The effect of air pollution on stone decay: the decay of the Drachenfels trachyte in industrial, urban, and rural environments—a case study of the Cologne, Altenberg and Xanten cathedrals

The three buildings are located in very different environmental settings (Fig. 1). Cologne cathedral (53 m above NN) is located in a metropolitan center with one million inhabitants next to the river Rhine. Xanten cathedral (22 m above NN) is the Catholic church of a small city on the Lower Rhine with 18,000 inhabitants and with smaller industrial effects (Arnhem, NL), while Altenberg cathedral (149 m above NN) is situated in a greenfield setting surrounded by forested mountains in the “Bergisches Land”.

Middle Europe belonging to the mid-latitudes has a damp, cool temperate climate. The western German climate shows a maritime influence due to its geographical proximity to the North Sea and the Atlantic, and therefore the Gulf Stream has an impact through the West Wind Drift (Lauer and Bendix 2004). This geographic position provides mild winters and moderate summers. Temperature, relative humidity, and rainfall at the three locations differ only a little (Table. 1). The mean annual temperature is 11.4 °C in Cologne, 10.6 °C in Xanten, and 7.1 °C in Altenberg. Warmest months are July and August; coldest months are January and December, with temperatures around −10 to +30 °C and relative humidity ranging between 65 and 95 %. The relative humidity shows highest annual values of 90–95 % in winter (November–February) and lowest annual values for April–August from 65 to 75 % (Table 1). Climate data from Hungary show less maritime but more continental influence.

The buildings are exposed to ubiquitous air pollutants of anthropogenic origin. These are mainly gaseous pollutants like SO₂ and NO_x. Concentrations of both of these have decreased over the last decades, causing an increase in precipitation pH (Fig. 2). The effect of particulate matter in the form of settling dust shows continuously steady values. SO₂ shows highest annual values in the winter months (December–March) and lowest annual values in August.

Data for the SO₂-fluxes show a strong decrease over the past 30 years, with a comparable impact for Cologne and Xanten and a lower influence in Altenberg. The particulate matter (PM10) has been monitored since 2003/2004 and shows the highest values for Xanten and the lowest for Altenberg. The relatively high values for Xanten in comparison to Cologne may be explained by a certain pollution impact from the bigger city of Arnhem, NL, from which pollution fluxes are transported with west winds. Also Altenberg is subject to west winds and therefore, to a certain impact by the pollution fluxes of the industrial area of Cologne, Leverkusen, and Düsseldorf. In general, the comparably low values for Altenberg reflect a low impact region of an arborous rural area. The Cologne cathedral is located in the city center next to the main railway station, which has served as an active traffic interchange since the industrial era. Within the city of Cologne there are four power plants, and 15 km south west of Cologne is a larger coal-fired power plant.

To understand the context of building stone diversity and the different deterioration features, a sample field at Cologne cathedral was mapped analogous to the system developed by Fitzner et al. (1995). Samples from crusts and dust as well as unweathered and weathered rock samples were collected from the Cologne, Xanten, and Altenberg cathedrals from the different building stones (Drachenfels trachyte, Obernkirchner sandstone, Krensheimer Muschelkalk).

Sample preparation and sample analyses were performed on the different sample types using overlapping techniques in mineralogical and geochemical analyses whenever possible. Thin sections perpendicular to the exposed surface of the rock were prepared and textural analysis of thin sections was performed by polarizing microscopy. The chemical data presented focus on C, Na, Mg, Si, S, K, Ca, Al, Ti, Mn, Fe, Zn, As, Sb, Pb, Bi contents of the host rocks and crusts.

The chemical composition was obtained by XRF spectroscopy. Major element oxides and the trace elements Ba, Cr, Ga, Nb, Ni, Rb, Sr, V, Y, Zn, and Zr were analyzed by XRF on 105 °C-dried samples, prepared as fused disks of lithium tetraborate-metaborate (FLUXANA FX-X65, sample-to-flux ratio 1:6). A PANalytical Axios Advanced wavelength-dispersive spectrometer and matrix correction programs were used to calculate concentrations. H₂O⁺ and CO₂ were determined using a Vario EL III (Elementar Analysensysteme GmbH, Hanau/Germany). An ELTRA CS 2000 (ELTRA GmbH Neuss) is used for measuring sulfur.

