ORIGINAL PAPER

Non-destructive mapping of dampness and salts in degradedwall paintings in hypogeous buildings: the case of St. Clementat mass fresco in St. Clement Basilica, Rome

Valeria Di Tullio & Noemi Proietti & Marco Gobbino &

Donatella Capitani & Roberto Olmi & Saverio Priori &Cristiano Riminesi & Elisabetta Giani

Received: 26 October 2009 /Revised: 9 December 2009 /Accepted: 11 December 2009# Springer-Verlag 2010

Abstract As is well known, the deterioration of wallpaintings due to the capillary rise of water through thewalls is a very widespread problem. In this paper, a study ofmicroclimate monitoring, unilateral nuclear magnetic reso-nance (NMR), and evanescent-field dielectrometry (EFD)was applied to map non-destructively, in situ, and in aquantitative way the distribution of the moisture in anancient deteriorated wall painting of the eleventh century.Both unilateral NMR and EFD are quite new, fullyportable, and non-destructive techniques, and their combi-nation is absolutely new. The approach reported here isproposed as a new analytical protocol to afford the problemof mapping, non-destructively, the moisture in a deteriorat-ed wall painting in a hypogeous building such as that of thesecond level of St. Clement Basilica, Rome (Italy), wherethe use of IR thermography is impaired due to theenvironmental conditions, and the gravimetric tests areforbidden due to the preciousness of the artifact. Themoisture distribution was mapped at different depths, from

the very first layers of the painted film to a depth of 2 cm. Ithas also been shown how the map obtained in the firstlayers of the artwork is affected by the environmentalconditions typical of a hypogeous building, whereas themaps obtained at higher depths are representative of themoisture due to the capillary rise of water from the ground.The quantitative analysis of the moisture was performed bycalibrating NMR and EFD signals with purposely preparedspecimens. This study may be applied before and afterperforming any intervention aimed at restoring and improv-ing the state of conservation of this type of artwork andreducing the dampness or extracting salts (driven by thevariation of moisture content) and monitoring the effective-ness of the performed interventions during the time. Thisprotocol is applicable to any type of porous material.

Keywords Unilateral NMR . Evanescent-fielddielectrometry . Rising damp .Wall paintings .

Porous materials

Introduction

As is well known [1–3], the moisture diffusion is a majorcause of decay to masonry materials such as stone, brick,mortar, and plaster. An accurate diagnosis of the causes andextent of the moisture is a fundamental step in theconservation of wall paintings; in fact, the porous mortarbacking may provide an easy route for the flow of dilutedsalt solutions. The volumetric expansion associated with thecrystallization of these salts affects the plaster-pigmentadhesion and leads to the disintegration of the surface.Besides, salts which crystallize on the wall surface causethe whitening of the painted layer (a phenomenon known as

V. Di Tullio :N. Proietti (*) :M. Gobbino :D. CapitaniLaboratorio di Risonanza Magnetica “Annalaura Segre”,Istituto di Metodologie Chimiche, CNR,Area della Ricerca di Roma, Via Salaria Km. 29,300,00016 Monterotondo Scalo,Rome, Italye-mail: noemi.proietti@imc.cnr.it

R. Olmi : S. Priori : C. RiminesiIstituto di Fisica Applicata Nello Carrara, CNR,Via Madonna del Piano 10,50019 Sesto Fiorentino, Florence, Italy

E. GianiIstituto Superiore per la Conservazione ed il Restauro,Piazza San Francesco di Paola,900184 Rome, Italy

Anal Bioanal ChemDOI 10.1007/s00216-009-3400-x

efflorescence) and, in the worse cases, fretting andcrumbling of the wall. Deterioration of wall paintingsresults from the open, porous nature of their support (walland mortar) and their interaction with the surroundingmicroclimates [4]. As a rule, before performing anyrestoration or conservation treatment, the moisture distri-bution in a wall painting should be known. Nevertheless,the knowledge of the moisture distribution within a wallpainting is a difficult task.

