• Keine Ergebnisse gefunden

5.2 Data Management

5.2.2 Data Acquisition and Data Flow

The data for the analysis is coming from two main sources:

• Parameters from the OSC production process

• Data from the OSC characterisation

Additionally there is auxiliary data with e.g. important material constants or device settings, which complement OSC preparation parameters and mea-surement data. Various types of data have to be managed and despite their dierent sources, all data is gathered and integrated into one central struc-ture, the ELN.

Manufacturing Data

During the OSC production process, the recording of the various dierent process steps and their parameters (see section 3.2) is necessary. Starting with paper-based route cards to standardise the OSC preparation parame-ters, the whole recording of the production was moved directly to an

elec-98 CHAPTER 5. DATA ANALYSIS METHODS AND ENVIRONMENT tronic protocol. The production data now is entered via a webinterface into a relational database (MySQL with Ruby on Rails, [80, 81]) which maps the complete production process for both substrates and solutions. Both the webinterface and the underlying database have been designed such that they can easily be adapted to new parameters or process steps. Two screenshots are shown in gure 5.2. When entering a process step in the webinterface, a set of mandatory elds always has to be lled. Basic syntax checks val-idate the type of entry to increase the data quality. Pull-down menus are available for many parameters, ensuring a consistent and standardised way of entering information into the database. The production data can be entered and accessed real-time by more people at the same time from any networked computer in the group.

Figure 5.2: Two screenshots from the webinterface, which is used to record all production details of the OSCs. The substrates for the next process step (here only F 580) is selected for an OSC-annealing step (1). Depending on the selected process, its process details can be entered the next page appearing after pressing Go! (2). The production data and auxiliary information can be entered and accessed from any networked computer in the group.

5.2. DATA MANAGEMENT 99 The MySQL-database also handles all auxiliary information necessary dur-ing production. This includes details e.g. on the materials available for the absorber solutions and for evaporation, as well as the settings of the equip-ment, e.g. speed proles of the spin-coating programs. All this can also be accessed and maintained through the same webinterface.

The measurement data from the characterisation is not included in the MySQL-database. When handling experimental data of various kinds, a re-lational database has several shortcomings, e.g. information on the measured data has to be stored separately from the raw data and the queries for retriev-ing data can become very complex. However, for recordretriev-ing the production parameters through an user-friendly webinterface, the MySQL-database with Ruby on Rails was the quickest realisable and most pragmatic solution for moving from a paper based to an electronic protocol. The relational database now replaces paper based notes in the laboratory for the standard processes making instead all information available in electronic form, accessible from any computer in the group.

Because the measured data would be separate from the production data, the MySQL-database therefore is only an intermediate step on the way to com-bining all necessary data. The subsequent data transfer from the MySQL-database to the ELN is done via the extended markup language (XML, [82]).

XML is a platform independent general purpose markup language, which makes XML very exible and powerful for data exchange between dierent systems.

Measurement Data

The second major source of data is the measurement data. Prior to the au-tomation of the main characterisation methods (section 4.2), measured data had been stored dierently for every measurement method. The output les then generally only contained raw data without extra information, making the attribution of a data le to a particular measurement and OSC often dicult.

The standard measurements are now all controlled by one central LabView program described in section 4.2.4, which stores the measured data in a standardised way independent of the actual measurement method. However, the used LabView version (7.0) so far lacks sucient support for writing its output les directly as XML les [82], which would have been the best way to

100 CHAPTER 5. DATA ANALYSIS METHODS AND ENVIRONMENT facilitate the data transfer to the ELB. Hence an alternative but yet exible way of storing the data was developed. The output les that LabView writes consist now each of two parts. The rst part contains detailed information about how the measured data was obtained, i.e. measurement method, all device settings, date, time, what raw data to expect, etc.. This data is called meta-data, because it is describing other data (the raw data, in our case). The second part of the le then contains the actual measured data in raw form. This le structure allows the necessary exibility for including new measurements while ensuring that each measurement le contains all relevant details and is self-describing.

The LabView program employs features like mandatory elds, pull-down menus, and basic syntax checking to ensure the standardisation and quality of the meta-data entered when a measurement program le is created (see section4.2.4).

Figure 5.3: Excerpts from the les that the LabView program controlling the automated characterisation setup writes. The raw measured data (right) alone does not make much sense, because it contains no information on what is contained in the columns. Thus the raw data is complemented with meta-data (left), i.e. meta-data describing the raw-meta-data. Those two parts are joined into one output le, which is then self-explanatory.

With the standardised structure of the output le from LabView, the data from the dierent experiments is now self-explanatory. This makes the trans-fer of the data into the ELN straight-forward.

5.2. DATA MANAGEMENT 101 Measured data obtained by measurement methods, which are not integrated in the automated setup, are only used to complement the data from the standard experiments of the automated setup. Because the data les are not handled by the central LabView program, their data format depends on the measurement equipment, usually plain text les without much meta-data (e.g. meta-data from impedance measurements) or even proprietary binary le formats (e.g. AFM images). There are too many non-standardised data structures and this data either is manually being imported into the ELN or administrated separately.