The mineralogical composition of the samples (black crusts and dust) was determined by XRD. Powder X-ray patterns were obtained using a PANalytical Empyrean powder diffractometer with Cu Kα radiation, automatic divergent and antiscatter slits and a PIXcel^3D detector. The diffraction data were recorded from 5° to 85° 2θ via a continuous scan with a step-size of 0.013 and a scan time of 60 s per step. The generator settings were 40 kV and 40 mA. The Rietveld algorithm BGMN was used for quantitative analysis (Bergmann et al. 1998).

Laser ablation inductively coupled plasma mass spectrometry analyses were performed on thin slices of the samples. The laser used was a Compex 110 Excimer (ArF 193 nm) by Lambda Physic (Goettingen, Germany), a GeoLas optical bench by MikroLas (Goettingen, Germany), a small volume sample chamber, an ablation pit with a diameter of 120 μm, 10 Hz repetition rate for the laser pulses, and about 3 J/cm² available energy. The mass spectrometer was a Perkin Elmer DRC II (Siex, Canada). Calibration was done using NBS610 (NIST, USA), internal Standard 43Ca, dwell time 10 ms/isotope, 0.925 s per sweep, and a total of 250 sweeps giving a total measurement time of 3 h and 50 min. Measured isotopes were: Li7, Na23, Mg24, Mg25, Al27, Si29, P31, S34, Cl35, K39, Ca43, Sc45, Ti47, Ti49, V51, Cr53, Mn55, Fe57, Co59, Ni60, Cu63, Cu65, Zn66, Ga71, Ge73, As75, Rb85, Sr88, Y89, Zr90, Nb93, Mo9.

To visualize the microfabric of the crust and the host rock and to detect elemental-mineralogical composition of samples, SEM–EDX techniques were applied on thin sections as well as on small fragment samples (LEO GEMINI SEM 1530 and LEO 1455 Gemini as well as AMRAY 1630). EDX-analyses were performed on a Quantra 200F (Fei) with a field emission cathode with an initial voltage of 20 kV (Department of Crystallography, Geoscience Center University of Goettingen, Germany).

Wavelength dispersive microprobe analyses were performed with a JEOL JXA 8900 RL instrument (Department of Geochemistry, Geoscience Center University of Goettingen, Germany). For quantitative measurements, 15 kV acceleration voltage, 15 nA beam current on the Faraday cup, a defocused beam of 3.5 μm, and counting times between 15 s on the peak for Na, Mg, Al, Si, K, Ca, and Fe and 30 s for P, S, Ti, and Ba were chosen. Data processing was done with the CITZAF routine in the JEOL software, which is based on the Φ (ρZ) correction method (Armstrong 1991, 1995). The following standards were used for the analysis: Albite for Na, MgO (synthetic) for Mg, anorthite for Al, wollastonite for Si and Ca, sanidine for K, apatite for P, baryte for S, TiO₂ (synthetic) for Ti, rhodonite for Mn, haematite for Fe and celsian for Ba. Detection limits are calculated from the error propagation of the two measurements of the background signals of each X-ray line and are given as a 2-sigma value. The element distribution of Mg, Al, K, Ca, Fe (WDS) and S, Si (EDS) was mapped using an acceleration voltage of 15 kV and beam current of 30 nA. The acquisition time was set to 50 ms per step. The scan grid was spaced at 0.5 μm per step, covering in total 400 × 400 steps. Simultaneous acquisition of the backscatter signal in composition mode was performed.

