Search in book...
Toggle Font Controls
Create new playlist

Name your new playlist

Playlist description (optional)
Sign In

Email address

Password

Forgot Password?

or

Continue with Facebook

Continue with Google
Sign Up

Full Name

Email address

Confirm Email Address

Password

or

Continue with Facebook

Continue with Google

Photovoltaic and fuel cells in power microelectromechanical systems for smart energy management

Juan García¹, Francisco J. Delgado¹, Pablo R. Ortega², and Sandra Bermejo² ¹University of Seville, Seville, Spain ²Polytechnic University of Catalonia, Barcelona, Spain

Abstract

Photovoltaic mini-generators and fuel cells are the most feasible alternatives for applications where a high-density micropower supply is needed. This chapter will focus on photovoltaic and fuel cell technologies in powering microelectromechanical systems for smart energy management, such as bioimplantable devices or portable electronic devices. Sun sensors are devices that are being newly introduced into photovoltaic systems to improve their performance. The chapter ends with a discussion of the operating principle of these sensors and their role within the photovoltaic system is described.

Keywords

Energy harvesting; Fuel cell; Photovoltaic mini-modules; Solar energy; Sun sensor

17.1. Introduction

There is a growing need for small energy sources for applications such as portable electronic devices (Koeneman et al., 1997; Dehlinger, 2000; Liu et al., 2010), distributed sensors, and telesupply, and in other order bioimplantable devices. Among them, microelectromechanical systems (MEMS) supplying with a near-integrated micropower supply is a hot topic in research. All these applications demand the use of small energy sources, each one of them trying to fulfill the electrical requirements of the specific application. The different options for mini–micropower supplies found in the literature can be mainly summarized into four types of devices: vibration-based (piezoelectric, electrostatic) (Sodano et al., 2004; Andò et al., 2010; Sari et al., 2010), thermoelectric (Muanghlua et al., 2000; Watkins et al., 2005), fuel cells (FCs) (Kelley et al., 2002; Luque et al., 2010), and photovoltaic mini-generators (Lee et al., 1995; Shreve et al., 1996; Keller et al., 2000; Brendel and Auer, 2001; Ortega et al., 2001, 2008; Bermejo et al., 2005). Fig. 17.1, which is based on some of the results shown in the review in Cook-Chennault et al. (2008), shows a density power versus voltage map of all these options.

As can be seen, there are mainly two types of device, which clearly show higher density power values: FCs and photovoltaic mini-generators. Of these two technologies, the photovoltaic approach has developed a wider voltage range. This can be important in applications such as the driving of MEMS switches, where relatively high voltage values are required. This chapter will focus on photovoltaic and FCs as power supplies for MEMS for smart energy management. However, MEMS is a quite broad term, which includes a significant number of applications. This chapter will briefly describe some of these applications, starting with MEMS switches control and bioimplantable devices (Rabdill et al., 2010).

Sun sensors are devices that are being newly introduced into photovoltaic systems to improve their performance. The operating principle of these sensors and their role within the photovoltaic system are discussed at the end of the chapter.

Figure 17.1 Power density and voltage map of different small-sized power supplies. The photovoltaic sources include only silicon technologies. MNT, Micro and Nano Technology.

17.2. Photovoltaic mini-generators

Photovoltaic energy conversion transforms light energy into electrical energy. Typically, this energy conversion is carried out in an illuminated semiconductor p-n junction diode (i.e., photovoltaic cell) in two steps. First, photons with energy greater than the semiconductor energy gap are absorbed, creating electron–hole pairs inside the device (photogeneration). Second, electron and holes pairs are separated by the electrical field existing in the p-n junction—electrons to the negative terminal and holes to the positive electrode—thus generating electrical power. Solar cells are, in fact, photovoltaic devices, which are especially designed to work with solar light as optical energy source.

In this section, the working principles of photovoltaic cells and modules will be explained. Next, several technological approaches to the fabrication of small photovoltaic modules based on crystalline silicon will be described. And finally, a number of photovoltaic applications in the fields of MEMS and autonomous systems will be explored.

17.2.1. Photovoltaic working principles

The electrical behavior of a photovoltaic cell can be studied using the superposition principle: the current–voltage I–V characteristics can be obtained from the corresponding characteristic of a p-n junction diode in the dark, adding a current shifting by I_ph (photocurrent), as shown in Fig. 17.2. Notice that current leaves the device by the voltage positive terminal, the usual convention for electrical power generators.

Photovoltaic cells can work in three possible regions depending on the actual voltage: the photovoltaic, photoconductive, and diode regions, as it is shown in Fig. 17.2:

• Photovoltaic region (first quadrant): this is relevant for voltages ranging from 0 to V_oc (open-circuit voltage)—this region is where the device really can deliver electrical power to a connected load.
• Photoconductive region (second quadrant) for negative voltages—current is almost constant in this region and equal to the short circuit current I_sc corresponding approximately to I_ph (I_sc ≅ I_ph); in this region, solar cells work like photodiodes (PR) being the photocurrent term I_ph proportional to the light irradiance and the device area. Sun sensors (described in Section 17.4) based on PR (solar cells) are biased in this working region.
• Diode region (fourth quadrant) for V > V_oc, where the device becomes a passive component.

17.2.1.1. Photocurrent and spectral response

Photocurrent I_ph is the most important parameter in a photovoltaic cell. It can be calculated using the spectral response, denoted by SR in units of (A/W). Spectral response is a light wavelength λ dependent function that takes into account how much photogenerated charge leaves the contacts per second related to the incident optical power (assuming monochromatic light). Then SR considers any collection losses, namely: light front reflectance, bulk and surface recombination, and partial photon absorption inside the device. These losses depend considerably on the technology and material (semiconductor) used to manufacture the device (see Fig. 17.3).

Figure 17.2 I–V characteristics of a photovoltaic cell for dark and light conditions. The device can work in three regions depending on the voltage: the photovoltaic region (quadrant I), photoconductive region (quadrant II), and diode region (quadrant IV).

Figure 17.3 Spectral responses versus wavelength for different photovoltaic semiconductor materials: monocrystalline silicon c-Si, CIGS, CdTe, and amorphous silicon a-Si.

In the case of a monochromatic light of wavelength λ₀, I_ph can be calculated using Eq. (17.1), where S is the optical incident irradiance—normal to the cell plane and A the device area.

I_{ph} = A \times S \times S R (λ_{o})

$I_{ph} = A \times S \times S R (λ_{o})$

(17.1)

Alternatively, if a spectral light is used (e.g., solar light), I_ph is given by

I_{ph} = A \times S \times \frac{\int_{0}^{\infty} S_{λ} (λ) S R (λ) d λ}{\int_{0}^{\infty} S_{λ} (λ) d λ}

$I_{ph} = A \times S \times \frac{\int_{0}^{\infty} S_{λ} (λ) S R (λ) d λ}{\int_{0}^{\infty} S_{λ} (λ) d λ}$

(17.2)

where S_λ is the spectral irradiance of the light in (W/m² μm) units, being the light irradiance given by

S = \int_{0}^{\infty} S_{λ} (λ) d λ

$S = \int_{0}^{\infty} S_{λ} (λ) d λ$

(17.3)

The spectral irradiance S_λ for three-light spectrums is shown in Fig. 17.4, where the extraterrestrial AM0 (air mass) (S = 1.366 kW/m²) and the terrestrial AM 1.5G (S = 1 kW/m²) standard solar spectrums are compared with a flash lamp light (S = 1 kW/m²) modeled by a black-body emitting source (3600K color temperature).

