Design and Fabrication of Thermopile Infrared Detector Based on Carbon Black Nanoparticle Absorption Layer

Abstract

This study demonstrates a high-performance thermopile infrared detector that incorporates a carbon black nanoparticle (CBNP) absorption layer. To overcome the limitations associated with conventional infrared-absorbing materials—including high cost, complex fabrication, and constrained spectral response—a highly porous CBNP thin-film absorption layer was deposited onto the thermopile sensing area using inkjet printing. Combined with an optimized microcavity design, this approach significantly enhances the photothermal conversion efficiency of the device. Experimental results indicate that the detector equipped with the CBNP absorption layer achieves a responsivity of 47.9 V/W and a detectivity of 1.14 × 10⁸ cm·Hz¹/²·W⁻¹. These values represent improvements of 34.55% in responsivity and 34.28% in detectivity, respectively, compared to a reference device without the CBNP layer. This work provides a promising strategy for the development of low-cost yet high-performance infrared detectors.

1. Introduction

Infrared detection technology is of significant value in applications such as security surveillance, medical diagnostics, and industrial inspection [1,2,3]. Thermopile detectors are widely utilized as core components in non-contact infrared sensing due to their ability to operate in a cryogen-free environment, broad spectral response, and excellent stability [4,5,6]. However, their performance is often constrained by the efficiency and cost of the infrared absorption layer [7,8]. Conventional metallic thin films (e.g., platinum black) exhibit high absorption rates but require complex fabrication processes and are prone to environmental oxidation [9,10]. Organic polymers, meanwhile, are constrained by limited thermal stability and narrow response bandwidth [11].

In recent years, carbon-based nanomaterials have gained considerable interest owing to their exceptional photothermal properties [12,13]. Carbon black nanoparticles (CBNPs) demonstrate considerable potential for photothermal applications due to their high specific surface area, broad spectral absorption (from UV to far-infrared), and low cost [14]. Theoretical studies indicate that the porous structure of CBNPs enhances light trapping capabilities through multiple scattering effects [15,16]. Nevertheless, their integration into infrared detectors remains challenging, primarily due to difficulties in optimizing the absorption layer structure and achieving efficient thermal coupling with thermopile elements [17].

This study innovatively proposes the integration of CBNPs as an infrared absorption layer in thermopile detectors. Combining microcavity resonance effects with a nanoporous structure, the proposed design significantly enhances photothermal conversion efficiency compared to conventional materials [18]. Comparative analyses of pure ink and CBNP-doped ink formulations were conducted to evaluate their impact on device performance. The experimental results validate that the proposed detector exhibits high responsivity, detectivity, process compatibility, and environmental stability, providing a feasible pathway for the low-cost and scalable manufacturing of infrared detectors.

2. Structural Design

A thermopile is a thermoelectric energy conversion device based on the Seebeck effect, composed of multiple thermocouple units connected in series. Each unit consists of two thermoelectric materials with different conduction types—typically p-type polysilicon and aluminum. When a temperature gradient exists between the hot and cold junctions of a thermocouple, carriers (electrons and holes) diffuse from the hot region to the cold region, generating a potential difference proportional to the temperature gradient. The mathematical expression for the potential difference produced by the thermopile is as follows [19,20,21]:

$$U=N\cdot \mathsf{\Delta }T\cdot {\alpha }_{\mathrm{a}\mathrm{b}}$$

(1)

where U is the potential difference generated by the thermopile, N is the number of thermocouple pairs, $\mathsf{\Delta }T$ is the temperature difference between the hot and cold ends, and ${\alpha }_{\mathrm{a}\mathrm{b}}$ is the difference in the Seebeck coefficients of the two thermoelectric materials. According to Equation (1), the potential difference depends on the Seebeck coefficient difference of the paired materials and the magnitude of the applied temperature gradient.

In addition to the output voltage, the performance of a thermopile detector is usually characterized by the responsivity R_v and detectivity D*.