For PAH analysis, the pulverized rock samples (1 g) were weighted into Teflon vessels, internal standards (acenaphthen D 10, phenanthren D 10, pyren D 10, chrysen D 12, perylen D 12, and bBenzo (g,h,i) perylen D 12) and 5 mL isohexane/acetone (3:1 v:v) were added and subsequently extracted on a MARS XPRESS microwave system at 130 °C for 20 min. After cooling, the supernatant was removed and transferred into purging vials and after the addition of 500 μL toluene purged in a gentle stream of nitrogen until dry. The remains were redissolved into 500 μL toluene, transferred into autosampler vials and centrifuged at 5,000 rpm for 10 min. They were then immediately analyzed on a GC–MS (Agilent 7890A, 5975E). Separation of the analytes was achieved using a Varian VF5-ms column (30 × 250 × 0.25) and a steady temperature gradient of 6°/min up to a final temperature of 325 °C. Data were recorded in single ion monitoring mode (quantifier in parenthesis) with a dwell time of at least 50 ms per amu for the following analytes: acenaphthylen (152), fluorene (166), phenanthrene (178), anthracenene (178), pyrene (202), 7H-benzo-fluorene (216), cyclopenta(c,d)-pyrene (226), benzo(a)anthracene (228), chrysene (228), 5-methyl-chrysene (242), benzo (b,j,k) fluoranthene (252), benzo(a)pyrene (276), indeno(1,2,3-cd)-pyren (278), dibenzo(a,h)-anthracene (276), benzo(g,h,i)-perylene (302), dibenzo(a,l)-pyrene (302), dibenzo(a,e)-pyrene (302), dibenzo(a,i)-pyrene (302), dibenzo(a,h)-pyrene (302). Quality was assured by simultaneous scanning 50–350 amu. The method quantitation limit was 1 μg/kg.

The Drachenfels trachyte, used for construction in the Rhineland since Roman times (Berres 1996; Wolff 2004), was the main construction material for the three investigated buildings. The trachyte has a medium porosity of 11.9–13.4 % and is of light gray color (Graue et al. 2011). Two different varieties of Drachenfels trachyte can be distinguished macroscopically: a partially pale grayish one and a yellowish reddish porphyritic trachyte (Simper 1990). The fabric of the trachyte can be divided into three different fabric elements: the very large phenocrysts of sanidine, up to 7 cm in size, are a characteristic feature of the stone. Plagioclase phenocrysts are observed as well. Secondly, a micro- to cryptocrystalline groundmass consists of mainly feldspar laths and quartz. Thirdly, altered volcanic glass fractions as interstitial material between the small feldspar laths of the groundmass form a mesostasis. These interstitial areas of recrystallized glass are much fractured with a high and distinct porosity (Fig. 3a). The Drachenfels trachyte has a classical porphyry fabric: the large-oriented phenocrysts are embedded in a very fine crystalline fluidal groundmass. The sanidine phenocrysts show a preferred orientation, tracing the flow fabric due to the former flow direction of the magma. The small lath-shaped feldspars flow around the large idiomorphic crystals, with the result that several textural domains with preferred orientation of the groundmass feldspar laths exist (Fig. 3b). The sanidine as well as the plagioclase phenocrysts show significant crack formation, which can be considered as a secondary porosity. This probably traces back to geological relaxation processes during the rise of the magma.

The rock mainly consists of sanidine (50 wt. %), plagioclase (24 wt. %) and quartz (13 wt. %). Other components are augite (5 wt. %) and biotite (5 wt. %) and common accessory minerals are ore (2 wt. %) and apatite, zircon and sphene (1 wt. %) (Vieten 1961). The sanidine phenocrysts often show Carlsbad twins and oscillatory zonation. Biotite is often strongly altered. The micro- to cryptocrystalline groundmass (64 wt. %) consists of feldspar and quartz (Simper 1990). The plagioclases are zonal structured and contain about 30 mol % anorthite (Ab₇₀An₃₀) (Kraus 1985a, b). The interstitial volcanic glass fractions are often recrystallized and altered to montmorillonite (Koch 2006) (Fig. 3c). Vieten (1961) detected 1–5 wt. % montmorillonite in samples from different outcrops at the Drachenfels. In places, calcite occurs in contiguity with Fe-oxides (Fig. 3d) indicating alteration processes.

