Filtern
Erscheinungsjahr
Dokumenttyp
- Vortrag (8)
- Posterpräsentation (6)
- Zeitschriftenartikel (3)
- Beitrag zu einem Tagungsband (3)
- Forschungsdatensatz (2)
- Video (1)
Schlagworte
- Data analysis (23) (entfernen)
Organisationseinheit der BAM
- 6 Materialchemie (14)
- 6.5 Synthese und Streuverfahren nanostrukturierter Materialien (11)
- 6.0 Abteilungsleitung und andere (3)
- 5 Werkstofftechnik (2)
- 5.1 Mikrostruktur Design und Degradation (2)
- 1 Analytische Chemie; Referenzmaterialien (1)
- 1.5 Proteinanalytik (1)
- 6.3 Strukturanalytik (1)
- 6.6 Physik und chemische Analytik der Polymere (1)
- 7 Bauwerkssicherheit (1)
Eingeladener Vortrag
- nein (8)
Currently research in chemical manufacturing moves towards flexible plug-and-play approaches focusing on modular plants, capable of producing small scales on-demand with short down-times between individual campaigns. This approach allows for efficient use of hardware, a faster optimization of the process conditions, and thus, an accelerated introduction of new products to the market [1]. Driven mostly by the search for chemical syntheses under biocompatible conditions, so-called “click” chemistry rapidly became a growing field of research. The resulting simple one-pot reactions are so far only scarcely accompanied by an adequate optimization via comparably straightforward and robust analysis techniques. Here we report on a fast and reliable calibration-free online high field NMR monitoring approach for technical mixtures. It combines a versatile fluidic system, continuous-flow measurement with a time interval of 20 s per spectrum, and a robust, automated algo-rithm to interpret the obtained data. All spectra were acquired using a 500 MHz NMR spectrometer (Varian) with a dual band flow probe having a 1/16 inch polymer tubing working as a flow cell. Single scan 1H spectra were recorded with an acquisition time of 5 s, relaxation delay of 15 s.
A round-robin study has been carried out to estimate the impact of the human element in small-angle scattering data analysis. Four corrected datasets were provided to participants ready for analysis. All datasets were measured on samples containing spherical scatterers, with two datasets in dilute dispersions and two from powders. Most of the 46 participants correctly identified the number of populations in the dilute dispersions, with half of the population mean entries within 1.5% and half of the population width entries within 40%. Due to the added complexity of the structure factor, far fewer people submitted answers on the powder datasets. For those that did, half of the entries for the means and widths were within 44 and 86%, respectively. This round-robin experiment highlights several causes for the discrepancies, for which solutions are proposed.
How much do we, the small-angle scatterers, influence the results of an investigation? What uncertainty do we add by our human diversity in thoughts and approaches, and is this significant compared to the uncertainty from the instrumental measurement factors?
After our previous Round Robin on data collection, we know that many laboratories can collect reasonably consistent small-angle scattering data on easy samples1. To investigate the next, human component, we compiled four existing datasets from globular (roughly spherical) scatterers, each exhibiting a common complication, and asked the participants to apply their usual methods and toolset to the quantification of the results https://lookingatnothing.com/index.php/archives/3274).
Accompanying the datasets was a modicum of accompanying information to help with the interpretation of the data, similar to what we normally receive from our collaborators. More than 30 participants reported back with volume fractions, mean sizes and size distribution widths of the particle populations in the samples, as well as information on their self-assessed level of experience and years in the field.
While the Round Robin is still underway (until the 25th of April, 2022), the initial results already show significant spread in the results. Some of these are due to the variety in interpretation of the meaning of the requested parameters, as well as simple human errors, both of which are easy to correct for. Nevertheless, even after correcting for these differences in understanding, a significant spread remains. This highlights an urgent challenge to our community: how can we better help ourselves and our colleagues obtain more reliable results, how could we take the human factor out of the equation, so to speak?
In this talk, we will introduce the four datasets, their origins and challenges. Hot off the press, we will summarize the anonymized, quantified results of the Data Analysis Round Robin. (Incidentally, we will also see if a correlation exists between experience and proximity of the result to the median). Lastly, potential avenues for improving our field will be offered based on the findings, ranging from low-effort yet somehow controversial improvements, to high-effort foundational considerations.
This is a remote presentation I gave at the 2022 Small-angle Scattering conference in Campinas, Brazil. The video has been obtained from the conference organisers with their explicit permission for use on YouTube. I've tried to spruce up the audio from the remote recording the best I could.
The conference abstract for this talk was:
"How much do we, the small-angle scatterers, influence the results of an investigation? What uncertainty do we add by our human diversity in thoughts and approaches, and is this significant compared to the uncertainty from the instrumental measurement factors?
