2.1 Entity framework core

The Entity Framework (EF) Core is a framework for storing data that allows developers to work with data at a higher level of abstraction to eliminate the programming need for most of the data-access codes. The main advantage of the EF core is its ability to utilize built-in classes in the framework and store instances in different types of data stores. To be specific, the data model, a collection of classes in object-oriented programming (OOP), is the center of the application. The data model describes how the XPS data are stored along with the related logic for the application and data validation constraints. Object-to-data store mapping technology coined EF Code is a cross-platform data-access framework for .NET, while an object-relational mapper (ORM) is designed to handle data access. ORM applies a mapping definition to associate columns in database tables with properties of the various Classes. There are two kinds of EF Code:

• Database-First: if there is an existing database, reverse engineering can be used to build the data model and context.
• Code-First: if a database does not exist, one must define a model by creating classes in C# and then use the EF Code to generate a new database that matches the structure and features of model classes.

As this was an update to the previous version (Lee et al., 2002), the Database-First approach was selected to build a data model. It is important to note that our data model comprises a series of classes with each one corresponding to a table in the relational database (Codd, 1970). All data model classes must interact with a built-in database context class (DbContext) in the EF Code. Database context class represents a session with the database, understands how to communicate with the database, and dynamically generates SQL (structured query language) statements by executing a language-integrated query (LINQ) to query and manipulate data.

2.2 User interface

The .NET Core Blazor is the web framework of the .NET Core ecosystem for developing user-interfaces (UIs). For speedy development, the open-source Radzen⁵ Blazor component library (Materials and Methods: 3) was chosen to build interactive web UIs. Figure 1 shows an overview of the Blazor Server architecture. SignalR is an open-source, real-time communication library that moves data back and forth between the web server and the client-side browser bidirectionally as well as manages real time UI and data updates. The Blazor components are rendered on the Blazor Server which can directly access the database with an EF Core.

To use Blazor in the application, three items were needed:

1. A reference to blazor.server.js,
2. Balzor services in the service container (Dependency injection, DI), where DI is ‘a technique for accessing services configured in a central location’.
3. Blazor’s SignalR hub

The database context, XPSContext, connects the XPS relational database directly between the Blazor components and database access. There is a small custom-built Data Service layer called XpsService which implements the Interface, IXpsService, so that we can reuse our developed data access codes for the entire SRD 20 development.

2.3 Search strategies and implementations

The NIST XPS App has a modern UI design (see Supporting information (SI) Figure S1) with several new features that will now be briefly described. The first new feature is a custom-built data validator. If a user wished to identify an unknown line in a measured XPS spectrum, they would click on the first option of the menu, select an energy type, and enter an observed energy and its estimated uncertainty, as shown in Figure 2. The observed energy is then checked with the custom-built data validator (Section 1 in the SI) to ensure that it is within the allowed data range for the selected energy type. If one or more matches are found, the relevant data records are listed that show the element, spectral line, chemical formula, and energy value as well as two hyperlinks (see Figure S2). One hyperlink (Details) provides metadata from the data record with information on the specimen material, the measurement conditions, and the literature citation (shown in Figure 3). The second link (Other Data) can be used to immediately find other data in the database from the same paper, as shown in Figure 4. A user can click on the column headers in the displays of Figures S2 and 4 to sort the information alphabetically (element, formula, and spectral line) or numerically (atomic number and energy).

Figure 3 shows an example of the Details Summary screen that displays selected metadata from a chosen data record. This Detail screen has five tabs that provide information on the specimen material, citation details for the paper from which the data originated, data processing procedures, measurement conditions, and the specimen material. One new feature of the Citation tab is a link entitled ‘All Records in this Publication’ which, when clicked, pulls a list of compounds and other data in the database from the same publication, thus providing a convenient means for accessing related data, as shown in Figure 4. Figure S2, Figures 3 and 4, and Figure S2 have the same layout for displaying a list of compounds and associated data, thus providing an excellent example for our use of the same reusable and custom-built component.

The component-based software-development approach (Basha and Moiz, 2014) was applied throughout the application development. For example, digital object identifiers (DOIs) (Gahan, 2023) were added to most of the references where XPS data were obtained. A component was built to advise a user that they would then be leaving the NIST website. Clicking on the DOI link in Figure 3 (right) first opens an alert window to advise users that they would then be leaving the NIST website, as illustrated in the upper right of Figure 3. Users would then be directed to the external site recorded in the DOI for the reference of interest.

2.4 Chemical shift plots

A useful new feature of the XPS App is the addition of chemical-shift plots. This type of plot displays the variation in measured binding energies, Auger-electron energies, or Auger parameters for a specified line representative of a selected element in different chemical states. We utilized the third-party Blazor Extensions library (Materials and Methods: 4) that contains the HTML5 Canvas graphic element for this purpose. We previously used a similar technology to display Wagner plots (Rumble et al., 1992; Wagner, 1977) in earlier versions of SRD 20. A new algorithm was designed and implemented (SI Section 6) to draw the chemical-shift plot using the HTML5 Canvas interface. After selecting chemical shift from the plot options on the home page, a user interface with a periodic table is displayed for selecting an element and energy type, as shown in Figure S3 (see SI). The user must next choose a specific line for their plot. The database then generates a list of materials that meet the selected requirements. Figure 5 provides an example screen showing the materials in the database that contain niobium and that have binding energy data for the 3d_5/2 line. The user can then select up to 12 of these materials and click ‘Graph’ to produce the chemical shift plot shown in Figure 6 (left). If more than 20 compounds were identified in the search, a menu for further filtering compounds by selection of desired elements in the compounds is displayed in the right panel of Figure 5. The horizontal lines in Figure 6 (left) represent the range of reported binding energies for the Nb 3d_5/2 photoelectron lines for each of the selected Nb compounds. The vertical lines indicate values of the median binding energy for each compound with the values presented below the vertical lines. A Summary Table is also produced (Figure 6 (right)) that lists the median binding energies for each compound and provides further information on the data source. A user could then click on a ‘Details’ link in the right most column of Figure 6 (right) to obtain information from that record similar to that shown in Figure 4.

In physical chemistry, molecular formulas and spectral data are traditionally presented with subscripts. Since this information is stored as simple text in the database, two new components for formatting displays of molecular formulas (SI Section 3) and spectral lines (SI Section 4) were developed. This was accomplished by first introducing HTML5 Canvas markup tags and then processing them. The molecular formatter converts a molecular formula from plain text into a standard chemical formula such as those shown in Figure 6 (right) while Figure S4 (see SI) is an example of the orbital formatter converting orbital text into a physically meaningful format for display.

Ideally, there should be little dispersion of the measured binding energies for each compound. For the example chemical-state plot for Nb compounds in Figure 6 (left), however, the 12 reported binding energies for the Nb 3d_5/2 photoelectron line of Nb₂O₅ range from 206.6 eV to 208.2 eV with a median value of 207.35 eV. It is difficult in retrospect to explain a spread of 1.6 eV in reported binding energies for the same compound. Possible explanations include drifts in the calibration status of each XPS instrument after its reported calibration. While there are now formal procedures for calibration of the binding-energy scales of XPS instruments (Powell, 2004), these were not available at the time that most data were acquired for the XPS App. Another possible explanation for the spread in reported binding energies for Nb₂O₅ is that the charge-neutralization procedures were inadequate for this nonconductive oxide. Again, improved procedures are now available for this purpose (Powell, 2004).