Introduction

To detect ingredients from an unknown sample with higher sensitivity and specificity is crucial especially regarding food safety and chemical detection. To achieve such a goal, a detection method without labelling procedure should be developed. There appeared many different techniques for detection of chemical structures such as mass spectroscopy, nuclear magnetic resonance spectroscopy, infrared spectroscopy, Raman spectroscopy, etc., and infrared absorption and Raman scattering provide footprints of compositions of a sample and are both applied in so many different fields such as bio-, chemical- and gas-sensors. Besides, signals from both the two methods could be further enhanced through surface modification, also known as surface enhanced infrared absorption (SEIRA)1,2,3and surface enhanced Raman scattering (SERS)4,5,6. The advance of SEIRA and SERS is quick with the development of different nanostructures7, integration with other materials, e.g., graphene/ graphene oxide, iron oxide, etc8,9,10,11,12,13. , different fabrication procedures for stronger hot spots14,15,16, field enhancement from surface plasmon polariton17,18and excitation of Fano resonance3,19. Among these different methods, one promising kind of SEIRA is to couple a narrow absorption band from molecules’ vibration modes to a broadband absorption often introduced by a metamaterial for excitation of Fano resonance that revealed high quality resonance, extended interaction time due to slow group velocity and spectrum reversal for a better signal to noise ratio. To test the performance of Fano resonance based SEIRA substrates, several works have been published, for example, directly printed spiky gold flakes for the coupling between localized surface plasmon absorption and functional groups’ absorption from protein–bovine serum albumin with absorbance difference of 8%16, double Fano resonance based on complementary plasmonic metamaterials and PMMA with a signal strength DR equal to ~ 0.320, and also an electrically tunable graphene metal hybrid sensor with an electromagnetic introduced transparency like behavior, demonstrating a multiple functional groups’ sensing ability9.

Still, the metamaterial introduced in the above-mentioned methods are narrow bands and only suitable for pre-determined molecular absorption bands, thus limiting their practical use in many different fields. Also, once the absorption peak frequency deviates from the functional groups’ absorption frequencies, the sensing abilities of the published sensors were compromised. To tackle this issue, researchers employed strategies such as multiple band metamaterial designs2, active tunable metamaterials21, and metal and thickness-dependent metamaterials22 to enrich the available detection frequency range. However, these methods are insufficient due to their drawbacks such as still limited detection frequency ranges, smaller hot spot areas, difficult fabrication process, and weaker field enhancement. Therefore, in this study, we would like to design an oblique-wire-bundle (OWB) metamaterial absorber accomplished by oblique deposition for its broad absorption band and strong hot spot with larger hot spot areas; with the proposed OWB MA, one might detect a sample composed of a mixture of different kinds of chemicals with a higher sensitivity and a higher signal to noise ratio and without a custom-made sensor.

Fabrication and characterization

Based on our previous work23,24, we designed an oblique wire bundle (OWB) metamaterial absorber on a silicon substrate with a parallelogram periodicity of 2\(\:\overrightarrow{x}\) and 0.5\(\:\overrightarrow{x}\)+\(\:\overrightarrow{y}\) where |\(\:\overrightarrow{x}\)| and |\(\:\overrightarrow{y}\)| equal to 900 an and 970 nm, respectively. On top of the Si substrate, a 150-nm-thick aluminum and a seed layer of MgF2 with a base width of 550 nm, a base length of 740 nm, and a thickness of 100 nm are deposited on top of the substrate by sophisticated E-beam lithography and an E-gun evaporator. Note that in experiments, the MgF2revealed a shape of the frustum of a rectangular pyramid due to the undercut of the resist in the lithography procedure. Then, the sample was put on a stage in an e-gun evaporator with a tilted angle of 86-degree which promised a coverage length of 1430 nm25, ensuring the growth of metallic nanowires only on top of the seed layer during deposition. Here, to make a comparison, a planar metamaterial absorber with the same periodicity and dimension is fabricated. Note that the thickness of metal in the planar metamaterial absorber is reduced to 50 nm instead of 460 nm, detected thickness of aluminum when fabricating the OWB metamaterial absorber, for better yielding of the lithography process. The entire fabrication procedure of the OWB and planar metamaterial absorbers was detailed in Methods. It is worth mentioning that since the OWB metamaterial absorber possessed a broadband absorption spectrum, we expected that the OWB metamaterial absorber could reveal strong Fano resonance without a specific design in advance. Here, to make a comparison, Fig. 1 depicted the schemes and absorbance spectra of the OWB and planar metamaterial absorbers under both the x- and y-polarization incidences from µ-FTIR. The OWB metamaterial absorber possessed a broad absorption band from 100 to 180 THz under x-polarization and an even wider bandwidth 111 THz (69 to 180 THz) under y-polarization incidence. On the contrary, the planar metamaterial revealed an absorption peak of 0.72 with a bandwidth of 15.6 THz under the x-polarization and an absorption peak of 0.6 with a bandwidth of 15.65 THz under the y-polarization. It is worth pointing out that the small peak/dip shown at the frequency of 70.5 THz in the spectra of the OWB and planar metamaterial absorbers could be attributed to the stretching vibration mode of C = O from CO2. Here, it is an absorption peak in the spectrum of the OWB metamaterial absorber but a dip for the planar absorber, which might stem from the CO2 concentration variation for the background and device measurement. Still, due to larger hot spot areas and stronger field enhancement, the OWB metamaterial absorber enhanced the vibration mode from CO2, thus resulting in stronger absorption compared to the planar metamaterial absorber.

