Bachelor Thesis Scheikunde
Optimisation and testing of the water droplet method for
identifying unstable glass in museum collections
door
Jelle Tromp
23 december 2020
Studentnummer 11886897 Onderzoeksinstituut HIMS Onderzoeksgroep Analytical Chemistry Verantwoordelijk docent Prof. Dr. M. R. van Bommel BegeleiderAbstract
As glass deteriorates, its composition and appearance change and as a result historical infor-mation and cultural value is lost. The process is irreversible which makes restoration methods difficult. A method would be needed to predict the stability of the glass so that preventive con-servation methods can be applied before any visible deterioration happens.
Guus Verhaar has developed such a method and uses a polyester swab method to remove deterioration products from the surface of the glass. He was able to analyse these samples using ion chromatography and successfully determine the stability of vessel glass, mainly using the potassium and sodium concentrations found. An intern of Guus, Danny Verschoor, found a method which might be able to determine the stability of the glass as well: the water droplet method. This method exists out of five main steps: the cleaning of the glass surface, the place-ment of the water droplets on the glass surface, the removal of the water droplets from the glass surface, extraction of the deterioration products from the swab and analysing the sample using IC. Multiple droplets are placed on the glass surface, these are removed at different exposure times. This method looks at the exchange rate between ions within the glass and a water droplet which is placed on top of it over time, this exchange is believed to be primarily diffusion con-trolled. For unstable glass a large concentrations of ions will be present in the glass and so a correlation would be expected between the concentration and the exposure time. For stable glass only a small concentration of ions is expected in the glass, so either a small increase is expected or no correlation at all between the concentration and the exposure time. Promising results were obtained, but he did not have time refine the method an further test it. In this thesis I investigated the promising water droplet method on its reliability and reproducibility and tested it on its capabilities of identifying unstable glass in museum collections. I also looked into the homogeneity of the distribution of ions on the glass surface using both the optimized water droplet method and the validated method developed by Verhaar. For the water droplet method 4 different glass types were used, two of which were stable, 19AY and Quartz, and two of which were unstable, KVL and JXS.
To investigate the reproducibility and reliability of the water droplet method the five main steps were examined. The first step of the water droplet method, cleaning the glass surface, was not looked into as there was not enough time available on the lab due to COVID-19. The second step, the placement of the water droplets on the glass surface, was investigated by plac-ing the droplets at different locations on the different glass types. The third step, the removal of the water droplets from the glass surface, was also looked into. The completeness of the removal was tested by placing water droplets on the surface of KVL glass and extracting these using two swabs. The first swab would remove most of the water while the second swab would collect the residue. The fourth and fifth step had already been validated by Verhaar.
The capability of the water droplet method to identify unstable glass was tested by plac-ing water droplets at different locations on the surface of the four different glass types and removing these at different exposure times.
The homogeneity of the distribution of ions on the glass surface was investigated by placing water droplets at different locations on the surface of the four different glass types and removing these at the same exposure time. The homogeneity of the distribution of ions on the glass surface was investigated using the validated method by Verhaar by sampling on various locations on glass objects from the Stedelijk Museum Amsterdam.
The results indicate that the droplets spread out and differ significantly in size for the KVL and 19AY glass, thus an alternative method should be found. However, none of the pro-posed alternatives seemed to work so the original method was used in the finalized procedure. For the JXS glass and Quartz the difference in size was acceptable. Furthermore, the removal
of the water droplets with one swab was found to be sufficient as only 5% of the total concen-tration was obtained from the second swab.
In the results correlations were found between the potassium/sodium concentrations and the exposure times for the unstable glass types, KVL and JXS. While for the stable glass types, 19AY and Quartz, no correlation was found. It was found that the distribution of ions on the surfaces of KVL and 19AY might not be homogeneous while the distribution of ions on the surfaces JXS and Quartz were homogeneous.
It was found that all of the unstable and most of the stable glass objects of the SMA did not have a homogeneous distribution of ions on the surface.
In conclusion, the method could not be made entirely reliable and reproducible as there was not enough time to look into the first step, cleaning the glass surface. Furthermore, no alternative method was found for the second step. The water droplet method was found to be capable of predicting the stability of glass as for the two unstable glass types correlations were found between the sodium/potassium concentrations and the exposure times while for the two stable glass types no such correlations were found. Lastly, it was found that the distribution of ions on the glass surface is not homogeneous but place dependent.
Samenvatting`
Glas wordt al een hele lange tijd door de mens geproduceerd, het eerste bereide glas dateert terug naar 2500 voor christus. Vele glazen ob-jecten uit verschillende periodes zijn heden-daags in musea te vinden. Naar schatting is 10% van al het museale glas instabiel, dit is het resultaat van een ongunstige chemische samenstelling. Instabiel glas kan degraderen wanneer het ongunstige condities ondervindt, denk bijvoorbeeld aan een hoge luchtvochtig-heid of temperatuur. Ten gevolge van degra-datie veranderen de samenstelling en het ui-terlijk van het glas en komen er ionen vanuit de structuur vrij. Hierdoor verliest het glas historische informatie en culturele waarde en zal het object (in het slechtste geval) niet meer kunnen worden tentoongesteld, dus actie moet worden ondernomen om dit tegen te gaan. Restauratie methodes zijn heel moeilijk om toe te passen aangezien het degradatie proces onomkeerbaar is. Preventieve
conser-veringsmethodes zouden kunnen worden toegepast maar vereisen veel tijd, geld en moeite. Mu-seum collecties kunnen uit honderden of zelfs duizenden objecten bestaan. Aangezien niet alle glazen objecten instabiel zijn zal er een selectie moeten worden gemaakt. De instabiele objecten zullen moeten worden geïdentificeerd, Guus Verhaar heeft een methode ontwikkeld die dit kan doen. Hij gebruikt bevochtigde polyester wattenstaafjes om het glas te bemonsteren. Deze mon-sters analyseert hij met een ion chromatograaf, dit is een apparaat dat de hoeveelheid ionen in een oplossing kan meten. Aan de hand van de hoeveelheid gevonden natrium en kalium ionen kon Verhaar bepalen of het glas vatbaar was voor degradatie.
Een stagiair van Verhaar, Danny Verschoor, bedacht een andere methode genaamd: de waterdruppelmethode. Deze methode bestaat uit vijf stappen: het schoonmaken van het
Figuur 0. Een gedegradeerd schenkkan
glasoppervlak, het plaatsen van waterdruppeltjes op het glasoppervlak m.b.v. een pipet, het verwijderen van de waterdruppeltjes van het glas oppervlak m.b.v. een polyester wattenstaafje, het extraheren van de ionen uit het polyester wattenstaafje en het analyseren van het monster. Er worden hierbij meerder waterdruppeltjes op het glas geplaatst, die vervolgens bij verschil-lende tijden van blootstelling van het glas worden verwijderd. Op deze manier wordt de snel-heid van het uitwisselingsproces van de ionen in het glas en de waterdruppel bepaald. De ge-dachte hierachter is dat de snelheid van dit uitwisselingsproces vervolgens ook iets kan zeggen over de stabiliteit van het glas en daarmee of het glas vatbaar is voor degradatie. In instabiel glas zullen zich meer vrije ionen bevinden, gezien deze vrijkomen bij het degradatie proces. Ten gevolge hiervan zal er een correlatie ontstaan tussen de concentratie ionen in de waterdrup-pel en de tijd van blootstelling. In stabiel glas zullen zich weinig vrije ionen bevinden, gezien er geen degradatie proces heeft plaats gevonden. Hierbij wordt dus geen correlatie verwacht tussen de concentratie ionen in de waterdruppel en de tijd van blootstelling. Gezien Verschoor pas in de laatste weken van zijn stage op deze methode kwam, had hij niet de tijd om deze uit te werken of uitgebreid te testen. Echter waren de behaalde resultaten veelbelovend.
