Ex situ Spectroscopic Characterization of Residual Effects of Thermomechanical Loading on Polyurea

Despite more than two decades of persistent research efforts involving several prominent researchers, polyurea elastomers still attract significant analytical, computational, and experimental investigations spanning over a wide range of length and time scales [1]. Polyurea is an elastomeric material with exceptional physical properties, including high stretchability, high tear and tensile strength, and moisture and chemical resistance [2,3]. Polyurea has been actively investigated for numerous applications [1,4–6], of which impact mitigation in a variety of environmental conditions has been a prime focus. For example, polyurea adhesive was found to increase the fracture toughness of composite marine joints by 1.3-fold compared to Hysol epoxy [7]. A polyurea monolithic coating layer has been shown to improve the blast resistance of armor steel and aluminum plates [8,9], composite sandwich panels, joints and structures [3,10], and sports gears [11,12]. In an attempt to further enhance the impact performance of polyurea, several reports investigated the mechanical properties of reinforced polyurea matrix composites using carbon nanotubes [13,14], fly ash [15], and polyurea microspheres [16–18], reporting various degrees of improvement of the local deformation resistance but with no appreciable overall enhancements. In general, the performance of polyurea in impact scenarios is attributed to (1) the strain rate sensitivity [19–22] and (2) the microstructure composing of interpenetrating hard and soft segments [23–27]. Recently, Shahi et al. performed an exhaustive experimental investigation on the role of the hard and soft segments dispersion and of the chain length on the thermomechanical properties of polyurea with different compositions of the constituents [28]. They reported that the composition plays a role in the high strain response of polyurea by altering the post-yield hardening behavior. Alternatively, Gupta et al. and Youssef et al. were successful in foaming polyurea after recognizing the positive effect of adding a thin layer of polyurea to several sports gears [29,30]. Polyurea foams were effective in mitigating biomechanical impacts due to low-velocity loading [17,30–33]. In all, polyurea remains a fascinating material from the basic and applied research perspectives.

The mechanical behavior of polyurea has been heavily investigated in response to different environmental and loading conditions experimentally, including a wide range of strain rates (∼10⁻³ s⁻¹ to 10⁷ s⁻¹) [4,19,20], temperatures (−100 °C–50 °C) [6,34,35], salinity [3,6], and ultraviolet radiation [6,36–39]. As the tensile strain rate increases, polyurea exhibits phase transition from the rubbery to the glassy regime passing through the leathery region, depending on the testing rate, where the strain-to-failure decreased while the yield stress and ultimate failure stress increased [20–22,40,41]. Our group also examined the effect of extended exposure to ultraviolet radiation on the elastic, hyperelastic, and viscoelastic properties of polyurea [36–39]. It was found that the hyperelasticity of polyurea remained unchanged, while the ultraviolet radiation increased the stiffness and resulted in the formation of surface microcracks extending 45 µm into the samples after 15 weeks of continuous exposure [39]. Che et al. investigated the response of polyurea in a marine environment, including ultraviolet radiation, for up to 150 days [42]. Che and collaborators attributed the aging response of polyurea to an increase in phase separation and chain scission of N–H, C=O, and C–O–C, and the hydrogen-bonded urea carbonyl group [42]. More recently, Mforsoh et al. also examined the compressive response of polyurea after exposure to aggressive marine environments, including salinity, temperature, and ultraviolet radiation [6]. They found that the strain energy decreased after extended exposure to saline water but increased after ultraviolet radiation. However, despite extensive research on the mechanical response of polyurea under different loading conditions, there is a gap in the knowledge about the residual effect of mechanical and thermal loadings on polyurea microstructure, potentially compromising its long-term reliability in repeated loading scenarios. This knowledge gap is the primary motivation of the research leading to this paper.

