Abstract
A heterodyne radiometer (LHR) based on the balanced-detection method has been developed to evaluate system performance under both balanced-detection and single-channel-detection modes. Experimental results demonstrate that the balanced-detection LHR exhibits superior suppression of laser-induced noise compared to the traditional single-channel-detection LHR. The developed system achieved a noise equivalent power (NEP) of 2.53 × 10−15 W/√Hz, which is only 1.7 times of the theoretical quantum limit. To further assess system performance, measurements of atmospheric carbon dioxide (CO2) column abundance were conducted. The measurement precision of the balanced-detection LHR was determined to be approximately 0.6%, which is 5.67 times higher than that of the single-channel-detection LHR.
1 Introduction
Laser heterodyne radiometry (LHR) is a powerful technique for high-precision retrieval of vertical profiles of atmospheric greenhouse gases [1–3]. For measurements at higher altitudes, where pressure broadening diminishes and Doppler-dominated spectral lines become narrower [4, 5], an LHR system with hundreds of MHz spectral resolution is highly desirable [6]. In LHR, the spectral resolution is twice of the passband width of the radiofrequency (RF) circuit, enabling straightforward implementation by incorporating an RF circuit with a narrow bandwidth on the order of tens of MHz [7, 8]. However, increasing the inevitably reduces the signal-to-noise ratio (SNR) of the detected heterodyne signal [9, 10], thereby degrading the precision of gas concentration measurements.
Currently, various approaches have been employed to enhance the SNR in heterodyne detection via amplifying the collected solar radiation or suppressing laser-induced noise. Recently, Rodin et al. reported a differential signal detection method to enhance LHR’s SNR by incorporating additional heterodyne detection channels, which has since been extended to measurements of atmospheric vertical wind profiles [11]. However, these approaches require extra detection channels, leading to increased system complexity and cost. More recently, we proposed a simple and effective method to enhance the power of collected solar radiation for high-spectral-resolution LHR by employing a semiconductor optical amplifier (SOA) [12], achieving a 9-times improvement in detection sensitivity. Nevertheless, the laser current-scanning mode commonly used in existing LHR systems introduces additional relative intensity noise (RIN) due to fluctuations in laser intensity [13, 14]. To address this issue, Li et al. applied erbium-doped fiber amplifier (EDFA) and SOA to stabilize laser output power [15], successfully reducing RIN and achieving a fivefold SNR enhancement. However, these amplification techniques may introduce excess noise from spontaneous emission in the SOA or EDFA [16], adding further complexity to the system design. Subsequently, we proposed a laser temperature-scanning wavelength-tuning method to enhance the LHR SNR, which is more compact than the above method [17]. Yet, this method entails a relatively long acquisition time of 7 minutes. In contrast, balanced-heterodyne detection provides a straightforward and effective means of improving the performance of high-resolution LHR, as independently reported by Rodin et al. [18] and Kurtz et al. [19]. However, the effectiveness of this method in suppressing laser-induced noise remains to be fully evaluated.
Therefore, in this paper, we report the development of balanced-detection LHR and evaluate its performance in suppressing laser-induced noise during high-resolution measurements of atmospheric column absorption.