I_ph has been calculated, for both monochromatic and spectral light, assuming normal light incidence γ = 0 degree (see Fig. 17.5). In other cases, normal S irradiance must to be replaced by S × cos(γ) in Eqs. (17.1) and (17.2). Not to compromise photocurrent, and thereby decrease photovoltaic efficiency, is important to maintain γ close to 0 degree using, for instance, a light tracking system (as discussed in Section 17.4). This aspect is critical in systems based on light concentration.

Figure 17.4 Typical light spectrums. Extraterrestrial AM0 (S = 1.366 kW/m²) and terrestrial AM1.5G (S = 1 kW/m²) solar spectrums. A flash lamp—assumed a black body 3600K—normalized to 1 kW/m² irradiance. The visible light (0.4–0.7 μm range) is also depicted in the graph.

17.2.1.2. Main photovoltaic electrical parameters

Current I and delivered power P versus voltage V characteristics corresponding to an illuminated cell working in the photovoltaic region are shown in Fig. 17.6. The main photovoltaic electrical parameters are pointed out in the I–V and P–V plots. Of these, short circuit current I_sc and open-circuit voltage V_oc have already been introduced. Additionally, other important parameters are the maximum delivered power P_m and the current and the voltage at the maximum power point (mpp point), I_m and V_m, respectively.

Another important parameter is the fill factor given by

FF = \frac{P_{m}}{V_{oc} I_{sc}}

$FF = \frac{P_{m}}{V_{oc} I_{sc}}$

(17.4)

Figure 17.5 Definition of the light incidence angle γ. γ = 0 corresponds to normal light incidence.

Figure 17.6 Characteristics of current–voltage I–V and power–voltage P–V. I–V (left) and power–voltage P–V (right). Ideal I–V and P–V characteristics are also shown (dotted lines). The main photovoltaic parameters are pointed out in the graphs, namely: V_oc, I_sc, V_m, I_m, and P_m.

This parameter indicates how far a photovoltaic cell is from ideal electrical performance (see the ideal curves represented by dotted lines in Fig. 17.6). Finally, the photovoltaic energy conversion efficiency η can be calculated taking into account incident light power (P_in) or, alternatively, considering irradiance S and cell area A, by

η = \frac{P_{m}}{P_{in}} = \frac{FF \times V_{oc} \times I_{sc}}{S \times A}

$η = \frac{P_{m}}{P_{in}} = \frac{FF \times V_{oc} \times I_{sc}}{S \times A}$

(17.5)

17.2.1.3. Photovoltaic modules

To scale up current and voltage to fulfill electrical requirements, cells are combined electrically in series to raise the voltage and in parallel to increase the current (see Fig. 17.7). Table 17.1 summarizes the scaling rules of an ideal module formed by N_p branches in parallel with N_s identical cells connected in series.

Module current can be increased by enlarging the cell area and then increasing photocurrent I_sc (see Eq. 17.1) or by connecting several branches in parallel (i.e., N_p > 1). However, scaling up voltage can only be carried out by combining cells in series (N_s > 1). For instance, if monocrystalline silicon material is used, a single solar cell provides ∼0.6 V in open-circuit conditions (T = 25°C AM1.5G 1 kW/m² spectrum) or ∼0.45 V at the mpp point. Thus, a typical module arrangement of 32 cells in series is required to feed a 12 V battery. As can be seen from Table 17.1, ideally, module conversion efficiency is the same as that of the single solar cell. However, this is not true of an actual module because of a useless module area (A_mod), from the perspective of photovoltaic conversion, which is necessary to perform isolation and electrical interconnections between cells (i.e., A_mod > N_p × N_s × A).

Figure 17.7 An ideal module consisting of a connection of N_p branches in parallel. Each branch is formed by N_s identical cells in series.

Table 17.1

Scaling up rules of an ideal module consisting of N_p branches in parallel of N_s identical cells in series
V_{oc_mod}	I_{sc_mod}	V_{m_mod}	I_{m_mod}	P_{m_mod}	FF_mod	η_mod	A_mod
N_s × V_oc	N_p × I_sc	N_s × V_m	N_p × I_m	N_p × N_s × P_m	FF	η	N_p × N_s × A

17.2.2. Mini-modules technologies

Compact and efficient small area photovoltaic arrays (<1 cm²) can be fabricated using different materials and technologies: (1) gallium arsenide (GaAs) substrate (Beaumont et al., 1991), (2) thin-film amorphous silicon (a-Si) on a dielectric substrate (Lee et al., 1995), or (3) crystalline silicon (c-Si) (Ortega et al., 2001, 2002, 2008; Bermejo et al., 2005). Table 17.2 summarizes the results published so far in the literature, where N_s is the number of cells connected in series, V_oc is the module open-circuit voltage, v_oc and p_m are the “specific open-circuit voltage” and “specific maximum power,” respectively, first introduced here and defined as the ratios of V_oc and P_m to the total module surface, respectively. These are indicators of performance per unit area.

Although GaAs mini-modules can be very efficient (both v_oc and p_m levels are high), they are more expensive to produce compared with mainstream silicon technology. Alternatively, a-Si arrays can be subject to long-term stability problems. Therefore, c-Si material appears to be a good choice for the fabrication of very small photovoltaic arrays.

Devices based on c-Si can be fabricated in a number of ways—for example, nonmonolithic integration using microelectronic packaging techniques such as multichip module (MCM) technology (Ortega et al., 2001). More interestingly, especially where integration and compatibility with other silicon devices is involved, a monolithic design (Bermejo et al., 2005; Ortega et al., 2008) reduces fabrication steps and reliability is enhanced. Generally, monolithic integration is achieved by means of total (Bermejo et al., 2005) or partial (Ortega et al., 2008) anisotropic etching of the silicon substrate, forming channels that isolate individual cells from each other. As can be seen in Table 17.2, a monolithic integration using silicon-on-insulator (SOI) technology is one of the best alternatives (using c-Si) to obtain very high specific voltages with relatively high specific powers.

Table 17.2

Comparison between mini-module fabrication technologies
Technology	Area (cm²)	N_s	V_{oc_mod} (V)	v_oc (V/cm²)	p_m (mW/cm²)	References
GaAs^a	0.01	6	6.5	650	1204	Beaumont et al. (1991)
a-Si	1.00	100	150	150	0.21	Lee et al. (1995)
c-Si/MCM	0.70	9	5.1	7.3	7.7	Ortega et al. (2002)
c-Si/fusion bonding	1.40	9	4.1	2.9	2.7	Bermejo et al. (2005)
c-Si/SOI	0.43	169	100	240	7.3	Ortega et al. (2008)

In the following subsections, the silicon-based photovoltaic mini-module technologies developed since 2000 by the Micro and Nano Technology Group are introduced and details are given regarding certain technological and electrical results.

17.2.2.1. Multichip module technology

This approach uses c-Si solar cells, batch-produced, in combination with MCM flip-chip technology (Al-Sarawi et al., 1998). The MCM approach (Ortega et al., 2001) is based on the interconnection of c-Si back-contact photovoltaic cells in series by means of a common c-Si substrate wafer, as shown in Fig. 17.8(a).