In deze scriptie is er kritisch gekeken naar de waterdruppelmethode en uitgezocht welke stappen kunnen worden verbetert op het gebied van betrouwbaarheid en reproduceerbaarheid. Het is namelijk van groot belang dat een analytische methode betrouwbaar en reproduceerbaar is, anders zou je voor elke meting willekeurige resultaten krijgen waaruit je geen conclusie zou kunnen en mogen trekken. Verder is er ook onderzocht of de methode kan worden gebruikt om instabiel glas te identificeren. Verkregen resultaten leken een interessant verband weer te geven tussen ion concentratie en locatie. Zodoende is de homogeniteit van de verdeling van ionen op het glasoppervlak onderzocht. Voor de waterdruppelmethode werden telkens 4 verschillende glas types gebruikt, hiervan waren er twee stabiel, 19AY en Kwarts, en twee instabiel, KVL en JXS.
De vijf stappen van de waterdruppelmethode werden kritisch geëvalueerd. De eerste stap, het schoonmaken van het glasoppervlak, werd niet geëvalueerd gezien er slechts een be-perkte hoeveelheid tijd op het lab beschikbaar was door COVID-19. De tweede stap, het plaat-sen van waterdruppeltjes op het glasoppervlak m.b.v. een pipet, werd daarentegen wel kritisch geëvalueerd. In deze stap is het belangrijk dat de spreiding en daarmee de grootte van de wa-terdruppel zo beperkt mogelijk blijft voor verschillende locaties op één object. Dit werd getest door waterdruppels op verschillende locaties van de vier verschillende glastypes te leggen. Er werd ook gekeken naar de derde stap, het verwijderen van de waterdruppeltjes van het glas oppervlak m.b.v. een polyester wattenstaafje, hierbij is de volledigheid van het verwijderen van het waterdruppeltje belangrijk. Dit werd getest door waterdruppels op het KVL glasstuk aan te brengen en deze te verwijderen m.b.v. twee polyester wattenstaafjes. Het eerste polyester wat-tenstaafje absorbeerde het grootste gedeelte van het water, terwijl het tweede polyester watten-staafje slechts het residu absorbeerde. De vierde en de vijfde stap uit de procedure, het extrahe-ren van de ionen uit het polyester wattenstaafje en het analyseextrahe-ren van het monster, zijn hetzelfde als in de al gevalideerde procedure van Guus Verhaar, deze hoeven dus niet meer te worden onderzocht.
Om te testen of het mogelijk is om met de waterdruppelmethode instabiel glas te iden-tificeren is deze gebruikt om de vier verschillende glastypes te bemonsteren op verschillende locaties van het glasoppervlak, deze druppels zijn vervolgens van het glasoppervlak verwijdert bij verschillende tijden van blootstelling.
Om de homogeniteit van de verdeling van ionen op het glasoppervlak te onderzoeken is de waterdruppelmethode toegepast op de vier verschillende glastypes. Nadat de druppels op verschillende locaties van het glas type waren toegebracht werden ze na een gelijke tijd van blootstelling (30 minuten) verwijdert. Daarnaast werd de homogeniteit van de verdeling van ionen op het glasoppervlak ook getest met de al gevalideerde methode van Guus Verhaar. Zijn
methode werd getest op verschillende glasstukken van het Stedelijk Museum Amsterdam (SMA).
De spreiding van de waterdruppels bleek voor JXS en Kwarts controleerbaar te zijn, gezien het verschil in druppelgrootte nihil was. De spreiding van de waterdruppels op het KVL en 19AY glas was daarentegen oncontroleerbaar, het verschil in de druppelgrootte was dras-tisch. Een alternatief zou moeten worden gevonden om de methode alsnog te valideren. Ver-schillende alternatieven werden uitgeprobeerd, maar zonder succes. Uiteindelijk is er toch be-sloten om de waterdruppels met een pipet op het oppervlakte te plaatsen in de procedure. Ver-der, werd er gevonden dat het verwijderen van de waterdruppel gedaan kan worden met één polyester wattenstaafje, gezien het tweede polyester wattenstaafje slechts 5% van de totale con-centratie verwijderde.
Er werd een correlatie gevonden tussen de natrium/kalium concentratie en de tijd van blootstelling van de instabiele glastypes, KVL en JXS. Verder werd er geen correlatie gevonden tussen de natrium/kalium concentratie en de tijd van blootstelling van de stabiele glastypes, 19AY en Kwarts.
De de verdeling van ionen op het glasoppervlakte leek voor KVL en 19AY niet homo-geen te zijn, voor JXS en Kwarts wel. Verder werd er gevonden dat alle instabiele objecten en de meeste stabiele objecten van het SMA geen homogeniteit hadden voor de verdeling van ionen op het glasoppervlak.
Er kan geconcludeerd worden dat de waterdruppelmethode niet volledig kan worden gevalideerd. Dit komt doordat de eerste stap, het schoonmaken van het glasoppervlak, niet is geëvalueerd en gezien er geen goed alternatief kon worden gevonden voor de tweede stap, het plaatsen van waterdruppeltjes op het glasoppervlak m.b.v. een pipet, waarbij de druppelgrootte controleerbaar zou zijn. Daarnaast is het mogelijk om met de waterdruppelmethode de stabiliteit van het glas te bepalen. Instabiel glas zal een correlatie geven tussen de natrium/kalium con-centratie en de tijd van blootstelling terwijl stabiel glas geen correlatie zal geven. Verder is de verdeling van de ionen op het glasoppervlak voor de meeste objecten niet homogeen maar plaats afhankelijk.