In a recent paper [5], we reported a study of microcli-mate monitoring, IR thermography (IRT), gravimetric tests,and portable unilateral nuclear magnetic resonance (NMR),aimed at mapping the moisture distribution in a verydeteriorated wall painting before performing any emergen-cy intervention. In the framework of this study, IRT andgravimetric methods allowed the validation of portableunilateral NMR as a new analytical and fully non-destructive tool for mapping in situ the distribution andthe amount of moisture in wall paintings as well as in anyporous material. Nevertheless many cases also exist, suchas that reported in the present paper, where this studycannot be successfully applied because the environmentalconditions impair the use of IRT [6, 7] and the drilling ofsolid cores for gravimetric tests is highly forbidden due to thepreciousness of the ancient wall paintings to be investigated.Under these circumstances, an alternative study must beapplied to map in situ and in a fully non-destructive way themoisture distribution in deteriorated wall paintings. At thisaim, a fully new study consisting of microclimate monitoring,unilateral NMR, and evanescent-field dielectrometry (EFD)techniques was carried out to obtain useful, quantitativeinformation on the moisture distribution on the very preciousand badly deteriorated St. Clement at mass and the legend ofSisinnius wall painting located in the second level of St.Clement Basilica, Rome (Italy).

It has been previously shown that unilateral NMR is asuitable non-destructive tool for assessing the state ofconservation of wall paintings [8, 9] and for quantitativelymapping the moisture distribution [5, 10]. As previouslyreported [1, 10], the intensity of the signal detected byunilateral NMR is directly proportional to the watercontent. Therefore performing a suitable number ofmeasurements and processing the experimental data in aproper way, it is possible to obtain a precise map of themoisture content of the wall painting under investigation.

EFD is a diagnostic technique based on a specializationof dielectric spectroscopy at microwave frequencies [11,12]. The EFD measuring instrument, developed andpatented by IFAC of the National Research Council [13],is a portable resonant microwave device for mapping in anon-destructive way the moisture content and salinity onwalls [14–16] up to a depth of 2–3 cm with a real timeelaboration. The high dielectric contrast between a dry wall

and water makes it possible to detect also small amounts ofmoisture by investigating the dielectric properties of thewall. The presence of soluble salts is detected andquantified by looking at the resonance line width, or thequality factor Q of the resonant probe. EFD has been used,up to now, in in situ monitoring and diagnostic campaigns,in particular for (1) determining the map of sub-superficialmoisture content, (2) detecting and mapping the presence ofsoluble salts, and (3) evaluating the effect of consolidationand salt extraction techniques, e.g., poultice applications.EFD has been also proved to be useful as a support toconservation interventions, as it allows to investigate thedynamics of water absorption, diffusion, and evaporation ina wall.

In this paper, the new study consisting of microclimatemonitoring, unilateral NMR, and EFD is reported for thefirst time. The purpose of this paper is to show how thisstudy is used as an analytical tool to monitor in a fully non-destructive way and in situ the distribution of moisture atdifferent depths in a seriously damaged wall painting. Aquantitative analysis of the moisture content has beenobtained, performing a proper calibration of both NMR andEFD signal with suitable specimens.

This study is particularly useful when the environmentalconditions impair the use of IRT and the preciousness of theartifact does not allow gravimetric tests to be performed.

Experimental

The wall painting

In Fig. 1a, the photo of St. Clement at mass and the legendof Sisinnius wall painting is shown. According to thecriteria and interests of the preservation officers in chargeof conservation of the fresco, two regions were chosenwhere performing NMR and EFD measurements. Inparticular, EFD measurements were performed over aslightly larger area for mapping both the moisture and saltscontent. Whereas NMR measurements were performed on aslightly more restricted area because a very detailedinformation on the moisture content both in very firstlayers of the wall painting and at a depth of 0.5 cm wasrequired by restorers.

The solid framework (see Fig. 1a) evidences the area onwhich both NMR and EFD measurements were carried out,whereas the dashed framework evidences the area on whichonly EFD measurements were carried out. In particular, thematrix of points where NMR and EFD analysis wereperformed is shown in Fig. 1b. Triangles represent thepoints measured by both NMR and EFD techniques,rectangles represent the points where only NMR measure-ments were performed, and circles represent the points

V. Di Tullio et al.

where only EFD measurements were performed. It is worthto note that the region where both NMR and EFDmeasurements were performed was the most degraded areaof the wall painting due to the presence of moisture.

