Non-cryogenic infrared (IR) microbolometers represent a cutting-edge technology with numerous advantages at room temperature operation. New thermally isolated micro-nano architectures equipped with integrated temperature sensors should outperform their predecessors regarding various critical aspects, such as ultra-broadband detection from ultraviolet (UV) to terahertz radiation , increased responsivity, unwavering reliability, ultrafast response time, and a substantial reduction in device size. The introduction of large, two-dimensional (2D) detector arrays has further increased the complexity of the detector optimization. In response to these multifaceted challenges and the quest for non-cryogenic IR microbolometer technology, carbon nanotubes (CNTs) have emerged as highly promising candidates with broadband blackbody absorption , high resistance perpendicular to the CNT orientation , low lateral thermal conductivity , a high temperature coefficient of resistance (TCR) , and high mechanical stability .

Vertically aligned CNTs (VACNTs) are considered to be nearly ideal blackbody absorbers because of their ability to absorb light over a broad spectral range (0.2–200 μm), closely resembling the theoretical blackbody that absorbs all wavelengths of light . This performance is based on the high absorption of VACNTs coupled with their low reflectance over this broad spectrum. Vertical alignment and distribution of the CNTs are critical to achieving such efficient absorption, making VACNTs effective blackbody analogs for broad-spectrum applications . The device resistance perpendicular to the CNT orientation is mainly determined by the high junction resistances of the neighboring aligned CNTs . This configuration allows for high device resistance, low thermal conductivity, and high temperature coefficient of resistance, thus, enabling customized specific requirements for various applications. While the thermal conductivity along the CNT orientation is large, it decreases dramatically in the perpendicular direction . This thermal behavior varies significantly depending on structure, density, and fabrication of the CNTs. Single-walled CNTs (SWCNTs) have achieved thermal conductivities as high as 6600 W/mK in the CNT orientation, in sharp contrast to multiwalled CNT (MWCNT) bundles, which can have thermal conductivities as low as less than 0.1 W/mK perpendicular to the CNT orientation, making them viable as thermal insulators.

At the core of non-cryogenic IR microbolometers is the fundamental principle of absorbing incident IR radiation and converting it to heat. This conversion process results in a change in the resistance of the active material, which, in turn, is related to the temperature coefficient of resistance (TCR) of the material. The electrical conductivity of a vertically aligned CNT structure is defined by the intrinsic conductivity along the CNTs and the tunneling at the CNT junctions. The vertical electrical conduction is based on long conduction paths along the CNTs and few junction contacts, while the lateral electrical conduction is based on short conduction paths along the CNTs (length between junction contacts) and many junction contacts. Experimental studies have shown that the TCR is negative for aligned CNTs . The main factor influencing the TCR of a network is the TCR of the CNTs themselves. In particular, larger negative TCRs can be achieved if the conduction along the CNTs in the effective conduction path remains small, as is the case for lateral electrical conduction paths . VACNTs with mainly lateral electrical conduction paths can satisfy this condition. These properties make CNTs highly attractive options for non-cryogenic IR microbolometer and thermal detection technology.

The sensitivity of a bolometer is quantified by its responsivity, which is given by the expression :

where ΔV is the corresponding voltage change, ΔP represents the incident power on the active detection area, Ibias is the bias current, R is the resistance of the bolometer, α = (dR/dT)/R is the TCR, η is the absorption efficiency, G is the thermal conductance to the substrate, ω is the angular modulation frequency, τ = 1/(2πfcut−off) is the time constant of the detector, and fcut−off is the frequency at 70% responsivity. Based on this equation, efficient bolometers require high optical absorption, a high TCR, good thermal isolation from the environment (characterized by a small value of G), and a small time constant.

The research presented in this paper provides a 3D CNT design with micro-nano integrated lateral contacts. The lateral and vertical dimensions of the M-shaped CNT architecture can be adjusted independently to achieve low thermal conductivity, high lateral resistance, and small response time to achieve a higher bolometric effect. Our first responsivity measurements with similar CNT wall structures were presented in the conference proceedings . This configuration 1 (called sample 1) had additional parasitic contact resistances to the M-shaped CNT block. The new configuration 2 (called sample 2) used direct Au whisker contacts to the CNT block with much lower contact resistances for reliable measurements. The true performance of the original M-shaped CNT block with higher TCR, higher voltage, and higher current responsivity could be demonstrated. In this work, new noise measurements were performed and compared for both configurations. The new configuration 2 is superior regarding the optimization of the M-shaped bolometer design. The final design with a micro-nano integration using optimized contacts between the Cr/Au pads and the M-shaped CNT block can then be achieved with additional CNT blocks on the Cr/Au pads as shown in .

Furthermore, this research embarks on a comprehensive exploration of the potential for large-scale production of CNT-based non-cryogenic IR microbolometers and thermal detectors, promising cost-effective solutions ready for widespread adoption across multiple industries. It represents a significant step forward in addressing the growing demand for advanced thermal sensing solutions and sets the stage for transformative advances in non-cryogenic IR microbolometer and thermal detector technology. The following sections provide a comprehensive overview of the 3D micro-nano design, fabrication methods, performance characteristics, and potential applications. Simultaneously, emphasis will be placed on the effective conversion of heat into electrical signal, which is the most critical effect contributing to the overall success of non-cryogenic IR microbolometer and thermal detector design .