The different building stones of the Cologne cathedral show a large variation of weathering phenomena. In particular, the Drachenfels trachyte shows severe deterioration in the form of cracks, surface deterioration, and back-weathering coexisting with flaking, exfoliation, and structural disintegration to crumbling and the massive formation of gypsum crusts (Fig. 4a) (Graue et al. 2011). The flaking can occur in a very pronounced fashion (von Plehwe-Leisen et al. 2007), which eventually leads to structural disintegration and complete material loss (Fig. 4c). Back-weathered areas often display stronger further decay in terms of microcracks, crumbling to total fabric collapse (Fig. 4d). Scaling is observable and very often shows a granular disintegrated zone on the reverse side of the scale, whereas the original stone surface generally still exists (Fig. 4b). Formation of fissures may also propagate many centimeters in depth into the stone. The flow direction of the Drachenfels trachyte, which is indicated by the orientation of the large crystals of sanidine, has a certain impact onto the weathering behavior of the stone. In the Drachenfels trachyte, the deterioration is more intense when the flow fabric is parallel to the visible surface of the building stone, i.e., the preferred orientation of sanidines is surface parallel. Sanidine crystals and the groundmass matrix often show different weathering behavior depending on the mounting direction of the building stone. The large crystals are either weathered-out or are protruding due to the loss of the surrounding matrix. A third variation is the surface parallel weathering of components, matrix and phenocrysts (surface parallel-oriented) show a simultaneous back weathering.

On the Drachenfels trachyte, the formation of thin laminar crusts as well as thick framboidal crusts is observed (Fig. 4a). Black framboidal crusts tend to bulge out and detach from the stone surface. The stone structure in the background of these crusts is strongly weakened and disintegrates in the form of multiple flaking, exfoliations, and further crumbling. Thin laminar crusts often build up on structurally intact stone surfaces but contour scaling often accompanies this. These surface parallel scales show a thickness of a few millimeters to 1–2 cm with the formation of a brittle disaggregated zone on the back.

The deposition of airborne particulates and the formation of crusts can be observed on surfaces over the entirety of Cologne cathedral. These surface deterioration features appear in different manifestations from patina-like grayish black surface depositions and soiling to thin black laminar crust and thick framboidal- or cauliflower-like black crusts. Especially in context with framboidal crusts surface detachment and loss as well as further disintegration is observed (Fig. 5a). These cauliflower-like weathering crusts vary extremely in thickness from 2 to 15 and even 30 mm (Török et al. 2011). Their specific morphology also describes them as dendritic (Török et al. 2011), globular (Bonazza et al. 2007), ropey (Antill and Viles 1999), or framboidal (Török 2003, 2008). They display globular or rosette-like formations of gypsum crystals. Calcareous and also quartz particles cover the stone surface and organic as well as inorganic particles from the pollution fluxes are incorporated into the crusts. Framboidal crusts generally build up in sheltered to moderately exposed areas as well as in cavities on vertical stone surfaces. They are often observed at different sites and for different building stones (Amoroso and Fassina 1983; Camuffo 1995; Sabbioni 1995; Bonazza et al. 2004, 2007).

Thin laminar black crusts trace the stone surface and may cover complete sections of the building’s structure not necessarily preferring protected sites (Fig. 5b). This kind of crust does not change the morphology of the stone surface (Fitzner et al. 1995) and seems to have very strong bonds with the stone surface (Török et al. 2011; Siegesmund et al. 2007). Scaling is often observed in context with laminar crusts.

Other investigated building stones of Cologne cathedral also show crust formation. On the Schlaitdorfer sandstone massive gypsum crusts build up due to the carbonate cement (app. 14 %) leading to characteristic disintegration in the form of scaling, flaking, and granular disintegration into sand (Lukas 1990). The main deterioration phenomenon of Stenzelberg latite is a typical formation of scales with a thickness of 2–3 mm (Graue et al. 2011). Obernkirchner sandstone has a high weather resistance, but shows black crusts and the formation of gypsum crusts in posterior and sheltered areas. Krensheimer Muschelkalk as a carbonate building stone shows massive gypsum crust formations. This is visible in rain-protected areas, while on surfaces exposed to rain, solution phenomena can be observed, e.g., microkarst. Unlike the Drachenfels trachyte, the limestone shows no structural disintegration with crust formation. At first, it has almost a consolidating or inhibiting function.

In order to describe the variations of the different building stones and the specific deterioration features, a representative survey area has been mapped in accordance with the classification by Fitzner et al. (1995) (Fig. 6). The building material and the deterioration phenomena-surface deterioration, gypsum crusts, scaling, flaking, cracks and depositions-have been mapped displaying their distribution within the selected wall area. The detailed map illustrated in Fig. 6 shows a section on the northern pillar of the flying buttresses of the choir, indicating that the major deterioration is on the Drachenfels trachyte.

