A porous silicon JFET gas sensor: experimental and modeling G. Barillaro, G. M. Lazzerini, L. M. Strambini Dipartimento di Ingegneria dell’Informazione: Elettronica, Informatica, Telecomunicazioni Università di Pisa Via G. Caruso, 56126, Pisa, Italy
[email protected] the surrounding environment. On the other hand, an effective modeling would greatly help the comprehension of the physical/chemical phenomena behind gas sensors and assist in the sensor optimization, in terms of sensitivity, selectivity, reliability, life time, but also, size, power dissipation, etc.
Abstract—In this work, electrical measurements and modeling of an integrated porous silicon (PS) JFET (PSJFET) gas sensor are reported. The experimental IDS-VDS curves were measured for different VGS voltages, both in synthetic air and in presence of 300 ppb of NO2. The modeling was carried out by taking both the PS layer and the FET structure into account. By best fitting the experimental data with the given model, a quantitative evaluation of the effect of NO2 on both the PS layer and the FET structure was performed. Firstly, an indirect estimation of the PS resistance values in air (RPS_AIR~31 kΩ) and in NO2 (RPS_NO2~3.2 kΩ) was achieved, in agreement with data reported in literature and acquired by direct electrical measurements on PS. Moreover, it was also estimated that exposure to 300 ppb of NO2 resulted in a reduction of the FET channel section of ~0.2 μm, with respect to air.
I.
PS is one of the most interesting materials for the fabrication of integrated gas sensors for environmental monitoring. For instance, its conductance changes of several order of magnitude when exposed to NO2 at a concentration of hundreds ppb [5, 6]. Recently, it has been demonstrated that PS can be also exploited to modify the electrical properties of an integrated solid-state device, such as a FET structure, upon exposure to NO2 [7-8]. For example, in Ref. 8 an integrated p-channel JFET was modified by integration of a PS layer between its drain and source terminals (Porous Silicon JFET - PSJFET). This approach allowed to integrate PS-based sensors together with electronic circuits by using an industrial BCD (Bipolar+CMOS+DMOS) process, as reported in Ref. 3.
INTRODUCTION
In last years a great effort has been devoted to the development of integrated, CMOS-compatible gas sensors for environmental applications. A number of sensing materials, i.e. metal oxides [1], polymers [2], porous silicon [3], carbon nanotubes [4], etc., are today under investigation. The challenge is the fabrication of small size, low cost, high reliable integrated sensing-platforms containing both the sensors and the necessary driving-readout electronic circuits on the same chip. The difficulty is related to: i) the actual compatibility of both sensing materials and sensor structures with industrial CMOS processes and, in turn, to their integration within a CMOS flow; ii) the modeling and simulation of both materials and structures and, thus, to the actual knowledge of the chemical/physical principles ruling the sensor behavior.
In this work, the electrical characterization and modeling of a PSJFET sensor are reported. The experimental IDS-VDS curves were measured for several VGS voltages, both in synthetic air and in the presence of NO2 at a concentration of 300 ppb. The modeling was performed by taking both the PS layer and the JFET structure into account. By best fitting the experimental curves with the theoretical model, a quantitative evaluation of the effect of the NO2 on the PS layer (resistance value) and on the JFET structure (p-channel section) was performed, with respect to air. II.
The PSJFET is an integrated p-channel JFET having two independent gates: 1) an electrical gate, that is the n-type substrate; 2) a floating, sensing gate, that is a PS layer between the drain and source terminals. A schematic crosssection of the PSJFET, along with the typical polarization conditions used for measurements, is shown in Fig. 1. Actually, changing the reverse polarization of the n-type
In fact, despite the huge number of papers published on gas sensors, only a few cases of CMOS sensing platforms integrating both gas sensors and electronic circuits have been reported so far [1-3]. Moreover, the sensor modeling is often overlooked, mostly because of the intrinsic difficulty of modeling both the sensing material and its interaction with
1-4244-2581-5/08/$20.00 ©2008 IEEE
WORKING PRINCIPLE AND FABRICATION
494
IEEE SENSORS 2008 Conference
substrate is a feasible way to modulate the depletion zone w2 of the p-n junction and, in turn, the actual section of the pchannel. On the other hand, adsorption of molecules in the PS layer modifies the PS unbalanced charge – as a function of the pollutant concentration – and, once more, the pchannel section, as a result of the space-charge width modulation w1 at the p-Si/PS interface.
III.
