Open Access
Add to BibSonomy
Abstract
Absorption lines of atmospheric vapor commonly appear in terahertz (THz) spectra measured in a humid air environment. However, these effects are generally undesirable because they may mask critical spectroscopic information. Here, a self-adaptive method is demonstrated for effectively identifying and eliminating atmospheric vapor noise from THz spectra of an all-fiber THz system with the Hilbert-Huang transform. The THz signal was decomposed into eight components in different time scales called the intrinsic mode functions and the interference of atmospheric vapor was accurately isolated. A series of experiments confirmed the effectiveness and strong self-adaptiveness of the proposed system in vapor noise elimination.
© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Terahertz (THz) lies in the frequency gap between the millimeter wave and the infrared, has attracted extensive attention of considerable researchers owing to its great technological potential in different fields, such as material evaluation [1,2], cultural heritage [3], IC authentication [4,5], transmission imaging with enhanced resolution [6–10]. One of the most appealing features is to identify material signatures through their spectroscopic fingerprints that is unavailable in conventional spectroscopy, since the vibration and rotation frequency of a large variety of molecules are within this band, such as polar gases [11] and biological molecules [12]. In order to obtain accurately THz spectroscopic information of the substance, THz time-domain spectroscopy technology (THz-TDS) [13] based on femtosecond lasers [14] has been well demonstrated and extensively employed as an excellent technique in characterizing many organic compounds [15,16]. Recently, high power and high resolution THz imaging and spectroscopy systems have been obtained based on semiconductor materials with special performance, such as gallium nitride [17,18], which further promotes the development of THz technology. THz-TDS allows us to measure the time-resolved and high-precision THz electric field directly and then obtain the spectral information by applying a Fourier transform on the time-domain waveform. However, in spite of these advantages, THz-TDS has not been widely applied in real-world due to the fact that the THz signals are strongly absorbed by the atmospheric vapor, and Grischkowsky et al. have reported a number of researches about water vapor absorption in the THz range, especially from 0.2 to 2 THz [19,20]. This is because many different pure rotational transitions of atmospheric water molecules would be excited coherently during the propagation of THz pulses and reradiate a free-induction decay signal [21], which causes additional fluctuations after the main pulse and sharp absorption lines in corresponding spectra. Such undesirable effects will result in the attenuation of the THz radiation, increase strongly interference components for the process of identification and characterization of materials, and then obscure the true spectral data. Therefore, it is imperative to identify and eliminate the water vapor noise from THz spectroscopy so that the more accurate feature information of materials can be obtained.
In general, the effect of water vapor noise on THz radiation is removed primarily by purging the propagation path with dry air or such a non-polar gas as nitrogen, which does not have transition energy levels in the THz region [22], and alternatively a vacuum is sometimes used. Undoubtedly, the treatment increases the complexity and cost of the system and is not suitable for all conditions. For example, it is only feasible under laboratory conditions rather than in the field. Also it is not suitable for the applications where the stand-off detection is required [23]. Therefore, it is expected that the signal postprocessing method can address the issue, however, it is difficult to distinguish whether a ripple is caused by water vapor noise or materials in THz time-domain signals, and the spectral information may also overlap in the frequency domain [24]. In the deconvolution method [25], the ripple of the time-domain signal caused by the external noise was fitted, and then the noise effect was removed by deconvolution. However, this method is constrained by many factors including the dynamic range [26], frequency resolution [27], and measurement uncertainties [28]. Kong and Wu recently proposed an artificial neural network model, while the trained network relies on specific atmospheric conditions [29]. In [30], Wang et al. modelled the absorption lines by using the Lorentz line and the spectral parameters extracted from the HITRAN database [31] and then iteratively subtracting them in the frequency domain. Recently, Huang et al. used the numerical method based on the transfer function to realize the elimination of water vapor noise [32], which was also only limited to be effective in a low-humidity environment. These methods can only exclude water vapor noise for a certain extent under the specific experimental conditions. It is extremely important to develop a self-adaptive THz de-noise method without the limited external conditions.
In this paper, a strongly self-adaptive THz signal postprocessing method is introduced based on the Hilbert-Huang transform (HHT) that could effectively identify and remove water vapor noise from THz spectra and successfully retain valuable sample information. The HHT decomposes the THz time-domain pulse into eight components at different time scales called the intrinsic mode functions (IMFs). The detailed time-frequency analysis demonstrates that the first two IMFs carry the most sample information while the other IMFs correspond to the external noise, and the interference of atmospheric vapor noise is accurately identified in the third IMF. To test the strong self-adaptiveness of the HHT, the algorithm is performed to processing on the environments with the different relative humidity (RH 13.2%, RH 38.4%, 51.2% and 70.4%) and time windows (30 ps, 50 ps, 90 ps and 140 ps). Furthermore, the samples of the polymer materials polytetrafluoroethylene (PTFE) with no significant absorption and the graphene with THz responses were also measured by our system. The experimental results verified the strong self-adaptiveness of the HHT based THz-TDS and confirmed its applications in various environments. In the following section, an all-fiber THz-TDS is demonstrated and the interaction of atmospheric vapor and THz waves is analyzed through the time-domain signals and their spectra. Sec. III shows the solution and implementation process based on HHT to eliminate water vapor noise. In Sec. IV, the HHT is introduced and the de-noise results are initially presented. Through the detail time-frequency analysis, the effectiveness and strong self-adaptiveness of our proposed system for removing water vapor noise are experimentally demonstrated in Sec. IV. Finally, the conclusions on our major results are drawn in Sec. V.