2 Experimental details
A near infrared LHR based on a balanced photodetector was developed to investigate its performance on the heterodyne detection and noise suppressing, as illustrated in Figure 1. A distributed feedback (DFB) laser (NEL, NLK1L5GAAA) operated near 1.57 μm was used as the local oscillator to detect atmospheric CO2 column absorption. A ramp signal was produced by a 16-bit data acquisition card (NI, USB-6363, 2 MHz bandwidth) to control the laser’s injection currents. A solar tracker (EKO, STR-32G) was applied to track the position of the sun, where a fiber collimator (LBTEK, RFCAG-1.8-APC) was mount on it to collect the sunlight into a single-mode fiber. A variable optical attenuator (VOA) was utilized to precisely regulate the laser power within the shot noise dominated region. A fiber switch (Agiltron, NanoSpeed 1 × 1) was employed to modulate the sunlight intensity. The modulation mechanism of the fiber switch is amplitude modulation, with a modulation depth of 90%, a sawtooth waveform, and a modulation frequency of 832 Hz. Then a 2 × 2 fiber coupler with a power splitting ratio of 50%:50%, was used to couple the laser beam and the modulated sunlight beam into two output beams. Then the beams were superimposed on the active area of a balanced pair of photodetectors (THORLABS, PDB480C-AC), input (+) and input (−), with a response bandwidth of 1.6 GHz, respectively, to produce radio frequency heterodyne beat signals. The ac-coupled signal passed through a two-stage RF amplifier (Mini-Circuits, ZX60-4016E-S). In order to obtained high resolution solar absorption spectra, the effective bandwidth of the combined photodiode-amplifiers for the heterodyne beat signal was set to be 58 MHz via a bandwidth filter circuit (Mini-Circuits, ZABP-59-S+) with a passband of 30–88 MHz. The power level of the amplified RF heterodyne beat signal was measured by a circuit based on a Schottky diode (Herotek, DHM124AA). The raw heterodyne signal was recorded at 2 MSamp/s via the data acquisition card and then finally demodulated by a software-based lock-in amplifier (LIA) programmed by LabVIEW.
FIGURE 1
3 Results and discussion
We used a commercial frequency spectrograph to analyze the distribution of the LHR noise on the radiofrequency region ranging from 10 to 150 MHz, as shown in Figure 2. The black line represents the output noise from the balanced-detection operating mode. As a comparison, two noises output from the single photodetector are also depicted in this figure, which are represented by red and blue lines, respectively. From this figure, we can obviously see that the noise with an amplitude of 0.0046V from the balanced-detection operating mode is equally distributed in the RF region, while the noise from the single-channel-detection operating mode focuses in the passband of the inserted RF filter circuit, of which the maximal amplitude is 3 times higher than that of the balanced pair. It also proves the effectiveness of the used balanced-detection method on LHR noise suppression.
FIGURE 2
In LHR, the dominant noise source in the absence of laser radiation arises from the radiofrequency (RF) detection circuit. When laser radiation is applied, two additional noise contributions are introduced: laser shot noise and laser relative intensity noise (RIN). The background noise performance of the heterodyne system was experimentally evaluated under both balanced detection and single-channel detection configurations, as shown in Figure 3. The black curve represents the system noise floor measured with no laser illumination, yielding a standard deviation of 0.019 V. The blue curve corresponds to the system noise under laser illumination using single-channel detection, with a standard deviation of 0.372 V. The red curve shows the system noise under identical laser illumination but employing balanced detection, resulting in a standard deviation of 0.043 V. Comparison of the red and blue curves reveals that balanced detection achieves substantial suppression of laser RIN. Specifically, the ratio of the single-channel to balanced-detection noise levels quantifies the RIN component suppressed by the balanced scheme. This corresponds to a RIN suppression ratio of 18.74 dB relative to single-channel detection.
FIGURE 3
An Allan variance technique was used to analyze their noise level to retrieve the optimal data-averaging time of the LHR, respectively, as plotted in Figure 4. The black line represents the noise analysis of the balanced-detection operating mode, while the blue and red lines represent the noise analysis of single-channel-detection operating mode. The results show that the noise levels reach their minimums at the averaging times of 2314 s and 3026 s for the single one, and 1960 s for the balanced pair, respectively.
FIGURE 4
Moreover, in order to evaluate the developed LHR performance, another 2 MHz linewidth of DFB laser operating at a fixed frequency of approximately 6369.408 cm-1 was used to replace the sunlight as the input signal light, of which the output optical power was adjusted to be 0.2 nW by another VOA. 0.2 nW is the minimum incident power detectable by the heterodyne mixing system. Subsequently, the local oscillator laser was swept across this target wavelength. The heterodyne signal generated by the photomixing of the two lasers was detected by a photodetector and subsequently amplified by an RF amplifier. The output from the RF power detector was demodulated using the LIA in conjunction with the fiber switch. Figure 5 illustrates the demodulated heterodyne-detected signals from the two operating modes, respectively. Figures 5a–c show the heterodyne beat frequency signals containing background, obtained from single channel 1, single channel 2, and balanced detection, respectively. A nonlinear least-square fit method was applied to obtain the background baseline. The background-corrected heterodyne signals for single channel 1, single channel 2, and balanced detection were obtained, as depict in panels (d), (e), and (f).