Responsivity (R_v) is defined as the ratio of the output voltage of a thermopile detector to the incident infrared radiant power, which represents the detector’s ability to convert thermal energy into electrical signals. The expression for the response rate is as follows [22]:

$$R_v={\displaystyle \frac{\mathsf{\Delta }U}{P_0}}={\displaystyle \frac{\mathsf{\Delta }U}{{\phi }_0{A}_d}}$$

(2)

where P₀ is the infrared radiation power incident on the detector, A_d is the area of the absorption zone of the thermopile chip, and φ is the infrared radiation power density, which is expressed as follows:

$${\phi }_0={\displaystyle \frac{C_r\cdot \sigma \cdot {\epsilon }_1\cdot ({T_1}^4-{T_0}^4)\cdot A_s}{\pi \cdot {d_0}^2}}$$

(3)

where C_r is the root mean square factor of the chopper, σ is the Stefan-Boltzmann constant, ε₁ is the emissivity of the radiation source, T₁ is the temperature of the infrared radiation source, T₀ is the ambient temperature, A_s is the infrared radiation source area, and d₀ is the distance between the thermopile chip and the infrared radiation source.

Detectivity (D*) is defined as the responsivity normalized by the root-mean-square noise voltage, referred to a unit bandwidth and unit detector area. This parameter characterizes the detector’s ability to detect weak infrared signals—a higher detectivity indicates a better capability to detect faint signals. The detectivity is expressed as [23,24,25]:

$$D^{\mathrm{*}}={\displaystyle \frac{R_v}{V_n}}\cdot \sqrt{A_d\mathsf{\Delta }f}$$

(4)

where Δf is the bandwidth and V_n is the noise voltage, derived from thermal and Johnson-Nyquist noise components. The noise voltage is given below [26]:

$$V_n=\sqrt{4k{R}_0{T}_0\mathsf{\Delta }f}$$

(5)

where k is the Boltzmann constant and R₀ is the electrical resistance of the thermopile.

The MEMS thermopile chip presented in this work employs a suspended membrane structure. The overall chip dimensions are 1800 μm × 1800 μm × 400 μm, with a suspended membrane radius of 600 μm. The thermoelectric materials are configured such that one end is located on the suspended membrane (hot end) and the other end is anchored to the silicon substrate (cold end), thereby establishing a temperature gradient [27,28]. In order to maximize the performance of the thermopile and minimize the resistance of the thermopile, in this study, p-type polysilicon and aluminum with a length of 553 μm were employed as thermoelectric materials.

A schematic diagram of the MEMS thermopile structure is illustrated in Figure 1. Seventy-two thermocouple pairs are arranged radially on the suspended membrane, with a central circular region serving as the infrared absorption area. Carbon black nanoparticles (CBNPs) are prepared in the infrared absorption region to absorb infrared thermal radiation and transfer heat energy to the thermocouple hot ends. The integration of CBNPs enhances the temperature difference between the hot and cold ends, thereby increasing the output voltage of the thermopile.

3. Fabrication

3.1. Fabrication of CBNPs

The carbon black nanoparticle (CBNP) suspension was primarily composed of 50 nm carbon black nanopowder. Several formulations with varying mass ratios of carbon black to ink were prepared. The formulation results indicated that at a mass ratio of 1:10 (carbon black to ink), the ink was insufficient to fully submerge the carbon black powder. Initial stirring resulted in agglomeration of the carbon black, which exhibited high volatility and rapid solidification, leading to a cement-like mixture unsuitable for further processing. At a mass ratio of 1:12, the black ink barely submerged the carbon black powder. Although the agglomeration was mitigated, slight clustering persisted after adequate stirring, and the solidification time remained relatively short. Thus, further increasing the ink proportion was necessary. When the mass ratio reached 1:15, the ink completely submerged the carbon black, forming a liquid mixture. After sufficient stirring, no agglomeration was observed, and the solidification time extended due to the higher ink content. Therefore, the optimal formulation for the drop-casting solution was determined to be a carbon black-to-ink mass ratio of 1:15. Figure 2 displays the three mass ratio formulations.