Table of contents
1. Introduction ... 1
1.1 Research context ... 1
1.2 Theoretical Background... 2
1.2.1 The structure of glass... 2
1.2.2 The mechanism of glass deterioration ... 4
1.2.3 The result of the deterioration process ... 5
1.2.4 The factors that influence the deterioration of glass... 6
1.2.5 Analysing unstable glass ... 7
1.2.6 Investigation into the stability of glass using ion chromatography ... 8
1.3 Research questions and methodology ... 9
1.4 Thesis structure...10
2. Equipment, materials & methods ...11
2.1 Equipment ...11
2.2 Materials ...11
2.3 Existing analytical procedures ...12
2.3.1 Preparation and use of the internal standard...12
2.3.2 The swab method by Guus Verhaar...12
2.3.3 The water droplet method by Danny Verschoor ...13
2.4 Research approach and experiments...13
2.4.1 Optimisation of the swab method...13
2.4.2 Optimisation of the water droplet method ...13
2.4.3 Application of the optimized water droplet method ...15
3. Results and discussion ...19
3.1. Optimisation of the swab method ...19
3.2 Optimisation of the water droplet method...20
3.3 Application of the optimized water droplet method ...24
3.4 Place dependency ...28
Conclusion...34
5. Follow-up research ...36
5.1 Optimisation of the water droplet method...36
5.2 Validation of the optimized water droplet method ...36
5.3 Place dependency ...36
Acknowledgements ...37
References ...38
Appendix 1. Pictures of the water droplets on the different glass surfaces. ...40
Appendix 2. Objects sampled from Stedelijk Museum Amsterdam. ...56
Appendix 3.1 Sodium concentrations of different locations on the different stable objects sampled
from Stedelijk Museum Amsterdam...63
Appendix 3.2 Potassium concentrations of different locations on the different stable objects
1
1. Introduction
1.1 Research context
Even though the exact origins of man-made glass remains unknow, it is generally agreed upon that its discovery was around 2500 BC in Iraq.[1] In these times glass was used mainly as
deco-ration and jewellery.[2] The Egyptians produced glass objects as well and even had a term for
it: iner en wedeh or in English ‘stone of the kind that flows’. Core-forming was the main method used with the early production of glass. It was based on a core of removable material around which the glass was moulded, after cooling down the core would be removed and a piece of decorative glass was obtained. This process was not very efficient and required great skill. Dur-ing the followDur-ing centuries minor developments were made in the glass production until around 100 BC when glass blowing was invented in modern day Syria. This made the production pro-cess a lot easier and more efficient which led to mass production and the wide availability of glass, these times may also be referred to as the first Golden age of glass. Furthermore, the use of glass switched from solely decorative to functional purposes as well, like storage. Glass con-tainers had many advantages as they were reusable, left no taste on their contents, were rela-tively light and transparent as well. As the central Roman imperial power declined , the general technique of the glass making also declined. It was not until the 13th century that the glass
production reached yet another golden age in Venice. During this golden age, the production technology was improved ultimately leading to the development of the world -famous Cristallo glass. This golden age endured until around the 17th century. In the 18th century the industrial
revolution took place. Glass began to play an important role in scientific advancements such as lenses, microscopes, chemical glass ware and mirrors. In modern day, glass is indispensable as it has become part of every ay life as packaging, architecture and scientific equipment just to name a few. Currently, glass objects from all periods of glass production are held in museums for their historical information as well as cultural value is gained.
Roughly 10% of the glass objects in museum collections are deemed to be unstable. The instability of historical glass is mainly caused by an unfavourable chemical composition. If unstable glass encounters unfavourable environmental factors the glass can deteriorate. The deterioration process of the glass might lead to weeping, which is the formation of a moist layer, and or crizzling, which is the formation of a network of hairline cracks. These phenomena cause the transparency of the glass to be diminished and the overall appearance of the glass to be altered As a result of these events the objects might no longer be suitable for display. Unfor-tunately, these processes are irreversible as the structure of the glass itself is altered, making restoration methods possible but very complex. On the contrary preventive conservation and preservation methods can be applied to a certain extent.
The deterioration process cannot be prevented entirely as the stability of the glass is largely dependent on its composition which cannot be changed. However, preservation and conservation procedures can be applied to control the environmental factors and slow down the deterioration process, postponing the final date of display. Priorities must be made since it would be inexecutable to apply such procedures to an entire museum collection given the fact that these might consist of hundreds or even thousands of glass objects. Currently, many muse-ums screen their collections just by inspecting their objects on crizzling and weeping. Consid-ering that the deterioration process is irreversible the alteration of appearance cannot be re-versed. It would be more practical if the stability of the glass could be determined before it showed any change in appearance.
The research presented in this thesis focusses on the development of a method which can screen a collection and pick out the most vulnerable objects before they show any signs of deterioration. This method should be reliable, reproducible and non-damaging to the objects. A deeper understanding of the glass structure as well as the deterioration mechanism is needed to
2 find a method that determines the stability, hence these will be discussed in the upcoming sec-tion 1.2.
1.2 Theoretical Background
1.2.1 The structure of glass
Glass is see-through and brittle. As a result, one might expect that glass has a rigid crystalline structure consisting of silicon and oxygen atoms as depicted in figure 1.1. However, its structure is considered amorphous, in other words, the atoms do not have a definitive arrangement but resemble that of its liquid state. This is a result of the manufacturing process and in particular the rate at which a material is cooled down after melting (figure 1.2).[3] A material can be cooled
down slowly and steadily, during the first part the volume will decrease steadily until the melt-ing temperature is reached. Here the atoms will have the time to rearrange into a structured crystalline form which causes a sudden decrease in volume. On the contrary, the liquid can be cooled down rapidly as well, in this case the atoms will not have the time to rearrange into a crystalline structure and no sudden volume decrease can happen. The volume will decreases with the same rate it did when cooling down to the melting temperature consequently the amor-phous structure is formed.
Figure 1.1 Two-dimensional structures of a Crystal (left) of a certain oxide AO2 and its
corre-sponding glass structure (right). [2]
Figure 1.2 The volume of the oxide plotted against the temperature during the heating and
3 The earliest recognition of the amorphous structure of glass was by Zachariasen in 1932, who developed the Zachariasen-Warren network theory.[5] This theory had a set of rules to
which an oxide should suffice to form three-dimensional random networks. Oxides A2O3, AO2
and A2O5 satisfy these rules. As to be expected the main ingredient of most historical glasses,
SiO2, is such an oxide. SiO2 or silica can be found all around the world as it is the main
com-ponent of sand. Silicon has a coordination number of four thus it can be bound to four oxygen atoms at once, the resulting molecular structure is a tetrahedron. When these tetrahedra are linked together they form a crystal structure known as quartz.[6] However, as explained in the
last section, when silica is melted and cooled down quickly these tetrahedra’s will have varying bond angels resulting in the amorphous structure known as glass.
Silica melts at 1710 degrees Celsius which is relatively high compared to other historical used materials like bronze which has a melting point at 950 degrees Celsius. Reaching the melting point of silica without a modern day blast furnace is impossible, thus people had to lower the temperature at which the glass was made. To achieve this other materials were added to the silica. This resulted in the introduction of different cations into the glass net-work (Figure 1.3).[2] The relatively large cations caused the
amor-phous structure to alter by partially breaking silicon oxygen bonds. The oxygen atoms would be bound to one silicon atom instead of two thus a negative charge would be created on the oxygen, this type of oxygen is called a non-bridging oxygen or NBO. Cations with a single positive charge, most commonly Na+ and K+, could
bind monovalently to an oxygen atom. While cations with two positive charges, most commonly Ca2+ and Mg2+ could bind
diva-lently to oxygen atoms. Furthermore, experimentation with the colour and clarity of the glass also lead to adding other compounds into the mixture. Cobalt and copper containing compounds were used to change the colour of the glass while compounds containing arsenic and manganese were added to increase the clarity.
