A very sensitive porous silicon based humidity sensor

A very sensitive porous silicon based humidity sensor

Sensors and Actuators B 111–112 (2005) 135–139 A very sensitive porous silicon based humidity sensor G. Di Francia, A. Castaldo, E. Massera ∗ , I. Na...

322KB Sizes 1 Downloads 218 Views

Sensors and Actuators B 111–112 (2005) 135–139

A very sensitive porous silicon based humidity sensor G. Di Francia, A. Castaldo, E. Massera ∗ , I. Nasti, L. Quercia, I. Rea RC-ENEA Loc. Granatello, 80055 Portici, Napoli, Italy Available online 8 August 2005

Abstract A relative humidity (RH) sensor device based on n-type porous silicon has been designed, fabricated and characterized. The simple device is a diode reverse dc biased, exhibiting an extremely high sensitivity in ambient air and a large operating range (30–70% RH). Aging and NO2 interference on the sensor performance have been also investigated. © 2005 Elsevier B.V. All rights reserved. Keywords: RH; Humidity; Porous silicon; Diode; Aging

1. Introduction Humidity detection in the surrounding atmosphere is of great interest for industrial, scientific and environmental applications [1]. Measurement of air humidity under various engineering processes, for example, in the electronic industry, is a sophisticated technological problem because of the low concentration of moisture and the small size of packages of integrated circuits and semiconductor devices [2]. Most moisture related reliability problems in such devices boil down to formation of water layers on critical surfaces and interfaces, which results in soft and/or hard equipment failure [3]. Practically, all moisture related failure modes are related to the relative humidity (RH) that, therefore, requires an accurate control with suitable sensors [4]. At present, most humidity sensors, designed to detect humidity through changes of electrical properties, such as electrical resistance, use electrolytes [5], metal oxides, organic polymers [6] and porous semiconductors [7] as sensitive materials. Several research groups have explored the possibility to use porous silicon (PS) as the basic material for RH sensing in both resistive and capacitive device [8]. Currently, the effect of water vapour adsorption on the porous silicon physical and chemical properties has been already investigated, ∗

Corresponding author. Tel.: +39 081 7723359; fax: +39 081 7723344. E-mail address: [email protected] (E. Massera).

0925-4005/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.snb.2005.06.048

mainly with respect to the influence of relative humidity on the dc electrical conductance of p-type porous silicon [9]. Interesting results have been already reported although the proposed sensor design is often not simple at all. Here, we report on the fabrication and characterization of a very sensitive porous silicon based RH sensor device, obtained using as starting material n-type crystalline silicon. Sensor area is about 3 cm2 and its quite simple design is the result of an optimisation process, that has taken into account all the various fabrication steps and specifically the morphology of the porous layer. The sensor operating principle is based on a Au/PS/Si multi-junction, that exhibits remarkable rectifying properties [10]. Sensor hysteresis, stability and drift will be also addressed.

2. Experimental Single-crystal 1 0 0 phosphorus-doped, n-type silicon wafers, with 1  cm resistivity, 525 ␮m thick have been used ˚ thick in the current study. Before anodization, a 1500 A indium tin oxide (ITO) back contact has been deposited on substrates by e-beam evaporation. Porous silicon samples have been fabricated by electrochemical etching, using a solution of hydrofluoric acid:water:2-propanol (35:40:25, wt%), at a constant current density of 90 mA/cm2 , under the light of a 300 W Hg lamp positioned at 15 cm from the sample. The etching time was 2 min. After the etching, samples have been rinsed in pentane and dried using N2 . We have measured

136

G. Di Francia et al. / Sensors and Actuators B 111–112 (2005) 135–139

simultaneously mix other gases, to study sensor selectivity. Flow and mixing rate of air, electrical readings, temperature and humidity have been adjusted and recorded by a personal computer. Software based on a database engine (Microsoft Access) acquired all the experimental parameters in database records of the data table; the same software could modify several chamber parameters via time-resolved records (step) in the experimental plan. Operator could easily write or modify records to make different step types for a standard and repeatable device calibration. For aging studying, we have tested some devices for several days at room atmosphere, recording current-device, humidity and temperature. At least once a month, the device dynamic response to humidity has been measured to check variations in sensitivity and hysteresis.

