Uncooled infrared focal plane array (IRFPA) is the core component of uncooled infrared detectors, which can be subdivided according to different detection mechanisms: thermistor type, pyroelectric type, thermopile type, diode type, field effect tube type, Optomechanical type etc. Among them, the diode-type IRFPA is conducive to achieving high array uniformity and high spatial resolution because the manufacturing process is compatible with the complementary metal-oxide-semiconductor (CMOS) process.
As the pixel size continues to shrink, the diode is the core sensitive element of the diode-type IRFPA. On the one hand, its electrical parameters affect the design of the readout circuit; on the other hand, its performance directly determines the noise equivalent temperature difference (NETD) of the device. ). For this reason, how to optimize the diode in the case of small pixels has become the focus of research. It is necessary and important to comprehensively investigate the influence of diodes on circuit and device performance, and then guide the design of diode structures in small pixels.
According to Memes Consulting, recently, the joint scientific research team of Institute of Microelectronics, Chinese Academy of Sciences, University of Chinese Academy of Sciences and Wuxi Internet of Things Innovation Center Co., Ltd. Diode structure optimization research" as the theme of the article. The first author of the article is Qu Fan, who is mainly engaged in the research of key technologies of intelligent MEMS infrared sensors; the corresponding author is associate researcher Fu Jianyu, who is mainly engaged in the research of key technologies of intelligent MEMS infrared sensors and thermal detectors.
In this paper, for the diode-type uncooled infrared focal plane array, the influence of the sensitive element diode on the readout circuit and device performance is theoretically analyzed. While determining the optimal operating current of the diode, the number of series connection and the junction area in the diode structure are extracted as the main parameters. performance-influencing factors.
Working Principle
Taking a diode with p-region and n-region doping concentrations of 10¹⁶ cm⁻³ and 10¹⁸ cm⁻³ as an example, Figure 1 shows the temperature characteristic curve simulated by Sentaurus TCAD simulation software. It can be seen from the figure that as the temperature increases, the diode I-V curve shifts to the left. When the diode is biased with a constant current, the voltage across the diode will decrease with the increase of temperature, thus realizing the conversion of thermal signal to voltage signal.
Figure 1 Temperature characteristic curve of diode
The schematic diagram of the structure of a single pixel in IRFPA is shown in Figure 2, and the sensitive area is suspended above the thermal insulation cavity through the support beam. The sensitive area consists of an infrared absorbing layer and a heat sensitive element, namely a diode connected in series. When the infrared absorbing layer absorbs infrared radiation and converts light energy into heat energy, the output voltage of the series diode under constant current bias changes, and the changed voltage is connected to the readout circuit through the signal line in the support beam, and finally realizes detection.
Figure 2 Schematic diagram of a single pixel structure
Circuit performance
Fig. 3 is a relationship curve of total capacitance Ca and leakage current Is as a function of summary area Aa obtained according to relevant theoretical formulas. It can be seen from the figure that the total capacitance and leakage current increase with the increase of the junction area. When the summed area is less than 1000 μm², the leakage current value is below 1.7 nA, and the capacitance value is kept below 120 nF, both of which are kept in a very small magnitude range, so the impact on the readout circuit is small.
Figure 3 Theoretical relationship between total capacitance Ca, leakage current Is and junction area Aa
Device Performance
Device performance mainly includes: NETD and response time. Since the response time is mainly determined by the thermal parameters of the structure, the diode, as a thermally sensitive element, mainly affects NETD. NETD is the temperature change when the temperature of the detected target changes in the field of view of the imaging system with the focal plane array (FPA), causing the output signal-to-noise ratio and the minimum unit change of the readout circuit signal.
Figure 4 shows the relationship between the response-to-noise ratio f₀ and the forward bias current If. It can be seen from the figure that with the increase of If, f₀ first increases and then decreases, and there is a maximum value at 10 μA. It is mainly due to the fact that the noise voltage Vn decreases first and then increases with the increase of If. Therefore, 10 μA is the optimal forward bias current for diode operation.
Figure 4 Theoretical relationship between response noise ratio f₀ and bias current If
Assuming that the junction area of a single diode in the structure is Ai, Fig. 5 shows the change trend diagram of TCV with the junction area Ai of a single diode and the number N of diodes in series in the structure. It can be seen from the figure that when N is constant, increasing the junction area of the diode will increase the TCV; when the junction area is constant, increasing the N value will also increase the TCV, and the effect is more obvious than increasing the Ai value.
Fig.5 Variation trend of TCV with the junction area Ai and the number N of diodes in series
As shown in Figure 6, six diode structures were investigated, namely: (a) traditional diode structure, (b) back-type diode structure, (c) p⁺n-n⁺p two-in-one diode, (d) and (e ) are the 2n⁺p-p⁺n three-in-one diode and the 2p⁺n-n⁺p three-in-one diode directly expanded on the basis of the p⁺n-n⁺p two-in-one diode, and (f) p⁺n-pn-n ⁺p 3-in-1 diode. Among them, structure b was proposed by Zhang Qiang and others in our research group; structure c was proposed by Mitsubishi; structure d is based on structure c, adding an n-type heavily doped region structure, consisting of p⁺n, n⁺p, n A three-in-one diode structure composed of ⁺p; structure e adds an n-well and p-type heavily doped region structure on the basis of structure c, forming a three-in-one diode structure composed of p⁺n, p⁺n, and n⁺p Diode structure; structure f is to add an n well to the region between the n well and the n-type heavily doped region of structure c, and form a new pn junction with the p-type Si substrate, thus forming p⁺n, pn, n ⁺p The structure of three diodes in series, the n-type heavily doped region in the n-well is used to form an ohmic contact. According to this method, it can continue to expand on the basis of structure f to form an N-in-one diode.