ELECTRICAL MEASUREMENTS
The flow-through technique was used to test the PSJFET behavior both in synthetic air and in NO2, the latter at a concentration of 300 ppb. Synthetic air was used as carrier gas. The electrical characterization was carried out by using a source-measure unit to record the IDS-VDS curves of the sensors. The polarization voltage VDS was varied between 0 V and -5 V, with step of 0.1 V, while simultaneously monitoring the IDS current flowing between drain and source. The effect of the electrical gate on the sensor current was evaluated by changing the VGS voltage between 0 V and -2 V, with step of 0.5 V. All the measurements were performed
The fabrication process of the PSJFET is sketched in Fig. 2 and consists of the following main steps: 1) boron implantation and diffusion (in an n-type silicon substrate with doping of 1015 cm-3) to form a p-type layer, 4 μm deep with a surface doping of 1017 cm-3 (Fig. 2a); 2) aluminum deposition on the front-side and patterning (1st mask) to define two interdigitated electrical contacts for the drain and source terminals; 3) aluminum evaporation on the back of the sample to define the electrical gate contact (Fig. 2b); 4) source and drain contacts protection by using a photoresist mask (2nd mask) hardbaked at 140 °C for 30 minutes; this step is essential in order to make the resist layer able to withstand the electrochemical etching (anodization) in HF, which is necessary to the PS formation (Fig. 2c). The final step was the selective anodization of the p-type material through the photoresist-free spaces to produce the PS layer. A schematic section of the sensor at the end of the fabrication sequence is shown in Fig. 2d. The thickness of the p crystalline layer underneath the PS, and thus the section of the p-channel, depends on the anodization time. On each chip several sensors with different dimensions were integrated.
a)
The whole fabrication process is straightforward and requires only two masks and few technological steps. The more critical step is undoubtedly the PS layer formation but: i) it is the last step of the process and ii) it was already demonstrated that such a step is compatible with industrial integrated circuit (IC) processes [3].
b)
c)
d) Figure 2. Fabrication process: a) p-type dopant implantation and diffusion; b) aluminum deposition and patterning for source, drain and gate terminal formation; c) photoresist deposition and patterning for source and drain terminal protection; d) porous silicon (PS) formation throughout the photoresist-free spaces and photoresist removal.
Figure 1. Schematic cross-section of a PSJFET sensor along with its polarization voltages. The two space-charge regions occurring in the pchannel – due to the unbalanced charge in the PS layer (w1) and at the p/n junction (w2) – are also highlighted.
495
in a temperature-stabilized sealed chamber, at room temperature.
gives rise to an increase of the sensor current, for any VDS and VGS value. For instance, for VDS= -5 V and VGS= 2 V a variation of the IDS current of about an order of magnitude occurs upon injection of 300 ppb of NO2, with respect to synthetic air. Moreover, the saturation of the FET structure in NO2 is not as much evident as in air.
Fig. 3 shows some typical experimental IDS-VDS curves of a PSJFET in synthetic air, for different VGS voltages. According to the behavior of a p-channel JFET: the currentvoltage curves show a linear behavior for low VDS voltages (linear region) and a saturation for high VDS voltages (saturation region), for any gate polarization voltage; moreover, the conduction current reduces as the absolute voltage value of the gate increases.
IV.
MODELING THE PSJFET
The behavior of the PSJFET in air and in the presence of NO2 can be explained if one supposes that NO2 adsorption of in PS produces: i) a variation (reduction) of the unbalanced charge of the PS-itself and, in turn, of the depletion region w1 at the PS/p-channel interface; and/or ii) a variation (increase) of the PS conductance, according to literature data [5, 6]. Both these effects give rise to an increase of the sensor current, with respect to synthetic air. The depletion region w1 at the PS/p-channel interface (see Fig. 1) can be thought as due to a positive unbalanced charge in the PS layer. Therefore, a reduction of the unbalanced charge in the PS layer, as a consequence of NO2 adsorption, gives rise to an enlargement of the actual cross-section of the p-channel of the JFET and, thus, to an increase of the IDSJ FET current. On the other hand, the increment of the PS conductance upon exposure to NO2, with respect to air, produces an increase of the IPS current flowing through the PS layer. Therefore, the PSJFET current IDS= IDSJ + IPS increases accordingly. It is worth noting that, the IPS current flowing throughout the PS layer can be supposed always lower than the IDSJ current of the FET structure, as the FET behavior (linear and saturation regions) is still visible both in air and in NO2. However, the less evident saturation of FET in NO2 is compatible with an increment of the IPS current upon NO2 exposure.
Fig. 4 shows the experimental IDS-VDS curves of the same PSJFET of Fig. 3 upon injection of 300 ppb of NO2 in the test chamber, for different VGS voltages. The sensor still behaves as a p-channel JFET, with both a linear and a saturation region, for any VGS value. However, by comparing Fig. 3 and Fig. 4 it is possible to infer that NO2 injection
In order to perform a quantitative analysis of the effect of the NO2 on the PSJFET, with respect to air, a simplified, model has been implemented. The PSJFET was schematized as a p-channel JFET with two series resistance RD and RS at the drain and source terminals, respectively. The JFET was modeled by assuming an abrupt p-n junction between the gate and both the drain and source terminals. The effect of the PS layer on the JFET was taken into account by modeling the PS layer as a floating gate with an unbalanced positive charge, the latter depending on the gas type and concentration in the measurement environment. The depletion region width w1 in the p-channel, depending on the PS unbalanced charge concentration, was supposed constant
Figure 3. Experimental IDS-VDS curves of the PSJFET for different VGS voltages in synthetic air.