Unilateral NMR

The campaign of measurements was performed in situ witha unilateral NMR instrument from Bruker Biospin, which isa variant of NMR-MOUSE [17–24].

Measurements were performed on the very first layers ofthe surface of the wall painting and at a depth of about0.5 cm in the plaster. In particular two probeheads wereused. In the former case, a 1-mm probehead, operating at18 MHz, was used, allowing the measurement to beperformed on a 0-0.1 cm slice of sample. In the latter casea 5-mm probehead, operating at 16 MHz, was used,allowing the measurement to be performed on a 0.45–0.55 cm slice of sample fully disregarding the signal of thesurface.

Since in an inhomogeneous magnetic field the NMRsignal decays very quickly, the NMR signal must berecovered stroboscopically [18]. Therefore the integral ofthe NMR signal is actually the integral of the signalobtained applying a Hahn echo pulse sequence [25, 26].

In the case of the 1-mm probehead, the maximum echosignal corresponding to the π/2 pulse was obtained with apulse width of 3 μs, whereas in the case of the 5-mmprobehead the pulse width corresponding to the π/2 pulsewas 10.4 μs. In both cases the dead time was 15 μs. In theHahn echo pulse sequence the echo time τ was set at 12 μswhen measuring with the 1-mm probehead, whereas τ wasset at 20 μs when measuring with the 5-mm probehead. Inboth cases the τ value was set as short as possible.

Contour plot of the moisture distribution in the wallpainting by unilateral NMR

In Fig. 1a the solid framework evidences the area on whichNMR measurements were carried out (about 200×100 cm2). In particular Hahn echo measurements wereperformed choosing a matrix of ‘‘points’’ on the framedarea (see Fig. 1b). Actually, each experimental ‘‘point’’covered an area of 2×5 cm2, that corresponded with thearea of the probehead. This area is the one where theflux lines generated by the radio frequency coil areperpendicular to the flux lines generated by the staticmagnetic field.

A matrix of 50 points (see rectangles and triangles inFig. 1b) was measured. Afterward experimental data wereprocessed to obtain a contour plot applying an algorithmfor smoothing sharp variations of the dependent variablevalues within the 3D data set. A fraction of 40% of thetotal number of data points was used to compute eachsmoothed value. A Gaussian weight function and aquadratic fit were used to weight the data. A non-linearfunction (4th degree polynomial) was then applied to theweighted data for computing each smoothed value [27,28]. As previously reported, a contour plot is a graphicalway of obtaining a 2D representation of a 3D surface,where x and y were the coordinates of the measured areaof the wall painting, and z was the integral of the NMRecho signal.

Fig. 1 a St. Clement at mass and the legend of Sisinnius wallpainting. The solid framework indicates the area investigated byunilateral NMR and EFD, and the dashed framework indicates thearea investigated only by EFD. b Detail of the area of the investigatedwall painting; in the figure, the measured matrix of “points” is alsoshown: triangles represent the points measured by both NMR andEFD, rectangles represent the points where only NMR measurementswere performed, and circles represent the points where only EFDmeasurements were performed

Non-destructive mapping of dampness and salts

Analytical procedure to establish the relationbetween the integral of the NMR echo signaland the gravimetric moisture content

To determine the exact level of moisture in the wall painting, acalibration with suitable specimens was performed. Fourspecimens, namely S1, S2, S3, and S4, were purposelyprepared by the restorers. The size of the specimens was 5.5×5.5×3 cm. These specimens were made of bricks with a5-mm layer of mortar on the top. The bricks constituting thespecimens were of the same type as those constitutingthe backing of the investigated wall painting. The mortar onthe top of the specimens, a mixture of lime and pozzolana,was prepared according the ancient original recipe toreproduce the same type of mortar used in the wall paintingunder investigation. The so-prepared specimens were used tocalibrate the integral of the NMR echo signal.

