Processing Glass Property Data with Sci Glass Information

Processing Glass Property Data with Sci. Glass Information System Alexander Priven Corning Korea privenai@corning. com * ITC is the name of the company who brought Sci. Glass to world scientists – including Corning. © 2015 Corning Incorporated . .

What Sci. Glass can do Usually, a database is considered as a tool for data search. Also Sci. Glass can: Sci. Glass is the world largest database of glass properties. • make the data visible • validate the found data • predict the unknown data • … and more, that the Sci. Glass can do. . © 2015 Corning Incorporated . .

Sci. Glass: search with confidence • • • Basic options Visualization tools Property calculation Data validation Patent search . © 2015 Corning Incorporated . .

How Sci. Glass looks • Extended data search options • Multiple visualization tools • Property prediction tools • Indexes • And many more tools… . © 2015 Corning Incorporated . .

Basic data search You specify composition and property requirements… • • • Viscosity Density Thermal Mechanical Optical Etc. etc. …. © 2015 Corning Incorporated . .

Basic data search … get the results… By systems: >270 000 oxide >35 000 chalcogenide >16 000 halide. © 2015 Corning Incorporated . .

Basic data search … and look at a source data table. Experimental details Reference Unique ID of source data table Kind of percent By references: Unique ID of glass >40 000 refs >25 000 authors >18 000 patents >1100 trademarks Compositions Properties. © 2015 Corning Incorporated . .

Unit conversion You can use convenient unit system. . © 2015 Corning Incorporated . .

Input of user’s data A Sci. Glass user can import his/her own data into the database. The user’s data are processed in the same manner as the Sci. Glass data. . © 2015 Corning Incorporated . .

Property calculation in Sci. Glass Calculator: simplest tool that calculates a property of your glass from its chemical composition. Result of calculation Composition (userdefined or pasted from database) Property and method of calculation . © 2015 Corning Incorporated . .

Glass formation diagrams Using more than 4000 ternary diagrams of glass formation, a specialist can diminish the scale of experimental research required for developing new glass compositions. . © 2015 Corning Incorporated . .

Optical spectra Three versions of the same spectra ü > 20 000 spectra ü 105 versions of spectral coordinates ü Absorption transmission conversion ü Superposing spectra reported in different units. © 2015 Corning Incorporated . .

Sci. Glass: search with confidence • • • Basic options Visualization tools Property calculation Data validation Patent search . © 2015 Corning Incorporated . .

Binary plot In binary systems, a user of Sci. Glass can find the data, plot a dependence of property on composition and approximate it with a polynomial. . © 2015 Corning Incorporated . .

Composition-to-property relationships Just specify components and a property and run the query. . © 2015 Corning Incorporated . .

Composition-to-property relationships … get the list of glasses… . © 2015 Corning Incorporated . .

Composition-to-property relationships … approximate the dependence with polynomial… . © 2015 Corning Incorporated . .

Composition-to-property relationships … and copy the formula to the Clipboard: -2. 6293 E-02 *( x -27. 567)^3 1. 5644 E-01 *( x -27. 567)^2 8. 9629 E+00 *( x -27. 567) +1189. 7 You can use this formula in your calculations – for example, to verify your own data. . © 2015 Corning Incorporated . .

Composition-to-property relationships Results of property calculation Grouping by system Tg in chalcogenide system As-Se Calculations by author’s method. © 2015 Corning Incorporated . .

Ternary diagrams of properties A user can specify three components, property and method of calculation. Sci. Glass program builds the diagram. Density of sodium borosilicate glasses. 2675 experimental points. Thick lines results Curves are results of of calculations by the Priven 2000 model Curves are results of polynomial Thin lines are results of approximation ploynomial extrapolation. © 2015 Corning Incorporated . .

Ternary diagrams of properties You can use the formula of polynomial in an external application, e. g. MS Excel. . © 2015 Corning Incorporated . .

Ternary diagrams of properties Is it for simple system only? – No! Let’s try to plot 5 component system: Na 2 O-K 2 O-Al 2 O 3 B 2 O 3 -Si. O 2… … and we can see the dependence…. © 2015 Corning Incorporated . .

Ternary diagrams of properties … and you are also able to use the formula in your applications. . © 2015 Corning Incorporated . .

Ternary diagrams of properties Too much closely spaced points? Use Zoom option! To see the glass composition just click on the plot. . © 2015 Corning Incorporated . .

Ternary diagrams of properties • >100 property characteristics • >100 methods of calculation • Up to 70, 000 experimental data points in a single diagram • Unlimited number of components – oxide, chalcogenide, halide, mixed systems – Molecular, weight, atomic % • Approximation with 1 to 6 order polynomials • Use of polynomial formulas in your applications. © 2015 Corning Incorporated . .

Sci. Glass: search with confidence • • • Basic options Visualization tools Property calculation Data validation Patent search . © 2015 Corning Incorporated . .

Property-to-property correlations Simplest way of calculation is the use of correlations. Correlations of glass properties with each other are important to know when developing new glass compositions with desired combination of properties. Sci. Glass allows easy visualization and approximation of such correlations. . © 2015 Corning Incorporated . .

Property-to-property correlations Some correlations are general enough. Click on an “outlier” gets an idea of compositions with unusual combinations of properties. . © 2015 Corning Incorporated . .

Property-to-property correlations Some other correlations may appear only in a specific composition area. . © 2015 Corning Incorporated . .