MCM mini-module fabrication is divided into three phases: fabrication of the solar cells, fabrication of the substrates, and flipping of the solar cells:

• Fabrication of the solar cells. These are made using c-Si wafers (device wafer); single solar cells are obtained by dicing the device wafer using a diamond circular saw.
• Fabrication of the substrates. These are used for the interconnection between solar cells and are made from c-Si wafers (substrate wafer).
• Flipping of the solar cells. The solar cells are flipped, picked, and placed on top of a substrate to be assembled with an alignment accuracy of 5 μm. A separation between dice (solar cells) in the substrate of around 100 μm is necessary to perform the pick and place stage adequately. Before placing the chips into the substrate, a solder paste is deposited onto the substrate contacts using screen printing technology. Many different solder pastes are available in the market. They can be lead-based such as Sn/Pb and Sn/Pb/Ag and lead-free such as Sn/Ag, Sn/Sb, In, Pd, and Sn/Ag/Cu. The metallic paste is converted into a microbump (∼50 μm high) after a reflow step at a temperature well above the melting point, ranging from 180 to 320°C depending on the selected alloy. Once the solder paste has melted, the chips are attached to the substrates. Finally, the substrate wafer has to be diced to obtain the MCM mini-modules.

Flip-chip technology requires that the cell contacts (emitter and base contacts) are placed in the same surface (rear surface) so that the device and substrate contacts are face-to-face. As a consequence, light must enter into the device by the opposite surface to the contacts (the front). Therefore, photogenerated minority carriers have to travel the thickness of the wafer to arrive at the p-n junction at the rear of the unit. In this type of solar cell, known as a back-contact solar cell (Van Kerschaver and Beaucarne, 2006), it is important to have a very well-passivated front surface (i.e., low recombination surface velocities) and high bulk lifetimes to guarantee a high carrier collection (i.e., high photocurrents).

Figure 17.8 (a) A sketch of the multichip module mini-module concept showing the interconnection of two adjacent solar cells and (b) a fabricated module, before dicing, of nine solar cells electrically connected in series with a total area of 2.8 mm × 2.8 mm. On the right-hand side of the image there is a substrate without solar cells showing the solder microbumps.

Fig. 17.8(b) depicts a fabricated MCM square module of nine solar cells connected in series with an area of 2.8 mm × 2.8 mm, obtaining an open-circuit voltage of almost 5.1 V with a conversion efficiency of 10.6% under a standard solar spectrum (AM1.5 1 kW/m²). One of the main drawbacks of using this technology to fabricate small photovoltaic modules is the minimum solar cell size necessary to perform the pick and place stage, which is restricted to ∼1 mm². However, in applications where very small mini-modules are not required, merging silicon solar cells with mass production flip-chip assembly can significantly reduce the mini-module fabrication cost.

17.2.2.2. Silicon-on-insulator technology

In this approach, solar cells and mini-modules are fabricated monolithically together using off-the-shelf SOI c-Si wafers. As Fig. 17.9(a) shows, photovoltaic cells are processed on an active layer of a relatively small thickness (3–50 μm) on top of a handle wafer, which acts as a mechanical support; in between the active layer and the handle wafer there is a thick SiO₂ buried oxide layer (BOX), typically 1 μm thick. The top active layer is where individual solar cells are fabricated (see Ortega et al., 2008 for details). An electrical isolation between devices is mandatory; this is carried out by means of channels created by anisotropic etching of the whole active layer. Both the emitter and base contacts are on the front of the wafer, making series interconnection between cells easy; this is performed by metal evaporation and patterning. A thin thermal oxide is grown on the front surface of the device to passivate and to obtain an antireflective coating film. An important difference with respect to MCM technology is that light enters into the device on the same face where the base and emitter contacts are placed. Thus, the photogeneration mechanism occurs near the p-n junction, decreasing carrier recombination losses. However, metallization of the front should be reduced to a minimum to avoid excessive shadow light losses.

Figure 17.9 (a) Top view of an silicon-on-insulator (SOI) solar cell (left) indicating the cross-section that was sketched to show the concept of the SOI mini-module (right) and (b) a fabricated module of 81 solar cells monolithically connected in series with a total area of 4.5 mm × 4.5 mm.

SOI wafers present a design challenge: to guarantee an acceptable photogeneration under either monochromatic or spectral light, the active layer should be designed to be sufficiently thick. On the other hand, thick active layers require deep channels, thereby causing possible interconnection problems; there is therefore a trade-off between the thickness of the active layer and the fabrication yield. This technology is very well suited to UV and visible light (monochromatic or spectral) due to the high absorption coefficient of silicon in the 350–700 nm wavelength range; in this instance, photons are absorbed very close to the front surface. Fig. 17.9(b) shows a fabricated mini-module consisting of 81 cells, each one being 0.225 mm², and the total module area being 0.45 cm² × 0.45 cm². In this mini-module, an open-circuit voltage of 57 V has been measured with a conversion efficiency of 9.7% using standard solar spectrum AM1.5 1 kW/m². It is worth noting that, in this mini-module fabrication technology approach, very high specific open-circuit voltages (∼240 V/cm²) can be achieved.

17.2.2.3. Fusion-bonding technology

This concept is similar to the SOI technology explained in the previous section, though in this case the active layer is 350 μm thick. Like in the SOI approach, two wafers are used: a silicon wafer (also known as the device wafer), where the solar cells are fabricated, and the handle wafer. The device wafer is bonded or stuck to the handle wafer using fusion-bonding or adhesive-bonding techniques. Fusion bonding has been chosen as an appropriate alternative, using a second silicon wafer as a substrate material. The handle wafer provides electrical isolation and mechanical stability to the devices placed in the silicon wafer on the top. Electrical isolation between solar cells is essential because they share the same substrate, so individual cells are isolated by channels created by anisotropic etching of the silicon, and series are connected to the neighboring cell by means of the metal pattern. The schematic view is similar to that shown in Fig. 17.9(a).

Monolithic mini-module fabrication has three main steps: the bonding process of the device and the substrate wafer, the fabrication of the isolation channels, and the serial connection between individual cells by metal tracks. Fig. 17.10 presents a closed view of a 2″ processed wafer exhibiting different topologies before the deposition of the metallization tracks. Fig. 17.11 shows a photovoltaic mini-module with nine cells connected in series. The total area is 1.4 cm²; the total open-circuit voltage is 4.1 V and the specific power density is 2.7 mW/cm².

The main advantage of this technology compared with the SOI approach is the improvement in light absorption, thus increasing the available current density. The main drawback of this technology is the challenging fabrication procedure, where critical points such as the photolithography and metal deposition over 350 μm depth grooves may need to be resolved.

17.3. Applications of photovoltaic mini-generators

There are many applications, which demand autonomous power supplying. Wireless telesupplying is mandatory in harsh environments or in places where electrical conduction is not possible, such as in satellites or in bioimplantable systems. Consumer electronics will also benefit as well from a close, suitable mini-energy source. MEMS switches demand a relatively high voltage and low current values, which make the use of a specific mini-energy source an interesting choice. In this chapter, we show two example applications: (1) telesupplying bioimplantable devices and (2) MEMS driving. In these applications, the introduction of a small-sized photovoltaic mini-module significantly modifies the energy balance of the system, whether energy harvesting solar light or using monochromatic light.