To further explore the difference between the OWB and planar metamaterial absorbers, we conducted simulation based on finite integration method with a boundary condition of unit cell along the x- and y-directions and open boundary conditions along the z-direction. Since the MgF2 is in the shape of frustum, in simulation, its dimensions are modified with a base width and length of 550 and 740 nm and a width and a length of 492.7 and 652.7 nm for the top face. Multiple nanowires with randomly distributed positions, widths, lengths, and thicknesses were generated on top of the seed layer. Indicated by the blue curves of Fig. 1, both the OWB and planar metamaterial absorbers revealed similar absorption bands for the x- and y-polarization between the measured and simulated results. Moreover, from simulation, the absorption bands for the x- and y-polarization incidences started from 85 to 67 THz, respectively, and extended to 384 THz, the upper band edge of the near infrared regime; therefore, our proposed OWB metamaterial absorber revealed bandwidths of 299 and 317 THz under the x- and y-incidences, respectively, which covered almost all the vibrational modes of chemicals for the both polarizations. It is worth mentioning that we expected the OWB metamaterial would reveal stronger absorption and larger bandwidth under the y-polarization incidence due to various optical paths below the metallic wire bundles. Thus, the OWB metamaterial absorber is free from re-design procedure for the detection of most chemicals. In contrast, the planar metamaterial absorber only possessed the limited absorption bandwidth, thus limiting its coupling with molecular vibration modes.

Fig. 1
figure 1

Schemes of (a) an oblique wire bundle (OWB) and (b) planar metamaterial absorbers. Measured and simulated absorption spectra of (c, d) oblique wire bundle (OWB) and (e, f) planar metamaterial absorbers under the x- and y-polarization incidences without PMMA. Insets show SEM images of the fabricated absorbers. It is worth pointing out that there appeared small peaks in the measured results at 70.5 THz for the spectra of the OWB metamaterial absorber, which could be attributed to the vibrational stretching mode of C = O. These peaks indicated the stronger field enhancement and larger hot spot areas, thus facilitating occurrence of Fano resonance from the OWB metamaterial absorber than the planar one. Note that the absorption dip in (c, d) might stem from the fluctuation of CO2 concentration for the measurement of the background and the absorber.

Full size image

To further characterize the OWB metamaterial absorber and its ability to molecule-plasmon coupled SEIRA, we spin-coated a layer of A5 PMMA on the absorbers with a rotation speed of 1000 rpm for 10 s and then 4000 rpm for 60 s. The expected thickness is around 380 to 460 nm. Here, PMMA possessed three molecular absorption modes within the targeted frequency range, including a C = O stretching mode at 51.9 THz, a CH2 asymmetric stretching mode at 88.5 THz and a CH3 stretching mode at 90 THz. Meanwhile, the stretching vibration mode of C = O from CO2 also existed. As portrayed in Fig. 2, all the four molecular vibration modes could interact with the OWB metamaterial absorber and revealed strong Fano resonance under both the x- and y-polarizations. Note that when PMMA was applied to the metamaterial absorbers, the resonance frequencies of the OWB metamaterial absorber down shifted to the lower range, facilitating a stronger coupling between the stretching vibrational mode of O = C = O from CO2, indicated by the excitation of the Fano resonance. In contrast, it is difficult to observe the asymmetric line shapes from Fano resonance for the planar metamaterial absorber; instead, the superposition of absorption is observed with larger absorption values at the molecular vibrational frequencies in the spectrum.

Fig. 2
figure 2

Measured and simulated absorption spectra of the (a, b) OWB and (c, d) planar metamaterial absorbers under the x- and y-polarizations with PMMA. The OWB metamaterial absorber could couple to all the four molecular vibration modes and developed Fano resonance while the planar metamaterial absorber could only couple to the vibrational mode of CO2 with a superposition behavior for all the other three modes from PMMA.

