Full Information Acquisition

Table 2 : G-Mode Limitations Problem or Limitation

Large data volumes: classical 2D SPM image is 1 MB, an equivalent full G-Mode image is 4 GB


Compression with minimal information loss reduces the data volume to ~40 MB depending on information content in data.

Due to the emergence of cloud storage and high-density disk arrays, storage has become a much smaller issue, even in comparison to the scenario 5 years ago; this is a trend that is expected to continue in the future.

Information captured in G-Mode is simply inaccessible using classical methods, making the technique worthwhile.

Lossless compression allows full reconstruction of signal for continuous analysis with a wide variety of methods.

Sparse methodology using intrinsic data structure provides a pathway to a much higher degree of compression than currently available.

Information theory methods such as PCA ignore relevant physics and hence can be difficult to interpret

Physics-based compression methods (lossy compression) have been demonstrated in other communities and can be adapted for SPM dynamics.

More robust Bayesian inference, as well as supervised learning methods with built-in physical constraints, are available to analyze G-Mode data.

No current feedback system in G-Mode: precludes realization of frequency-tracking modes

Presently, feedback can be run in parallel to classical SPM processing.

Fast FPGA processing will allow real-time feedback. T e tools are available and are being implemented in next iterations of the G-Mode techniques.

Information overload: full cantilever trajectory as a function of force or voltage excitation often contains many more components than are captured by classical methods, overloading the user

T is information is relevant to material behavior and is rooted in real phenomena that is otherwise lost or ignored.

Broad-based theory support is developing a mathematical framework for extracting material properties from probe trajec- tories, mixed harmonic signals, etc. However, these are generally limited due to the novelty of the technique.

Synergy with the global open-source soſt ware development community (for example, through GitHub distribution) will allow eff ective deployment and a dialogue in the scientifi c community.

Serendipitous phenomena: transients, single events, etc. complicate true material behavior reconstruction

Information without clear interpretation can always be analyzed as new theories are developed.

G-Mode can always be reduced to classical methodology of lock-in, or Band Excitation techniques.

Practically, this new technique can elucidate material behavior coupling, or help characterize the measuring tool itself, decoupling real physical properties from instrumentation bias.

and the corresponding loading map shows characteristics of the transient cantilever response induced by the edges of topographical features, resembling the error signal in the force feedback loop. Appropriate signal de-mixing algorithms can be used to decouple domain contrast from topographic information. T ese G-PFM data can also be analyzed using clustering algorithms, independent component analysis, Bayesian linear unmixing methods, and correlational analysis techniques [ 48 , 81 – 86 ]. T e G-PFM approach can also be extended to study polarization switching in ferroelectric materials, as implemented in G-Mode voltage spectroscopy (G-VS) [ 87 ]. In G-VS, the amplitude of the G-PFM excitation signal is increased beyond


the coercive bias of the sample. G-VS uses data-driven adaptive signal fi ltering techniques to reveal strain loops that are indicative of polarization switching in ferroelectric materials, as shown in Figure 5a . T e raw data itself is used to calculate an appropriate noise-fl oor for the signal at each pixel. A band-pass fi lter only retains the signal from the excitation frequency to 10–12 harmonics of the drive frequency, and any signal below the calculated noise fl oor is rejected. T e fi ltered signal reveals numerous bias-induced strain loops. Such extraction of hysteresis loops would not be possible without the complete data or by using the traditional heterodyne detection schemes since the response frequencies are not known a priori . Compared to the current state-of-art • 2017 July

Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40  |  Page 41  |  Page 42  |  Page 43  |  Page 44  |  Page 45  |  Page 46  |  Page 47  |  Page 48  |  Page 49  |  Page 50  |  Page 51  |  Page 52  |  Page 53  |  Page 54  |  Page 55  |  Page 56  |  Page 57  |  Page 58  |  Page 59  |  Page 60  |  Page 61  |  Page 62  |  Page 63  |  Page 64  |  Page 65  |  Page 66  |  Page 67  |  Page 68  |  Page 69  |  Page 70  |  Page 71  |  Page 72  |  Page 73  |  Page 74  |  Page 75  |  Page 76