Figure 17.10 Closed view of a 2″ processed wafer with different cell topologies before deposition of the interconnection path.

Figure 17.11 Mini-module of nine cells in series processed with fusion-bonding technology. The total area is 1.4 cm².

17.3.1. Telesupplying bioimplantable devices

There are many applications where the use of a complementary small-sized energy source that is capable of wirelessly giving an extra energy contribution is of interest. Bioimplantable devices such as pacemakers (Dinesh et al., 2010; Ashraf and Masoumi, 2016), implantable defibrillators (Andrew et al., 2010; Khanna, 2015), drug deliverer systems (Nisar et al., 2008; Cobo et al., 2016), neurotransmitters (Yoshimi et al., 2004), implantable sensors (Christensen and Roundy, 2015; Basaeri et al., 2016), or implantable communication systems (Nageswara and Raghavan, 2016), for example, would benefit from the extra energy supplied by the photovoltaic mini-module, thus increasing the useful life of the implantable battery.

The design of a suitable energy source goes through the definition of the specific voltage and current requirements of the whole system. In an autonomous device, the initial capacity of the battery can be improved by means of an external energy source. Calculating the energy balance of the total system, the total capacity of the autonomous device is given by Eq. (17.6).

{C (t) |}_{t = M} [Wh] = C (0) [Wh] + (P_{PV} - P_{L}) [Wh / year] M [year]

${C (t) |}_{t = M} [Wh] = C (0) [Wh] + (P_{PV} - P_{L}) [Wh / year] M [year]$

(17.6)

where C(0) is the initial battery capacity, P_L is the power consumption, P_PV is the power given by the photovoltaic mini-module, and M is the number of autonomy years. The power produced by a photovoltaic source is defined as

P_{PV} [Wh] = \int_{0}^{t} S [W / {cm}^{2}] A_{\mod} [{cm}^{2}] η d t

$P_{PV} [Wh] = \int_{0}^{t} S [W / {cm}^{2}] A_{\mod} [{cm}^{2}] η d t$

(17.7)

where S is the irradiance, A_mod is the photovoltaic mini-module surface, and η the photovoltaic efficiency. Taking S, A_mod y η constant in time, the stored energy as a function of the exposition time, H_exp, is

P_{PV} [Wh / year] = S [W / {cm}^{2}] A_{\mod} [{cm}^{2}] η H_{\exp} [h / week] \frac{1 week}{7 days} \frac{365 days}{1 year}

$P_{PV} [Wh / year] = S [W / {cm}^{2}] A_{\mod} [{cm}^{2}] η H_{\exp} [h / week] \frac{1 week}{7 days} \frac{365 days}{1 year}$

(17.8)

Substituting Eq. (17.8) in Eq. (17.6) and imposing the final capacity as being null, the surface can be evaluated by

A_{\mod} = (P_{L} - \frac{C (0)}{M}) \frac{7}{S η H_{\exp} 365}

$A_{\mod} = (P_{L} - \frac{C (0)}{M}) \frac{7}{S η H_{\exp} 365}$

(17.9)

where M can be expressed as a function of the increase of battery life P and the nominal duration of the battery without any provision of external energy, following the expression:

M [years] = N [years] (1 + \frac{P [%]}{100})

$M [years] = N [years] (1 + \frac{P [%]}{100})$

(17.10)

Figure 17.12 Current consumption of a commercial pacemaker with a Wilson Greatbatch WG 8402 lithium iodine battery.

Table 17.3

Photovoltaic area necessary to increase the battery lifetime up to P of the pacemaker consumption shown in Fig. 17.12
Light source	η (%)	S (kW/m²)	Exposition time (h/week)	P (%)	Area (cm²)
Solar AM 1.5	4.3	1	1	60	1
Monochromatic λ = 830 nm	10.5	0.18	5	60	1.2
Low consumption bulb OSRAM 23 W (distance 5 cm)	5	0.115	2.5	20	1

Fig. 17.12 shows the real consumption of a commercial pacemaker with a Wilson Greatbatch WG 8402 lithium iodine battery (C(0) = 2.8 Wh, V_L = 2.8 V, f = 60–70 Hz). Table 17.3 shows the photovoltaic area necessary to fulfill the consumption of the pacemaker for different illumination conditions, photovoltaic efficiencies, irradiance, and exposition time to increase the battery lifetime to P. For instance, if we consider a standard AM1.5 solar spectrum (1 kW/cm²) and 4.3% of photovoltaic efficiency, a 1 cm² solar cell can increase the battery life by up to 60%, using an exposition time of only 1 h/week.

17.3.2. Driving microelectromechanical systems switches

MEMS electrostatic actuators are currently used for a number of applications such as optical switching, electrical relays, or radio frequency variable capacitors, for example. Driving such devices is usually performed by applying a voltage, which produces an electrostatic force that is nonlinearly related to the deflection of the moveable part of the actuator and usually requires higher voltage than conventional standard processing electronics. The nonlinearity involved makes movement unstable, leading to the collapse of a device at a so-called “pull-in” voltage (Nathanson et al., 1967). It has been observed (Castañer and Senturia, 1999) and experimentally demonstrated (Castañer et al., 1999) that a “charge drive,” instead of a voltage drive, has a number of benefits because the total power consumed in the switching can be drastically reduced, the kinetic energy at the impact of the two plates is reduced by orders of magnitude and the movement can be analog-controlled across the gap (Castañer et al., 2001). It has also been shown that a charge drive usually requires high-voltage compliance (Seeger and Crary, 1997).

Actually, a photovoltaic mini-array can be considered as a new DC/DC step-up converter without the need of conventional charge pumps of switching converters, avoiding the use of switched-capacitor high-frequency circuits or bulk inductors. The series connection of solar cells allows the scaling up voltage in the array to levels more suitable for MEMS actuation (typically more than 5 V). In this section, we will briefly show the potential use of a photovoltaic source as the direct drive of electrostatic actuators. Because the I–V characteristic of a photovoltaic array has the shape of a current source limited in voltage, the actuation process can be that of a charge drive, voltage-limited. The main potential benefit we foresee is that a direct drive with galvanic isolation can be achieved at high voltages without using any typical DC/DC converter. This photovoltaic drive therefore becomes a new inductor-less DC/DC type of converter “opto-activated.” The number of individual solar cells can be chosen to drive specific MEMS switches of high pull-in voltage value.

A PSpice model has been developed (Bermejo and Castañer, 2005; Bermejo et al., 2008) including the photovoltaic mini-module and an electrostatic switch. Fig. 17.13 shows the main parameters of the transient. The transient starts in t = 0 s in Fig. 17.13. As soon as the irradiance pulse is turned ON, the MEMS device capacitance starts charging with a current of value I_sc and the moveable arm starts moving because of the increase of charge and, hence, of the attractive force. For a while, the current remains reasonably constant while the voltage rises (Region I in Fig. 17.13), and the operating point will move from left to right following the I–V characteristics (Fig. 17.13(a)). When the operating point goes beyond the flat part of the stationary characteristic in Fig. 17.13(a), the voltage (Fig. 17.13(c)) will still be increasing, but more slowly, and the current (Fig. 17.13(b)) will start dropping (Region II). At some point, the charge in the MEMS device reaches the value of the pull-in charge, which is required for the actuator to collapse. At this moment, the moveable arm moves drastically producing a large change in the value of the actuator capacitance. As the current has a limiting value, the voltage across the MEMS device has to drop (this is seen in Regions III and IV) because the total charge cannot increase further. This drop in the voltage forces an increase in the current and the operating point now changes direction and moves backward over the I–V curve finally reaching a hold-on voltage considerable smaller than the maximum drive-in voltage.