FIGURE 5
The standard deviations of the baseline noises in Figures 5d–f were calculated to be 0.002, 0.002, and 0.001, respectively, while the corresponding peak values of the demodulated heterodyne-detected signals were 11.07, 10.7, and 36.49, respectively. The SNRs, defined as the signal peak amplitude divided by three times the standard deviation of the baseline noise, were calculated for the two operating mode, yielding values of 1845, 1783.3, and 12,163.3, respectively. Therefore, given that the lowpass filter bandwidth of 42.3 Hz from the software-based lock-in amplifier, the noise equivalent power densities of the system for SNR = 1 in the single-channel-detection operating mode were evaluated to be 1.67 × 10−14 W/√Hz and1.72 × 10−14W/√Hz, which are 11 times and 11.4 times of the theoretical quantum limit [20, 21]. The noise equivalent power densities of the system for SNR = 1 of the system in the balanced-detection operating mode was estimated to be 2.53 × 10−15 W/√Hz, which is 1.7 times of the theoretical quantum limit.
The experimental measured CO2 heterodyne absorption via balanced-detection method after 50 spectra-averaging times has been displayed in Figure 6. As a comparison, the two absorptions measured by single-channel-detection operating mode are also shown in this Figure, respectively. It can be observed that the heterodyne intensity acquired via balanced-detection operating mode is approximately 2.3 times higher than that from single-channel-detection operating mode and the corresponding SNR achieved an enhancement of 5 times.
FIGURE 6
Two consecutive clear and cloudless days were selected to measure atmospheric CO2 column absorption from 10:00 a.m. to 1:00 p.m. The acquired raw absorptions were normalized by dividing by the baseline, followed by application of an optimal estimation algorithm to fit the transmittance. The resulting measured absorptions, fitted curves, and fitting residuals are presented in Figure 7. The fitting residuals for the single-channel detection spectrum range within ±0.04, whereas those for the balanced detection spectrum are confined to ±0.02.
FIGURE 7
The retrieved CO2 column abundances over three consecutive hours from 10:00 a.m. to 1:00 p.m. are presented in Figure 8 [22]. The red and black dots represent the CO2 column abundances retrieved by using the balanced detection method and the single-channel detection method, respectively. The average values of the CO2 column abundance obtained via balanced detection and single channel detection are 8.79 × 1021 molecules/cm2 and 8.77 × 1021 molecules/cm2, respectively, with corresponding standard deviations of 5.3 × 1019 and 3.0 × 1020. The ratio of the standard deviation to the mean value was used as a metric for measurement precision, yielding values of 0.6% and 3.4% for balanced and single channel detection, respectively. The measurement precision of balanced detection is thus 5.67 times higher than that of single channel detection, demonstrating that the developed balanced detection laser heterodyne radiometer effectively suppresses laser-induced background noise and enhances the accuracy of atmospheric CO2 column abundance measurements in remote sensing applications.
FIGURE 8
4 Conclusion
The LHR based on balanced-pair photodetectors operating in the near-infrared region has been developed to evaluate the system performance of both the balanced-detection and single-channel-detection operating modes. Analysis of the noise levels in these two modes shows that the noise-equivalent power (NEP) of the balanced-detection LHR is 1.7 times the theoretical quantum limit, which represents a 6.6-fold improvement over the single-channel-detection LHR. Furthermore, measurements of atmospheric CO2 column abundances were conducted in Hefei, China, using both the balanced-detection LHR and the conventional single-channel-detection LHR. The measurement precisions for the retrieved CO2 column abundances were determined to be 0.6% and 3.4% for balanced detection and single-channel detection, respectively, indicating that the precision of balanced detection is 5.67 times higher than that of the conventional method. These results demonstrate that the balanced-detection LHR exhibits significantly superior performance compared to the traditional single-channel scheme in terms of both noise suppression and measurement accuracy. The dual-channel differential architecture of balanced detection effectively suppresses common-mode noise, thereby enhancing stability for long-term field deployment of laser heterodyne radiometers.