2. Water vapor responses
The all-fiber THz-TDS has been built based on the basic structure of the free space THz system in order to reduce the system space volume and enhance system stability, which is similar to that in [33]. The schematic diagram is shown in Fig. 1. The fiber-coupled photoconductive antennas (FC-PCA) from BATOP Company, consisting of an LT-InGaAs absorber layer and a bow-tie antenna, were utilized as the transmitter and receiver in our system. The pump source was a 1550 nm femtosecond fiber laser with 50 fs pulse width and 50 MHz repetition rate from MENDOCINO Systems. The TDS set-up with collimating TPX (Polymethylpentene) lens (CTL-D25mm) emits parallel THz beams. The method of averaging multiple measurements was adopted to get the high accuracy and signal-to-noise ratio. The obtained time domain pulse was transformed into the frequency spectrum by the fast Fourier transform (FFT). The obtained peak frequency and band width of the THz spectrum were 0.3 THz and 0.9 THz, respectively. The frequency resolution of all-fiber THz-TDS is up to 7 GHz corresponding to the maximum time window of 140 ps. The THz radiation path was sealed by the enclosure and the humidity was measured and monitored in real-time with a humidity sensor (GM1360A) to ensure the stability of the measurement conditions. The pure nitrogen and humid air were filled through two holes left on the enclosure while the humidity level could be changed by adjusting the filling time.
To confirm the influence of water vapor on THz radiation, the FFT spectral information under the different humidity conditions was measured by our all-fiber THz-TDS shown in the Fig. 2. It is clear that two strong absorption peaks (0.56 THz and 0.75 THz) appear in the spectrum of atmospheric vapors. Furthermore, the positions of the absorption peaks do not change with the increasing ambient humidity from RH 4.3% to RH 70.4%. The results are completely consistent with theses in [19] with the frequency resolution of 20 GHz. Therefore, we firmly believe that these two characteristic peaks are caused by the water vapor. It can be clearly observed that the THz time-domain waveform contains a main pulse and the tail oscillations in the lower left inset of Fig. 2. The time window is selected as 50 ps and the mainly pulse waveform from 5 ps to 12 ps are enlarged in order to easily observe the tiny distinctions caused by the humidity. The THz pulse amplitude attenuates gradually and the waveform is delayed according to the increasing environment humidity. The increasing air humidity causes the change of refractive index of THz propagation path, which influences the optical path, forms the phase differences, and leads to the time delay in time domain. The attenuation of the amplitude is caused by the different intensity of reflection, scattering and absorption of water vapor, resulting in the energy loss. Since the length of the THz radiation path is only 7 cm, causing a weak interaction between the THz wave and water vapor [34], the intensity is gradually strengthened and the absorption bandwidth is also increased, which further confirm the strong absorption of water vapor on THz wave. The experimental results also verified the different rotational transmissions of atmospheric water vapor molecules in the THz band. Therefore, the above experimental tests illustrated that the atmospheric water vapor noise has caused great interference to the THz-TDS system.
Fig. 2 THz time-domain pulses (lower left inset) and FFT spectra of relative humidity 4.3% – 70.4%. The absorption peaks of water vapor are clearly identified and the absorptions get more intense with the increasing relative humidity.
3. HHT
Atmospheric vapor noise causes great interference to THz waves, which greatly limits the development of THz technology. So far, there is still no effective THz de-noise method which is suitable for all situations, such as atmospheric humidity, THz systems with different resolution and characterization of different types of samples. Therefore, we introduce a new THz signal post-processing method based on the HHT, which was developed by Huang in 1998 and used to study physical insight of ocean waves [35]. Different from the past methods, such as the traditional Fourier transform, the wavelet transform, the HHT is self-adaptive to improve the signal processing. Therefore, the HHT has a very wide range of applications and development potentials in different fields of physics, such as optically-sectioned images [36], digital holographic microscopy (DHM) for phase imaging [37], searching for gravitational waves [38], detection of defects and displacements generated during material processing [39], and fringe pattern phase demodulation [40]. Since the HHT is especially suitable for analyzing the nonlinear and non-stationary signals, the time and frequency resolution are very promising to be simultaneously increased when the HHT is introduced into the THz-TDS.