Analyzing the energy at the impact, a decrease of up to three orders of magnitude, for the same switching time, can be obtained by using this current-controlled method.

Figure 17.13 PSpice simulation results of the photovoltaic source and the microelectromechanical systems switch models. (a) I–V of the photovoltaic source, (b) current delivered to the switch, (c) switch voltage, and (d) deflection of the moveable arm.

17.4. Microfuel cells

An FC is a device that converts chemical energy into electrical energy. Like in photovoltaic cells, this energy conversion is direct, so no thermodynamic cycle is involved. Unlike a battery, an FC produces electric energy as long as fuel is supplied and has a higher density of stored energy. Both properties make the FC a very interesting power source for portable equipment.

17.4.1. Fuel cell principles and classification

An FC has a very simple electromechanical structure, so it is easy to implement using MEMS technology. The basic structure of an FC consists of two electrodes (cathode and anode) connected through an electrolyte. The oxidant and the fuel are combined internally to produce electrons, which are driven through an external electric load. A distribution chamber provides the fuel and the oxidant to the electrodes and drives electrons through an external load (see Fig. 17.14).

Figure 17.14 Basic structure of a fuel cell.

An electrolyte is a chemical medium with high ionic conductivity but low electronic conductivity. The most widely used electrolyte is a membrane of polytetrafluoroethylene: Nafion. This membrane has good mechanical and chemical properties such as high mechanical strength, ease of handling, and high stability. Other usual electrolytes are aqueous alkaline solution, phosphoric acid (PO₄H₃), lithium potassium carbonate (C0₃ Li K), or yttria-stabilized zirconia.

Electrodes are in direct contact with both surfaces of the membrane. In the anode, fuels such as hydrogen are dissociated in ionic species (in this case, protons) and electrons.

H_{2} \to 2 H^{+} + 2 e^{-}

$H_{2} \to 2 H^{+} + 2 e^{-}$

(17.11)

Protons are driven to the cathode through the membrane, whereas electrons travel to the cathode through the external load. In the cathode, the oxidant (typically oxygen), protons, and electrons react to produce water, in this case.

2 H^{+} + 2 e^{-} + \frac{1}{2} O_{2} \to H_{2} O

$2 H^{+} + 2 e^{-} + \frac{1}{2} O_{2} \to H_{2} O$

(17.12)

The desirable properties of electrolytes include their high porosity and the catalyst power. Porosity gives the protons higher mobility, and the catalyst power shifts the reaction to the product side. Electrodes are usually composed of a mixture of platinum, carbon, and Nafion.

Figure 17.15 Channeled structure of an electrode.

The last components of an FC structure are the distribution chambers. They constitute the external shield of the system and are in charge of providing fuel and oxidant to electrolytes, the removal of the product of reaction (water) and connecting the cell electrically to external devices. The distribution chamber requires high electrical and thermal conductivities, mechanical strength, and corrosion resistance. Graphite is a good material because it has good electrical and thermal conductivities but unfortunately is too fragile. To drive the fuel and the oxidant efficiently, electrodes have to be channeled. Those channels are also used to remove the resulting water from cathode; this is an important target in the effort to improve FCs (see Fig. 17.15).

The most common fuel is hydrogen, but several hydrocarbons are used too, such as methanol or ethanol. Despite the wide variety of fuels, the classification of FCs is based on the type of electrolyte used in their structure. Table 17.4 shows the most widely used FCs, their properties, and their operational conditions.

The voltage of a single FC is ∼0.7 V in normal operating conditions, so several cells have to be stacked in series to increase the output voltage or in parallel to produce more electrical current. The efficiency of an FC typically ranges from 40% to 60% depending on how efficient several processes are; for instance, wastewater removal, humidity of the membrane, heat evacuation, and fuel supply. All these factors must be taken into consideration in the design and manufacturing technology of an FC.

Table 17.4

Properties of the fuel cells (FCs) most usually employed
FC type	Electrolyte	Electrolyte state	Charge carrier	Operating temp (°C)	Power range (kW)
Polymer electrolyte membrane	Sulfonated polytetrafluoroethylene	Solid	H⁺	50–80	5–250
Alkaline	aqueous alkaline solution	Liquid	OH⁻	90–100	5–150
Direct methanol	H⁺ interchange membrane	Solid	H⁺	80	5
Phosphoric acid	PO₄H₃	Liquid	H⁺	160–220	50–1 × 10⁴
Molten carbonate	CO₃ Li K	Liquid	CO₃²⁻	620–660	100–2 × 10³
Solid oxide	yttria-stabilized zirconia	Solid	O²⁻	800–1000	100–250

17.4.2. Microelectromechanical systems–based polymer electrolyte membrane FC

At the time of writing the trend is to miniaturize power sources. The use of MEMS technology for FC design allows decreased dimensions and reduced losses, thereby improving energy conversion and fuel storage efficiency. Many efforts have focused on developing an MEMS-based process to miniaturize FC structure.

Fig. 17.16 shows the structure of a polymer electrolyte membrane FC in which an integrated silicon distribution chamber has been developed (Luque et al., 2010). The use of silicon technology has several advantages because it uses well-known fabrication processes. Silicon is a conductive material, which will collect generated current from the anode, and the resulting monolithic structure could be integrated with auxiliary electronics.

Fig. 17.17 shows the structure of the fuel distribution chamber. Two channels have been implemented to supply hydrogen to the membrane and to evacuate the waste hydrogen not consumed. The distribution chamber has many holes to connect it with the membrane electrode assembly (MEA). A catalyst, made of platinum, is in contact with the MEA and promotes hydrogen dissociation. The generated electrons are collected by a silicon structure, and protons travel through it to reach the “MEA.”

Two types of structures have been implemented in the distribution chamber: pillars and walls. Pillars were built to support the membrane and prevent it from failing when hydrogen flows below it. Walls are used to force the hydrogen to follow a particular path when flowing and thus force it to cover a greater area of the membrane. Both were made by patterning the sacrificial oxide of silicon and a particular distribution of holes, as shown in Fig. 17.18.

Figure 17.16 Polymer electrolyte membrane fuel cell (FC) design. (a) Overhead view of the printed circuit board (PCB) bottom cover with the FC chip, (b) overhead view of the top cover with the current collector, and (c) cross-sectional view of the complete assembly. MEA, membrane electrode assembly.

Figure 17.17 General structure of a silicon fuel cell with the integrated fuel distribution chamber. MEA, membrane electrode assembly.

Figure 17.18 Detail of mask patterning for pillars design.

The fabrication process, based on a previous work (Kelley et al., 2002), is described by Luque et al. (2010). Fig. 17.19 shows how the process starts with a p-doped double-sided polished Si wafer. The steps in the fabrication process are summarized in points 1–6:

1. the growing of a thermal oxide used to be used as a sacrificial layer;
2. the growing of a polysilicon layer by low-pressure chemical vapor deposition to support the MEA;
3. the patterning and removing of polysilicon by reactive ion etching;
4. the opening of input and output holes for hydrogen feeding;
5. the etching of oxide to form the reaction chamber (Schilp et al., 2001);
6. the deposition of a Pt layer to form the current collector.