3. Results and discussion Fig. 1. Top view photo and draft of the sensor device and the electrical connections in the testing apparatus. The diameter device is about 2 cm. (1) Porous silicon; (2) gold contact; (3) Au wires bonded to the contacts by silver paste.

porous thickness by a TENCOR profilometer and porosity via gravimetric method. Morphology was also investigated by SEM. Photoluminescence spectrum and intensity have been measured by ccd Spectrometer. Device fabrication has been completed by interdigitated gold contacts deposition, by e˚ of gold trough a suitable mask, as gun evaporating 3000 A shown in Fig. 1. A volt–amperometric technique at constant bias has been employed for sensor dc electrical characterization in a controlled gas-flow environment. The device was placed in a stainless steel test chamber at controlled temperature under a gas flow of 500 sccm. For all the measurements here reported, we have used a Precision Humidity Generator (Fig. 2) based on the two-flow method. Briefly, test chamber was fed by two streams of air, one of these was saturated by bubbling with water at known temperature. Flow rate has been regulated by MKS flow-meters, affected by 5% tolerance full range. Test chamber has been equipped with KEITHLEY 6517-RH Humidity Probe sensor, so that RH reading could be compared with the device response. It was also possible to

Fig. 2. Schematic illustration of test chamber with humidity generator.

Porous silicon samples chosen for the realization of our sensor are 10 ␮m thick and show a porosity of about 30%. Pores diameter and distance between pores are 90 nm and 400 nm, respectively (see Fig. 3). Photoluminescence spectrum (Fig. 3) suggests that the material has a sponge-like morphology, characterized by nanoporous phase with nanostructure dimension less than 5 nm [11]. Over these samples, Au contacts have been deposited in the shape shown in Fig. 1. In the same figure, a draft of our device is reported. It con˚ ITO back contact, the n-type silicon bulk, sists of a 1500 A the porous layer and the gold contacts. All the characterized devices exhibited nice sensibility towards RH in all the humidity range. However, at high humidity levels (RH > 70%), device operation is hampered by water condensation phenomena especially with respect to measurement recovery time. On the other hand, at humidity levels lower than 30%, device response becomes so slow that a drastic change in working conditions is essential. In Fig. 4, we report dynamic current response versus time at different humidity steps for a typical device. The step is 40 min

Fig. 3. SEM photograph and device PL spectrum pumped under a blue light (<500 nm). Pores diameter and distance between pores are 90 nm and 400 nm, respectively.

G. Di Francia et al. / Sensors and Actuators B 111–112 (2005) 135–139

137

Fig. 4. Device dynamic response to relative humidity changes (grey line). Fig. 6. Hydrovoltaic effect. Device short circuit current (SCC) vs. relative humidity.

long compared to the 80 min of the whole cycle. A complete measurement lasts more than 20 h. The n-type based porous silicon device has a fast response, quite similar to the commercial humidity probe, and it exhibits very short recovery time within each step, so that the baseline remains almost fixed in each cycle. The measurement also shows that the device is characterized by quite low currents. Device working point has been set as a result of the log(I)–V characteristic study of our device in test chamber at different humidity levels, as shown in Fig. 5, for dry air and 30% RH. The rectifying behaviour is evident. The forward currents values are several order of magnitude greater than those measured at reverse voltage. These properties are strongly dependent on ambient humidity: a very large increase of the reverse current at low values of the reverse bias voltage is observed for a comparatively small increase of RH. In the forward region, an increase of the current with RH is still observed, but current is now much more noisy.

Hydrovoltaic effect is also evident at least in the best performing devices [12]. This effect has so slow dynamic that it can be measured only after leaving the device in test chamber for 24 h at constant humidity and temperature. In Fig. 6, the short circuit current has been reported at three different humidity levels. Even at the highest RH level, the effect does not seem to reach any saturation. In Fig. 7, the response, (I − I0 )/I0 × 100, where I is the reverse current of device for increasing RH and I0 is the current at 30% RH, is reported versus RH cycling in the range 30–70% RH. The low device hysteresis and its exponential response to humidity changes are features of particular relevance. The graph also shows how sensitivity and hysteresis vary under device aging. After 2 months of aging, the device sensitivity lowers; however, a soft reconditioning treatment (the device is kept 8 h in a oven at low pressure and 40 ◦ C) restores device performance.