Future efforts will be directed toward advancing the balanced-detection LHR along four complementary avenues: (1) extending its measurement capability to enable simultaneous detection of key atmospheric trace gases—including CO2, CH4, and N2O—thereby enhancing its applicability for quantitative monitoring of atmospheric composition; (2) integrating adaptive optics to mitigate the effects of atmospheric turbulence in ground-based telescopic configurations, thus improving measurement stability and accuracy under realistic field conditions; (3) pursuing system miniaturization through photonic integration and compact radio-frequency design, facilitating deployment on airborne and spaceborne platforms to expand operational flexibility and spatial coverage for remote sensing; and (4) optimizing real-time signal processing by incorporating machine learning–enhanced algorithms, specifically targeting improved signal-to-noise ratio estimation and robust spectral feature extraction. Collectively, these research directions aim to fully exploit the inherent advantages of balanced detection—namely common-mode noise rejection and high dynamic range—thereby strengthening the technical foundation for next-generation atmospheric remote sensing.
Statements
Data availability statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
Author contributions
RL: Writing – original draft, Data curation, Writing – review and editing, Software, Investigation. HD: Funding acquisition, Validation, Supervision, Writing – review and editing. CY: Resources, Formal Analysis, Writing – review and editing. TS: Formal Analysis, Validation, Writing – review and editing. ZX: Methodology, Writing – review and editing. NC: Writing – review and editing, Visualization, Supervision. RK: Writing – review and editing, Validation, Project administration.
Funding
The author(s) declared that financial support was received for this work and/or its publication. This work was funded by the National Natural Science Foundation of China (6247034053) and the National Key R&D Program of China (2023YFC3705300).
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
References
1.
LiJTanTShenFWangGLiuKChenWet alRemote sensing of H2O/HDO in the atmospheric column based on a near-infrared laser heterodyne radiometer suppressing local oscillator relative intensity noise. Sensors Actuators B: Chem (2024) 419:136405. 10.1016/j.snb.2024.136405
2.
SchiederR. High resolution diode laser and heterodyne spectroscopy with applications toward remote sensing. Infrared Phys and Technology (1994) 35(2-3):477–86. 10.1016/1350-4495(94)90104-x
3.
DengHYangCXuZLiMHuangAYaoLet alDevelopment of a laser heterodyne spectroradiometer for high-resolution measurements of CO2, CH4, H2O and O2 in the atmospheric column. Opt Express (2021) 29(2):2003–13. 10.1364/OE.413035
4.
EshelBRiceCAPerramGP. Pressure broadening and shift rates for Ar (s–p) transitions observed in an Ar–He discharge. J Quantitative Spectrosc Radiative Transfer (2016) 179:40–50. 10.1016/j.jqsrt.2016.03.006
5.
SimmsPAMcCabeSPertGJ. Comparison of neon-like germanium inner shell and nickel-like gadolinium lasing at 62 Å and 66 Å by picosecond heating of a preformed plasma. Opt Communications (1998) 153(1-3):164–71. 10.1016/s0030-4018(98)00129-1
6.
HuangJMengYHuangYLuXWuPCaoZet alDesign and inversion study of a 1.6 µm high resolution non-modulated laser heterodyne radiometer (NM-LHR). Opt Express (2024) 32(21):37355–68. 10.1364/OE.535246
7.
WeidmannDWysockiG. High-resolution broadband (>100 cm-1) infrared heterodyne spectro-radiometry using an external cavity quantum cascade laser. Opt Express (2008) 17(1):248–59. 10.1364/OE.17.000248
8.
WeidmannDTsaiTMacleodNAWysockiG. Atmospheric observations of multiple molecular species using ultra-high-resolution external cavity quantum cascade laser heterodyne radiometry. Opt Lett (2011) 36(11):1951–3. 10.1364/ol.36.001951
9.
BlaneyTG. Signal-to-noise ratio and other characteristics of heterodyne radiation receivers. Space Sci Rev (1975) 17(5):691–702. 10.1007/bf00727583
10.