Figure 5. Equivalent electrical circuit of the PSJFET.
Figure 4. Experimental IDS-VDS curves of the PSJFET for different VGS voltages in the presence of 300 ppb of NO2, using synthetic air as carrier gas.
496
and RPS are unknown parameters, and λ (whose value is zero for an ideal FET) takes the slope of the current-voltage curves in the saturation region into account. The values of RD and RS can be estimated on the basis of the device geometrical factors: RD=RS ≈ 50Ω, due to the device symmetry. A guess for λ can be deduced by extrapolating the interception of the saturation region curve in synthetic air with the voltage axis, under the assumption that the leakage current trough the PS layer in air is negligible with respect to the FET current IDSJ: λ ≈ 0.025 V-1.
along the FET structure (first order approximation). A complete depletion of the space charge region was supposed both at the PS/p-channel interface and at the p-n junction. Finally, in order to take account of the leakage current throughout the PS layer, a variable resistance RPS, whose value depends on the NO2 concentration, was also included in the model. The equivalent electrical circuit of the implemented model is reported in Fig. 5. On the basis of the equivalent circuit of Fig. 5 and of the physical assumptions made above, an analytical, closed form for the PSJFET current IDS can be found [9]. The sensor current resulted a non-trivial function of the following parameters: IDS=f(VDS, VGS, RD, RS, λ, w1, RPS), where only w1
Fig. 6 and Fig. 7 show the theoretical IDS-VDS curves of a PSJFET calculated by best fitting the experimental data with the proposed model. The calculated curves are in good agreement with the measured characteristics and allow to perform quantitative estimation on the effect of NO2 on both the PS layer and the FET structure. For instance, it was found that RPS value changes from RPS_AIR~31 kΩ in air to RPS_NO2~3.2 kΩ in the presence of 300 ppb of NO2. According to data reported in literature [5, 6], usually obtained by performing direct electrical measurements on PS, a resistance variation of one order of magnitude was obtained. Moreover, the sensor modeling also allowed for a quantitative evaluation of the PS unbalanced charge variation upon NO2 exposure, with respect to air, and, in turn, on the variation of the p-channel section of the FET. A reduction of the channel section of ~0.2 µm upon exposure of the sensor to 300 ppb of NO2 was estimated. REFERENCES [1]
Figure 6. Theoretical IDS-VDS curves of the PSJFET for different VGS voltages in synthetic air, obtained by best fitting the experimental data of Figure 3 with the proposed analytical model. [2]
[3] [4]
[5]
[6]
[7]
[8]
[9]
Figure 7. Theoretical IDS-VDS curves of the PSJFET for different VGS voltages in the presence of 300 ppb of NO2, obtained by best fitting the experimental data of Figure 4 with the proposed analytical model.
497
M. Y. Afridi, J. S. Suehle, M. E. Zaghloul, M. E. Zaghloul, D. W. Berning, A. R. Hefner, R. E. Cavicchi, S. Semancik, C. B. Montgomery, C. J. Taylor, “A monolithic CMOS microhotplatebased gas sensor system”, IEEE Sensors Journal, vol. 2, pp. 644-655, 2002. C. Hagleitner, A. Hierlemann, D. Lange, A. Kummer, N. Kerness, O. Brand, H. Baltes, “Smart single-chip gas sensor microsystem“, Nature, vol. 414, pp. 293-296, 2001. G. Barillaro, L. M. Strambini, “ An integrated CMOS sensing for NO2 detection”, Sensors Actuators B: Chemical, in press J. Zang, A. Boyd, A. Tselev, M. Paranjape, P. Barbara, “Mechanism of NO2 detection in carbon nanotube field effect transistor chemical sensors”, Appl. Phys. Lett., vol. 88, 123112, 2006. V. Yu, Timoschenko, Th. Dittrich, V. Lysenko, M. G. Lisachenko, F. Koch, “Free charge carriers in mesoporous silicon”, Physical Review B, vol. 64, 085314, 2001. L. Boarino, F. Geobaldo, S. Borini, A. M. Rossi, P. Rivolo, M. Rocchia, E. Garrone, G. Amato, “Local environment of Boron impurities in porous silicon and their interaction with NO2 molecules”, Physical Review B, vol. 64, 205308, 2001. G. Barillaro, A. Diligenti, A. Nannini, L. M. Strambini, E. Comini, G. Sberveglieri, “Low-concentration NO2 detection with an adsorption porous silicon FET”, IEEE Sensors Journal, vol. 6, pp. 19-23, 2006. G. Barillaro, A. Diligenti, L. M. Strambini, E. Comini, G. Faglia, “FET-like silicon sensor with a porous layer for NO2 detection” Proceedings of IEEE Sensors 2005, pp. 121-124, Irvine (CA). The analytical expression of the PSJFET current IDS=f(VDS, VGS, RD, RS, λ, w1, RPS) will be given in a future work, also with details on the mathematical steps.