Zachariasen divided the cations found in glass into three different categories.[5] Firstly,
network formers, if the cation had a coordination number of 3 or 4. Secondly, network modifi-ers, if the cation had a coordination number equal to or larger than 6. Lastly, int ermediates, if the cation could not be sorted within the first two categories. As mentioned previously silica has a coordination number of four thus it can be classified as a network former. The cations introduced into the glass like Na+, K+, Ca2+ and Mg2+ have coordination numbers of within a
range of five to seven, thus they can be classified as network modifiers. The cations which bind monovalently to oxygen atoms cause the structure to open up, these network modifiers are called network looseners.[7] On the other hand, cations which bind divalently cause the structure
to be more rigid, these are called network stabilisers. The quantity of each of these cation com-ponents make up the glass network. In the table below commonly used compositions of glass are presented.[8] The used composition is dependent on the location of the production site which
determines the materials available there.
Table 1. Common compositions of glass[8]
Glass type Network-former Network loosener Network stabiliser Soda-lime-silica SiO2 Na2O CaO
Potash-lime-silica SiO2 K2O CaO
Mixed-alkali SiO2 Na2O, K2O CaO
Figure 1.3
Two-dimen-sional structure of a glass network with network-modifiers and non-bridg-ing oxygen atoms.[6]
4
1.2.2 The mechanism of glass deterioration
The deterioration of glass is the change of its composition, ultimately causing the formation of an altered surface layer.[7] The layer that has formed plays a big role in the appearance of the
glass as it may change it, mostly for the worse. The change in appearance can be divided into two main types: crizzling and weeping. Crizzling is the formation of a network of small cracks while weeping is the formation of a moist surface layer or droplets on the glass.[9] The exact
underlying mechanisms of glass deterioration are not yet fully understood but vital for predict-ing the stability. Multiple models have been made to describe set mechanism, with the ion ex-change model being the most prevalent one in the case of historical glass. Most other proposed models are concerned with the storage of radioactive waste using glass. Even though this branch of research is very interesting it is out of scope of this thesis and so will not be discussed.[10][11]
Ion exchange model
The ion exchange model covers the following three processes occurring at the surface of the glass: hydration, hydrolysis and ion-exchange.[12] During the hydration, water enters the
glass structure through the voids that have formed in the amorphous structure. Once inside, the water molecules may hydrolyse the metal-oxygen bonds forming hydroxyl groups (Reaction 1). Since hydrolysis is a reversible process an equilibrium is established with both water mole-cules and hydroxyl groups being present. The hydroxyl groups that form can react destructively with the amorphous silica structure breaking the Si-O-Si bonds, this is called glass corrosion (Reaction 2), this ultimately leads to dissolution of the glass network. The water that enters the glass may also undergo the ion-exchange process. Here alkali ions in the glass structure are exchanged with hydrogen or hydronium ions, the alkali ion is excelled and a hydroxyl ion is formed (Reaction 3).
The rate of the hydration process is dependent on the structure of the glass as they de-termine the size of the voids through which the water may penetrate, with larger voids allowing more water to pass through. The amount of NBO/ion exchange sites is also dependent on the structure of the glass. The more of these available sites the higher the rat e of ion-exchange, hydrolysis and rupture of Si-O-Si bonds since all these processes take place at such sites.
The rates of the rupture of Si-O-Si bonds and ion exchange are directly dependent on the pH of the attacking solution.[9] At a low pH value more hydronium and hydrogen ions are
available for the reaction thus the rate is higher. Considering that the hydronium and hydrogen ions migrate into the glass the concentration in the attacking solution decreases which increase s the pH value. At a high pH value more hydroxyl ions are present resulting in more Si-O-Si bonds being ruptured. This leads to an increase in nonbonding oxygen/ ion exchange sites, again
5 leading to higher ion-exchange and hydrolysis rate. In the process hydroxyl anions are formed and the pH is increased (Figure 1.4). Furthermore, ion exchange creates extra space in the network by replacing the large alkali ions with smaller hydronium ones allowing for easier hydration.
1.2.3 The result of the deterioration process
Weeping
As a result of the ion-exchange process, cations are freed from the glass structure and leach out of the glass.[9] These cations may form salts with anions
pre-sent in the atmosphere and form deterioration prod-ucts on the surface. The deterioration prodprod-ucts may attract water from the atmosphere, ultimately result-ing in a moist layer on top of the glass, this process is called weeping (figure 1.5). As more water is drawn to the surface of the glass, the rate deterioration pro-cess will increase, leading to the formation of more salts etc.
Crizzling
Glasses containing an insufficient amount of network formers, may form a silica gel layer containing up to 20% by weight of water. Inside this layer an equilib-rium has been established between the water inside and the relative humidity and temperature of the en-vironment. If such an object would be moved to a dif-ferent environment which has a relatively lower hu-midity and/or higher temperature, the established equilibrium would be disturbed. Consequently, the water inside of the silica gel layer would evaporate, leading to a volume decrease and the formation of hair cracks. This process is called crizzling (figure 1.6).
Figure 1.4 positive feedback loop of a low pH.
Figure 1.5 Weeping on the glass
sur-face (Photo credit: Corning museum of glass).
Figure 1.6 Crizzling on the glass
sur-face (Photo credit: Corning museum of glass).
6
1.2.4 The factors that influence the deterioration of glass
As described in the previous section the attacking solution and its corresponding pH play a big role in the deterioration of glass as it is the cause of the dissolution of the network and leaching of the alkali ions. Other environmental factors such as the temperature and the presence of other compounds also play a role.[2] Most importantly, the composition of the glass determines
its intrinsic stability and so a ‘bad’ composition may lead to a vulnerable glass structure. How-ever, such a structure will only deteriorate in the presence of certain environmental factors. In short, the deterioration of glass is a very complex process with the two main factors, the envi-ronment and the composition, cohesively determining the extent of the deterioration process.
Composition of the glass
The composition of glass is determined by the reactants used to make the glass. As mentioned, before the existence of modern day blast furnaces organic compounds were added into the mix-ture of reactants to lower the melting point of the batch or to add colour and other desired properties. As a result, the compositions of glasses may vary and so does its intrinsic stability. Glasses that are deemed unstable will have an unfavourable ratio of network formers, looseners and stabilizers and may have other destabilizing additives as well.
A lot of research has been done on the correlation between the composition of glass and its susceptibility to deteriorate. An early example of such research includes a paper by Brill.[8]
In 1975 he discovered that glasses containing a low concentration of the network stabilizer CaO and a relatively high concentration of network looseners showed severe crizzling. He tested his hypothesis by using replica glasses from which similar result were obtained.
Dr. Jerzy J. Kunicki-Goldfinger is another scientist who has done a lot of research on the relationship between composition and stability of historical vessel glass from central Eu-rope. In a recent paper he compared a large number of glass vessels with respect to their com-position and stability.[13] He deemed the glasses containing relatively low amounts of the
net-work stabilizer, CaO, and high amounts of the netnet-work looseners, K2O, Na2O and As2O3, to be
susceptible to crizzling. The ratio between As2O3 and CaO could serve as an indicator for the
stability of the glass. The ratio was determined by analysing the glass with XRF.
A study by Rodrigues et al. shows results relevant to the subject as well by comparing three types of glass with different compositions.[14] It was concluded that the composition of
the glass influences the structure of the surface (the size and number of the lattice voids and the porosity) as well as the hygroscopicity. Furthermore, under the same conditions different com-positions were found to have different rates of glass surface alteration.
