Radiant Technologies, Inc. 2835D Pan American Freeway NE Albuquerque, NM 87107 Tel: 505-842-8007 Fax: 505-842-0366 www.ferrodevices.com e-mail: radiant@ferrodevices.com Chapter 3: Theory of Paraelectric Capacitance Joe T. Evans, Jr. Introduction: Linear capacitors are not very interesting: they just make straight lines on a graph of charge and voltage. Fortunately, we can put highly complex materials between the plates of a capacitor and generate highly complex behaviors from those capacitors. These nonlinear capacitors are critical to the operation of our society where they are used for sensors, variable tuning devices, filters, and transducers. The discussion below will contrast linear capacitors with a type of commonly used non-linear device. A Review of Linear Capacitance: All capacitors can be charged by the application of a voltage. Q = CV eq(1.3) We can rewrite the equation as a ratio where the capacitance C is the “slope” of the function: Q/V = C eq(3.1) For the purpose of this discussion the distinction between linear and non-linear devices is simple: a linear device has a constant capacitance at all voltages at a constant temperature and pressure. A non-linear capacitor does not. Hysteresis of a 1nF Ceramic Disk Capacitor 0.005 0.004 C = ∆Q/∆V 0.003 Charge (µC) 0.002 ∆Q1 0.001 ∆V1 0.000 -0.001 ∆Q3 -0.002 ∆Q2 -0.003 ∆V2 ∆V3 Linearity means: C1 = C2 = C3 -0.004 -0.005 -5 -4 -3 -2 -1 0 Voltage 1 2 3 4 5 Figure 3.1 The slope of a linear capacitor is constant at all voltages. It is important to notice that the conditions here are for constant temperature and pressure. The reason for this is that there is no truly linear dielectric except a vacuum. All materials change their relative dielectric constant at higher electric fields. They also change their capacitance at higher temperatures and higher pressures. The easiest way to visualize this concept is to remember the equation for capacitance: ∴Capacitance = C = εo • A / t eq(1.2) The thickness of the capacitor is part of the equation. If we compress a capacitor with even a millionth of a newton of force (about the force of the spoken voice), “t” changes and the capacitance increases a very small amount. If we increase the temperature, the capacitor increases its size due to the coefficient of thermal expansion, again changing “t” in the denominator of equation (1.2). The situation is even further complicated by friction. For example, a coat hanger heats up when you bend it back and forth. A capacitor has a little “friction” when you charge it up and discharge it. Its temperature will rise, changing its capacitance a little bit. You can even make the case that the temperature of the capacitor will increase even as you are increasing the voltage across the capacitor, causing its charge vs voltage curve to arc downwards slightly at each higher voltage. Of course, the capacitor will lose excess heat to its package and the air around it so that if you charge it slowly, the average 2 temperature will remain roughly constant and the change in capacitance effect of temperature from “friction” will be less. We have now reached an amazing level of theoretical complexity while analyzing the charge vs voltage curve of a linear capacitor. We just proved that any capacitor with an ε greater than εo is not truly linear. In reality, it is nearly impossible to measure the theoretical non-linearity because there are real limitations to measurement accuracy due to noise, time, and sensitivity. As far as we are concerned in this discussion, a traditional linear capacitor is linear because our best measurements indicate that it is so. Our concern is with capacitors having dielectric materials that have large non-linearities by their nature. These materials show large changes to their dielectric constants that are easy to see in real devices. Non-Linear Capacitors: Let’s jump right into it. Figure 3.2 is a plot of a non-linear capacitor called a paraelectric capacitor. Radiant 9/65/35 PLZT [ 1700A ] 60 50 ∆Q2 ∆V2 40 30 Polarization 20 10 ∆Q1 0 ∆V1 -10 -20 -30 -40 Non-linear Capacitance: C1 ≠ C2 -50 -60 -30 -25 -20 -15 -10 -5 0 Volts 5 10 15 20 25 30 Figure 3.2 A Paraelectric Capacitor How did this happen? Why does the capacitance change with voltage? Remember that in the lesson on Chapter 1 - Linear Capacitance, I described how the relative dielectric 3 constant arising from the electric field inside the capacitor causes the electrons and atoms of the dielectric material between the plates to separate a little bit. Well, the material used for the dielectric material in the capacitor of Figure 3.2 has a lot of atoms in its crystal lattice stuffed into a very small space, about 4 angstroms on a side. An angstrom is equal to 1/10th of a nanometer. An atom usually occupies a sphere about 1 angstrom in diameter and this material has an entire titanium or lanthanum atom, ½ of six oxygen atoms each, and 1/8th of eight lead atoms occupying that 4Å x 4Å x 4Å cube. There is not a lot of room so that when the electric field is applied, the atoms in their entirety move to compensate. This gives the material a high relative dielectric constant, ranging from a value of 100 on the low side for some materials to 30,000 on the high side. Note that there are no units given for the relative dielectric constant. εr has no units because it multiplies εo which does have units. When the atoms move in an electric field, some move towards the positive side of the field and some move away. The material stretches. But, it can only stretch so far. Its initial movement is large, giving us C1 in Figure 3.2. But, it runs out of room to move at higher voltages, giving us C2 in Figure 3.2. Capacitor Plate E Dielectric material The dielectric material expands in one direction and shrinks in the other. Figure 3.3 Non-linear Capacitor Another issue in Figure 3.2 is that the loop is open, not closed. The cause for this effect is very complex. Suffice it to say that this is one of those “friction” effects mentioned in the earlier section of this chapter about linear capacitors and we will see it again during our experiments. 