Shielding effects in thin films of carbon nanotubes within microwave range

The electromagnetic shielding properties of thin ﬁlms comprising diﬀerent types of carbon nanotubes (CNTs) were analysed in the microwave frequency range (26–36 GHz). A comparative analysis of the shielding properties was achieved for ﬁlms based on long and short single, double-and multi-walled CNTs. The experimental results proved that long-length single-walled CNTs demonstrate the highest interaction with the electromagnetic (EM) ﬁeld, thereby providing the best shielding eﬃciency. At the same time, double-walled CNTs demonstrate a higher level of absorption ability (50%) along with the overall high EM shielding eﬃciency (88%), which makes them attractive for using in nanoelectronics screens as they produce the smallest secondary EM pollution.


Introduction
Progresses in nanoscience and nanotechnology continuously allow developing new applications, including communications and electronics.e development of new ultra-light and low-cost shielding nanomaterials for the microwave frequency range is a current task for improving the reliability of electronic equipment.is is the reason why intensive investigations of the shielding e ect are carried out for new types of composite materials based on graphite, graphene [1], carbon nanotubes [2][3][4] and other llers.
Due to a unique combination of high electrical conductivity, resistance to corrosion, exibility, lightness and mass-production ability, composites based on polymer matrices containing conductive llers are among the most attractive materials for shielding applications.As for various kinds of available conductive inclusions, carbon nanoparticles (amorphous carbon, graphene, fullerenes, carbon nanotubes, etc.) are considered to be the most promising for developing ecient electromagnetic shielding materials.As already shown in our previous research works [5][6][7][8], carbon nanotubes (CNTs) are among the best candidates for designing EM shielding polymer materials, owing to their high aspect ratio (~1000) and unique electronic properties.Shielding e ectiveness (SE) of CNT-based composites depends on the type of polymer matrix and on the characteristics of CNTs such as length [9], diameter [5,10], and orientation [11].However, composite production with a homogeneous distribution of CNTs in the matrix is still a non-trivial problem, due to the strong tendency of CNT to agglomerate.Such e ect limits the CNT concentration in the composites and lowers the total EM e ectiveness of the material.
In addition to composites prepared by means of traditional processes, e. g. by mixing polymers with CNT suspensions, other hybrid materials were recently developed [12].us, based on the idea that higher EM shielding might be obtained by using high concentration of separated CNTs (or their bundles), Wu et al. avoided the agglomeration problems and fabricated CNT micro lms on a polymer substrate (like sandwich structures) presenting high shielding e ectiveness (SE up to 61-67 dB) in the X band [13,14].Wang et al. presented a screen printing technique for the commercial fabrication of low-cost thin CNT lms on a polymer substrate with high shielding performances [15].Unlike usual composites, for which EM properties mostly depend on those of CNT agglomerates and of small amounts of isolated CNTs, the shielding e ectiveness of thin lms is more sensitive to CNT geometrical parameters.In spite of numerous publications related to traditional CNT-based polymer composites, the proper choice of optimal parameters is still a problem to be solved and it motivated our investigations.In the present paper, we compared microwave properties of thin lms comprising CNTs of di erent lengths and di erent natures (single-walled, double-walled and multi-walled CNTs).
e average diameter of individual SWCNTs was 0.8-1.2nm and their length was ~1 μm.Doublewalled CNTs (DWCNTs) were synthesized at CIRI-MAT (Toulouse, France) by catalytic chemical vapour deposition (CCVD) through decomposition of CH 4 over a Mg 1-x Co x O solid solution containing small amounts of molybdenum [16,17].
Short-length CNTs were produced by cutting long-length CNTs using the procedure detailed elsewhere [18].Atomic force microscopy (AFM, Solver P47 PRO, NT-MDT Inc) showed that intensive ultrasonication in an acid mixture (95% H 2 SO 4 and 59% HNO 3 at volume fractions of 8:1, respectively) at low temperature (<8 °C) allowed reducing the CNT length down to 100-300 nm for all kinds of tubes, without a signi cant change of their diameter distribution or modi cation of their electronic properties (Fig. 1).
1 wt.% of sodium dodecyl sulfate (Sigma-Aldrich) in water by ultra-sonication (44 kHz) for 1 h.en, in order to separate them from undissolved CNTs bundles and from impurities (e. g. small amounts of amorphous carbon and catalytic particles), centrifugation was carried out for 10 min at 12000 g.
lected and then ltered with a cellulose acetate membrane (Millipore, 0.22 μm pore size).During the ltration process, the CNTs accumulated on the membrane surface, forming a homogeneous lm.
ter.Finally, the ltration membrane was dissolved in acetone, the lm was washed with water, and then transferred onto 10 μm-thick Te on substrates.
e substrate is almost transparent in the microwave frequency range and does not contribute to the total electromagnetic response of the investigated samples.In order to measure the thickness of obtained lms, a small part of each one was also transferred onto a silica substrate.e average thickness of each lm was then determined with a digital pro lometer Veeco Dektak 6M and is given in Table 1.e microwave measurements of transmission/ re ection coe cients in the Ka-band (26-36 GHz) were performed with a scalar network analyzer R2-408 (ELMIK into a transmission line (waveguide of cross-section 7.2 × 3.4 mm) perpendicular to wave propagation [23].
e EM response of samples was measured as ratios of transmitted to input (S 21 ) and reflected to input (S 11 ) signals.