The basic idea of the HHT is the empirical mode decomposition (EMD) method [41], which decomposes the THz signals into a series of frequency components in different time scales called intrinsic mode functions (IMFs), and the number of decomposed IMFs mainly depends on the signal length. By comparing the IMF component with the original signal including the THz time domain and the FFT spectrum, the multi-scale oscillatory features in the THz signal can be obtained, such as sample information and background noise, so the frequency component of the noise can be easily eliminated and the sample features are retained. Therefore, the HHT algorithm is very suitable for narrow-band filtering of THz pulse signals with picosecond magnitude.
The specific decomposition process of the HHT algorithm is summarized as follows. First, the THz signal is decomposed into a summation of single component signals of finite number, that is,
where is the ith IMF single-component signal, that is to say, the local only contains one frequency component or a narrowband signal centered on a certain frequency, and is a residual component, which is monotonic or constant. Each IMF component is obtained by sieving, the finest mode contained in the THz signal is gradually separated, that iswhere, is the mean function of the upper and lower envelopes of the kth IMF, is the result of the kth decomposition, and is the residual signal after the (i–1)th decomposition, which is used as the original signal of the further decomposition. The entire decomposition process ends when the residual is a monotonic function or constant.4. Eliminating atmospheric vapor noises
The polytetrafluoroethylene (PTFE) polymer of 960 μm thick is chosen as the test sample. The THz time domain signal of the PTFE was obtained by using our all-fiber THz-TDS incorporating the HHT algorithm. The experimental condition is in the control of the ambient humidity RH 38.4% and room temperature. The decomposed results based on HHT are shown in Fig. 3, in which the original signal was decomposed into the eight IMFs and the residual. It is clear that the pulse amplitudes of the lower-order modes (IMF1 and IMF2) are close to the original signal, the other high-order modes (IMF3-IMF8) decrease in sequence, and the indecomposable residual is relatively small and almost a DC type function. By comparing the position and amplitude of each IMF component with the original signal waveform, in an error type of analysis, it can be preliminary determined that the first two IMFs best fit the original signal, which may carries the most sample information and the smaller amplitudes of the other IMF components (IMF3-IMF8) are caused by external noise. The experimental results in Sec. II have confirmed the strong interference of water vapor noise on the THz signal. Therefore, in the remaining IMF components, the largest amplitude of IMF3 component may correspond to water vapor noise while IMF4-IMF8 are mainly caused by the thermal and background noise.
Fig. 3 THz original time-domain signal of PTFE (black line) and its decomposed eight IMF components (red line) and residuals (green line).
In order to prove that HHT can effectively eliminate the interference component of water vapor noise and no loss of the intrinsic information of sample, the time domain signal of the decomposed IMF components and their corresponding FFT spectra are analyzed in detail, which are shown in Figs. 4(a) and 4(b), respectively. To clearly observe the differences of the time domain signals, the recorded experimental data of the main pulse (6 ps – 12 ps) and tail oscillating waveforms (15 ps – 50 ps) are depicted in the inserts of Fig. 4(a). It can be found that the measured THz main pulse amplitude through the atmospheric vapor (RH 38.4%) is lower than that through the nitrogen environment (RH 4.3%) while the waveform is also delayed. The excited rotational transitions of atmospheric water vapor molecules lead to the attenuation of THz energy and change the refractive index and optical path of THz radiation. Since main pulse of the IMF1 + IMF2 changes a little while the amplitude of the tail oscillation (the blue line) is greatly reduced by approximately two orders of magnitude in the IMF1 + IMF2 (the red line), it is inferred that the IMF1 + IMF2 contains most of the sample information without interference component from external noise. With the addition of the IMF3 (the green line), the amplitude of the main pulse does not increase much while the amplitude of the tail oscillation increases sharply, indicating that the IMF3 component contains a large amount of water vapor noise.
Fig. 4 THz time-domain waveforms (a) and their corresponding FFT spectra (b) of the PTFE. The original signals of the PTFE in the nitrogen and the air with RH 38.4% are respectively presented by the black dotted and blue dash-dot lines, so do their spectra. The Fourier spectra of IMF1 + IMF2 (the red solid line) and IMF1 + IMF2 + IMF3 (the green dashed line) demonstrate the elimination and identification of the water vapor, respectively.
To further verify the above predicted results, the FFT have been performed for the above time domain signals and then obtained the corresponding spectral information as shown in Fig. 4(b). The PTFE spectrum measured under nitrogen environment (RH 4.3%) is denoted by the black dotted line and there is no obvious characteristic absorption peak in the THz band, indicating the transparency of PTFE polymer materials to THz waves [42]. However, two significant absorption peaks appear in the spectrum measured in the humid air with RH 38.4%. The position (0.56 THz and 0.75 THz) of the peak frequency coincide with the absorption lines of water vapor. It is no doubt that these two false characteristic peaks were caused by the strong interference of atmospheric water vapor noise, which mistake the actual material characteristics. By observing the original signal (the blue dash-dot line) and IMF1 + IMF2 (the red solid line) in Fig. 4(b), it is confirmed that the interference of external noise is easily eliminated in the IMF1 + IMF2. However, when the IMF3 component is added (the green dashed line), the two characteristic peaks at 0.56 THz and 0.75 THz appear again, indicating the water vapor noise carried by the IMF3. Thus, through above detailed time-frequency analysis, based on the first two IMFs, the interference of external noise can be effectively exclude and the sample intrinsic information is retained. Furthermore, it is also found that the atmospheric vapor noise exists in the IMF3. The proposed HHT based THz-TDS does efficiently eliminate the vapor noise and the obtained spectrum confirms the transparency of PTFE materials.