After our previous Round Robin on data collection, we know that many laboratories can collect reasonably consistent small-angle scattering data on easy samples[1]. To investigate the next, human component, we compiled four existing datasets from globular (roughly spherical) scatterers, each exhibiting a common complication, and asked the participants to apply their usual methods and toolset to the quantification of the results (https://lookingatnothing.com/index.ph....
Accompanying the datasets was a modicum of accompanying information to help with the interpretation of the data, similar to what we normally receive from our collaborators. More than 30 participants reported back with volume fractions, mean sizes and size distribution widths of the particle populations in the samples, as well as information on their self-assessed level of experience and years in the field.
While the Round Robin is still underway (until the 25th of April, 2022), the initial results already show significant spread in the results. Some of these are due to the variety in interpretation of the meaning of the requested parameters, as well as simple human errors, both of which are easy to correct for. Nevertheless, even after correcting for these differences in understanding, a significant spread remains. This highlights an urgent challenge to our community: how can we better help ourselves and our colleagues obtain more reliable results, how could we take the human factor out of the equation, so to speak?
In this talk, we will introduce the four datasets, their origins and challenges. Hot off the press, we will summarize the anonymized, quantified results of the Data Analysis Round Robin. (Incidentally, we will also see if a correlation exists between experience and proximity of the result to the median). Lastly, potential avenues for improving our field will be offered based on the findings, ranging from low-effort yet somehow controversial improvements, to high-effort foundational considerations."
This poster deals with improvements and characteriztion of small-angle scattering limitations, by looking at the trifecta of Data collection and uncertainty propagation, data analysis methodologies, and real-world tests. It is found that - with appropriate care and instrumentation - accuracies of 1% on mean nanomaterial sizes, and 10% on the size distribution width as well as the volume fraction can be achieved.
Small-angle Scattering Data Analysis Round Robin: anonymized results, figures and Jupyter notebook
(2023)
The intent of this round robin was to find out how comparable results from different researchers are, who analyse exactly the same processed, corrected dataset.
This zip file contains the anonymized results and the jupyter notebook used to do the data processing, analysis and visualisation. Additionally, TEM images of the samples are included
These are four datasets that were made available to the participants of the Small-angle Scattering data analysis round robin. The intent was to find out how comparable results from different researchers are, who analyse exactly the same processed, corrected dataset.
In this repository, there are:
1) a PDF document with more details for the study,
2) the datasets for people to try and fit
3) an Excel spreadsheet to document the results.
Datasets 1 and 2 were modified from: Deumer, Jerome, & Gollwitzer, Christian. (2022). npSize_SAXS_data_PTB (Version 5) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.5886834
Datasets 3 and 4 were collected in-house on the MOUSE instrument.
Data analysis of SAS measurements has been dominated by the classical curve fitting approach. This method finds optimal parameters of a scattering model composed of analytical expressions. SASfit represents such a classical curve fitting toolbox: it is one of the mature programs for small-angle scattering data analysis and has been available and used for many years. The latest developments [1] will be extended by improving the interoperability of the extensive data base of models with third-party analysis software. An updated format of model definitions is also presented, which allows model function plug-ins to be used with the Python language.
To complement the classical curve fitting method, the user-friendly opensource Monte Carlo regression package McSAS was developed. Most importantly, the form-free Monte Carlo approach of McSAS means that it is not necessary to provide any further mathematical restrictions to the Parameter distribution. Future developments include separating the core optimization from the GUI (allowing 'headless' integration), as well as parallel computing which reduces the computing time proportional to the number of available computing cores. The headless mode is presented by an example of Operation within interactive programming environments such as a Jupyter notebook.
The promising results of Monte Carlo based data analysis for determining form-free Parameter distributions motivated the evaluation of the method with dynamic light scattering (DLS) data. For this purpose, the method was adapted for analyzing correlation curves such as those from multi-angle dynamic light scattering (DLS) data. The development of McDLS intends to overcome limitations of existing methods at reliably determining the modality of size distributions. An example of Monte Carlo based data analysis of multimodal DLS measurements will be presented.
To ensure a high operational reliability of offshore wind turbines (OWEC) with economically acceptable repair and maintenance efforts, comprehensive diagnosis and supervision concepts are required. Automatic monitoring Systems will be an essential part of such concepts. Because of the fact, that during Operation there will be static and dynamic interaction between the components ‘structure’, ‘machinery’ and ‘blades’ it is necessary to develop the monitoring techniques in an overall concept. These monitoring Systems are supposed to be applied for the design and testing as well as for the Operation and maintenance phases. The knowledge of the dynamic behavior of wind turbines is important both for the design and for a safe Operation. The available monitoring data from a period of three years, allow first conclusions on the long-term Operation of such Systems in terms of quality requirements to the instrumentation to the structure and the rotor blades