Property-to-property correlations In this composition area, it is difficult to increase thermal expansion coefficient without making the glass heavier. Use of property prediction models built in Sci. Glass can help to resolve the problem. . © 2015 Corning Incorporated . .

Property calculation in Sci. Glass How to reach this combination of properties? . © 2015 Corning Incorporated . .

Search for predicted properties Let us search for predicted property values! Any combinations of • 64 oxides • 55 property characteristics. © 2015 Corning Incorporated . .

Search for predicted properties Search results: 51 glass composition with prescribed combination of properties Prediction are made by using the method Priven-2000 • 270, 000 oxide glasses with pre-calculated properties – Viscosity, density, optical, elastic, acoustical properties, thermal expansion, heat capacity, surface tension • Quick search for any combination of these properties. © 2015 Corning Incorporated . .

Property calculation in Sci. Glass One click – and we see: • How to synthesize this glass • Reference to source document • Information about known properties of the glass But… how reliable the calculations are? Let’s estimate the accuracy of predictions with Sci. Glass Premium. . © 2015 Corning Incorporated . .

Sci. Glass: search with confidence • • • Basic options Visualization tools Property calculation Data validation Patent search . © 2015 Corning Incorporated . .

Validation of calculation results Sci. Glass Premium compares • results of prediction – by multiple methods • with experimental data – for multiple similar compositions Results of calculation: • most probable property value • confidence interval The feature is available in Sci. Glass Premium only. Algorithm of calculation is patented by ITC. . © 2015 Corning Incorporated . .

Validation of calculation results Confidence limits for CTE: 33 ± 6. 10 -7 K-1 Density: 2. 27 ± 0. 06 g/cm 3. © 2015 Corning Incorporated . .

Data validation in Sci. Glass We found a really prospective glass composition. Asterisk specifies confidence limits for properties. . © 2015 Corning Incorporated . .

Validation of experimental data But what about experimental data points with unusual combinations of properties? Not always these data are correct… Let’s validate an experimental data point. Asterisk specifies confidence limits for properties. . © 2015 Corning Incorporated . .

Validation of experimental data Coefficient of thermal expansion is exact… . © 2015 Corning Incorporated . .

Validation with Sci. Glass Premium … but density value seems to be wrong – let’s check more… . © 2015 Corning Incorporated . .

Search for similar compositions A special query returns the data about the same property for glasses with closest compositions. Density value does not consist with alternative data. . © 2015 Corning Incorporated . .

Comparison with calculation results Also, reported value does not consist with the results of calculations… Reported value: 2. 31 g/cm 3 Calculation results: 2. 76 to 2. 88 g/cm 3 Alternative measurement: 2. 76 g/cm 3 9 independent calculation methods and one alternative measurements give similar results. Thus, the reported date is not correct. Use of Sci. Glass might preserve the authors from publishing erroneous data. . © 2015 Corning Incorporated . .

Property calculation: Summary • Prediction of multiple properties including: – – – viscosity density thermal properties electrical properties elastic & acoustical properties etc. • 100+ calculation methods • Oxide, fluoride, chalcogenide glasses • Calculations available in plots and diagrams • The most general calculation methods are available in Sci. Glass only . © 2015 Corning Incorporated . .

Sci. Glass: search with confidence • • • Basic options Visualization tools Property calculation Data validation Patent search Forecasting. © 2015 Corning Incorporated . .

Prior art analysis Patent analysis of glass compositions is not an easy task. Often patent offices grant a patent for already known compositions. How to find prior art for such patents? This table presents the first claim of a real patent with priority date after 2005. . © 2015 Corning Incorporated . .

Prior art analysis Just to specify the claimed limits in Search for Glasses query, and get a table of prior art. 25 compositions are found, most of them being published before 2005. . © 2015 Corning Incorporated . .

Prior art analysis This and other data sources were omitted by the patent office… … but the claimed composition was known almost 20 years before being patented! . © 2015 Corning Incorporated . .

Prior art analysis Here are only 2 of ~20 contesting examples taken from 8 prior publications. Oxide Compared patent Prior art Kuznetsova (1988) Paulson, US 2004043887 (2004) Min Max Fact Si. O 2 50 70 60. 01 68. 94 Al 2 O 3 10 20 17. 23 13. 58 B 2 O 3 8 12 10. 70 8. 62 Mg. O 0 3 0 0 Ca. O 4 15 12. 06 6. 34 Sr. O 0 10 0 0 Ba. O 0 1 0 0 Zn. O 0 5 0 2. 52. © 2015 Corning Incorporated . .

Sci. Glass today Characteristics Total number of glasses Patent sources Ternary diagrams of glass formation Spectral curves Number >400 000 >18 000 >4000 >20 000 Prof. Oleg Mazurin and his colleagues started this work in 1960 th. After half a century, we have the world largest collection of glass property data. . © 2015 Corning Incorporated . .

Summary: Why we use Sci. Glass? What Sci. Glass can do: • Data search • Visualization (tables, plots, diagrams) • Unit conversion • Property prediction • Importation of user’s data • Exportation of data and graphs to external applications What we can do with Sci. Glass: • Get more information • Make information clear • Verify the data • Predict unknown property values • Find new dependences What we want from Sci. Glass: • Find prior art • Find prospective compositions • Get an idea of improvement of glass compositions Sci. Glass: search with confidence!. © 2015 Corning Incorporated . .

Processing Glass Property Data with Sci. Glass Information System Alexander PRIVEN privenai@corning. com . © 2015 Corning Incorporated . .