The post-loading microstructure of polyurea has been examined using X-ray scattering and Brillouin scattering spectroscopy [43,44]. The X-ray studies performed on polyurea-metal bilayer plates elucidated the strain hardening response of polyurea and its contribution to the overall energy dissipation of the composite structure [44], which was attributed to the sample preparation process. The hard segments failed under the influence of the high stress associated with high strain testing, hence increasing strain hardening. On the other hand, Brillouin scattering spectroscopy was used to extract the longitudinal and shear sound velocities of polyurea at pressures up to 13.5 GPa [43]. Since polyurea vitrifies under extreme strain hardening, conventional mechanical testing and characterization techniques cannot capture the subtle changes in material properties upon solidification. The material properties of polyurea under high pressure, high strain rate conditions were then determined by correcting the high-frequency Brillouin data to account for viscoelastic effects, resulting in the extraction of better estimates of the change in the density from the Brillouin sound velocity data [43]. Another promising spectroscopy technique in the area of ex situ or in situ characterization of loaded-polymers is terahertz time-domain spectroscopy since it is suitable for noncontact, noninvasive, and non-ionizing interrogation of polymers. This research examined the optical and mechanical response of elastomeric material under mechanical stresses using a novel in situ terahertz time-domain spectroscopic technique [45]. The resulting time-domain signals at different applied stresses were used to elucidate subtle changes in the elastomer microstructure by investigating the temporal characteristics of the terahertz waves interacting with the samples. In other words, terahertz time-domain spectroscopy was used to provide physical evidence of mechanically stress-induced conformational changes of polymers, signified by changes in the complex index of refraction. While being a macroscopic property, the latter can be used to reveal microscopic properties such as the change in the polarizability using Lorentz–Lorenz and Clausius–Mossotti relations.

This research aims to elucidate the residual effects of thermal and mechanical loading on polyurea elastomers after submersion in cryogenic and quasi-static loading conditions, respectively. Cast polyurea disks with two different thicknesses were first extracted from two cast sheets and fully submerged in liquid nitrogen (LN₂) to reach isothermal conditions for up to 3 h. The additional specimens were extracted from the same sheets for quasi-static tensile testing at a strain rate of 0.01 s⁻¹ up to 100% strain using a ±1 kN load frame (Instron 5843). Therefore, one set of samples with a thickness of 1661 ± 25 µm was separately tested under uniaxial tension and cryogenic isothermal condition. Another set of polyurea samples with a thickness of 497 ± 65 µm was also tested under the same testing conditions. The choice of this range of thickness was based on a priori work, showing thin film polyurea with <2 mm in thickness to be effective in mitigating impacts in several applications [9,29]. Moreover, the thickness range was selected based on the specification of the terahertz spectrometer developed in our laboratory. The tested samples (thermally or mechanically) were then allowed to rest in ambient conditions for 4 h. The optical properties of the samples were assessed using terahertz time-domain spectroscopy (THz-TDS) operating in the transmission mode, reporting the complex index of refraction as a function of frequency ranging from 0.4 to 1.3 THz.

Polyurea sheets with different thicknesses were manufactured using a slab-molded technique, where the chemical constituents, diisocyanate (DOW Industrial, Isonate^® 143L, Midland, MI), and oligomeric diamine (AirProduct Inc., Versalink^® P1000, Allentown, PA), were mechanically mixed with a 1:4 weight ratio, respectively. The sheets were first cured in ambient conditions for 24 h, and then in a vacuum oven at 80 °C for an additional 24 h. The importance of the full curing cycle on the overall performance of the polyurea sheet was previously discussed in Ref. [46]. The average thickness of one sheet was 1661 ± 25 µm, while the average thickness of the second sheet was 497 ± 65 µm, in what is referred to hereafter as “thick” and “thin” samples, respectively. The samples thickness was based on the bandwidth and capabilities of our developed in-house terahertz-wave spectrometer. The notation of “thin” and “thick” is therefore descriptors and not categorical. Discs of 25.4 mm in diameter were die cut using a hydraulic press from an area with consistent thickness from each sheet. Another disc-shaped specimen from each sheet was neither mechanically nor thermally loaded, referred to herein as the virgin samples. Thick and thin polyurea discs were soaked in liquid nitrogen (LN₂) for 180 min and 60 min, respectively, to where the discs reached isothermal conditions. The long soaking time was chosen to ensure thermal equilibrium even though the time required to reach isothermal condition was calculated to be only a few seconds based on the submersion condition and the thermal conductivity of polyurea (taken to be 0.16 W·mK⁻¹) reported by Shahi et al. using one-dimensional heat conduction [28]. Notably, this thermal loading condition resulted in a thermal strain (ɛ_th) of 4.38% using a thermal expansion coefficient (α) of 2 × 10⁻⁴ K⁻¹ based on ɛ_th = αΔT, the ΔT is the difference between room temperature and the LN₂ bath [34]. Upon removal from the LN₂ bath, the samples were left to naturally reach ambient conditions for 240 min before being investigated using THz-TDS.