The HHT-based THz-TDS not only restores real sample information, but also accurately extracts optical parameters, such as refractive index. Figures 5(a) and 5(b) shows the measured time domain waveforms and the FFT spectrum of the PTFE sample and the air reference, respectively. Obviously, the sample signal produces a time delay of 1.218 ps, which is due to the acquisition of the equal optical path of the THz system. The phase difference is calculated based on the FFT spectrum and then the refractive index [43]
where is the sample thickness, is the speed of light in vacuum, and is the THz circular frequency. By observing the PTFE refractive index curve (blue dashed line) measured in the atmospheric vapor in Fig. 5(c), distinct artifacts associated with water vapor absorption appear on the curves in the frequency range higher than 0.5 THz. The unphysical oscillations appear in the refractive index curve. However, it is very impressive that the HHT algorithm effectively eliminates water vapor interference and the quality and resolution of the refractive index curve (red solid line) was improved dramatically. The oscillations disappear and the refractive index fluctuates a little with frequency. The obtained refractive index is even better than that measured in the nitrogen environment (green dotted line), where the small oscillations can be attributed to the leakage of our enclosure as the measured humidity is still 4.3%. It is evident that the optical parameters extracted from improved HHT signals are more accurate.Fig. 5 THz time domain signal with RH 38.4% (a) and the corresponding FFT spectrum (b), where the black dotted line and the red solid line represent the air reference and the PTFE sample, respectively. Refractive index curve of PTFE is obtained in (c), where the black dotted, blue dashed, and red solid lines depict the frequency dependent refractive index for nitrogen, air humidity (RH 38.4%), and HHT algorithm, respectively.
5. Self-adaptiveness
Although it have been shown that the proposed method can efficiently improve the all fiber THz-TDS system in shielding atmosphere vapor interference, its applicability to different circumstances and systems is still a question. To demonstrate the self-adaptiveness of our HHT based THz-TDS, a large number of experiments in different humidity and scanning lengths are performed. The PTFE THz time domain signals are firstly measured in different atmospheric vapor humidity (RH 13.2%, RH 51.2%, and RH 70.4%) as shown in Fig. 6(a). In order to clearly observe the differences between the signals, a time window from 6 ps to 14 ps is presented, and the overall signals (50 ps) are presented in the up right corner. As the atmospheric vapor humidity increases, it can be seen that the amplitude of the THz pulse gradually attenuates and time delay increases significantly. The corresponding FFT spectra are presented in Fig. 6(b), and the spectrum obtained by nitrogen filling method is also depicted by the black dotted line as reference. The experimental results show that the different atmospheric vapor humidity has caused different degrees of interference to the spectrum at two frequencies of 0.56 THz and 0.75 THz, and especially in high humidity environments, the absorption peaks greatly mask the intrinsic transparency information of PTFE materials. According to the above described process, the HHT de-noise results are presented in Fig. 7. The IMF1 + IMF2 can still effectively remove the noise interferences and retain sample information in the all humidity. As the external environment humidity increases, the two interference peaks in the IMF1 + IMF2 + IMF3 are also gradually strengthened, and the positions (0.56 THz and 0.75 THz) of the absorption peak coincide with the water vapor absorption lines. Thus, the HHT based THz-TDS can effectively identify and characterize samples under different humidity conditions and eliminate the strong interference from atmospheric vapor noise. It is well self-adaptive to the humidity and can be used in humid environments without more techniques.
Fig. 6 THz time-domain waveforms (a) and their corresponding Fourier spectra (b) of PTFE sample in different ambient humidity of RH 4.3%, RH 13.2%, RH 51.2%, and RH 70.4%.
Fig. 7 Fourier spectra of PTFE sample in different humidity: (a) RH 13.2%, (b) RH 51.2%, and (c) RH 70.4%. The original signal, IMF1 + IMF2, and IMF1 + IMF2 + IMF3 are denoted by the blue dash-dot, red solid, and green dashed lines, respectively. The time domain signals are depicted in the insects and the PTFE spectra in the nitrogen are also shown as reference.