Fig. 5. Plot of output current vs. bias voltage of the device in test chamber. Positive on Au contact.

Fig. 7. Device sensitivity vs. relative humidity, considering increasing and decreasing humidity. Different sensitivity curves for the same device during 4 months of aging.

138

G. Di Francia et al. / Sensors and Actuators B 111–112 (2005) 135–139

centration up to 360 ppb during measurements at different humidity levels, causes only a quite negligible perturbation on the sensor response, as reported in Fig. 9. As far as the sensor operating mechanism is concerned, it has been proposed that device performance depends both on the material nanostructure and on the semiconductor properties [14]. In our opinion, a second aspect that could also play a role in the sensor response and that requires further investigations is the morphology of the interface bulk silicon/PS [15].

4. Conclusions

Fig. 8. Comparison between device output current and humidity level in a window time of 5 days during 4 months aging at room atmosphere.

Sensitivity during the third month increases and after the fourth month, it is quite similar to the as prepared device. The soft treatment probably expels water from the porous surface restoring the active surface even if the material requires a long time to find equilibrium with the surrounding atmosphere. At this stage, in order to investigate a possible drift in the output current, the device has been biased at room atmosphere and its output current recorded, continuously for a maximum of 5 days, and compared to a commercial RH sensor (KEITHLEY 6517-RH Humidity Probe sensor). In this time window, device output current shows a good agreement with the measured relative humidity level, as reported in Fig. 8. However, considering time window of weeks, a drift appears in the output current. Finally, since the high sensitivity of PS towards NO2 is known [13], we have investigated the interfering effect of this gas on the sensitivity of our device towards RH. The introduction in test chamber of NO2 at con-

Fig. 9. NO2 360 ppb perturbation on device humidity response at different values of RH, compared to commercial RH probe.

We propose a novel and simple way to use n-type porous silicon as active material for humidity sensing. The sensor described is very simple in design and shows a significant sensitivity towards RH, not previously obtained with porous silicon, in our knowledge. Morphological and electrical characterizations of the device are shown. Its dynamic response and hysteresis has been followed over 4 months and, although a loss of sensitivity is recorded, a possible reconditioning treatment has been proposed. In the device characterization, we studied hydrovoltaic effect and we would like to value how it influences device performance. Device miniaturisation and completely electrical reconditioning treatment are the next goals. At this purpose, a second series of devices is at present being produced. The new devices are square with 5 mm side and will be mounted in a TO-08 case. Work is in progress to understand diode nature and the device operating mechanism, in order to reduce the device impedance.

References [1] Z.M. Rittersma, Recent achievements in miniaturised humidity sensors—a review of transduction techniques, Sens. Actuators A 96 (2002) 196–210. [2] M. Tencer, J.S. Moss, Humidity management of outdoor electronic equipment: method, pitfalls and recommendation, IEEE Trans. Components Packaging Technol. I (25) (2002) 66–72. [3] G.I. Vorobets, O.P. Zhuk, O.E. Ilarianov, V.S. Tanasyuk, P.M. Shpatar, Measurement of air humidity inside electronic devices, Instrum. Exp. Tech. 44 (2) (2001) 266–268. [4] G.J.W. Visscher, in: J.G. Vebster (Ed.), Measurement, Instrumentation and Sensors Handbook, CRC Press LCC, 2000, Chapter 72. [5] K. Carr-Brion, Moisture Sensors in Process Control, Elsevier Applied Science Publishers, London/New York, 1986. [6] J.M. Ingram, M. Greb, J.A. Nicholson, A.W. Fountain III, Polymeric humidity sensor based on laser carbonised polyimide substrate, Sens. Actuators B 96 (2003) 283–289. [7] L.H. Mai, P.T.M. Hoa, N.T. Binh, N.T.T. Ha, D.K. An, Some investigation results of the instability of humidity sensors based on alumina and porous silicon materials, Sens. Actuators B 66 (2000) 63–65. [8] G.M. O’ Halloran, W. van der Vlist, P.M. Sarro, P.J. French, Influence of the formation parameters on the humidity sensing characteristics of a capacitive humidity sensor based on porous silicon, in: Proceedings of the Eurosensors XIII, The Hague, The Netherlands, September 12–15, 1999, pp. 117–120.