Results and discussion
e measured electromagnetic response (S 11 and S 21 ) of thin lms based on long-length and short-length SWCNTs is presented in Fig. 2. Frequency dependences of S-parameters for both types of SWCNT lms were at and did not show any special feature such as absorption peak or other trends.e same behaviour was observed for other composites based on DWCNTs and MWCNTs, and therefore the electromagnetic response was analyzed only at one frequency (30 GHz).
Our results reveal that thin lms based on 200 nmlong CNTs are transparent to microwave radiation (T ~ 84-92%) and give a negligible attenuation level (SE ~ 0.35-0.75dB).It can be noticed that, in this case, such high transparency does not depend on the type of CNTs used for lm fabrication.Figure 3 shows that the SE of lms based on SWCNTs is 13 dB, and decreased down to 3.62 dB as the number of walls increased.Such attenuation level was achieved due to high levels of both re ectance (58%) and absorbance (36%).
is result agrees with our previous theoretical predictions [7].Indeed, as it has been shown earlier, the electromagnetic properties of composites are mostly determined by the polarizability α of the individual inclusions.e imaginary part of the polarizability α has a peak in the terahertz range (1.5-6 THz).e frequency position f p of this peak depends on both length and diameter of CNTs.As demonstrated in [7], the frequency f p splits the spectra into two di erent parts: quasistatic and dynamic regimes.In the quasi-static regime (f ≤ f p ), the CNT polarizability is strongly in uenced by the nite-length e ect, leading to a strong screening as outermost shells screen the internal ones.Such a strong screening e ect in the case of short-length tubes explains the transparency (T > 80%) of the corresponding lms as well as the absence of dependence on the type of CNTs.e increase in CNT length decreases the impact of such nite-length e ect, resulting in the increase in the electromagnetic absorption from a moderate level of 15% up to 36-49%.As it was mentioned, the real part of the polarizability of long-length MWCNTs is larger than that of SWСNTs, but the substantial difference in re ection of thin lms (58% against 1% for SWCNT-and MWCNT-based lms, respectively) and high shielding e ectiveness can be explained by a signi cant di erence in their number density.
To conclude, the strong in uence of the nanotube length and the number of walls on the electromagnetic response of thin CNT-based lms was evidenced in the microwave frequency range.It has been shown that the thin lms based on short-length nanotubes demonstrated the negligible attenuation level due to the strong depolarizing eld and screening e ect in CNT. e opposite was observed for the lms based on long-length CNTs.e analysis of the obtained dependence for different types of nanotubes allowed us to conclude that the lms based on long-length SWCNTs are more preferable in case we are looking for overall high attenuation ability (the transmission rate is less than 5% because of 60% re ection and 35% absorption for the lm being approx.500-600 nm thick).However, lms made of double-walled CNTs are much more absorptive in comparison with SWCNT ones (50 vs 35%), which makes them in some sense more attractive as they do produce much smaller additional EM interference.Finally, MWCNTs demonstrate the smallest EMI shielding e ciency out of the three families of tubes investigated here, at the level of 55%, but still absorptive (50%), which also makes them suitable.

Fig. 2 .
Fig. 2. Frequency dependence of S 11 and S 21 .Inset: scanning electron microscope image of a long-length SWCNT lm.As can be seen, ultracentrifugation removed large particles and agglomerates, providing high-quality CNT lms.

Fig. 3 .
Fig. 3. Absorbance A, re ectance R and transmittance T at 30 GHz of lms based on CNTs: (a) long-length CNTs, (b) short-length CNTs.e inset shows the corresponding SE.

Table 1 .
Measured average thickness for each CNT lm.