Concluding, the composition plays an important role in the stability of the glass. With high ratios for network loosener/network stabilizer leading to less stability in the glass.
Environmental factors
There are many different environmental factors which may play a role in the deterioration pro-cess. The most important ones are: relative humidity (RH), temperature, chemical properties of attacking solution and presence of other pollutants, other less important factors include: micro-organisms, storage history and the ratio between the attacking solution volume and the surface area of the attacked glass.[2] A lot of research has been done on environmental factors, especially
on the relative humidity or RH.
An example of such research is a paper by K. Cummings et al., here it was found that a high RH leads to an increased rate of hydration and if combined with atmospheric pollution the rate of the process is increased even further.[15] A different paper, written by Fearn et al. showed
that the depletion dept of sodium is increased as a result of a higher RH.[16] Other papers were
written on the effect of temperature like that of Rohanova et al. showed that an increase in temperature not only increases the dissolution of the glass network but also shifts the
7 mechanism through which it happens.[17] Finally, interesting articles can be found on
atmos-pheric pollutants as well, as these might form salts, increase the pH and/or influence the rate at which the deterioration processes happens.[15][18]
In short, many environmental factors can have a negative influence on the glass this may be by increasing the rates at which the deterioration processes happen or by forming products on the surface. Furthermore, it should be noted that these factors almost never act on their own and are found to influence and interact with each other.
Preservation and conservation methods
The composition of glass objects obviously cannot be changed, so in order to preserve and conserve the glass objects the environment must be controlled. First of all it is important that the RH is controlled, as this will prevent the interaction of atmospheric water with the surface of the glass. Robert Organ was one of the first researches to point this out as he recommended that the RH should be below 42%.[19] Furthermore, as stated in 1.2.3 large fluctuations in RH
will lead to crizzling so these fluctuations must be kept as low as possible.[2][20]
Another possible conservation method might be flushing the storage with nitrogen gas. This way the presence of an unfavourable micro climate might be prevented.[20] The method
has been looked into but currently no solid evidence has been found to prove that it indeed effective for conserving historical vessel glass.[21][22]
1.2.5 Analysing unstable glass
There are many ways to analyse archaeological and historical glass, Koen Janssens has dedi-cated an entire book to possible methods from which the most relevant ones will be dis-cussed.[23] These methods can be used to analyse the composition of the glass and its altered
surface layer, which can tell us about the intrinsic stability as mentioned above, or to analyse the deterioration products that have formed, which would give us information about the deteri-oration process itself. In many cases a combination of multiple techniques is used to gain more complete and insightful results.
Multiple method types can be used to analyse the composition. X-ray fluorescence anal-ysis (XRF) is the most important method based on X-ray. It is non- destructive, can provide specific and understandable information about the composition. XRF can be used in situ but elements lighter than aluminium cannot be detected which is a major drawback seen that glass generally contains those elements. XRF needs a homogeneous sample to obtain reliable results and in the case of deteriorated glass this is not always the case. Scanning electron microscopy (SEM), laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) and Sec-ondary ion mass spectrometry (SIMS) are other examples of methods which can analyse the composition of the glass. Even though these methods give good results, they cannot be per-formed in situ. A sample of the glass must be taken and put inside of a vacuum chamber in order for it to be analysed, resulting in irreversible damage.
Methods which can be used for analysing the deterioration products that have formed include: infrared reflection absorption spectrometry (IRRAS), Raman, SIMS and ion chroma-tography (IC). With IRRAS and Raman the bond vibrations can be studied in situ, but are not always able to paint the full picture. SIMS is not suitable for analysing the deterioration prod-ucts as well, because it is damaging to the object. IC however, seems to be a very promising candidate as it does not damage the object and can detect many anions and cations related to the deterioration process. Furthermore, the transportation problem can be avoided entirely, as the analytical samples can be taken in situ.
8
1.2.6 Investigation into the stability of glass using ion chromatography
Verhaar has developed an analytical method using IC to analyse the stability of vessel glass.[7][24] With the developed method the compounds found on the surface of the glass are
analysed. If certain deterioration products are present and have a relatively high concentration it can be concluded that that piece of glass is suspectable to degradation and in need of special treatment. However, if the presence of these deterioration products is missing or if the concen-tration of the present deterioration products is too low the glass is likely not suspectable and no further measures will have to be taken. He tested the quality of his method and the IC-analysis thoroughly. The IC gave a good quantitative and repeatable analysis of the ions regarding the deterioration of glass with the exception of chloride and carbonate. It was found that sodium and potassium ions were the most important cations and can be used to give an indication of the stability of vessel glass. Consequently, the focus of this thesis will be on quantifying the concentrations of these specific ions.
Description of the inner workings of the IC
Ion chromatography (IC) is an analytical method based on High Performance Liquid Chroma-tography (HPLC) and is used to separate ions form aqueous solutions. A schematic drawing of the inner workings can be found in figure 1.7. The analysis starts at the auto-sampler here a sample is injected into the sampler loop, eluent is pumped from another tube and is degassed. The eluents is then pumped through the sampler loop and into the system. The solution first passes a guard column were large particles and ions damaging to the system are filtered out in advance. Then the solution passes through the analytical column. The ICS-1100 has an acidic solvent containing hydronium ions, while the ICS-2100 has a basic solvent containing
9 hydroxide ions. The ions from the solvents compete with the dissolved ions from the sample to bind to the column which has negatively charged side groups in case of the ICS-1100 system and positively charged side groups in case of the ICS-2100 system. As the anions have varying sizes and charges they come out of the column at different retention times, thus a successful separation is achieved. However, for the ICS-2100 system some of the anion retention times overlapped, thus a gradient was used to space them out. After passing through t he columns the ions are passed through a suppressor, which suppresses the noise of the signal. A conductivity detector then measures the conductivity through a flowcell. The conductivity is measured in micro Siemens and is plotted on the y-axis while the corresponding retention time is plotted on the x-axis. The peak surface can be determined and on basis of the calibration line the ion con-centrations are calculated.
1.3 Research questions and methodology
The water droplet method
An intern of Guus Verhaar named Danny Verschoor did a follow up research on the optimisa-tion of the IC method.[25] In the final weeks of his project he found a different method to
exam-ine the deterioration process. Instead of moistening the cotton swap, water droplets would be placed on the surface of the glass and removed at different times. This way, the exchange rate of the sampled ions between the glass and the water droplet could be measured, which might give an indication of the stability of the glass. The deterioration products that are extracted will not be influenced by any environmental factors as the ions are extracted directly from the glass. Using this method the leaching behaviour of the glass would be studied while Verhaars method studies the deterioration products on the surface. Unfortunately, Verschoor did not have the time to work out this method and only reported preliminary results, which seemed to be prom-ising. Follow up research would be needed to collect more data on this method and find out if it would be reliable, reproducible and capable of correctly predicting the stability of the glass. The research done in this thesis is builds on the work done by Verschoor and the following questions will be answered:
1. Can the water droplet method be made reliable and reproducible?
In order for the water droplet method to be reliable and reproducible the placement and extrac-tion of the droplets would have to be done systematically. Different methods were tested and are described in methodology.