G. Di Francia et al. / Sensors and Actuators B 111–112 (2005) 135–139 [9] J.J. Mareˇs, J. Kriˇstofik, E. Hulicius, Influence of humidity on transport in porous silicon, Thin Solid Films 255 (1995) 272– 275. [10] D.G. Yarkin, Charge carrier transport in thermally oxidized metal/PS/p-Si and metal/PS/n-Si structures, Semicond. Sci. Technol. 19 (2004) 100–105. [11] G. Garcia, B. Garrito, P. Pellegrino, R. Ferre, J.A. Moreno, J.R. Morante, Size dependence of lifetime and absorption cross section of Si nanocrystals embedded in SiO2 , Appl. Phys. Lett. 82 (10) (2003) 1595–1597. [12] T.D. Dzhafarov, B.C. Omur, C. Oruc, Z.A. Allahverdiev, Hydrogen sensing characteristics of Cu–PS–Si structures, J. Phys. D: Appl. Phys. 35 (2002) 3122–3126. [13] E. Massera, I. Nasti, L. Quercia, I. Rea, G. Di Francia, Improvement of stability and recovery time in porous silicon based NO2 sensor, Sens. Actuators B 102 (2) (2004) 195–197. [14] K. Moln´ar, T. Mohacsy, A.H. Abdulhadi, J. Volk, I. Barsony, On the nature of metal-porous Si-single crystal silicon (MPS) diodes, Phys. Stat. Sol. (a) 197 (2) (2003) 446–451. [15] R.M. Vadjikar, A.N. Chandorkar, D. Sharma, S. Venkatachalam, Computational modelling of nanostructured porous silicon, Nanostruct. Mater. 5 (1) (1995) 87–94.

Biographies G. Di Francia was born in Naples (Italy) in 1958. He has a doctorate in Physics. He works at the ENEA Research Center in Portici (NA). At present he is in charge of the research activity on porous silicon gassensor devices, he had been previously in charge of the ENEA activity on GaAs and silicon solar cells.

139

Anna Castaldo received the MSc degree in Chemistry from the “Federico II” University of Naples, Italy (1998) and the PhD degree in Chemical Science from “Federico II” University of Naples, Italy (2002). She worked for Pirelli Cables and Systems Telecom SpA up to 2003. From January 2004 she is employed by ENEA, Portici Research Center, Italy, where she works on polymeric and porous silicon based sensors. Ettore Massera received his degree in Physics from the “Federico II” University of Naples in May1997. He has been working at the ENEA Research center in Portici (NA) from June 2003. At present he is in charge of research activity on gas sensor devices based on nano-structured materials. Previously he worked on the study of thermal and optical properties of porous silicon at the Physics Department in Naples. Ivana Nasti is a chemical technician employed in ENEA RC from December 1999. Her activity is based on fabrication of sensor devices and nanostructured materials using electrochemical and/or chemical etching, photolithography, FIB, e-beam evaporation. She also characterizes materials by BET and SEM analysis. L. Quercia was born in Naples (Italy) in 1961. He has a doctorate in Physics. He has been working at the ENEA Research Center in Portici (NA) since 1992. Presently involved in the research activity on porous silicon gas-sensor devices, has previously worked on CuInSe2 and amorphous silicon solar cells, high resolution spectroscopy, molecular beams and cluster formation Ilaria Rea graduated in Physics at University of Naples “Federico II” in July 2003. In 2003 and 2004 she worked in ENEA, Portici Research Center, Italy, on porous silicon based sensors. Since January 2005 she is a research fellow at Institute for Microelectronic and Microsystems of National Council of Research in Naples. Her current research interests are in fields of silicon optoelectronics and optical sensors.