The NMRmeasurements were performed on the side of themortar with the 1-mm probehead as well as with the 5-mmprobehead. Before measuring, the specimens were dried untilreaching a constant weight Psd (see Table 1, column 2)according to the NORMAL protocol [29] and then measuredby unilateral NMR. A very low value of the integral of theNMR echo signal Asd was recorded for all specimens due to avery low amount of residual moisture in the porous structure(see Table 1, columns 3 and 4). Then the specimens werefully saturated with water by total immersion according to theNORMAL protocol, up to constant weight Psw (see Table 1,column 5). In this case, a very high value of the integral ofthe NMR echo signal Asw was recorded (see Table 1,columns 6 and 7). Note that in Table 1, Asd values measuredon dry specimens with the 1-mm probehead and with the 5-mm probehead were respectively reported as As1 mm

d andAs5 mm

d . Analogously Asw values measured on water-saturatedspecimens with the 1-mm probehead and with the 5-mmprobehead were respectively reported as As1 mm

w and As5 mmw .

The average values of the weights of the dry (Psd) and ofthe water-saturated (Psw) specimens as well as the averagevalues of the integrals of the NMR echo signal of the dry(As1 mm

d ,As5 mmd ) and of the water-saturated (As1 mm

w ,As5 mmw )

specimens were calculated and reported in Table 1 alongwith the standard deviations.

The average values Psw and Psd were used to calculatethe imbibition coefficient ic defined as the maximumamount of water adsorbed by the specimen comparedto its dry weight and expressed as a percentage:

ic ¼ Psw�PsdPsd

� 100. The imbibition coefficient of the speci-

mens worked out to be 25%. Accordingly, a scale rangingfrom 0% to 25% was used to calibrate the level of moisturecontent in the wall painting.

It must be noted that the measurements on the specimenswere performed using the same experimental conditions

used in St. Clement Basilica for measuring the wallpainting. In this way, a correlation between the integralof the NMR echo signal and the weight of the adsorbedwater was established and summarized in Eq. 1. In thisequation, “p” is a label used to distinguish between datacollected with the 1-mm probehead and data collected withthe 5-mm probehead, namely p=1-mm probehead or 5-mmprobehead. Applying this equation, the values Ap

i of theintegral of the NMR echo signal reported in the contourplot were calibrated and related to the moisture contentvalues:

MCpi NMRð Þ ¼ Ap

i � Apmin

� � ic

Aspw � Aspd

!

ð1Þ

where Api is the integral of the NMR echo signal measured

on the wall painting with unilateral NMR, Apmin is the

lowest value of the integral measured on the wall painting,Aspw is the average value of the integrals of the NMR echosignal measured on the four water-saturated specimens,and Aspd is the average value of the integrals of the NMRecho signal measured on the four dried specimens (seeTable 1).

According to the calibration procedure, each area of thecontour plots reported in Fig. 4a, b corresponded to aprecise level of moisture. In particular, the maximum valueof the integral of the NMR echo signal measured on thewall painting using the 1-mm probehead corresponded toapproximately 13% of the moisture content, whereas themaximum value of the integral of the NMR echo signalmeasured on the wall painting using the 5-mm probeheadcorresponded to about 8% of the moisture content.

The maximum error on MCpi (NMR) values was

evaluated using the error propagation theory, and it wasfound to be always less than 10% of the nominal value.

Evanescent-field dielectrometry

EFD measures the parameters of interest, moisture, and saltconcentration by estimating the dielectric properties of awall that is viewed as a “binary” dielectric mixtureconsisting of bulk material (mortar, plaster, brick, and soon) and water. The high contrast between the dielectricconstant of a dry wall (ε′<4) and water (ε′≅80) allows thedetection of water inside a wall [12]. The real part ofpermittivity (ε′) depends on the moisture content (MC),whereas it is nearly independent on the saline content. MCis evaluated through the measurement of a frequency shift.The presence of soluble salts is detected and quantified bylooking at the resonance line width, or the quality factor Qof the resonant probe. A function called “salinity index”(SI), inversely related to Q (i.e., to the loss tangent of thematerial under investigation), has been devised. Such a

V. Di Tullio et al.

function strongly depends on the electric conductivity ofthe material, and it is weakly dependent on ε′, and as such,it allows the presence of electrolytes to be estimated.