Polyurea strips with dimensions of 10.16 cm long × 2.54 cm wide were cut out from the same sheets for mechanical tensile loading. The 10.16 cm × 2.54 cm samples were loaded using an Instron 5843 load frame with a ±1 kN load cell. The strain was measured using an Instron 2603-84 large-strain extensometer, where the samples were loaded in tension to a maximum strain of 100%. The corresponding maximum forces applied to the thick sample and thin sample were 258 N and 84 N. Upon removal from the load frame, three discs with a 25.4 mm diameter were removed from the loaded region of the stretched polyurea strips, as seen in Fig. 1(a). The straight notch in each sample (shown in Figs. 1(a) and 1(b)) was introduced to ensure the direction of loading coincided with the direction of terahertz propagation during the spectroscopic investigation. The virgin, thermally, and mechanically loaded discs were fitted into mounts with an inner diameter of 25.4 mm to eliminate any tilt or misalignment during the measurement acquisition using THz-TDS. Each of the three plugs extracted from each stretched sample was interrogated with terahertz waves. For the thermally loaded sample, a polyurea disc of each thickness was soaked in the LN₂ bath, and three terahertz signals were collected from each sample. The specimen discs were lightly clamped between the surface of the mount on one side and a retaining ring on the other. This mounting method facilitated the handling and placement of each specimen during THz testing, where the focal point of the terahertz beam was guided to the center of the specimen, avoiding any obstruction. A schematic of the sample situated in the terahertz time-domain spectroscopy can be seen in Fig. 2.

Figure 3(a) shows the THz time-domain signals of thermally and mechanically loaded samples compared to their virgin counterparts and the sample-free reference signals (collected when the samples were not present in the THz beam path). The primary wave characteristics of interest are summarized in Table 1, including the average thickness (post-loading thickness is reported for the thermal and mechanical samples, which were mechanically measured), the peak amplitude, and the delay time. The calculated optical and electrical properties are also included in Table 1. The notable reduction in amplitude and temporal shift in all the signals resulted from the convoluted effect of the thickness and the change in the index of refraction. For example, when comparing the reference (solid magenta line) with a thin (solid black line) and thick (dotted black line) virgin signal, the peaks appeared at 39.40 ps, 40.32 ps, and 43.13 ps, respectively, marking the temporal delay in the arrival of t_thin = 0.9167 ps and t_thick = 3.7333 ps. The average thickness of the thin samples was 476 µm, while the thick samples were 1686 µm, representing nearly a threefold increase (d_thin/d_thick ≈ 0.28), which is commensurable with the ratio of the time delay (t_thin/t_thick ≈ 0.25). The disparity between these ratios signifies the contribution of the thickness-dependent change in the index of refraction. The real part of the refractive index (n) is related to the delay time (t) between the main pulse of the reference and sample signals, the measured thickness (d) of the sample, and the speed of light (c) in vacuum, where $n=1+ctd$⁠. Therefore, the real refractive index of polyurea is estimated to be 1.65 and 1.52 for thick and thin samples, respectively, using the attributes of the time-domain signals in Table 1. Similarly, the rest of the temporal characteristics are used to estimate the imaginary part of the refractive index (a measure of the absorption coefficient), where κ = ln(2/(n + 1)A_R/A_S) is based on the amplitude of the sample (A_S) and the reference (A_R) signals. Here, κ was defined based on the logarithmic decrement concept while considering the Fresnel coefficient in the case of total transmission. The imaginary part of the index of refraction for thin and thick polyurea are 0.15 and 0.45, respectively, since the amplitude of the time-domain signals were found to be 0.445 V and 0.309 V while the amplitude of the reference was 0.634 V.

In terms of the complex refractive index, the optical properties are then used to compute the dielectric function (ɛ) of polyurea since ɛ = (n + iκ)². The dielectric constant (|ɛ|) of thick and thin virgin polyurea samples was found to be 2.93 and 2.34, respectively, which is in reasonable agreement with the constants measured using the electrical method and reported to range 3.5–4.2, depending on the preparation method (including the stoichiometric ratio and the type of diamine and isocyanate), temperature, and frequency [47,48]. In short, the difference in the values of the dielectric constant reported here using the optical properties, and that reported a priori using the electrical method is thought to be due to (1) the sample thickness used in the experiments and (2) the sample preparation approaches. Here, the samples were extracted from slab-molded polyurea sheets, while the samples used in the electrical testing experiment were prepared via vapor deposition polymerization or vacuum thermal vapor deposition techniques [47,49]. It is worth noting that the thickness plays a significant role in the final conformational arrangements within the sample, where an ultrathin configuration may be energetically favorable for a higher degree of crystallization [47,49]. The manufacturing process may also affect the conformations given factors such as the substrate clamping effect and confinement, resulting in mechanical anisotropy [50,51].