Figure 1: Fabrication process of the pixel-based CNT microbolometer. (a) AlOx deposition, (b) lithography process and metal contact evaporation, (c) iron catalyst evaporation, and (d) CNT growth.

Figure 2: (a) Scanning electron microscopy image and (b) 3D close-up of the pixel-based CNT microbolometer.

The resulting M-shaped VACNTs have an initial pixel dimension of 20 × 20 μm2 with a ≈1 μm thick CNT thermistor wall stabilized by complementary CNT block structures. Two strategically positioned metal electrodes on the Si substrate provide lateral contacts with the vertical CNT bolometer, facilitating effective resistance measurements. Importantly, this CNT structure exhibits remarkable thermal stability, retaining reliability at temperatures reaching up to 200 °C. This exceptional thermal resilience makes it ideally suited for demanding high-temperature applications, showing its potential across various industries.

Figure 3: (a) TEM image showing DWCNTs and (b) Raman spectrum of the VACNTs.

Compared to SWCNTs , for which the characteristic absorption peaks depend on the nanotube diameter, DWCNTs have a relatively flat spectral absorption due to the excitonic transition energies of the inner and outer tubes . The broadband absorption of our fabricated block structures has not yet been measured. However, the absorption of similarly grown VACNTs (water-assisted CVD on silicon substrates with ethylene as the carbon source) has been investigated by . They showed that vertically aligned SWCNTs can absorb light almost perfectly with a reflectance of 0.01–0.02 over a very wide spectral range (0.2–200 μm). The UV-to-mid-IR absorption characteristics showed little variation with CNT heights ranging from 2 to 460 μm . The authors attributed this blackbody behavior to the sparseness and imperfect alignment of the vertical SWCNTs, which suppressed reflection. Due to multi-reflection in the CNT forest and broadband absorption in the single CNTs , the effective reflection from the forest decreased very rapidly. MWCNTs have also been used as absorbers in the infrared region (2–20 μm) with an optical absorption coefficient above 90% . Similar values were achieved for different synthesis temperatures (500–700 °C) with different CNT diameters, qualities, and heights. So far, we have only investigated the IR response. Far-IR and THz measurements will be carried out in the future to prove the broadband property of our new bolometer architecture.

Figure 4: Schematic of the device under test (DUT) and measurement setups to characterize the CNT-based microbolometer. (a) Sample 1 with metal contact pads and (b) sample 2 with direct contacts using gold whiskers with a diameter of 20 μm. (c) Responsivity measurement setup using a preamplifier and (d) responsivity as well as noise measurement setup using a lock-in amplifier. Panels (a) and (b) do not represent the real dimensional relationships.

Figure 5: Temperature-dependent I–V curves (a) for sample 1 with contact pads and (b) for sample 2 using gold whisker contacts.

Figure 6: Resistance–temperature relationship (a) for sample 1 with contact pads and (b) for sample 2 using gold whisker contacts.

Figure 7: Response measurements for sample 1 using metal contact pads at (a) 90 Hz and (b) 1100 Hz (with an average of 10 measurements). The standard deviation of the peak σ is 18 mV.

Figure 8: Response measurements for sample 2 using gold whisker contacts at (a) 90 Hz and (b) 1100 Hz (with an average of 10 measurements). The standard deviation of the peak σ is 22 mV.

where ℜV is the voltage responsivity of the microbolometer. The noise voltages of sample 1 and sample 2 measured at 90 Hz were 40.6 and 16.6 μV, respectively. They were almost constant at chopper frequencies ≥60 Hz. The calculated NEP values for sample 1 and sample 2 were 3.12 and 0.55 μW·Hz−0.5, respectively. The decreased contact resistance in sample 2 with direct Au whisker contacts resulted in lower noise values, indicating the importance of the final contact resistance.

Table 1: Comparative analysis of microbolometer performance metrics. All values were obtained at room temperature (if not marked otherwise).

This work aims to advance the understanding and implementation of CNT-based non-cryogenic IR microbolometers, providing valuable insights for researchers, engineers, and industries seeking to harness the full potential of this new thermal sensing technology. A 3D micro-nano integrated CNT architecture has been utilized to develop a non-cryogenic IR microbolometer, enabling a minimum pitch dimension of less than 5 × 5 μm2 through a simple lithography process. This approach has significant potential for high-density integration. The device exhibited a rapid response time of ≈0.15 ms and a high negative TCR of −0.91 %/K. The new microbolometer operates reliably at temperatures even above 200 °C, demonstrating its robustness in extreme environments. We achieved a current (voltage) responsivity of 2 mA/W (30 V/W) to the absorbed IR power, which is already comparable in the context of current CNT-based microbolometer technology with high potential for miniaturized design and responsivity optimization. This study not only contributes to the current state of the art in microbolometer technology, but also opens avenues for further innovation and broader application in both scientific and consumer markets. Ongoing advances in CNT research and microbolometer technology hold the promise of efficient cost reduction as well as sensitive and versatile thermal detection devices suitable for a wide range of high-tech applications.