Secondly, the HHT program is integrated into the THz-TDS with different frequency resolutions to investigate its self-adaptiveness to systems. The PTFE THz time domain signals with different time window (30 ps, 90 ps, and 140 ps) are shown in Fig. 8(a), and it can be clearly observed that as the signal length increases, the more tail oscillation waveform is included, which leads to the stronger interference from external noises. The corresponding Fourier spectra are depicted in Fig. 8(b) and the frequency resolutions for the lengths of 30 ps, 90 ps, and 140 ps are 33 GHz, 12 GHz and 7 GHz, respectively. The spectral data measured in humid air with RH 38.4% has been disturbed by the water vapor noise and the intensity of the water vapor absorption peaks do not change with the increasing scanning length, but the frequency oscillations in the whole spectrum has increased significantly, indicating the introduction of stronger background noises. The HHT-based de-noise results for the signals of 30 ps, 90 ps, and 140 ps are also given in Figs. 9(a)–9(c), respectively. Due to the different signal lengths containing frequency oscillation modes in different time scales, the number of decomposed IMF components is quite different. The 30 ps signal is decomposed into seven IMFs while the 90 ps and 140 ps correspond to eight and nine IMFs, respectively. Whatever, the IMF1 + IMF2 can still effectively remove the interference of atmospheric vapor noise, which has been identified in the IMF3.
Fig. 8 THz time-domain waveforms (a) and their corresponding Fourier spectra (b) of PTFE sample with different scanning lengths (30 ps, 90 ps, and 140 ps).
Fig. 9 The PTFE THz time-domain waveforms (insert) and corresponding FFT spectra analysis of different time windows (30ps (a), 90ps (b) and 140ps (c)), including pure nitrogen (black dotted line), original signal (blue dot-dashed line), IMF1 + IMF2 (red solid line), IMF1 + IMF2 + IMF3 (green dashed line).
To further evaluate the performance of the HHT algorithm, we have introduced the mean squared error (MSE) as
where is a signal measured in the nitrogen atmosphere, is the signal through the air with or without the HHT, and is the number of data points. The evaluation results are shown in Tables 1 and 2 for different humidity environments and delay times, respectively. The MSEs of PTFE signals after the HHT are all greatly reduced, which proves that the proposed HHT based THz-TDS is strongly self-adaptive and can be easily applied to characterize materials in various humidity environments.
Table 1. MSEs of the Signals through the Humid Air with and without the HHT.
Table 2. MSEs of the signals obtained by different delay times.
6. Material characterization applications
We have performed the de-noise treatment on transparent polymer of PTFE in the THz band and a series of experimental results have proved the effectiveness and strong self-adaptiveness of our HHT based THz-TDS. However, in practices, we often characterize the material absorptions in THz frequency range and the intrinsic absorption peaks maybe very close to the vapor’s, such as the well-known graphene. To verify the applicability of the proposed system, we prepared the monolayer graphene (MG) sample and characterized in the high humidity environment (RH 63.7%). The MG sample was generated based on the CVD method [44] and then transferred onto the PTFE substrate by the polymethyl methacrylate (PMMA) assisted wet transfer technology. To guarantee the single-layer and high-quality of our MG sample, it is first characterized by the Raman spectroscopy with the position and amplitude of the G-band and 2D-band [45]. The sample and its Raman spectroscopy are illustrated in the lower left and upper right corners in Fig. 10(a). Due to the small size (30 mm × 25 mm) of our MG sample, the THz-TDS was rebuilt with the focused TPX lenses (FTL-f30mm) to provide a 30 mm focus at the sample position between the emitter and detector antenna and the waist diameter of THz radiation is about 1 mm.
Fig. 10 (a) MG/PTFE sample and its Raman spectroscopy. (b) THz time-domain waveforms of MG/PTFE, IMF1 + IMF2 (red solid line), and IMF1 + IMF2 + IMF3 (green dash line). The original signals of the PTFE in the nitrogen and the air with RH 63.7% are respectively presented by the black dotted and blue dash-dot lines. (c–e) Data of tail oscillations (20 ps–50 ps) and (f) main pulse (4 ps–12 ps).
Figure 10 shows that the amplitude of time domain pulse under ambient humidity are significantly lower than that in the reference nitrogen due to the presence of water vapor noise. Comparing the reference signal, the IMF1 + IMF2 component has no significant effect on the main pulse but greatly suppresses the amplitude of tail oscillations. Therefore it can be judged initially that the IMF1 + IMF2 effectively eliminate the interference of water vapor noise and preserved intrinsic information of MG. The IMF3 component increases the tail oscillation amplitude but has little contribution to the main pulse, illustrating that the IMF3 contains only water vapor noise and no sample information.
The corresponding Fourier spectra of the MG are shown in Fig. 11. The characteristic peak of MG near 0.7 THz [46] appears in the spectrum in the nitrogen atmosphere. The experimental results measured under ambient humidity indicate that the atmosphere vapor noise causes a great interference in the MG spectrum around 0.56 THz and 0.75 THz, so that the characteristic absorption peak of graphene is submerged in the spectrum and is difficult to be identified. Interestingly, the IMF1 + IMF2 component successfully retains the graphene characteristic peak near 0.7 THz and eliminates the extremely strong interference of atmosphere vapor noise. The IMF3 component only increases the interference of water vapor noise and has the effects on the graphene characteristic peak at 0.7 THz. The above measurements demonstrate that the all-fiber THz-TDS based on the HHT can effectively eliminate external noise interference and preserve the intrinsic information of the sample. Moreover, the atmospheric vapor noise is accurately identified in the IMF3. The experiments on the materials with characteristic peaks exhibit the high performance of our proposed THz-TDS, which can be applied in more practical environments.
