Theory and Techniques, vol. 55, pp. 504 - 510, 2007

work page 2007

[4] [4]

Analytical correction for direct detection in the retrieval of gas spectral lines measured with superconducting hot -electron bolometer mixer,

A. D. Semenov, M. Wienold, M. Sidorova et al. , "Analytical correction for direct detection in the retrieval of gas spectral lines measured with superconducting hot -electron bolometer mixer," Journal of Applied Physics, vol. 135, 2024

work page 2024

[5] [5]

Stabilized hot electron bolometer heterodyne receiver at 2.5 THz,

D. J. Hayton, J. R. Gao, J. W. Kooi et al., "Stabilized hot electron bolometer heterodyne receiver at 2.5 THz," Applied Physics Letters, vol. 100, 2012

work page 2012

[6] [6]

Resonant metal -mesh bandpass filters for the far infrared,

D. W. Porterfield, J. L. Hesler, R. Densing et al. , "Resonant metal -mesh bandpass filters for the far infrared," Applied Optics, vol. 33, pp. 6046 -6052, 1994

work page 1994

[7] [7]

Development of metal mesh based quasi -optical selective components and their application in high - power experiments at novosibirsk terahertz FEL,

S. A. Kuznetsov, V. V. Kubarev, R. V. Kalinin et al., "Development of metal mesh based quasi -optical selective components and their application in high - power experiments at novosibirsk terahertz FEL," 29th International Free Electron Laser Conference, FEL 2007, pp. 89-92, 2007

work page 2007

[8] [8]

Direct fabrication of terahertz optical devices on low - absorption polymer substrates,

Y. Ma, A. Khalid, T. D. Drysdale et al. , "Direct fabrication of terahertz optical devices on low - absorption polymer substrates," Optics Letters, vol. 34, pp. 1555-1557, 2009

work page 2009