The strain percentage due to thermal loading by supercooling the sample in an LN₂ bath and then bringing it back to ambient temperature was estimated to be 4.38%, based on the difference in temperature (ΔT) and the thermal expansion coefficient (α), i.e., the strain is due to thermal effects. This low strain level is within the linear elastic region of the mechanical response of polyurea, which can be confirmed by projecting this low strain percentage on the engineering stress–strain curve shown in Fig. 3(b). The loading is thought to be completely reversible, hence, the residual stress and strain effects can be considered minimal in the case of thermal loading. It is worth noting that concurrently observing the molecular transitions of polyurea during cooling is a focus of future research and outside the scope of this investigation, where an accurate determination of the glass transition can be identified [52]. The negligible residual effect was also captured in the THz time-domain signal by comparing the data of the virgin and thermally loaded samples. For the thick samples, the peak amplitude and time delay changed by 4.9% and 0.01%, respectively, indicating the reversibility of the deformation. Similarly, the THz signals of the virgin and supercooled thin polyurea samples were also comparable. The peak amplitude of the thin samples was 0.476 V and 0.466 V for the virgin and thermally loaded samples with a time delay of 0.83 ps and 0.80 ps, respectively. In all, and regardless of the sample thickness, the thermal loading had a negligible residual effect, signifying the utility of polyurea over a broad range of temperatures [1].

In the case of mechanically loading the polyurea samples up to 100% strain, analysis of the THz time-domain signals reported a notable difference, indicating that the residual effect is present while signifying irreversibility. It is worth noting that the samples were characterized using THz-TDS, on average, two hours post-loading. There are three noteworthy observations, following a close examination of the time-domain signals in Fig. 3(a) and the summary of the major attributes in Table 1. First, the thickness has exhibited a reduction of 18.8% and 11.5% for thick and thin samples, respectively. The remnant set in the thickness is attributed to the relatively short duration between testing and spectroscopic characterization, not providing ample time for full recovery based on the viscoelastic properties of polyurea. The mechanical loading at 100% strain corresponds to the engineering stress of 5.5 MPa, which is nearly twofold higher than the reported yield stress of the material [6]. Second, the amplitude of the terahertz time-domain signal increased, regardless of the thickness of the tested samples. On average, the amplitude of the thin samples increased by 14.6% and by 9% for the thick samples, compared with the signal amplitude of the virgin samples. The amplitude growth was accompanied by variations in the time delay. As discussed above, the amplitude and time delay changes are generally associated with the complex refractive index changes. Correspondingly, the latter was calculated to be 1.67 + i0.39 for the stretched thick sample and 1.66 − i0.05 for the thin sample, which is greater than the complex index of refraction for the virgin samples reported previously. It is then thought that the post-loading index represents a change in the molecular structure that resulted in less attenuation and dispersion of the THz electromagnetic waves as it propagates throughout the samples, which is realized by the reduction in the extinction coefficient (κ) for the stretched polyurea samples.

Figure 4(c) shows the quality factor results of thick and thin samples for the virgin, stretched, and supercooled samples, which succinctly summarizes all the conclusions elucidated via the consideration of the optical and electrical properties thus far. It is worth noting that the quality factor was calculated for each thickness and testing condition separately by comparing the signal collected from that sample to its reference data. Three overarching conclusions of time-domain data are concisely encoded in the Q values. First, the thickness plays a measurable role due to the attenuation and dispersion of the waves within the material, where the difference in Q was found to be 33% when comparing thin and thick virgin samples. The disproportionality of Q and thickness is attributed to the changes in the complex index of refraction, as discussed before. Second, mechanical loading tends to increase bond oscillation dampening since the Q values for the loaded thick and thin samples were smaller than their virgin counterparts by 11% and 44%, respectively. Third, and on the contrary, supercooling the thick sample resulted in a significant decrease in dampening, manifested in an increase of 10% and 23% in Q values comparative to virgin and mechanically loaded samples. The submersion in cryogenic temperatures is thought to freeze the chains, increasing the bond stiffness while reducing energy dissipation due to dampening. Future research can emphasize the interrelationship between the decrease in temperature and bond stiffness and recrystallization mechanisms using complementary methods such as X-ray diffraction, Raman Spectroscopy, and molecular dynamic simulations.