The EFD instrument works at microwave frequencies(1–1.5 GHz), when equipped with the standard measuringhead. The probehead has been studied to allow a reliablecontact with the surface of the investigated material [30,31]. Lower frequencies can be obtained by increasing thelength of the resonant line, while the spatial and depthresolutions essentially depend on the probe size. Details ofthe construction and the operating principle are given in[11]. The measured parameters, MC and SI, refer to aweighted-average over a depth of about 2 cm, the weightdecreasing with increasing the depth. The probed region isapproximately semispherical, with a radius of 2 cm.

The range of measurable MC is about 0–20%. The SI isusually in the range 1–10, the lower limit meaning“negligible salts.” To exemplify, an SI of 1 is related toan electrical conductivity of about 0.4 S m−1, that of tapwater, while an SI of 9 means 10 S m−1, the conductivity ofa 2 mol/l Ca(NO3)2 water solution.

The method is quite fast, and the measurement of asingle point requires a few seconds, about ten includingdata acquisition and processing.

Contour plot of the moisture distribution in the wallpainting by EFD

In Fig. 1a, the solid framework evidences the area on whichNMR and EFD measurements were carried out, whereas thedashed framework evidences the area on which only EFDmeasurements were carried out. In particular, in Fig. 1b,triangles represent the points where both EFD and NMRmeasurements were carried out, whereas circles representthe points where only EFD measurements were carried out.

The effective measurement area of the EFD probe is12 cm2; thus, to cover the region of interest (about 200×150 cm2), 36 measurements were carried out on a grid of

six rows by six columns. The first row was at a distance of50 cm from the floor, and the row and column spacing was30 and 40 cm, respectively.

A smoothing procedure similar to that applied to theNMR data allowed the obtainment of a filled contour plots(also known as level plots [32]) for both MC and SI. Thelevel plots were obtained by means of the graphics facilitiesof the R environment [33], which is a programminglanguage and a software package for statistical computingand graphics.

Analytical procedure to establish the relationbetween the EFD signal and the gravimetric moisturecontent

The EFD signal consists of a frequency shift Δf and of aresonance line-width Lw. These two primary quantities arerespectively related to the average dielectric constant of thematerial under measurement ε′ and to the dielectric lossesε″ (also expressible as an “equivalent” electrical conduc-tivity). The “secondary” parameters ε′ and ε″ are physicallyrelated to the water volume fraction in the mixtureconstituting the wall material [12] and to the salt concen-tration [11], i.e., to the “diagnostic” parameters.

An extensive theoretical and laboratory work has beenconducted to establish a quantitative relationship between“primary” and diagnostic parameters. The physical basis isreported in [12], where that relationship has been investi-gated by studying the dynamics of water mass transfer inporous media and determining its effect on the dielectricproperties. In the referenced paper, a coupled mass transfer/dielectric problem was introduced and solved numerically,based on a finite elements model, and the results werecompared to experimental dielectric measurements per-formed on plaster samples by the open coaxial method.

In particular, the relation between the percentage ofmoisture content and Δf has been demonstrated to bealmost linear in the range of MC of interest (about 0–20%),

Table 1 Gravimetric and NMR measurements carried out on four specimens

Specimen Psd gð Þ As1 mmd a:u:ð Þ As5 mm

d a:u:ð Þ Psw gð Þ As1 mmw a:u:ð Þ As5 mm

w a:u:ð Þ

S1 138.4539 8.3 10.1 172.5512 74.1 50.3

S2 142.8742 9.9 10.2 177.5539 73.1 49.7

S3 144.9520 7.2 10.0 179.8310 73.8 49.2

S4 138.8769 8.7 9.9 174.5584 74.2 50.9

Averagevalues

Psd ¼ 141� 3 As1 mmd ¼ 8:5� 0:9 As5 mm

d ¼ 10:0� 0:2 Psw ¼ 176� 3 As1 mmw ¼ 73:8� 0:4 As5 smm

w ¼ 50� 1

Psd and Psw are the weights (in grams) of the dry specimens and of the water-saturated specimens respectively, and Psdand Psw are the averagevalues. As1 mm

d and As5 mmd are the integrals of the NMR echo signal reported in arbitrary units (a.u.) measured on the dry specimens with the 1-

and 5-mm probehead, respectively, and As1 mmd and As5 mm

d are the corresponding average values. As1 mmw and As5 mm

w are the integrals of the NMRecho signal measured on the water-saturated specimens with the 1- and 5-mm probehead, respectively, and As1 mm

w and As5 mmw are the

corresponding average values

Non-destructive mapping of dampness and salts

and the ionic conductivity has been shown to have anegligible influence on it. Here the MC is defined on a wetbasis, in terms of the weights Pw and Pd corresponding tothe “wet” and “dry” specimens, respectively:

MCw ¼ 100� Pw � Pdð Þ=Pd

The relation between MCw (wet) and MCd (dry) isstraightforward:

MCd ¼ MCw

1� MCw100

and allows to easily convert wet-based data (obtained withEFD) into dry-based ones.