Fig. 11 THz Fourier spectrum of MG/PTFE corresponding to the time domain signal. The original signals of the PTFE in the nitrogen and the air with RH 63.7% are respectively presented by the black dotted and blue dash-dot lines. The Fourier spectra of IMF1 + IMF2 (the red solid line) and IMF1 + IMF2 + IMF3 (the green dashed line) demonstrate the elimination and identification of the water vapor, respectively. Furthermore, the graphene is characterized by the absorption peak near 0.7 THz according to the red solid line.
7. Conclusions
We have demonstrated a self-adaptive THz spectroscopy from atmospheric vapor based on the HHT. With the all-fiber THz-TDS, the interactions between THz wave and atmospheric water vapor has been experimentally verified, therefore, water vapor noise have caused a great disturbance in THz systems and covered up the intrinsic material characteristics. We have integrated the HHT into the THz system and selected the PTFE polymer material as the test sample. The THz time domain pulse is decomposed into the eight IMF components with the HHT algorithm. By detailed time-frequency analysis, we can only use the first two IMF components to effectively eliminate the interference of atmospheric vapor noise, which has been precisely identified in the IMF3. The experiments in different atmospheric humidity and different delay lengths confirm the strong self-adaptiveness of our HHT based THz-TDS. Besides the PTFE sample, we have conducted the similar measurement and analysis on the MG/PTFE sample. The IMF1 + IMF2 component effectively eliminates the strong interference of high-humidity atmospheric water vapor and successfully retains the characteristic peak of MG near 0.7 THz. The proposed THz spectroscopy based on the HHT is simple, effective, and strongly self-adaptive from atmospheric vapor noises and can be successfully applied in various humid environments.
Funding
National Natural Science Foundation of China (NSFC) (11374071); Fundamental Research Funds for the Central Universities of China; the 111 project (B13015) to Harbin Engineering University; Natural Science Foundation of Heilongjiang Province, China (A2018004); the Opened Fund of the Key Lab of In-fiber Integrated Optics, Ministry Education of China.
References
1. H. Zhang, S. Sfarra, A. Osman, K. Szielasko, C. Stumm, M. Genest, and X. P. V. Maldague, “An infrared-induced terahertz imaging modality for foreign object detection in a lightweight honeycomb structure,” IEEE Trans. Ind. Informatics 1–8 (2018).
2. A. Woldegeorgis, T. Kurihara, B. Beleites, J. Bossert, R. Grosse, G. G. Paulus, F. Ronneberger, and A. Gopal, “THz induced nonlinear effects in materials at intensities above 26 GW/cm2,” J. Infrared Millim. Terahertz Waves 39(7), 667–680 (2018). [CrossRef]
3. H. Zhang, S. Sfarra, K. Saluja, J. Peeters, J. Fleuret, Y. Duan, H. Fernandes, N. Avdelidis, C. Ibarra-Castanedo, and X. Maldague, “Non-destructive investigation of paintings on canvas by continuous wave terahertz imaging and flash thermography,” J. Nondestruct. Eval. 36(2), 34 (2017). [CrossRef]
4. K. Ahi, S. Shahbazmohamadi, and N. Asadizanjani, “Quality control and authentication of packaged integrated circuits using enhanced-spatial-resolution terahertz time-domain spectroscopy and imaging,” Opt. Lasers Eng. 104, 274–284 (2018). [CrossRef]
5. M. M. U. Rahman, Q. H. Abbasi, N. Chopra, K. Qaraqe, and A. Alomainy, “Physical layer authentication in nano networks at terahertz frequencies for biomedical applications,” IEEE Access 5(99), 7808–7815 (2017). [CrossRef]
6. K. Ahi, “A method and system for enhancing the resolution of terahertz imaging,” Measurement, (2018). In Press.
7. I. N. Dolganova, K. I. Zaytsev, S. O. Yurchenko, V. E. Karasik, and V. V. Tuchin, “The role of scattering in quasi-ordered structures for terahertz imaging: local order can increase an image quality,” IEEE Trans. Terahertz Sci. Technol. 8(4), 403–409 (2018). [CrossRef]
8. K. Ahi, “Mathematical modeling of THz point spread function and simulation of THz imaging systems,” IEEE Trans. Terahertz Sci. Technol. 7(6), 747–754 (2017). [CrossRef]
9. K. Ahi and M. Anwar, “Modeling of terahertz images based on x-ray images: A novel approach for verification of terahertz images and identification of objects with fine details beyond terahertz resolution,” Proc. SPIE 9856, 985610 (2016). [CrossRef]
10. H. Zhang, M. Genest, H. Fernandes, J. Fleuret, and X. Maldague, “Numerical and experimental analyses for natural and non-natural impacted composites via thermographic inspection, ultrasonic C-scan and terahertz imaging,” Proc. SPIE 10214, 102140I (2018).