If the three cathedrals from the industrial (Cologne), urban (Xanten), and rural (Altenberg) locations are compared, they show clear differences in terms of deterioration gradients and crust formation. While Altenberg cathedral shows only very little stone deterioration, in Xanten and especially in Cologne the decay is significant. At Cologne cathedral, severe damage can be observed entailing static disturbances. At all three cathedrals, scaling and flaking are the main deterioration features, though the observed effect is less at Xanten and Altenberg. As the mapping shows (Fig. 6), scaling and flaking are mainly concentrated in the Drachenfels trachyte. At Xanten and Altenberg cathedrals the deterioration is often located in superficial areas of the building stones of Drachenfels trachyte, reaching into the stone material only for several centimeters in depth. Here, it seldom reaches such a significant depth where the structural disintegration comprises crumbling and total fabric collapse, as seen at Cologne cathedral. The same can be ascertained for the intensity of surface soiling and weathering crusts. At Cologne cathedral, weathering crusts on building stones become apparent whether as dark brownish–black surfaces of a mostly already detached stone surface or as framboidal crusts with a disintegrated stone matrix underneath (Fig. 5). In Xanten weathering, crusts are more to be addressed as thin laminar crusts or soiling on a mostly structurally intact original building stone surface. The current condition of Altenberg cathedral reflects the extensive restoration and repair works carried out since the 1990s. Nevertheless, previous investigations only detected black weathering crusts to a minor degree (IBS 1990).

Microscopic observation reveals high porosity of the framboidal crusts (Fig. 7a) and very small (10–50 μm) and evenly spread acicular gypsum crystals (Fig. 7b). Cavities within the crystals cause the high porosity of the crust. The crust contains a significant amount of widely spread organic matter (black opaque particles). The crust formation on the Drachenfels trachyte is characterized by a thin, dense black layer on the surface of the host rock (10–20 μm) mainly consisting of organic material or carbonaceous particles (containing elemental and organic carbon) (Saiz-Jimenez 1993; Turpin and Huntzicker 1995), which may function as catalyst for the formation of gypsum (Amoroso and Fassina 1983; Rodriguez-Navarro and Sebastian 1996; Ausset et al. 1992). On this surface layer, a porous framboidal crust builds with finely distributed gypsum crystals and crystal aggregates together with quartz and feldspar particles (<0.1 mm) (Fig. 7d). The brownish tanning derives from iron-oxides/hydroxides (Do 2000). Framboidal black crusts on Drachenfels trachyte can be described as a porous mixture of gypsum, organic compounds, iron oxides, quartz, and feldspar particles. A large number of siliceous as well as carbonaceous fly ash particles are commonly embedded in such crusts and high ratios of Fe-rich particles are detected. In the limestone samples, the fly ash particles are firmly incorporated into the gypsum crust, surrounded by gypsum crystals and overgrown with sub-micron size gypsum crystals (Török et al. 2011). However, although fly ash particles are found embedded in the trachyte structure in the samples of Drachenfels trachyte, their incorporation into the crust is not as strongly wedded to the growth of gypsum crystals as is the case in the crust on limestone. SEM–EDX analyses of the crust surface reveal a high gypsum concentration for framboidal crusts (Fig. 8a–d), whereas on laminar crusts a composition mainly of silicate and organic components is detected (Fig. 8e–f). The thin opaque black layer on the surface of the host rock marks the boundary of the stone, on which the crust forms. In some places, this defined line is distorted, which may be attributed to the structural disintegration of the stone material underneath. Surface parallel cracks are often observed in the host rock beneath the crust. These cracks not only run along grain boundaries, but also characteristically cut through larger grains and minerals (Fig. 7d). This structural degradation of the host rock finally leads to the detachment of the crust including the upper superficial region of the host rock. The width of this detached zone is not limited to the crust but reaches into the host rock with a thickness of 3–10 mm. The host rock shows further structural disintegration in the form of multiple flaking and exfoliation (Fig. 4).

Laminar crusts are very thin (5–15 μm, where bulging occurs up to 50 μm) and have a dense composition of mainly organic compounds with parts of gypsum, iron oxides, quartz, and feldspar particles (Fig. 7c). As observed with framboidal crusts, the host rock beneath laminar crusts also shows surface parallel cracking with a lower quantity and latitude of the cracks. In context with laminar crusts, the structural disintegration of the stone is often manifested in the form of scales 0.5–1.5 cm thickness.