Figure 6 Diode structure: (a) traditional diode; (b) loopback diode; (c) p⁺n-n⁺p two-in-one diode; (d) 2p⁺n-n⁺p three-in-one diode; (e) 2p⁺n-n ⁺p three-in-one diode; (f) p⁺n-pn-n⁺p three-in-one diode.
As discussed above, when the diode is working under the optimal constant current bias condition, the larger the summed area Aa will increase the leakage current and capacitance, but the increased leakage current and capacitance level will be smaller, which will affect the readout circuit. The effect of Aa is small, so the increase of Aa mainly affects the improvement of TCV. Here, let the ratio of the pn junction summary area Aa to the pn junction overall size Sa be the effective junction area ratio Z,
Simulation Data
Using Sentaurus TCAD simulation software to simulate the above six diode structures, the doping concentration of the n-type and p-type heavily doped regions is 10¹⁸ cm⁻³, and the doping concentration of the n-well and p-type silicon is 10¹⁶ cm⁻³.
Figure 7 shows the simulation results of TCV and quiescent operating voltage of six structures under the forward current bias of 10 μA. Figure 7(a) shows the TCV values of the six structures at different temperatures, from large to small are structures f, c, e, d, b, a. For structure d and structure e, although the N value is 3, the TCV values of the two are close to and slightly smaller than that of structure c with N=2. According to Figure 7(b), the forward voltage of the six structures varies with temperature It can be seen from the change curve that the forward voltage values of structures d and e are closer to those of structure c, and differ from the voltage value of structure f by about 1 diode voltage difference (about 0.6 V). When the temperature range is small, TCV can be regarded as a constant value. The TCV values of structure f are about 1.5 times that of structures c, d and e, 2.6 times and 3.7 times that of structures a and b, respectively. It is proved that among the six structures, the performance of structure f is the best. If the N-in-one diode structure is obtained on the basis of the structure f, the TCV value will also increase accordingly.
Figure 7 Sentaurus device simulation
The reason why structures d and e increase the number of series diodes compared with structure c, but the forward voltage and TCV exhibited do not increase with the increase of the number of series connection as theoretically analyzed, because in these two structures, Additional parallel resistors and parasitic transistors are introduced respectively. Figure 8(a) is the equivalent circuit diagram of structure d. The current all enters the p-type substrate region after passing through the p⁺n junction at the right end of the structure. The parasitic resistance r of the substrate under the region is small, and r is connected in parallel with the extended n⁺p₁ junction in the middle of the structure, so that most of the current flows out from the branch of the parasitic resistance r through the n⁺p₂ junction at the left end of the structure, resulting in n The ⁺p₁ junction does not function to elevate the structural TCV. Figure 8(b) is the equivalent circuit of structure e, the n-well at the right end of the structure, the npn-type parasitic transistor formed by the p-type substrate and the n-type heavily doped region at the left end of the structure, and the current passes through the p⁺n₁ junction at the right end of the structure The p⁺n₂ expanded through the middle reaches the substrate. At this time, the substrate potential VB is lower than the potential VC of the n-well at the right end of the structure, and higher than the potential VE of the n-type heavily doped region at the right end of the structure, so that the parasitic triode is turned on. Therefore, after the current flows through the p⁺n₁ junction, it flows out directly through the parasitic transistor, so that the p⁺n₂ junction does not play a role in improving the TCV of the structure.
Figure 8 Equivalent circuit diagram
Fig. 9 shows the simulation situation of TCV when the structure f decreases with the overall size. The X-axis represents the overall size area, and the Y-axis represents the size of the TCV value. It can be seen from the figure that the TCV decreases with the decrease of the overall size, which is due to the decrease of the effective junction area of the structure f due to the decrease of the overall size. However, after the junction area is reduced by about 700 μm², the TCV is only reduced by about 0.3 mV/K, that is, the structure f still has good performance under small pixel sizes.
Fig.9 TCV simulation results under different sizes of structure f
Based on the working principle of the diode in IRFPA, this paper analyzes the main structural factors of the diode from the perspective of circuit performance and device performance. When the overall junction area of the diode in the structure is within 1000 μm², the leakage current and parasitic capacitance of the diode are estimated Values kept within 1.7 nA and 120 nF have little effect on the circuit. The response-to-noise ratio f₀ has a maximum value near the bias current of 10 μA, so this current is the optimum bias current of the diode. TCV will increase with the increase of the number of diodes in series and the junction area, and the effect of increasing the number of diodes in series on TCV is obviously better than that of increasing the junction area. Therefore, the number N of series diodes in the structure and the overall junction area Aa are taken as the structural factors that mainly affect the performance of the device. Therefore, the p⁺n-pn-n⁺p three-in-one diode is designed. Through the comparative analysis of theory and Sentaurus TCAD simulation software, it is found that the p⁺n-pn-n⁺p three-in-one diode has the most diodes in the same area. The number and the largest junction area, and avoid the influence of parallel resistance or parasitic triode, so it is the optimal structure among the six diode structure designs. At the same time, expanding the N-in-one diode can further optimize the performance of the device, and provide a reference for optimizing the diode structure of a single pixel thermal element in IRFPA.
XINXIN GEM provides infrared windows for the uncooled infrared detectors, and we can package the uncooled infrared detectors according to customers' requirements.
Comments