2. To what extent is the water droplet method suitable for indicating the stability of histor-ical vessel glass?
The method was tested on two types of stable and two types of instable glass. The obtained concentrations was plotted against time. It is expected that for unstable glass the ion concentra-tion against time will give a flattening curve while for stable glass no curve or only a slight increase in ion concentration over time would be found.
During the research interesting results were found regarding the location of the droplets on the glass and a third question arose:
3. To what extent are the ion concentrations of a sample influenced by the location at which it was taken on the glass?
10 It is expected that the location on the glass vessel influences the concentration of the deteriora-tion products that is found. Firstly, the four different glass types (used to answer the second research question) were researched on their place dependency using the finalized water droplet method. Furthermore, Verhaars approach was used on a variety of glass vessels from Stedelijk Museum Amsterdam to test the hypothesis using a validated method as well.
1.4 Thesis structure
The thesis is divided into 5 chapters. The first chapter was the introduction of the subject, the research questions and methodology were discussed as well. In the second chapter, the equip-ment and materials are discussed in full, followed by the existing methods made by Verhaar and Verschoor. Finally, the research approach and the methods used for the research will be discussed. The obtained results are presented and discussed in the third chapter. In the fourth chapter the research questions are answered. In the fifth and final chapter different follow-up researches are proposed.
11
2. Equipment, materials & methods
2.1 Equipment
A Digital HIROX microscope was used to take pictures of the drop-lets. Making the internal standards and placing droplets on the glass surface was done with pipettes made by Thermo Scientific (Wal-tham, USA) and had the following capacities: 20-200 µL, 100-1000 µL and 500-5000 µL. The HPLC grade water (milliQ) that was used was made by following water purifying machine: Thermo Sci-entific Barnstead MicroPure ST (Figure 2.1). The VWR, micro star 17 was used as centrifuge and as vortex the VWR Reax top, VWRI444-1372.
Ion chromatograph
The Dionex ICS-1100 IC system was used to analyse cations, it was equipped with an Ionpac CS12-A 2x250 mm analytical column protected by a CG12-A 2x50 mm guard column, a Dionex DS6 conductivity detector and a CERS 500 Dionex cation
electrolyti-cally regenerated suppressor. Methane sulphonic acid (20 mM) was used as an eluent and was pumped through the system at a flow rate of 0.25 mL/min over a runtime of 15 minutes.
The Dionex ICS-2100 IC system was used to analyse anions, it was equipped with an Ionpac AS17-C 2x250 mm analytical column protected by a AG17-C 2x50 mm guard column, a Dionex DS6 conductivity detector, a Dionex
EGC-III eluent generator cartridge, a CR-ATC trap column and a AERS 500 Dionex anion electrolytically regen-erated suppressor. Potassium hydroxide (gradient ranging from 0.5 to 45 mM) was used as an eluent and was pumped through the system at a flow rate of 0.37 mL/min over a runtime of 22 minutes. A Dionex EGC-III eluent generator cartridge was used to gen-erate the gradient of the eluent. In order to remove unwanted anionic contaminations a CR-ATC trap column was installed.
Both of these systems were connected to an autosampler: Thermo Scientific Dionex AS-AP. In Figure 2.2 the entire setup can be found.
2.2 Materials
Swabs
Texwipe Alpha TX761 swabs were used which have been abbreviated in the procedure to pol-yester swab.
Stock solutions
The internal cation standards were made using milliQ and a Dionex 6 cation stock solution containing: lithium (50 mg/L), sodium (200 mg/L), ammonium (400 mg/L), potassium (200 mg/L), magnesium (200 mg/L) and calcium (1000 mg/L). The internal anion standard was made using milliQ and a Dionex seven-anion IC-standards solution containing: fluoride (20 mg/L), chloride (30 mg/L), nitrite (100 mg/L), bromide (100 mg/L), nitrate (100 mg/L), phosphate (150
Figure 2.1 Thermo
Sci-entific Barnstead Micro-Pure ST.
Figure 2.2 The complete setup of the IC, from left to right: terminal, au-tosampler, ICS-2100, ICS-2200.
12 mg/L) and sulphate (150 mg/L). A second anion standard was made using milliQ and Dionex IC-standard carbonate (1000 mg/L), acetate (1000 mg/L) and formate (1000 mg/L) solutions.
Glass types used
During the research of this thesis a total of four different glass types were used: two of which were unstable, KVL and JXS, and two of which were stable, 19AY and Quartz, their composi-tions can be found in Table 2. The KVL sample is a piece of an oval sheet of historical glass originating from the frame of a print or painting. It had been established that the glass was in a developed stage of degradation as crizzling can be seen on the surface. The JXS and 19AY glasses were not actual historical glasses but were produced by Robert Brill from the Corning Museum of glass in the 60’s and 70’s. The quartz (Alfa Aesar, Haverhill, USA) that was used was that of a microscope slide, these slides consists out of pure SiO2.
Table 2. Composition (wt. %) of the glass types used for this thesis.
Glass type SiO2 Na2O K2O CaO MgO Stability
KVL 68.8 5.7 19.8 2.9 0.1 Unstable
JXS 73.5 22.0 - 2.0 1.0 Unstable
19AY 66.0 17.0 2.0 10.0 10.0 Stable
Quartz 100 - - - - Stable
2.3 Existing analytical procedures
2.3.1 Preparation and use of the internal standard
The internal standard for the ICS-1100 was made by diluting the Dionex 6 cation stock solution 100 times this brought the sodium concentration down from 200 mg/L to 2.00 mg/L. From this solution six standards were made with the sodium concentration ranging from 0.040 mg/L to 2.0 mg/L. For the ICS-2100 the Dionex seven-anion IC-standards solution was diluted 10 times this brought the nitrite concentration from 100 mg/L to 10.0 mg/L. Six standards were made using this solution with the nitrite concentration ranging from 0.200 mg/L to 10.0 mg/L. An-other standard was made for the ICS-2100 system using the carbonate, acetate and formate solutions. The carbonate and acetate were diluted from 1000 mg/L to 6.000 mg/L while formate was diluted from 1000 mg/L to 5.000 mg/L. Five standards were made using the solution with the concentration of carbonate ranging from 0.12 mg/L to 6.0 mg/L. After preparation, the so-lutions were homogenized using a vortex. The samples were analysed by the ion chromato-graph. The peak surface areas obtained from the IC were plotted against the known concentra-tions of the solution and the calibration curve was obtained. The coherency of these values were expressed using the coefficient of determination, R2, if this value exceeded 0.990 it was
con-sidered accurate and the curve would be used. The exact ion concentrations in a sample could then be calculated by plotting the obtained signals from the injected samples against the cali-bration curve.