Figure 17.19 (a–f) Fabrication process for polymer electrolyte membrane silicon fuel cell. See text for explanation of steps.

Figure 17.20 Measurement of power density and output voltage generated by the polymer electrolyte membrane silicon fuel cell for a 4-mL/min H₂ flow rate. The figure shows the peak power of 15 mW/cm², under about 57-mA/cm² current load.

All the steps for this process use standard of MEMS devices manufacturing techniques, which make the industrial-level production easier and more cost-effective.

Experimental results can be seen in Fig. 17.20. A peak power density of 15 mW/cm² was obtained for a 4 mL/min H₂ flow rate at 1 atm. The maximum output voltage was 560 mV. Experimental tests were carried out at an ambient temperature of 25°C. The hydrogen flow rate of 4 mL/min was chosen to maximize generated power density.

17.5. Applications of microfuel cells

The initial applications of FCs were very diverse. Research teams were developing different electrolyte technologies, and the resulting FCs were applied, among other applications, in power generation for NASA space missions. Hydrogen and oxygen were used to produce electric power and water—products that were very much appreciated in space vehicles. The typical power of such FCs ranged between 5 and 10 kW. Larger FC systems were developed in cogeneration plant to produce both electric power and heat in domestic applications. In these cases, power was increased to an order of magnitude of 200–400 kW.

The use of FCs in portable devices first appeared in military applications. The necessity of personal communication systems with sufficient autonomy lead to the development of microfuel cells as power sources. The direct methanol fuel cell (DMFC) is the most suitable for personal applications because of its low operating temperature and the type of fuel used. DMFCs do not use corrosive liquids and have a very simple structure. The company SFC Energy Inc (http://www.sfc-defense.com) is developing the DMFC series with military applications. Jenny 1200S is the second item in that series, after the successful Jenny 600S. Its 3.3 kg weighs and a nominal power output of 50 W, and joint to a power management electronic system, increase the power range of soldier's equipment up to an energy capacity of 1200 Wh and reduce weight by 60% compared to conventional batteries usage.

The XX55 DMFC of Bren-Tronics Inc releases a similar power output (50 W) and is used in military applications including satellite communication, remote surveillance systems, and infrared vision systems.

In the area of civil applications, several companies introduced FC technology in small devices such as cell phones, MP3s, and laptop computers. For instance, the Samsung Q35 notebook was adapted to use DMFC technology, running for 10–15 h with a 100 cc methanol cartridge. In addition, Toshiba has developed a prototype FC for laptops, which is capable of generating 20–100 W of power. Moreover, the British company Intelligent Energy has developed a FC small enough to be inserted in an iPhone, although the fuel cartridge has to be placed outside of mobile.

Other manufactures, such as Micro Fuel Cell Inc, Motorola, or Fujitsu have developed rechargeable DMFCs for mobile applications. Most of them use methanol for fuel, delivering between 100 mW and 20 W with a capacity about 100 mL (Arunabha et al., 2007).

However, some drawbacks are arising recently in the marketing of these products. Although some companies, such as Samsung, have abandoned its FC development program, most of companies' portfolio includes nowadays only FC systems for nonportable devices, such as FC-based power generator systems for power grid failures or shutdowns, or similar products.

The low cost, easy charging, and low weight of lithium-ion rechargeable batteries make FC not so attractive to users of electronic portable devices. Most of countries have a well-developed power grid, and the recharging of batteries is not an annoying process to justify the current cost, weight, and marketing drawbacks of FCs. Only in emerging countries, such as India or Morocco, where a wall socket is not so accessible for users, the FC-based supply for portable devices may be currently attractive. In this way, some companies as Intelligent Energy is planning to develop the distribution infrastructure to offer the user enough charging points based in this technology.

In conclusion, FC technology has to be developed even more to be competitive all in the world and not only in places where electricity is not disposable.

17.6. Smart energy management with sun sensors

For decades, considerable effort has been devoted to research into the possibility of obtaining electrical energy in large power plants directly from sunlight by means of thermal concentration techniques. Solar collector and parabolic trough technologies have both had to overcome major technological problems to become efficient and profitable. Among them are the problems related to focusing the maximum radiation on the target; a minor deviation in alignment with the target causes a dramatic drop in system performance.

On the other hand, photovoltaic technology has been well known for a considerable time. The energy efficiency of electrooptic direct conversion is very high, but the use of static structures, to reduce installation and maintenance costs, makes impossible to get the target of collecting the maximum energy throughout the day.

Both types of system need a solar tracking technique to improve their performance. However, while the use of this technique is mandatory for concentrating solar power plant, for a photovoltaic power plant it is only optional.

Analytical methods have been used to predict the position of the sun. For instance, Mitchaltsky equations provide an analytical curve that describes the path followed by the sun in the sky; an established global positioning system coordinates with the date and time on Earth. These methods are quite accurate (error < 0.1 degree); however, they require a relatively complex implementation and periodic recalibration of each of the tracking structures. To reduce the frequency of this recalibration, robust and precise mechanical structures and engines are used, increasing the price of the system.

The use of sun sensors is an alternative to analytical methods for predicting the position of the sun. These sensors provide a direct measure of position, allowing closure of the control loop of the tracking systems. Many solar power companies include some kind of optical sensor in its parabolic trough collectors to track the solar position throughout the day. Fichtner Solar GmbH (www.fichtnersolar.com) uses a very simple photovoltaic two-cell structure in its 150 MW solar power plant in Kuraymat (Egypt). Flagsol GmbH (www.flagsol.com) has designed and engineered the solar field and heat transfer system for the AndaSol 1, 2, and 3 solar thermal power plants in Andalusia (Spain). In the AndaSol plant, the parabolic trough collectors track the sun from east to west using a high-precision optical sensor, thus collecting the maximum solar radiation. Bright Green Energy Ltd (www.wirefreedirect.com) uses a combination of optical solid states and sun sensors to align the modules to the arc of the sun and measure the solar radiation falling on the array.

Although the performance achieved by solar sensors is quite good (<0.1 degree), so far they have been made with conventional electromechanical technologies, obtaining large devices, unreliable and expensive. That is why sun sensors have not been widely used in this kind of system yet. However, the use of MEMS technology in this type of sensor overcomes these problems, so they are now being incorporated into the new tracking systems. Companies such as Solar MEMS Technologies (www.solar-mems.com) have developed an MEMS-based sun sensor for industrial applications (see Fig. 17.21), which is being incorporated into tracker structures such as parabolic trough and parabolic dish concentrators.

Moreover, sun sensors are being used in illumination applications to supply natural light to buildings. They are included in two-axe motorized mirrors to focus the solar beam in a fixed place on the building. Sun sensors allow closure of the illumination loop, providing a system with greater accuracy and flexibility and making easier the installation and maintenance procedures (Delgado et al., 2011).

In space area, the use of sun sensors has been limited to attitude control applications. Sun sensor is a reliable and common navigation instrument included in the attitude control system of satellite. All space vehicles have a solar power supply consisting of a solar cell array. Most of them may orientate it to sun, independent on the satellite body orientation. Because sun sensors are placed on the satellite body, it is not possible to use it to close a control loop for the solar cell array orientation.