In order to facilitate integration with the thermopile, carbon black nanopowder was mixed with oily black ink at the optimal 1:15 mass ratio. This combination improves adhesion between the CBNPs and the thermopile, while the black ink also enhances the infrared absorption rate of the CBNPs [29,30,31]. To ensure uniform dispersion, the mixture was sonicated at 30 °C for 5 min to achieve complete incorporation of the carbon black into the ink. The solution was then centrifuged sequentially at 800 rpm for 90 s, 1200 rpm for 60 s, and 2000 rpm for 30 s using a stirring and deaeration device, resulting in the final CBNP drop-casting solution. Scanning electron microscopy (SEM) images of the CBNP layer (Figure 3) reveal a highly porous and rough microstructure, which considerably enhances the infrared absorption rate of the device. The surface-to-volume ratio of the porous network transforms the coating into an efficient light-trapping medium. Incident radiation undergoes multiple internal reflections and scattering within the nano-pores, significantly prolonging the effective optical path length and maximizing the probability of energy absorption [32].

To characterize the infrared absorption performance of the CBNPs, three samples measuring 3 × 3 cm were prepared. To simulate the film covering the thermocouple hot junctions as closely as possible, all samples were fabricated on 400 μm silicon substrates coated with 300 nm silicon nitride. Sample #1 had no coating applied; Sample #2 was coated with ink only; and Sample #3 was coated with the CBNP mixture. Photographs of the three samples are presented in Figure 4.

The infrared absorbance of the samples was measured using a Frontier infrared spectrometer across the 2–20 μm wavelength range. The results are shown in Figure 5.

The results indicate that Sample #1 exhibits some infrared absorption capability, though with limited efficiency. With the addition of an infrared-absorbing ink layer on the silicon nitride surface, Sample #2 demonstrates significantly enhanced absorption, reaching 30–60% across the 2–20 μm infrared band, confirming the ink’s intrinsic infrared absorption properties. Sample #3, which features a composite layer of silicon nitride and carbon black nanoparticles (CBNPs), achieves an average absorption rate exceeding 75% over the 2–20 μm band, with values generally above 90% in the 8–12 μm band. This notable improvement is attributed to the porous, high-surface-area microstructure formed by the CBNPs, which enhances light trapping through multiple internal scattering and absorption mechanisms.

3.2. Fabrication of Thermopile

The fabrication process of the thermopile infrared detector with a carbon black nanoparticle absorber layer is illustrated in Figure 6. First, a 510 nm silicon oxide layer and a 240 nm silicon nitride layer were deposited on a silicon substrate to serve as the supporting layers. Next, a 750 nm polysilicon film was deposited on the support layer and doped to form p-type polysilicon, which was subsequently patterned into thermocouple strips via photolithography and etching. Thermocouple structures and metal electrodes were then formed using photolithography, sputtering, and lift-off processes, followed by annealing to establish ohmic contacts. A silicon nitride isolation layer, an aluminum reflective layer, and a silicon nitride passivation layer were successively deposited over the thermocouple layer. Contact windows were opened by etching to expose the aluminum electrodes. The suspended membrane was released by deep reactive ion etching (DRIE) of the silicon substrate from the backside. Finally, the prepared CBNP drop-casting solution was applied to the central absorption region of the thermopile. For comparison, a reference MEMS thermopile sensor without CBNPs was also fabricated.

Figure 7a presents an optical micrograph of the fabricated thermopile chip, which measures 1.8 × 1.8 mm. The dark circular region at the center corresponds to the deposited CBNP layer. Figure 7b shows a fully assembled detector based on the CBNP absorber, packaged in a TO-46 housing for subsequent performance testing. A packaged reference device without CBNPs is also shown in Figure 7c.

3.3. Measurement and Characterization

A dedicated test system was constructed to evaluate the performance of the thermopile detector, comprising a blackbody radiation source, a test mount, a high-precision multimeter, and data acquisition software. The system was used to characterize the I–V characteristics, temperature–voltage (V–T) response, and output voltage of the MEMS thermopile infrared detector. A schematic of the test setup is shown in Figure 8.