11. D. Bigourd, A. Cuisset, F. Hindle, S. Matton, E. Fertein, R. Bocquet, and G. Mouret, “Detection and quantification of multiple molecular species in mainstream cigarette smoke by continuous-wave terahertz spectroscopy,” Opt. Lett. 31(15), 2356–2358 (2006). [CrossRef] [PubMed]
12. T. Globus, D. Woolard, T. W. Crowe, T. Khromova, B. Gelmont, and J. Hesler, “Terahertz Fourier transform characterization of biological materials in a liquid phases,” J. Phys. D Appl. Phys. 39(15), 3405 (2006). [CrossRef]
13. P. Y. Han, M. Tani, M. Usami, S. Kono, R. Kersting, and X.-C. Zhang, “A direct comparison between terahertz time-domain spectroscopy and far-infrared Fourier transform spectroscopy,” J. Appl. Phys. 89(4), 2357–2359 (2001). [CrossRef]
14. M. Tokurakawa, A. Shirakawa, K. Ueda, H. Yagi, S. Hosokawa, T. Yanagitani, and A. A. Kaminskii, “Diode-pumped 65 fs Kerr-lens mode-locked Yb3+:Lu2O3 and nondoped Y2O3 combined ceramic laser,” Opt. Lett. 33(12), 1380–1382 (2008). [CrossRef] [PubMed]
15. J. Chen, Y. Chen, H. Zhao, G. J. Bastiaans, and X.-C. Zhang, “Absorption coefficients of selected explosives and related compounds in the range of 0.1-2.8 THz,” Opt. Express 15(19), 12060–12067 (2007). [CrossRef] [PubMed]
16. M. R. Leahy-Hoppa, M. J. Fitch, X. Zheng, L. M. Hayden, and R. Osiander, “Wideband terahertz spectroscopy of explosives,” Chem. Phys. Lett. 434(4–6), 227–230 (2005).
17. K. Ahi, “Review of GaN-based devices for terahertz operation,” Opt. Eng. 56(9), 90901 (2017). [CrossRef]
18. K. Ahi and M. Anwar, “A survey on GaN-based devices for terahertz photonics,” Proc. SPIE 9957, 99570A (2016). [CrossRef]
19. M. van Exter, C. Fattinger, and D. Grischkowsky, “Terahertz time-domain spectroscopy of water vapor,” Opt. Lett. 14(20), 1128–1130 (1989). [CrossRef] [PubMed]
20. Y. Yang, A. Shutler, and D. Grischkowsky, “Measurement of the transmission of the atmosphere from 0.2 to 2 THz,” Opt. Express 19(9), 8830–8838 (2011). [CrossRef] [PubMed]
21. H. Harde, R. A. Cheville, and D. Grischkowsky, “Terahertz studies of collision-broadened rotational lines,” J. Phys. Chem. A 101(20), 3646–3660 (1997). [CrossRef]
22. D. Grischkowsky, S. Keiding, M. van Exter, and C. Fattinger, “Far-infrared time-domain spectroscopy with terahertz beams of dielectrics and semiconductors,” J. Opt. Soc. Am. B 7(10), 2006–2015 (1990). [CrossRef]
23. J. F. Federici, B. Schulkin, F. Huang, D. Gary, R. Barat, F. Oliveira, and D. Zimdars, “THz imaging and sensing for security applications–explosives, weapons and drugs,” Semicond. Sci. Technol. 20(7), S266–S280 (2005). [CrossRef]
24. W. Withayachumnankul and M. Naftaly, “Fundamentals of measurement in terahertz time-domain spectroscopy,” J. Infrared Millim. Terahertz Waves 35(8), 610–637 (2014). [CrossRef]
25. M. Naftaly and R. E. Miles, “A method for removing etalon oscillations from THz time-domain spectra,” Opt. Commun. 280(2), 291–295 (2007). [CrossRef]
26. P. U. Jepsen and B. M. Fischer, “Dynamic range in terahertz time-domain transmission and reflection spectroscopy,” Opt. Lett. 30(1), 29–31 (2005). [CrossRef] [PubMed]
27. J. Xu, T. Yuan, S. Mickan, and X.-C. Zhang, “Limit of spectral resolution in terahertz time-domain spectroscopy,” Chin. Phys. Lett. 20(8), 1266–1268 (2003). [CrossRef]
28. W. Withayachumnankul, B. M. Fischer, H. Lin, and D. Abbott, “Uncertainty in terahertz time-domain spectroscopy measurement,” J. Opt. Soc. Am. B 25(6), 1059–1072 (2008). [CrossRef]
29. S. G. Kong and D. H. Wu, “Signal restoration from atmospheric degradation in terahertz spectroscopy,” J. Appl. Phys. 103(11), 113105 (2008). [CrossRef]
30. Y. Wang, Z. Zhao, Z. Chen, Y. Zhang, L. Zhang, and K. Kang, “Suppression of spectral interferences due to water-vapor rotational transitions in terahertz time-domain spectroscopy,” Opt. Lett. 33(12), 1354–1356 (2008). [CrossRef] [PubMed]