So far, the strategy of maximization of the collected sun power uses the same solar cell array to search the maximum power collecting point. Including sun sensors in the solar cell frame would allow increasing the collected power, but this increment is not compensated by the expensive cost of space sun sensors. The use of MEMS technology provides small and low cost devices, with good dynamic range and accuracy and makes it not only an interesting device for power control strategy but also a very competitive navigation instrument for nanosatellites.

Figure 17.21 Microelectromechanical systems–based sun sensor, ISS-DX, for industrial applications. Courtesy Solar MEMS Technologies.

Recently, same companies have developed tiny sun sensors to be inserted between solar cells. NewSpace Systems offers a two-axis sun sensor for CubeSat designed with position sensitive detector technology. Despite good properties as a field of view (FOV) of 114 degrees, with 0.5 degrees of accuracy, 5 gr weight, and 50 mW of power consumption, this CubeSat sun sensor does not use a microsystem technology but standard optical components.

On the other hand, the nanoSSOC-A60, developed by Solar MEMS Technologies S.L., has similar functional properties (FOV of 120 degrees, 0.5 degrees of accuracy, 4 gr, and 10 mW) and includes microsystems technology in the manufacturing process. In this way, the linking of the optical interface on the microelectronic integrate circuit makes use of a critical procedure to guarantee the required accuracy.

17.6.1. Principles and structure of a sun sensor

The objective of a sun sensor is to measure the angle of incidence of sunshine radiation; in other words, the relative position of the sun in the sky. The relationship between the real angle of incidence α and its components, θ_x and θ_y, in the axes x and y (see Fig. 17.22) is given by Eq. (17.13).

\tan^{2} (α) = \tan^{2} (θ_{x}) \tan^{2} (θ_{y})

$\tan^{2} (α) = \tan^{2} (θ_{x}) \tan^{2} (θ_{y})$

(17.13)

Figure 17.22 Decomposition of a real sunlight beam in both x and y components.

Figure 17.23 Operation principle of a sun sensor.

To measure these angles, it is necessary that the beam of sunlight crosses a window and a cavity to reach a sensing element. The incidence angle of radiation can be calculated from the relative position of the illuminated area in the sensing element (see Fig. 17.23). The angular range of the sky that can be measured by a sun sensor is known as the “FOV”.

The top window can be manufactured in a variety of ways. One of the most widely used procedures involves drilling a hole in the material that fits over the sensing element. Alternatively, the window can be obtained by metallization and patterning processes on the rear surface of a glass cover, which is then bonded to the silicon die.

Realizing the window by drilling a hole in the sensing element cover leaves an empty cavity between the window and the silicon. Thus, independence between axes is guaranteed. In other words, variations on a single axis do not affect the measurement made in the orthogonal axis; the system is therefore much more linear and the data processing simpler. On the other hand, the silicon surface is exposed to external conditions, either atmospheric weather (for terrestrial applications) or dangerous radiation (for space applications).

The silicon die is protected by metallization and patterning processes on the rear surface of a glass cover; the device is therefore more robust, giving it a longer life span and allowing its use in more hazardous environmental conditions. However, using glass in the window involves a change of material in the path of the light beam, bringing Snell's law into play.

n_{1} \sin (θ) = n_{2} \sin (θ^{'})

$n_{1} \sin (θ) = n_{2} \sin (θ^{'})$

(17.14)

where n₁ and n₂ are the refractive indices of air (or vacuum) and glass, respectively, θ is the angle of incidence and θ′ the refracted angle (see Fig. 17.24).

The main drawback to this interface is that the independence of the two angles is lost. The decomposition of both original and refracted angles produces different x and y components because of Snell's law. This means that the variation of the illuminated area in the silicon die, when moving in a single axis, also depends on the angular position on the orthogonal axis.

The way to process this information and obtain accurate measurements depends on the type of technology used. Traditionally, sun sensors are classified as either digital or analog.

Figure 17.24 Refraction of a light beam crossing an interface between two different materials.

17.6.1.1. Analog sun sensor

The simplest analog structure of a sun sensor consists of a pair of silicon PR and an opaque surface with a (L × W μm²) window placed at h μm height. The light beam enters the window and illuminates both PR, and the illuminated areas (A₁ and A₂) depend on the angle of the beam. Each photodiode will generate a current proportional to the illuminated area, which will be measured by the data acquisition system (Ortega et al., 2010).

Using this structure, it is possible to measure the angle of incidence on a single axis. To obtain the angle of incidence in two axes (and therefore precisely locate the light source), it is necessary to place two such structures orthogonally.

As Eq. (17.15) shows, by using a differential measurement of these currents a dimensionless function R_i is obtained. It depends on the angle of incidence and is independent of factors that affect all the PR equally (intensity of light radiation, temperature, etc.).

R_{i} = \frac{v_{1} - v_{2}}{v_{1} + v_{2}}

$R_{i} = \frac{v_{1} - v_{2}}{v_{1} + v_{2}}$

(17.15)

Figure 17.25 Four-quadrant analog sun sensor.

Figure 17.26 Angle θ_x as a function of the sensor's responses R₁ and R₂.

By combining the functions of both axes, the characteristic curves of the sensor θ_x = f(R₁, R₂) and θ_y = f(R₁, R₂) are obtained. During normal operation, the procedure consists in measuring the voltage and calculating the corresponding R functions, and then applying them to the characteristic curve to obtain the value of the angular position (see Fig. 17.26).

17.6.1.2. Digital sun sensor

In a digital sensor, the sensing element is either a charge-coupled device or active pixel sensor image detector onto which the beam of sunlight is projected. The small dimension of the pixels means that the projection of the beam of sunlight illuminates a set of them (Liebe and Mobasser, 2001; Mobasser and Liebe, 2003). Using image processing algorithms, it is possible to obtain the centroid of this projection—and, hence, the angle of incidence—with great accuracy (Enright and Godard, 2008).

The most common digital design consists of a silicon die coated with a thin layer of chrome and a layer of gold with hundreds of small pinholes. This structure is placed on top of the image detector at a distance of less than 1 mm. Images of the sun are formed on the detector when the sun illuminates the assembly. Software algorithms must be able to identify the individual pinholes on the image detector and calculate the angle to the sun (Mobasser et al., 2001). The accuracy of this kind of sensor depends on several factors, including the pattern implemented in the mask, the holes (number, size, and separation between them), and the algorithms used to obtain the angle of incidence.

Sun sensors are classified into two groups according to the accuracy of their sensor: coarse sun sensors, which have a large FOV but a low degree of accuracy and fine sun sensors, which provide a high degree of accuracy from a small FOV. In a complete detector system, the best performance is obtained by using a combination of both kinds of sensor. A comparison of typical values of accuracy for coarse and fine sun sensors is shown in Table 17.5.

Table 17.5

Usual accuracy and field of view (FOV) of digital and analog sun sensors
	FOV (degrees)	Accuracy
	FOV (degrees)	Analog	Digital
Coarse sun sensor	±60	1 degrees	0.1 degree
Fine sun sensor	±5	0.01 degree	0.005 degree

To summarize, an analog sensor provides robustness and simple signal processing, whereas a digital sensor is more sensitive to failure and has a more complex signal processing. However, the accuracy achieved with the digital sensors is greater than provided by analog sensors.

17.6.2. Analog sun sensor manufacturing

The analog design is simple and uses large PR (about 1 or 2 mm²), so analog sun sensors are very robust against environmental conditions, making them reliable as a navigation instrument in space applications.