ShiodaTFujiiKKashiwagiKKurokawaT. High-resolution spectroscopy combined with the use of optical frequency comb and heterodyne detection. J Opt Soc America B (2010) 27(7):1487–91. 10.1364/josab.27.001487
11.
RodinAVChurbanovDVZenevichSGKlimchukAYSemenovVMSpiridonovMVet alVertical wind profiling from the troposphere to the lower mesosphere based on high-resolution heterodyne near-infrared spectroradiometry. Atmos Meas Tech (2020) 13(5):2299–308. 10.5194/amt-13-2299-2020
12.
DengHLiRLiuHHeYYangCLiXet alOptical amplification enables a huge sensitivity improvement to laser heterodyne radiometers for high-resolution measurements of atmospheric gases. Opt Lett (2022) 47(17):4335–8. 10.1364/OL.468198
13.
SmithCREngelsholmRDBangO. Pulse-to-pulse relative intensity noise measurements for ultrafast lasers. Opt Express (2022) 30(5):8136–50. 10.1364/oe.450819
14.
GuoYLuHXuMSuJPengK. Investigation about the influence of longitudinal-mode structure of the laser on the relative intensity noise properties. Opt Express (2018) 26(16):21108–18. 10.1364/oe.26.021108
15.
LiJXueZShenFWangJLiYWangGet alErbium-doped fiber amplifier (EDFA)-assisted laser heterodyne radiometer (LHR) working in the shot-noise-dominated regime. Opt Lett (2023) 48(20):5229–32. 10.1364/ol.501761
16.
QiGYaoJSeregelyiJPaquetSBélisleJ. Effects of amplified spontaneous emission noise of optical amplifiers on the phase noise of optically generated electrical signals[C]//Photonic Applications in Nonlinear Optics, Nanophotonics, and Microwave Photonics, 5971. SPIE (2005). p. 611–8. 10.1117/12.628224
17.
DengHWuLYangCLiRKanRBaoMet alTemperature-tuned laser heterodyne radiometer. Opt and Laser Technology (2025) 192:113817. 10.1016/j.optlastec.2025.113817
18.
RodinAKlimchukANadezhdinskiyAChurbanovDSpiridonovM. High resolution heterodyne spectroscopy of the atmospheric methane NIR absorption. Opt Express (2014) 22(11):13825–34. 10.1364/oe.22.013825
19.
KurtzJO’ByrneS. Multiple receivers in a high-resolution near-infrared heterodyne spectrometer. Opt Express (2016) 24(21):23838–48. 10.1364/oe.24.023838
20.
ParvitteBZéninariVThiébeauxCDelahaigueACourtoisD. Infrared laser heterodyne systems. Spectrochimica Acta A: Mol Biomol Spectrosc (2004) 60(5):1193–213. 10.1016/j.saa.2003.07.006
21.
SiegmanAE. The antenna properties of optical heterodyne receivers. Appl Optics (1966) 5(10):1588–94. 10.1109/proc.1966.5122
22.
WangYZhangMWangJWangGCuiRDongLet alField measurement of atmospheric CO2 column abundance based on portable laser heterodyne radiometer. Front Phys (2025) 13:1553252. 10.3389/fphy.2025.1553252
Summary
Keywords
balanced detection, CO2, column abundance, laser heterodyne radiometer, measurement precision, system performance
Citation
Li R, Deng H, Yang C, Sun T, Xu Z, Cao N and Kan R (2026) High-sensitivity laser heterodyne radiometer based on balanced detection technology. Front. Phys. 14:1814739. doi: 10.3389/fphy.2026.1814739
Received
21 February 2026
Revised
16 March 2026
Accepted
01 April 2026
Published
30 April 2026
Volume
14 - 2026
Edited by
Dustin McIntyre, National Energy Technology Laboratory (DOE), United States
Reviewed by
Yong Wang, Shanxi University, China
Jingjing Wang, University of Shanghai for Science and Technology, China
Updates
Copyright
© 2026 Li, Deng, Yang, Sun, Xu, Cao and Kan.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Hao Deng, hdeng@aiofm.ac.cn; Zhenyu Xu, zyxu@aiofm.ac.cn
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.