31. L. S. Rothman, C. P. Rinsland, A. Goldman, S. T. Massie, D. P. Edwards, J. Flaud, A. Perrin, C. Camy-peyret, V. Dana, J. Y. Mandin, J. Schroeder, A. Mccann, R. R. Gamache, R. B. Wattson, K. Yoshino, K. V. Chance, K. W. Jucks, L. R. Brown, V. Nemtchinov, and P. Varanasi, “The HITRAN molecular spectroscopic database and HAWKS (HITRAN Atmospheric Workstation): 1996 edition,” J. Quant. Spectrosc. Radiat. Transf. 60(5), 665–710 (1998). [CrossRef]
32. Y. Huang, P. Sun, Z. Zhang, and C. Jin, “Numerical method based on transfer function for eliminating water vapor noise from terahertz spectra,” Appl. Opt. 56(20), 5698–5704 (2017). [CrossRef] [PubMed]
33. B. Sartorius, H. Roehle, H. Künzel, J. Böttcher, M. Schlak, D. Stanze, H. Venghaus, and M. Schell, “All-fiber terahertz time-domain spectrometer operating at 1.5 µm telecom wavelengths,” Opt. Express 16(13), 9565–9570 (2008). [CrossRef] [PubMed]
34. G. R. Kim, T. I. Jeon, and D. Grischkowsky, “910-m propagation of THz ps pulses through the Atmosphere,” Opt. Express 25(21), 25422–25434 (2017). [CrossRef] [PubMed]
35. N. E. Huang, Z. Shen, S. R. Long, M. C. Wu, H. H. Shih, Q. Zheng, N. C. Yen, C. C. Tung, and H. H. Liu, “The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis,” Proc. R. Soc. Lond. A 454(1971), 903–995 (1998). [CrossRef]
36. K. Patorski, M. Trusiak, and T. Tkaczyk, “Optically-sectioned two-shot structured illumination microscopy with Hilbert-Huang processing,” Opt. Express 22(8), 9517–9527 (2014). [CrossRef] [PubMed]
37. M. Trusiak, V. Mico, J. Garcia, and K. Patorski, “Quantitative phase imaging by single-shot Hilbert-Huang phase microscopy,” Opt. Lett. 41(18), 4344–4347 (2016). [CrossRef] [PubMed]
38. J. B. Camp, J. K. Cannizzo, and K. Numata, “Application of the Hilbert-Huang transform to the search for gravitational waves,” Phys. Rev. D Part. Fields Gravit. Cosmol. 75(6), 61101 (2007). [CrossRef]
39. S. Amar, M. Bahich, H. Dalimi, E. M. Barj, and M. Afifi, “Digital carrier superposition by Hilbert-Huang transform for optical phase recovery in speckle shearing interferometry,” Opt. Eng. 54(1), 13101–13107 (2015). [CrossRef]
40. M. Trusiak and K. Patorski, “Two-shot fringe pattern phase-amplitude demodulation using Gram-Schmidt orthonormalization with Hilbert-Huang pre-filtering,” Opt. Express 23(4), 4672–4690 (2015). [CrossRef] [PubMed]
41. P. Flandrin, G. Rilling, and P. Gonçalvés, “Empirical mode decomposition as a filter bank,” IEEE Signal Process. Lett. 11(2), 112–114 (2004). [CrossRef]
42. S. Wietzke, C. Jansen, M. Reuter, T. Jung, D. Kraft, S. Chatterjee, B. M. Fischer, and M. Koch, “Terahertz spectroscopy on polymers: A review of morphological studies,” J. Mol. Struct. 1006(1–3), 41–51 (2011). [CrossRef]
43. M. Naftaly and R. E. Miles, “Terahertz time-domain spectroscopy: A new tool for the study of glasses in the far infrared,” J. Non-Cryst. Solids 351(40-42), 3341–3346 (2005). [CrossRef]
44. X. Li, W. Cai, J. An, S. Kim, J. Nah, D. Yang, R. Piner, A. Velamakanni, I. Jung, E. Tutuc, S. K. Banerjee, L. Colombo, and R. S. Ruoff, “Large-area synthesis of high-quality and uniform graphene films on copper foils,” Science 324(5932), 1312–1314 (2009). [CrossRef] [PubMed]
45. A. C. Ferrari, J. C. Meyer, V. Scardaci, C. Casiraghi, M. Lazzeri, F. Mauri, S. Piscanec, D. Jiang, K. S. Novoselov, S. Roth, and A. K. Geim, “Raman spectrum of graphene and graphene layers,” Phys. Rev. Lett. 97(18), 187401 (2006). [CrossRef] [PubMed]
46. H. Cui, X. Zhang, P. Yang, J. Su, X. Y. Wei, and Y. Guo, “Spectral characteristic of single layer graphene via terahertz time domain spectroscopy,” Optik (Stuttg.) 126(14), 1362–1365 (2015). [CrossRef]
