Multispectroscopy generally refers to some kind of imaging or scanning done over a range of frequency (wavelength) broader than the range of one device. Dielectric spectroscopy refers to the measurement of impedance (in some form) over a frequency range. This is different. It’s very broad. It really is a “multi”-spectroscopic system. The idea is to be able to automatically apply a wide range of stimuli (for instance a spectrum in both frequency and/or electrical field magnitude and/or temperature and/or what have you) to a material or a process, measure the response (or responses), and then to be able to conveniently examine those responses in order to understand a property or behaviour.
For example: Permittivity
Most scientists are generally familiar with a material’s “dielectric constant” and many know that there’s not really any such thing because it’s not constant and a lot of “dielectrics” in fact conduct enough under a lot of conditions to forfeit the title “dielectric.” So materials are often described by the permittivity “epsilon,” a complex number whose real part is the dielectric “constant,” and imaginary part is related to the conductivity and which is known to change in response to many things including temperature, frequency of applied AC signal, magnitude of applied AC signal, superimposed DC signal, and time (real materials age from many processes). Furthermore the permittivity has frequency harmonics which stem from its nonlinearity in addition to contributions from “irreversible” processes involving charge dipoles and domain walls. Not really dielectric and not really constant.
To understand the physical processes in a material or to learn how well a material performs in real-life applications we would like to know the permittivity under lots of conditions, particularly those mentioned above. Most related processes manifest themselves most clearly when the permittivity is characterized over a particular set of conditions. For instance phase transitions are especially obvious in the temperature domain, but evidence may also be seen in the frequency domain. Conduction phenomena appear almost magically in the frequency domain and often have well behaved activation energies visible in the temperature domain. Rayleigh behaviour is apparent in the AC field magnitude domain, again influenced by temperature (isn’t everything). Aging processes present themselves over time, temperature, and DC bias. Nonlinear polarization phenomena show up as the magnitude of the AC and DC fields are changed. The characterization over a decent piece of some multidimensional space can be the work of years, even decades. Doctoral degrees have been awarded for doing just that. And even if you have the data for eps(freq, Temp, time, AC, DC, harmonic), forming a coherent picture is nearly impossible because what you’ve got is a pile of graphs from a thousand individual experiments done at different times under different conditions often by different people with different samples. Doctoral degrees have been awarded for that too, collating a ton of independently acquired information about some material and producing a single coherent picture (or at least reducing the number of conflicting pictures), especially by relating apparently unrelated phenomena.
Surprisingly, there don't appear to be any good solutions to automatically acquire such broad information about materials and systems. Some companies sell equipment specifically to measure, say, permittivity as a function of temperature and frequency but then fall short of being able to measure other electrical (and mechanical) properties (low values of resistance like superconductivity, leakage current, harmonics, polarization), vary fields at all but especially over a range of kilovolts, or change time durations and sequences, and certainly not be able to control repeated cycles of temperature and time. And then they can only represent the data as permittivity(temperature,frequency) so winkling out the behaviour as a function of field or time is up to you. And once you own their thing you have only that specific thing, you can't use it with other meters or temperature stages.
National Instruments sells Labview which is billed as a solution but, as anyone who uses it knows, requires a lot of knowledge in VI programming to do anything but the simplest measurement and is possibly the world's most inflexible environment. Changing a measurement to add the characterization of another variable requires reprogramming the VI because it changes the dimensional space, and substituting a different temperature controller or meter can be similarly complex.
Now a little hyperbole, but less than you might think
Wouldn’t it be great if you could, say, write down some directions like:
That’s scary close to what this system does. It's composed of two pieces of software and either some commonly-available measurement hardware like LCR meters, Network Analyzers, Lock-in Amplifiers, pico-Ammeters, and DMM's or some specialized hardware for charge measurements (or both). The first computer program (GADD) takes directions in the form of a short text file called a script file and controls the measurement hardware so as to do what you asked it to do. The second computer program (VIZ) reads the data file produced by the measurement and lets you interactively view the potentially multiple pieces of potentially 8-dimensional data in a coherent way, generally as a surface formed by two of the measurement parameters (which you pick interactively when you're examining the data, not locked into when you make the measurement) with any others held at fixed values (which you also pick). And it lets you save the (generally reduced to) 1 or 2 dimensional data as spreadsheets that the commonly available graphing programs like Excel, Origins, Kaleidagraph, Mathematica, and Matlab can read. At this point you might want to have a peek at the Block Diagram and the Examples which are real data files created by GADD that you can examine by downloading VIZ. At which point a lot of the last paragraph should be a lot clearer.
Oh yeah, almost forgot, here are scripts that could do the measurements in the wishlist above (with the right hardware, of course):
1
2
3
4
When you look at these I would appreciate it if you noticed a couple of things. First, all of the parameters are treated identically in the script and the viewing (except for a small setting applied to the temperature control, #2 ). For the scientist, measuring and viewing AC, DC, frequency, time, temperature, and harmonics are all treated identically. An aging experiment, a heating experiment, a frequency sweep, a field sweep, and more are all performed and processed identically (from the user's point of view). And the measured information is processed identically whether it is a real number, a complex number, or a list. You don't need different equipment or different programming to do very much different experiments. For instance, one wouldn't think of an aging experiment of nonlinear polarization and a frequency sweep of circuit gain as similar experiments to be done with the exactly the same control software, but you can do just that, and the scripts to do them aren't much different. Of course the hardware you would use would be a lot different, but nothing that doesn't work with this system.
Second, if you can ignore the //'s in the script files to help explain them, they are very short and very simple, as is viewing the results. Even when those results contain half a dozen different measured properties, each of which could have a different number of dimensions. It's entirely possible to do an experiment and end up with something that is measured as a function of temperature, something else that is measured as a function of temperature and frequency, something else that is measured as a function of temperature, frequency, and AC field... But the user doesn't have to handle any of that. A truly useful measurement system is simple at the beginning (measure X as a function of this, that, and the other thing) and simple at the end (show me X as a function of this and that, when the other thing is big/small/in between). The user should never experience the painful (I would like to say "backbreaking") complexity in the middle.