Full size image

Here, to elucidate the interaction between the metamaterial absorbers and chemicals, we conducted simulation of the OWB and planar metamaterial absorbers with PMMA and CO2 as well. A 460-nm-thick analyte was applied onto the absorber with a dielectric constant predicted by Lorentz model listed below,

$$\:{\upepsilon\:}={\epsilon\:}_{{b}_{Analytes}}+\sum\:_{j=1}^{4}\frac{{f}_{m}{\omega\:}_{0,j}^{2}}{{\omega\:}_{0,j}^{2}-{\omega\:}^{2}-i{\gamma\:}_{j}\omega\:}$$
(1)

where \(\:{\epsilon\:}_{{b}_{Analytes}}\) is the relative background permittivity of PMMA and CO2 and is equal to 2.25, the background permittivity of PMMA, fm resonance strength of functional absorption groups, \(\:{\omega\:}_{0,j}\) the vibrational angular frequency of the functional group and \(\:{\gamma\:}_{j}\) the damping frequency. Here, to infer the functional group absorption to the analyte, we included three resonance modes from PMMA, i.e., the C = O stretching mode (w0 = 51.9/g = 0.1 THz), the CH2 asymmetric stretching mode (88.5/0.7 THz) and the CH3 stretching mode (90/0.9 THz) and one resonance mode from CO2, i.e., the C = O asymmetric stretching mode (70.5/0.05 THz). All the resonance strengths are 0.001. From the blue curves of Fig. 2, the OWB metamaterial absorber facilitated occurrence of Fano resonance for all the four vibrational modes with the significant asymmetric line shapes, an indicator of Fano resonance; besides, the signals at the y-polarization incidence are stronger and observable at the 88.5 and 90 THz compared to the ones at the x-polarization incidence, another evidence of better enhancement for the y-polarization incidence. On the other hand, the planar metamaterial absorber only showed Fano resonance for the strongest vibrational mode with the smallest damping, i.e., the O = C = O stretching mode of CO2 and showed only superposition of absorption for all the other three modes. Still, the hot spots provided by the planar metamaterial absorber could boost the magnitude of absorption when compared to PMMA absorption spectrum with an enhancement factor of around 2 to 3. In addition, the OWB metamaterial absorber revealed profound asymmetric line shapes in experiments compared to the ones in simulation which could be attributed to the hot spots among nano-wire-bundles that are not included in simulation.

For the OWB absorber, under x-polarization incidence at 51.9, 70.5, 88.5 and 90.5 THz, the DR/R03 are 3.07, 1.71, 4.79, and 7.24 times higher when we compared the experimental and simulation results. Also, for the OWB absorber, under the y-polarization incidence at 51.9, 70.5, 88.5 and 90.5 THz, the DR/R0 are 2.13, 1.33, 1.05, and 1.01 times higher when we compared the experimental and simulation results. As for the planar metamaterial absorber under the x- and y-polarization incidences, the measured results showed 1.47, 0.27,0.85, 0.63 times and 2, 0.49, 0.50, and 0.51 times higher DR/R0 compared to the simulated results. The better enhancement between the measured and simulated results for the OWB metamaterial absorber could be attributed to the hot spots among nanowire bundles, which is not completely considered in the simulation.

Fig. 3
figure 3

Absolute field |E| distributions of (a, b) OWB and (c, d) planar metamaterial absorbers under x- and y-polarizations at 126.25 and 140.75 THz, respectively. The OWB metamaterial absorber revealed larger local maximum |E| compared to the ones of planar metamaterial absorber. Moreover, OWB metamaterial absorber showed much more hot spot area compared to the planar metamaterial absorber (not shown here). Also, the global maximum |E| for the OWB and planar metamaterial absorbers are 6.63 × 108/2.16 × 109 and 4.89 × 108/4.43 × 108 V/m under x- and y-incidences, respectively, evidencing that the OWB metamaterial absorber indeed provided stronger hot spots with a larger area, facilitating coupling with molecular absorption and occurrence of Fano resonance.

Full size image

To dig out the reasons for better coupling of molecular absorption by the OWB metamaterial absorber, we recorded the absolute field distribution |E| of the two absorbers at the frequency of 126.25 and 140.75 THz as shown in Fig. 3. Note that the chosen frequency was based on the maximum absorption from the planar metamaterial absorber under x- and y-polarizations. For the x-incidence of the OWB absorber, the local highest |E| at the 126.25 THz is 6.32 × 108 while for the y-incidence, the highest |E| at the 140.75 THz is 6.17 × 108, which are both approximately 1.46 times larger compared to the ones of the planar absorber. On the other hand, when considering the global highest |E|, the OWB absorber revealed a value of 6.63 × 108 and 2.16 × 109 for the x- and y-incidence. It is worth noting that the hot spot areas from the OWB absorber are also much larger than the ones of the planar absorber. Also, the maximum |E| differed at the different height of the OWB metamaterial absorber; still, the |E| also varied with respect to the different positions of the y-planes, thus facilitating the sensitivity enhancement for different thicknesses of the analytes.