While analysis of the THz time-domain signals revealed important preliminary attributes of the samples, frequency-domain analysis is typically applied to extract the complex refractive index as a function of frequency spanning over the bandwidth. The bandwidth is selected based on the power spectral response of the THz antennas as well as the stability of the frequency transformation algorithm. Here, the frequency range for the thick and thin samples were 0.4–1.3 THz and 0.7–1.3 THz, respectively. The range was truncated during the material extraction process to ensure algorithm stability and robustness. Figure 5(a) shows the real part of the refractive index while Fig. 5(b) is a plot of the absorption coefficient of the virgin, thermally, and mechanically loaded samples in two ranges of thicknesses. As concluded from the time-domain analysis, thermal loading results in an indifferent real part of the index compared to the virgin, where the difference is less than 0.2%, on average over the entire frequency range, for both thick and thin samples. Regardless of the testing conditions, the real part of the refractive index monotonically decreases with increasing frequency, while the absorption coefficient increases as a function of frequency. The frequency dependence of the refractive index is consistent with the description given in Eqs. (1) and (2), respectively, since the thickness was kept constant throughout the characterization using THz-TDS. On average, the refractive index of the thick and thin polyurea samples is relatively similar, where the latter is merely 0.02% greater than the former. In other words, the small difference between the index of refraction of the thick and thin polyurea samples indicates that they are thickness-independent in this scenario. The small variance in the index components values for a different regiment of samples at a given frequency is attributed to the threefold difference in thickness, as discussed before. Generally, the dispersion and attenuation of waves in a media are strongly coupled to the thickness through the delay time of arrival, commonly captured by the quality factor, as discussed earlier. The values of the index of refraction calculated from the frequency-domain analysis (Fig. 5) are in good agreement with the time-domain analysis discussed earlier.

The virgin thick polyurea samples reported spectral peaks at 0.56, 0.76, 0.95, 1.10, 1.17, and 1.21 THz, while the thin samples only showed peaks at the last three values. Zhao et al. investigated the intermolecular vibrational modes in crystalline urea in the terahertz regime, within the same range as the study herein, and observed three absorption peaks at 0.69, 1.08, and 1.27 THz [53]. Figure 6 compares the chemical structure between crystalline urea and urea-linkage in the aromatic polyurea investigated in this study. The latter is connected to a larger structure via aromatic benzene rings instead of hydrogen bonds, which may account for some of the spectral peaks that are seen in polyurea but not in the crystalline urea. Nonetheless, the good agreement of the 0.76, 1.10, and 1.21 THz absorption bands to the corresponding bands seen in Ref. [53] supports that those features seen in the polyurea refractive and absorbance spectra are ascribed to the urea linkages of the monomer chain. It is important to note that the 0.56, 0.76, and 0.95 peaks in the spectra of the thick virgin and supercooled samples are no longer apparent in the stretched sample. It is speculated that the molecular vibrational modes at the abovementioned frequencies are no longer visible due to the damping, excitation, or restriction of motions of the bonds as a result of the physical strain induced on the polyurea. Based on cluster calculations and the potential energy distributions analysis, the peaks at 0.69 and 1.08 THz are attributed to the bending and twisting of bonded urea crystals, respectively [53]. It is reflected that the elimination of the 0.76 THz feature can be attributed to the limitation of the bending mode of the urea linkages, thus, making that vibrational mode invisible in the stretched polyurea spectrum.

Polyurea has and continues to capture scientific interest, sparking computational, analytical, and experimental research to elucidate the property–structure–performance relationship. Polyurea samples with different thicknesses were mechanically or thermally loaded and interrogated with terahertz radiation to explicate the residual effect of the high-impact mitigating polymer after loading. The characteristics of the resulting averaged time-domain signals of six samples were used to calculate the change of optical and electrical properties of loaded and unloaded polyurea as a function of loading conditions and sample thickness. Such characteristics include the shift in time delay and change in amplitude of the main peak signal, which can be used to estimate the real refractive index and absorption coefficient of the investigated material. The quality factor was also used to relate the changes in the properties with loading conditions collectively. The dielectric constant extracted from the terahertz time-domain analysis was found to be in reasonable agreement with previous research [47,48]. To increase the utility of the collected data, the frequency spectra of the terahertz signals were used to calculate the frequency-dependent refractive index, reporting several spectra peaks associated with the fundamental vibrational modes of the investigated polyurea structure. These spectral peaks were in good agreement with previous investigations [53] and pointed toward the molecular structure change after mechanical loading. The latter was shown as the disappearance of the low-terahertz spectral peaks after large-strain mechanical loading. In all, the results reported here provide additional insights into the residual stress on the behavior of polymers while paving the way for more scientific investigations using terahertz waves for in situ and in operando testing.

The research leading to these results was supported in part by the United States Department of Defense under Grant Agreement No. W911NF1410039 and W911NF1810477. The authors are grateful for the guidance of Dr. Roshdy Barsoum of the Office of Naval Research. The research was also supported by the National Science Foundation under Award No. 1925539.

There are no conflicts of interest.

The data sets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request.