2.3.2 The swab method by Guus Verhaar
As discussed in the introduction, Verhaar has developed a method for the quantitative analysis of the degradation products. A Teflon template was placed on the piece of glass and a polyester cotton swab was moistened on one side with milliQ (50 µL). The swab was then systematically stroked in two perpendicular directions across the glass surface within the boundaries of the template, this was repeated with the dry side of the swab to collect any remaining solution. The polyester swab tips were chipped of and placed inside polypropylene centrifuge tubes here they were extracted in milliQ (1350 µL) for one hour. Afterwards, the swabs were removed and the
13 tubes were centrifuged for four minutes. From the solutions, 900 µL was pipetted into a vial which was analysed by the ICS-1100 and ICS-2100 systems.[7]
2.3.3 The water droplet method by Danny Verschoor
The surface of the glass was cleaned where after multiple droplets were placed on the glass surface using a pipette. These droplets were removed from the surface at different times using a polyester swab. The swabs were extracted in milliQ for one hour. Afterwards, they were re-moved and the solution was pipetted into a vial. This solution was analysed by the ICS-1100 system.[26]
2.4 Research approach and experiments
As mentioned in the introduction, Verhaar concluded that sodium and potassium could be used to give an indication of the stability of the glass. The main focus in this thesis will lay on the quantification of these ions. Therefore, most of the experiments in this thesis used solely the ICS-1100 for the analysis of the samples. An exception is made for the samples from Stedelijk Museum Amsterdam as these were analysed by both the ICS-1100 and ICS-2100 as described in 2.3.2 by Verhaars procedure.
2.4.1 Optimisation of the swab method
Verhaar wondered about the length of the extraction time of the polyester swab in milliQ. Thus different extraction times were tested. If a shorter extraction time would result in a similar or higher concentration of ions, the extraction time could be shortened making the procedure more efficient. However, if the ion concentration would be significantly lower for shorter extraction times the original 60 minutes must be used as this has already been proven to yield reliable results.[7]
A stock cation solution was diluted 100 times the sodium concentration decreased from 200 mg/L to 2mg/L. Polyester swabs were moistened with milliQ (50 µL) followed by the diluted stock solution (40 µL). These were then put in milliQ (1310 µL) and extracted for 5, 20, 30 and 60 minutes. These extractions were done in triplo, blanks were also taken with only milliQ (90 µL) on the polyester swaps. After the extraction time had passed the polyester swab tips were removed and 900 micro litre of the solution was transferred into a vial. The vials were analysed by the ICS-1100 system. This procedure was repeated with the extraction times of 30, 45 and 60 minutes.
2.4.2 Optimisation of the water droplet method
The different steps of the water droplet method
The method as described by Verschoor consists of five main steps: cleaning the surface of the glass, placement of the water droplets on top of the glass surface, removing the water droplets from the glass surface, extracting the deterioration products from the polyester swab and ana-lysing the sample that has been taken. The protocol will be examined on its reliability and re-producibility, see 3.2. If necessary, adjustments will be made to the protocol to improve it.
For the second step, placement of the water droplets on top of the glass surface, the size and spread of the water droplets are studied while for the third step, removing the water droplets from the glass surface, the completeness of the water droplet removal is looked into. The water droplet size and spread was studied for all glass types. The completeness of the water droplet removal was studied for the KVL and Quartz glass. The KVL glass is unstable and so a large ion concentration is expected, thus variation will be easier to detect. The Quartz glass is stable
14 and consists entirely out of SiO2 any other ions detected will be from the environment or the
milliQ.
Size and spread of the water droplets
On different locations of the glass surfaces 50 µL droplets were placed (the locations for each droplet on each of the glass types can be found in the templates below, figure 2.3). From these droplets pictures were taken with a microscope. The procedure was repeated with 25 µL drop-lets.
Figure 2.3 The templates used for the testing the water droplet size and spread, from top left to
bottom right: KVL, JXS, 19AY, Quartz.
Improving the method
In order to improve the water droplet method a different procedure for the placement of the water droplets was sought. Multiple methods were considered and include:
1. Placing the droplets on the glass surface with a pipette and extracting it by sucking it up with a pipette. The obtained solution might have to be diluted or can be injected straight into the IC. In order for this method to work, the droplets would have to retain their surface tension and not spread out.
2. Placing the droplets on the glass surface without breaking contact with the tip of the pipette while retaining pressure. Afterwards, the pressure is released resulting in the
15 suction of the droplet directly back into the tip of the pipette. The obtained solution might have to be diluted or can be injected straight into the IC.
3. Placing the droplets inside a Teflon template on the glass surface with a pipette and removing it with polyester swab. The deterioration products would have to be extracted from the polyester swab, before it could be analysed by the IC.
The following procedures were executed to see if the methods proposed above would be suita-ble. Quartz glass was used for every experiment as the droplet size and spread would not differ as much as for the other glass types.
1. On the surface of a slide of quartz glass a droplet of milliQ (50 µL) was placed with a pipette. After 5 minutes had passed the pipette was used to extract the droplet from the surface.
2. A droplet of milliQ (50 µL) was partially ejected onto the quartz glass surface while remaining contact with the pipette tip. It was held in this position for five minutes, where after it was sucked back up and ejected into a vial.
3. On the quartz glass surface a Teflon template was placed, inside this template a droplet of milliQ (50 µL) was deposited with a pipette. After 5 minutes a polyester swab was used to remove the droplet form the surface by swabbing the surface systematically in two perpendicular directions. The swab was flipped over and the same step was re-peated.
Removal of the water droplets
On the KVL glass surface two droplets of milliQ (50 µL) were deposited using a pipette on positions 2 and 8, see figure 2.3. After 30 minutes a polyester swab was used to remove the droplet form the surface by swabbing the surface systematically in two perpendicular directions. The swab was flipped over and the same step was repeated. Afterwards, a completely dry swab was used to wipe up any water droplet residue. On Quartz glass the same procedure was re-peated for location 1.
2.4.3 Application of the optimized water droplet method
The finalized water droplet method that was drafted from the results of the previous sections is given below:
Firstly, the outlines of the glass were roughly traced with a marker on a sheet of thin layer film. On this film circles were drawn using the Teflon template, these circles are marked with different exposure time. The surface of the glass was cleaned with an ethanol milliQ solu-tion (v/v 1:1) to remove any excess of impurity’s and atmospheric pollutants. The glass was then placed on top of the film and in each of the drawn circles a droplet of milliQ (50 µL) was placed. After the appropriate exposure time had passed the droplets were extracted with a pol-yester swab. This was done by swabbing systematically in two perpendicular directions across the glass where after the polyester swab is flipped and the same is done with the other sid e, which is still dry. The polyester swab tips were then chipped of and extracted in milliQ (1350 µL) for one hour. Afterwards, the polyester swabs were taken out and disposed of. 900 µL of the solution was pipetted into a vial which was then analysed by the IC.
Two unstable glass types were sampled using the finalized method in order to find the expected curve between the concentration and the time the droplets had been on the glass sur-face. The time the droplets have spent on the glass surface is referred to as ‘exposure time’ in the rest of the thesis. For comparison two stable glass types were sampled as well. As this glass would not be undergoing any deterioration process, the ion concentrations should be low and nothing more than a slight increase in concentration over time is expected.
16 The unstable glass type: KVL was tested first, to be as efficient as possible 8 different locations on the glass were used see figure for template. The other unstable glass type: JXS was tested secondly, again 8 different location were used see figure for template. Each of locations had a different exposure time starting at 0 minutes counting up to 70 minutes with 10 minute intervals. The templates can be found in figure 2.4 below.