Among the topologies developed by different research groups, there are sensors that consist of two basic structures arranged orthogonally (Ortega et al., 2010), an advanced version with the FOV divided into sectors (Delgado et al., 2010), a four-quadrant sensor in which the axes are identified by processing the voltages in pairs (Maqsood and Akram, 2010) or a version with triangular-shaped PR (Pedersen et al., 2003). An analog design with a different concept was developed by Bohnke and Stenmark (2005), where the sensor is composed of a semispherical photodetector element. The current is generated and distributed between different electrodes. The distribution of the photocurrent to the electrodes depends on the angle of incidence of the sunlight.

The fabrication of each of these types of sensor is similar and can be summarized in three main steps: silicon die fabrication, cover glass metallization, and the bonding process (the latter two, only if the sensor has a layer of glass). The silicon die is fabricated using p-type crystalline silicon wafers.

The fabrication process of the sun sensor developed by Ortega et al. (2010) consists of the following steps and can be seen in Fig. 17.27:

1. thermal oxidation;
2. definition of the emitter regions (N + regions) of the PR using standard photolithography;
3. SiO₂ etching at the front and back surfaces;
4. SiO₂ passivation;
5. definition of the emitter contacts in N + regions by photolithography and SiO₂ etching;
6. metallization using sputtering and the lift-off technique;
7. Al metallization on the rear surface and thermal annealing;
8. metallization in the pads and thermal annealing;
9. laser-fired contacts in the back surface.

Figure 17.27 (a–i) Fabrication process of a two-photodiode sun sensor. See text for explanation of steps.

The cover glass metallization is made as follows:

• photoresist deposition on the front and lithography;
• Al metallization using the sputtering technique;
• metal patterning using the lift-off technique.

Finally, a transparent and nonconductive epoxy resin is deposited carefully on the silicon die in four peripheral points (to avoid covering the electrodes and windows) to bond the silicon and glass cover. Fig. 17.28 shows a prototype of the analog sun sensor developed by Ortega et al. (2010).

17.6.3. Sun sensor strategy for energy management

A sun sensor is used in either of the following modes of operation:

• full operating range (full FOV);
• around a reference point, usually the origin (0.0).

Figure 17.28 Silicon-based prototype of an analog sun sensor.

Like most sensors, sun sensors can be used merely as an instrument for measuring the sun's position in a given range or they can be incorporated into a tracking system as an active element. When incorporated into a tracking system, the sensor usually works around the origin (0.0).

Sensor requirements are different for each of these applications. In the full operating range, one or more sensors with wide FOV (e.g., ±60 degrees) is necessary to cover the full angular field where the sun position has to be known. The system operates in an open loop, and the required accuracy is high throughout the operating range. To achieve this precision, the sensor has to be calibrated before it is deployed. In this way, manufacturing tolerances are compensated and no unacceptable errors occur in the measurement function.

Good calibration requires the use of a solar simulator and a positioning system, which range throughout the FOV and measure the response of this particular sensor in that range. This information should be included in a memory device so that it can be used later by a microprocessor that is responsible for performing the positional calculations. Both the microprocessor and the memory can be included in the device itself, as in the case of a digital sensor.

The relationship between the difference in optical power (P₁₂) received by PR 1 and 2 of the sensitive element and the difference in currents generated by both of them (I₁₂) is determined by the function I₁₂ = f(P₁₂). If the sensor operates close to the origin of reference, both differences are close to zero (P₁₂ ∼ 0), and the function f can be approximated, as shown by Eq. (17.16):

I_{12} = 0 + \frac{\partial f}{\partial P_{12}} P_{12} = S P_{12}

$I_{12} = 0 + \frac{\partial f}{\partial P_{12}} P_{12} = S P_{12}$

(17.16)

Eq. (17.15) shows that the differences of incident light and electrical current (P₁₂ and I₁₂) are related by a coefficient known as sensitivity S.

Figure 17.29 Equivalent model of a sensitive element.

Figure 17.30 Equivalent model of an actuator system.

Figure 17.31 Equivalent model of a closed-loop tracking system.

Fig. 17.29 shows the equivalent model of the sensitive element (S) whereby the incident optical power (P₁₂) is converted into electricity (I₁₂). Using an electronic process, represented by A, the angle of incidence of radiation is given as a voltage V. This electronic process includes both the current–voltage conversion and signal amplification stages.

A tracking system includes a set of motors that are driven by a control signal V moving the support structure, which causes a variation in the difference in optical power received by the PR. This operating mode is outlined in Fig. 17.30.

Including a solar tracking system, we can achieve a sensor that works around the origin (0.0). The system operation is shown in Fig. 17.31.

The relationships of different variables of the system are described by

V_{θ} = A I_{12}

$V_{θ} = A I_{12}$

(17.17)

I_{12} = S Δ P

$I_{12} = S Δ P$

(17.18)

Δ P = P_{12} - P_{R}

$Δ P = P_{12} - P_{R}$

(17.19)

P_{R} = R_{M} - V_{θ}

$P_{R} = R_{M} - V_{θ}$

(17.20)

where

V_{θ} = AS (P_{12} - R_{M} V_{θ})

$V_{θ} = AS (P_{12} - R_{M} V_{θ})$

(17.21)

(1 - {ASR}_{M}) V_{θ} = {ASP}_{12}

$(1 - {ASR}_{M}) V_{θ} = {ASP}_{12}$

(17.22)

V_{θ} = \frac{{ASP}_{12}}{(1 - {ASR}_{M})}

$V_{θ} = \frac{{ASP}_{12}}{(1 - {ASR}_{M})}$

(17.23)

Supposing that ASR_M ≫ 1, then

V_{θ} = \frac{P_{12}}{R_{M}}

$V_{θ} = \frac{P_{12}}{R_{M}}$

(17.24)

In conclusion, when the sun sensor works in a closed loop

• the scale factor is determined by the transducer, which is usually very linear, precise, and repetitive;
• the nonlinearity of the sensor (f) is removed because the sensor is working at all times around 0.0, so higher accuracy will be achieved.

The requirements of sun sensors used in this operating mode are typically those of a fine sun sensor, where the FOV and accuracy are usually less than ±15 degrees and 0.1 degree, respectively. The use of fine sun sensors around the origin (0.0), where accuracy is even greater, maximizes the performance of photovoltaic energy conversion by using tracking systems. In addition, a narrow FOV improves the problem of albedo. Although a fine sun sensor is the best option for the closed-loop control of a tracking system in transient situations, such as at sunrise or in partly cloudy weather, the sensor may be out of the FOV. In this situation, a coarse sun sensor works more robustly because of its wider FOV. A combination of both types is usually employed.

17.7. Conclusion

In this chapter, we showed photovoltaic and FC technologies as a suitable option for powering MEMS for smart energy management, such as bioimplantable devices or portable electronic devices.

We introduced the basic working principles of photovoltaics and followed this by describing the fabrication processes of different photovoltaic mini-module approaches based on silicon: MCM, SOI, and fusion-bonding technologies.

New photovoltaic devices and sensors are emerging continuously. The materials, structures, and strategies used have a fundamental role in achieving a successful application.

..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.

Table of Contents for 17. Photovoltaic and fuel cells in power microelectromechanical systems for smart energy management

Create new playlist

Sign In

Sign Up