The results of the XRF analyses show a significant relative depletion of SiO₂ (2–15 %) and of Al₂O₃ (2–19 %) in the crust samples in respect to the host rock (Table 2). Also K₂O shows an average decrease of 10–22 % and Na₂O of 0.5–19 %. The mean enrichment of Fe₂O₃ for the samples from Altenberg is 12 %, for the Xanten samples 16 %, and for the Cologne samples 11 %. The enormous increase of P₂O₅ concentration is striking, which is three times higher for the Cologne samples. An enrichment of CaO within the crust is clearly noticeable for the samples from Cologne (factor 2.8) and from Xanten (factor 1.7). The average CaO-enrichment in the Altenberg samples is within the measurement accuracy. The increase in SO₃, respectively, the sulfur concentration, shows a similar tendency: the Cologne samples shows an enrichment of sulfur with factor 139, the Xanten samples with factor 65, and the Altenberg samples with factor 7. The depletion of the oxides associated with silicate phases (SiO₂, Al₂O₃, Na₂O, K₂O) correlates with the increase in SO₃.

In crusts on limestone, where the substrate consists almost entirely of CaCO₃, a depletion of CaO and an enrichment of SiO₂ as well as aluminum and iron in the crust are detected (Török et al. 2011). The contrary is found for crusts on silicate stones: a relative depletion of oxides associated with silicate phases and an enrichment of CaO in the crust along with an enormous increase in sulfur, indicating high gypsum enrichment in the crust. The average sulfur concentrations normalized to the host rock correlate to gypsum contents for Altenberg (0.6 wt. %), Xanten (5.9 wt. %) and Cologne (12.6 wt. %). The sulfur concentration found in crust samples on limestone—investigated by Török et al. (2011)—correlates to an average gypsum amount of 22 wt. % in respect to the host rock. Gypsum concentrations in samples from marble and limestone from different sites range from 10.3 (Torfs and van Grieken 1997) to 23.3 wt. % (Fassina 1988) (Table 3).

The data show a clear discrimination of framboidal and laminar crusts (Table 4). A stronger depletion for SiO₂, Al₂O₃, and Na₂O as well as for K₂O is detectable. The increase of CaO and SO₃ concentrations from laminar to framboidal crusts is significant.

The calculated content of gypsum in the analyzed samples of framboidal crusts from Cologne is about 60 wt. %, and in the samples of framboidal crusts from Xanten about 29 wt. %. However, the samples of laminar crusts show a quite similar low gypsum concentration for the three locations: Altenberg 0.55 wt. %, Xanten 0.04 wt. %, and Cologne 0.88 wt. %. In terms of the mineralogical determination by the Rietveld method, the gypsum concentration of the laminar crust samples are below the detection limit for gypsum by XRD, but the framboidal crust samples clearly indicate gypsum content (Table 5; Fig. 9).

The enrichment of CO₂ in the samples for Altenberg (factor 5.6), Cologne (factor 4.2) and Xanten (factor 2.5) (Table 2) indicates a certain amount of carbonaceous fly ash particles in the investigated material as well as a partition of organic material, especially for the investigated samples from Altenberg. This again illustrates the importance of settling dust and particles in terms of weathering crust formation.

Laser ablation inductively coupled plasma mass spectrometry analyses show clear trends in terms of major and trace element distribution in the black crusts and the host rock. The microscale chemical investigation also reveals little change in oxides associated with the silicate phases. An enrichment of sulfur, lead, antimony, bismuth, and arsenic in black weathering crusts developed on Drachenfels trachyte is detected (Fig. 10). The crusts of the Cologne cathedral show an average concentration of 1,849 ppm Pb, those of Xanten 1,944 ppm Pb, while the crusts of Altenberg only show 890 ppm Pb—which corresponds to enrichment factors of 105, 110, and 50 in respect to the host rock (Table 6; Fig. 10).

The samples from the three locations are clearly distinguishable. Comparing industrial, urban and rural samples, data show high concentrations of heavy metals (e.g., Pb and Bi), As and Si in black crust samples collected from the industrial and urban sites (Cologne and Xanten). The samples from the rural area (Altenberg) contain significantly lower concentrations of heavy metals (Table 6). This clearly indicates a very strong pollution impact for the Cologne and Xanten samples, since the content of these elements is due to an intense impact of combustion emissions (Fig. 11).