To compare the performance of detectors with and without the CBNP layer, all tests were conducted at a stable ambient temperature of 25 °C. The blackbody source was set to 227.3 °C (500 K), and the detector was positioned 3 cm away from the source. Environmental conditions were monitored using calibrated temperature and humidity meters. Both detector types were tested under identical data acquisition settings. The I–V characteristics are presented in Figure 9, and the dynamic response behavior is shown in Figure 10.

As shown in Figure 9, the electrical resistance of the thermopile remains virtually unchanged after the addition of the CBNP layer. This is expected since the CBNPs are not in electrical contact with the thermocouple materials and serve only to enhance infrared absorption. However, Figure 10 reveals that the CBNP layer introduces additional thermal mass and modifies the heat conduction path, increasing the overall thermal resistance and capacitance of the sensing structure. As a result, the dynamic response time of the detector with CBNPs is longer, indicating a trade-off: the enhanced infrared absorption and responsivity are achieved at the cost of a slightly increased temporal response.

To evaluate the output voltage variation with temperature, the blackbody temperature was increased from an initial value of 30 °C in 5 °C increments. The detector was maintained at a fixed distance of 3 cm from the blackbody, and the output voltage was recorded at each temperature step. The resulting V–T response is shown in Figure 11.

The output voltage increases with rising blackbody temperature, consistent with the higher infrared power density incident on the detector surface. Moreover, at each temperature, the output voltage of the detector with the CBNP layer is significantly higher than that of the detector without CBNPs.

4. Discussion

The output voltage, responsivity, and detectivity of the thermopile detector were characterized before and after the incorporation of the CBNP layer. As summarized in

Table 1. Comparison of detector performance.

Parameter	Unit	Without CBNPs	With CBNPs	Improvement
Output Voltage	mV	17.049	22.926	34.47%
Responsivity	V/W	35.6	47.9	34.55%
Detectivity	cm·Hz1/2·W−1	0.849 × 108	1.14 × 108	34.28%

, the detector with the CBNP absorption layer achieved an output voltage of 22.926 mV, a responsivity of 47.9 V/W, and a detectivity of 1.14 × 10⁸ cm·Hz¹/²·W⁻¹. Compared to the detector without CBNPs, the responsivity and detectivity were enhanced by 34.55% and 34.28%, respectively.

A comparison with existing infrared detectors indicates that the proposed sensor exhibits superior performance over the other two sensors, as summarized in

Table 2. Comparison of key performance metrics among the three sensors.

Parameter	Unit	With CBNPs	[33]	[25]
Responsivity	V/W	47.9	14.522	34.2
Detectivity	cm·Hz1/2·W−1	1.14 × 108	1.59 × 107	1.02 × 108
Response Time	ms	34.88	/	26.9

.

As shown in Table 2, the thermopile detector developed in this work demonstrates superior responsivity and detectivity relative to other devices. Although its response time is slightly longer than that of the device listed in the third column of Table 2, it remains within the same order of magnitude. Furthermore, from a fabrication standpoint, the CBNP-based detector offers a comparatively simpler process than devices utilizing structures such as doped polysilicon nanocones. The latter relies heavily on semiconductor fabrication equipment, resulting in complex procedures and higher associated costs. In contrast, the raw materials employed in this work are low-cost and widely available, while the CBNP-based detector delivers significant performance improvements. These advantages in manufacturability and performance highlight the potential of the CBNP-based approach for scalable production.

5. Conclusions

In this study, a circular thermopile infrared detector was successfully designed and fabricated using MEMS technology. By incorporating a carbon black nanoparticle (CBNP) layer as an infrared absorber, the detector’s photothermal conversion efficiency was significantly improved. The CBNP–ink composite was drop-coated onto the suspended membrane of the thermopile, leading to a notable increase in infrared absorption and thus a greater temperature difference between the hot and cold ends. The resulting detector achieved an output voltage of 22.926 mV, a responsivity of 47.9 V/W, and a detectivity of 1.14 × 10⁸ cm·Hz¹/²·W⁻¹. Compared to a reference device without the CBNP layer, the responsivity and detectivity were improved by 34.55% and 34.28%, respectively. This work demonstrates that CBNP-based absorption layers offer a practical and effective route for developing high-performance, low-cost MEMS thermopile infrared detectors.