Fig. 4
figure 4

Simulated absorption spectra of the (a) OWB and (b) planar metamaterial absorbers under the x- and y-polarizations with 100-nm-thick PMMA. The OWB metamaterial absorber could couple to all the four molecular vibration modes while the planar metamaterial absorber only showed the superposition of absorption magnitude even for the O = C = O stretching vibrational mode of CO2.

Full size image

Furthermore, we would like to discuss two different cases for examining detection ability of the proposed metamaterial absorbers. The first of the two is to reduce the thickness of the analytes, i.e., from 460 nm to 100 nm. The absorption spectra are depicted in Fig. 4(a). Again, although small, the OWB metamaterial absorber supported the four asymmetric line shapes as shown in the inset of Fig. 4(a). In contrast, the planar metamaterial absorber shows the superposition behavior even for the O = C = O vibrational mode. The other is to reduce the resonance strength of the analyte from 0.001 to 0.0001 while the thickness was maintained. As illustrated in Fig. 5, the OWB metamaterial absorber supported four asymmetric line shapes as shown in the inset of Fig. 5(a). In contrast, the planar metamaterial absorber shows the superposition behavior even for the O = C = O vibrational mode.

Fig. 5
figure 5

Simulated absorption spectra of the (a) OWB and (b) planar metamaterial absorbers under the x- and y-polarizations with PMMA for the reduced resonance strength of 0.0001. The OWB metamaterial absorber could couple to all the four molecular vibration modes with the asymmetric line shapes while the planar metamaterial absorber only showed the superposition of absorption magnitude.

Full size image

Finally, to reinforce the detection ability of our proposed OWB metamaterial absorber, we plotted the DR with respect to different strengths of PMMA ranging from 0.1 to 0.00001, i.e., different concentrations at the three frequencies for the functional group absorption as shown in Fig. 6. Under the logarithm-logarithm plot, we could observe that all the curves revealed a characteristic of exponential decay. Furthermore, due to the strongest absorption for the O = C = O, the DR was the highest for the polarizations. Also, the DR would saturate once the resonance strength increased. From Fig. 6, all the DR started to saturate at a resonance strength of 0.01 for the two polarizations. Note that since the absorption is weaker for the CH2 asymmetric stretching modes, the smaller resonance strength (i.e., below 1×10-3) was not enough to excite Fano resonance.

Fig. 6
figure 6

Schemes of DR with the different resonance strengths for three different functional groups under the (a) x-polarization and (b) y-polarization incidences. The data points could be fitted by exponential decay with R-squares above 0.96.

Full size image

Conclusions

In this work, we have proposed an oblique-wire-bundle metamaterial absorber with broad absorption bandwidths of 299 and 317 THz under the x- and y-polarization incidences, respectively, which covered almost all the infrared absorption spectra of different chemicals. Such proposed OWB metamaterial absorber alongside with the planar metamaterial absorber are tested with PMMA and CO2 for the coupling with the vibrational modes of the different analytes. The OWB metamaterial absorber revealed strong interaction with the analytes even with smaller thicknesses or weaker resonance strengths; in contrast, the planar metamaterial absorber could not interact with the chemicals well and showed superposition of absorption for almost all the cases. We believed that this proposed OWB metamaterial absorber could help the development of rapid detection in the field of food safety and chemical detection.

Data availability

The data sets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Fabrication

To fabricate the oblique-wire-bundle metamaterial absorber and its comparison, i.e., the planar metamaterial absorber, we deposited 4-nm-thick titanium and 150-nm-thick aluminum on a silicon substrate. Then, spin coat photo resist of PMMA A5 with an initial rotation speed of 1000 rpm for 10 s and a final rotation speed of 4000 rpm for 60 s. Next, the sample was hard baked under 180 °C for 3 min. In the following, the electron beam exposure was carried out. After exposure, the sample was immersed into a solution of MIBK and IPA in a ratio of 1 to 3 for 50 s. In addition, 120-nm-thick magnesium fluoride was deposited onto the sample. In the following, to fabricate the oblique-wire-bundle metamaterial absorber, the as-fabricated sample was put into acetone for the lift-off process to remove residual photoresist; e-gun deposition was then conducted for 4-nm-thick titanium and for the growth of aluminum wire bundles on an 86-degree tilted stage. On the other hand, to fabricate the planar metamaterial absorber, 4-nm-thick titanium and 50-nm-thick aluminum were consequently deposited onto the sample. Finally, the as-fabricated sample was put into acetone for the lift-off process to remove residual photoresist.