4 The material shown in Figure 3.2 is called a “paraelectric” material and is identified by the classic “S” shape of its charge vs voltage curve. This particular material goes by the acronym PLZT, which stands for Lead (Pb) Lanthanate (La) Zirconate (Zr) Titanate (Ti). Its chemical formula is (PbLa)(ZrTi)O3. PLZT and related materials like Bismuth Titanate (BiTiO3) are ceramics descended from the Native American pottery of the American Southwest where Radiant makes its home. The various metal oxides are mixed together and heated to transform the mixture into a ceramic. The Native Americans used the ceramics to hold food and water. We use these ceramics in electronics and sensors. Small amounts of dopants are added to the mixture to adjust the properties of the final material. In the case of the PLZT in Figure 3.2, the lanthanum is a dopant to PZT that changes the capacitor from a “ferroelectric” material with memory, which we will discuss a little later, to a “paraelectric” material with no memory. The lanthanum makes the PLZT transparent. PLZT is very fast electrically. Its change in capacitance means that its dielectric constant changes with the applied voltage. As we learned in Chapter 2 - The Importance of Dielectric Constant, the speed of light in PLZT changes with the applied voltage so we can use PLZT to modulate light. Another paraelectric material, Lithium Niobate (LiNbO3), is used to make the electro-optic modulators that feed information into the optical fibers that form the backbone of the Internet from which you may have downloaded this document! Let’s introduce another method of plotting the measured data, a method that provides information about the inner workings of the capacitor. In Figures 3.1 and 3.2, the measured charge (or polarization) was plotted against the voltage. Both figures also show specific points where the slope of the measured data is calculated. As equation (3.1) shows, the slope of the measured data is the capacitance at that point of the measurement. On a paraelectric capacitor, the slope is not constant, meaning that its capacitance changes as a predictable function of the voltage profile. Using calculus, we can calculate the slope of the measured data at each point of the measurement and then make a plot of the slope vs the voltage, as in Figures 3.5 and 3.6. Figure 3.5 shows the hysteresis measurement of a commercial high voltage capacitor capable of withstanding 6,000 volts, a device used in control circuitry for manufacturing. High voltage capacitors are usually made with a form of PZT. A picture of this particular capacitor is in Figure 3.4. 5 Figure 3.4 6,000V 3.3nF Commercial High Voltage Capacitor High Voltage Paraelectric Capacitor [ 3.3nF 6000V Rating ] 10.0 7.5 5.0 Charge (µC) 2.5 0.0 -2.5 -5.0 -7.5 -10.0 -5000 -4000 -3000 -2000 -1000 0 1000 2000 3000 4000 5000 Voltage Figure 3.5 5,000V Hysteresis of the Commercial High Voltage Capacitor in Figure 3.4 6 Note that the maximum voltage of our test in Figure 3.4 is 5,000V. Now, in Figure 3.6, we plot the capacitance vs voltage function of the same sample. Figure 3.6 shows the slope of the hysteresis at each voltage in Figure 3.5. High Voltage Paraelectric Capacitor [ 3.3nF 6000V Rating ] 0.008 0.007 Capacitance (µF) 0.006 0.005 Linear Capacitance 0.004 0.003 0.002 0.001 0.000 -5000 -4000 -3000 -2000 -1000 0 1000 2000 3000 4000 5000 Voltage Figure 3.6 The Measured Capacitance vs Voltage Function of the Capacitor in Figure 3.5 Figure 3.6 is a magnificent demonstration of the paraelectric function. The dashed line across the plot represents what a linear capacitor would look like on the same plot. It is a horizontal line because its capacitance never changes. This paraelectric capacitor changes its capacitance by at least a factor of four over the ±5,000 volt test. It is rated as a 3.3n F capacitor. Note that it actually starts at 4nF (0.004µF) but then drops to 1nF by 5,000 volts. Why use a paraelectric material in place of a linear capacitor when the linear capacitor will be constant? The reason is that dielectric materials that act in a linear fashion usually have low dielectric constants. Paraelectric materials can have very large relative dielectric constants, sometimes approaching 30,000. To get the desired capacitance 7 value, a linear capacitor would have to be hundreds or even thousands of times larger in area than the capacitor made with paraelectric material. That my not be practical. Most ferroelectric materials derive their properties from the same crystal lattice characteristics as paraelectric capacitors. In the next chapter, we will see how adding “memory” to a paraelectric capacitor gives us the classic ferroelectric hysteresis curve. 8