In view of decreasing SO₂ fluxes, the ban on leaded petrol, and emission regulations, the high concentrations of heavy metals as anthropogenic combustion tracers in the samples from Cologne result not only from recent air pollution but show the long history of industrial development at this location. The high lead concentrations in the Xanten samples can be traced back to the strong impact of the industrial area of Arnhem (NL) over a long period of time.

The weathering crusts show high enrichment of the detected elements, whereas within the first 100–300 μm of the surface of the host rock a general decrease is detectable (Fig. 10). An accumulation can be observed in cracks and small cavities of the host rock (Fig. 10). This correlates well with the phenomenon observed by Török et al. (2011), that in the porous zone below the crust, an increase of Pb is observed, which marks the accumulation of lead not only within the crust but in the pores, too. The enrichment in the cracks demonstrates, that pollutants are not only superficially fixed to the crust, but also penetrate deeper into the stone and accumulate in structurally altered zones.

The plots of the S/Si, Ca/Si, and Pb/Si ratios show a sharp increase in the crust, indicating a high enrichment of Pb and S as well as of Ca in the crust (Fig. 10). The major atmospheric impact is indicated by the enormous enrichment of heavy metals, which, as anthropogenic combustion tracers, clearly refer to an atmospheric source. This is reaffirmed by the content of siliceous and carbonaceous fly ash particles in the crust.

If Pb concentrations are compared on an international scale, similar concentrations are found for Cologne, Xanten, and Halle (Table 7). Values from these sites as well as from Budapest, Milan, Rome, and Brussels can be related to higher combustion emission impacts for example from domestic heating, power plants, coal combustion, and vehicle exhausts (Sabbioni 2003).

Microscopic analyses reveal particulate matter is abundant on the stones’ surfaces and in the weathering crusts (Lefèvre and Ausset 2002). The detection of the PAH compounds reveals the organic fingerprint of the particulates of the settling dust. PAH as compounds indicative of air pollution are detected in selected samples of black weathering crusts, scales, and disaggregated material as well as from dust. PAH mainly arise from the incomplete combustion of fossil fuels, organic matter, or wood. The detected concentrations are referred to the values of the host rock. The range of the detected 21 PAH compounds are from acenaphthylene (m/z 152) to dibenzo(ah)-pyrene (m/z 302).

The total concentrations of the PAH compounds varies from 62 (laminar crust from Xanten cathedral) to 13,525 μg/kg OS (dust sample from Cologne cathedral) (Table 8). The high-end values belong to samples from Cologne cathedral and are typical for samples from an industrial area. The PAH concentrations of the samples from Xanten and Altenberg are in a close range (Table 8). A clear relationship can be drawn between PAH concentration and sample type. If the Cologne samples are compared to each other, samples of dust, framboidal weathering crusts, and strongly deteriorated stone material (scale and disaggregated stone) generally show higher PAH concentrations (1,362–13,525 μg/kg OS); samples of laminar crusts and slightly weathered stone show lower values (247–482 μg/kg OS).

When data from Török et al. (2011) are included, the higher PAH concentrations are in dust samples and from framboidal crusts from Cologne, Budapest, and Halle (Fig. 12; red ellipse). This group of samples also shows high SO₄ ²⁻ concentrations. A second group of low PAH but higher SO₄ ²⁻concentrations (Fig. 12; black ellipse) contains mainly laminar crusts from rural areas in Hungary as well as from Xanten and Altenberg besides four samples from Hungarian host rock. Except for one sample from Xanten (Drachenfels trachyte), these crust and host rock samples derive from limestone. A third group is formed by deterioration samples from Cologne cathedral on Drachenfels trachyte (Fig. 12; green ellipse). These samples show certain enrichment in SO₄ ²⁻, but a stronger increase of PAH. A fourth group with low SO₄ ²⁻ and moderate PAH concentrations (Fig. 12; blue ellipse) contains a host rock sample of the Drachenfels trachyte, and further consists of samples of light weathered stone and of laminar crusts from Cologne, Altenberg and, Naumburg as well as rural areas in Hungary.