Figure 2.4 The template used for the first and second run for the KVL glass (left), the template
used for the first run of the JXS glass (right).
Based on the obtained results from the first experiments the time intervals were altered and consists out of the following exposure time: 0, 2, 5 and 10 minutes. The KVL glass was tested at only four out of the eight possible locations, the reasoning can be found in 4.4. For the JXS glass the samples for 0, 5 and 10 minutes were taken in duplo while the samples for 2 minutes were taken in triplo. The templates that were used for these experiments can be found in figure 2.5.
Figure 2.5 The template used for the third run of the KVL glass (left), the template used for the
second run of the JXS glass (right).
The stable glass types: 19AY and Quartz, were also tested for this interval as a reference. The following exposure time were used: 0, 2, 5, 10 and 20 minutes. All of the samples were taken in triplo. The templates used can be found in figure 2.6. As the Quartz slides were relatively small multiple were used so that every measurement could be taken in triplo.
17
Figure 2.6 The template used for the 19AY testing (left), an example of the template used for
the Quartz testing (right).
2.4.4 Place dependency
KVL, JXS, 19AY and Quartz
In the previous section, the application of the finalized water droplet method, it was found that the place may also play a role in the concentration of the ions. To test this hypothesis, different locations on the four glass types were sampled using the finalized water droplet method (figure 2.7). However, for all of the different locations the time was kept constant at 30 minutes, which makes the location the only variable. If the difference in ion concentrations for different loca-tions on the same glass is significant it can be concluded that the place where the sample was taken also plays a role. If not, the place does not play a role and another factor will have caused the unexpected results in 3.3. As the 19AY and Quartz glass are stable no large difference in concentration is expected between different locations on the same glass. For KVL and JXS differences might be expected as the deterioration process might not have the same rate for each location.
18
Figure 2.7 The templates used for the testing place dependency of the ion concentration for
each of the glass types, from top left to bottom right: KVL, JXS, 19AY, Quartz.
Glass objects from SMA
In order to test the hypothesis with a validated method as well, samples from the SMA were sampled using the swab method of Verhaar. A total of 138 samples were taken from 22 different objects. A complete list of these objects can be found in Appendix 2. On different glass vessels different locations were sampled, most of these locations could be divided into: inside, outside, topside or bottom side. The remaining locations were labelled as ‘other’. The objects were screened on their physical appearance and 5 were classified as unstable while the other 17 ob-jects were identified as stable. However, as they were not screened using the swab method of Verhaar some objects classified as ‘stable’ at first sight might be unstable as well.
19
3. Results and discussion
3.1. Optimisation of the swab method
To find out if the time of extraction could be shortened, stock solutions were applied on poly-ester swabs which were extracted at different times in triplo. Using the internal standard the concentration of the cations was calculated in parts per million (ppm). The average concentra-tions were calculated (in ppm) and plotted against the corresponding time (in minutes). Lastly, error bars indicate the standard deviation.
Figure 3.1 The concentrations of the different cations plotted against extraction time.
0 20 40 60 80 0 0,1 0,2 0,3 0,4 Time (minutes) C on ce nt ra tio n (ppm )
Lithium
0 20 40 60 80 0 0,5 1 1,5 Time (minutes) C on ce nt ra tio n (ppm )Sodium
0 20 40 60 80 0 0,5 1 1,5 Time (minutes) C on ce nt ra tio n (ppm )Magnesium
0 20 40 60 80 0 0,5 1 1,5 2 Time (minutes) C on ce nt ra tio n (ppm )Potassium
0 20 40 60 80 0 2 4 6 8 Time (minutes) C on ce nt ra tio n (ppm )Calcium
0 20 40 60 80 0 1 2 3 Time (minutes) C on ce nt ra tio n (ppm )Ammonium
20 As seen in the graphs of figure 3.1, the concentrations of the different ions increase drastically from 5 to 30 minutes. After 30 minutes the graphs flatten out, with the exception of magnesium. The concentrations of the ions at 30, 45 and 60 minutes are very similar to each other, especially when considering the standard deviation and error bars, thus the extraction procedure could be shortened.
3.2 Optimisation of the water droplet method
The different steps of the water droplet method
The first step of the method, the cleaning of the surface of the glass, could be done using a simple protocol. The surface could be swept with a piece of torque that is moistened with an ethanol milliQ solution (v:v/1:1). Afterwards the surface can be dried with another piece of torque. The reliability and reproducibility of this first step would have to be tested to ensure it is reliable and reproducible. However, as the protocol for this step is rather simple and due to COVID-19 only a limited amount of time is available on the lab it is left out of this thesis.
The reliability and reproducibility of the second step, placement of the water droplets on top of the glass surface, is challenging as the size and the spread of the water droplets needs to be the roughly the same. Large variation in droplet size and spread would lead to a difference in effective surface areas, ultimately causing the concentrations for a measurements on the same spot to vary, this would make the placement of the water droplets using a pipette insufficient.
In the third step, removing the water droplets from the glass surface, the droplets would have to be removed as complete as possible every time. If the completeness of the water droplet removal would vary so would the final measured concentration.
The fourth and the fifth step, extracting the deterioration products from the polyester swab and analysing the sample that has been taken, are identical to the steps used in the procedure of Verhaar, which have already been validated.
Size and spread of the water droplets
To investigate the size and spread of water droplets, different volumes of water were pipetted on the surfaces of different glass types. Beneath in figures 3.2-3.8, the prime examples of the droplets are presented, the complete set of pictures can be found in the appendix. Using the scale found in the left bottom of the pictures, an estimation was made of the surface areas of these prime examples, see table 3.
21
Figure 3.3 KVL glass, 25 µL droplet, location 1 (left) and location 3 (right).
Figure 3.4 JXS glass, 50 µL droplet, location 1 (left) and location 3 (right).
Figure 3.5 JXS glass, 25 µL droplet, location 1 (left) and location 3 (right).
22
Figure 3.7 19AY glass, 25 µL droplet, location 1 (left) and location 4 (right).
Figure 3.8 19AY glass, 50 µL droplet (left), 25 µL droplet (right).
Table 3. Surface areas of the different droplets on the different glass types.
Glass type KVL JXS 19AY Quartz
Sample lo-cation 1 3 1 3 1 4 1 2 Surface area (mm2) 50 µL 38.5 150 100 105 138 105 44 48 Surface area (mm2) 25 µL 18.1 49.7 36.3 57.0 38.5 112 28 33
It can be seen in figures 3.2-3.8 and table 3 that the size and spread of water droplets with a volume of 50 µL at different locations on the KVL and 19AY glass differ a lot while the dif-ference droplet size and spread for JXS and Quartz is small. When decreasing the size of the to 25 µL the difference in size and spread increases for KVL, JXS and 19AY while the difference is nearly the same for Quartz.
From these results it can be concluded that pipetting the droplets onto the glass surface would not suffice for the glass types KVL and 19AY for both 25 µL and 50 µL volumes. The JXS and Quartz glass can be sampled using the pipetting method as the difference in droplet spread and size is small for a 50 µL volume. Quartz can also be sampled with droplets having a 25 µL volume while JXS cannot. Overall, an alternate method should be looked into which can control the size and spread of the water droplets for all of the different glass types.