The calibration procedure is analogous to that describedfor unilateral NMR, using specimens saturated with waterand in dry condition as well. The EFD instrument has beencalibrated on various materials (e.g., mortar, bricks, andMaastricht stone), as the dielectric response slightlydepends on the material density.

The salinity index is almost independent on the moisturecontent, as long as MC is higher than 3 (soluble saltsrequire water to go in solution). For lower values, SIlinearly increases from 1 (for dry samples) to a plateauvalue.

Result and discussion

The study case

St. Clement Basilica in Rome is built on three differentlevels. The middle level is located at a depth of about 12 mon respect the road level. This level contains severalprecious ancient frescoes, including that of St. Clement atmass and the legend of Sisinnius of the eleventh century.This wall painting is rapidly deteriorating due to thecapillary raise of water from the ground below. Thereduction of the rising damp is a very necessary interven-tion prior performing any restoration treatment.

The wall paintings of St. Clement Basilica show thetypical degradation processes induced by rising damp: saltdeposition, biological patina, and whitening of the pictorialsurface. In particular, on the surface of St. Clement at massand the legend of Sisinnius wall painting, wide areas of saltefflorescence are present and labeled as zones [A1, A2] inthe sketch reported in Fig. 2. In the central low part of thewall painting (zone [B], Fig. 2), a biological patina,possibly due to biological microorganisms, is also present.Thick encrustation is also present and labeled as zone [C](Fig. 2).

The wall painting, which had already undergone tradi-tional cleaning in the 1990s, still shows degradation, and

consequently, the degradation process is not resolvable byrepeating the same cleaning treatment. The problem of therising damp in this church had already been studied for over10 years since the first research on environmental con-ditions of the church began. A long-term monitoringcampaign in the church had been carried out, includingmicroclimatic behavior, drillings on non-painted walls, andsoluble salt analyses [34, 35]. The results concerning themicroclimatic monitoring in the period November 2006–July 2007 showed that close to the masonry, the relativehumidity was always over 90%, and the temperature rangedbetween 12 and 22 °C. The microclimatic behavior of theunderground level of the Basilica is typical of hypogeousbuildings, with very high levels of relative humiditythroughout the year. In addition, the presence of awatercourse flowing under the foundation of the Basilicais one of the causes of the rising damp in the walls. In asuch environment, IRT measurements are not reliable. Infact, the ambient is rather dark, no air heating, high andconstant RH, and the increase of the temperature before thescanning is not sufficient to dry the air near the surface ofthe wall painting to increase the evaporative flux and,therefore, to decrease the temperature of damp zones. Inthese environmental conditions, the evaporative flux isinhibited, and the IRT images are not representative of theeffective humidity conditions of the wall [6, 7].

The monitoring campaigns of the environment of thesecond level of the Basilica performed over 10 years clearlypointed out the importance of knowing the distribution ofthe moisture content and the level of moisture prior to planany suitable and lasting restoration project of the wallpaintings. At this aim, a campaign of measurements byportable unilateral NMR and EFD was carried out.

In Fig. 3, the microclimatic monitoring of the periodJanuary 1–15, 2007 is reported. This is the period duringwhich NMR and EFD measurements were performed. It isimportant to note that during this period, the temperaturewas constant, between 13 and 14 °C, and the relativehumidity was always higher than 95%. Therefore, themicroclimatic condition was rather constant and did notaffect the measurements.

Unilateral NMR investigation

Unilateral NMR technique was used to monitor directly insitu and in a non-destructive way the distribution of themoisture.

As previously reported [1, 5], the integral of the NMRecho signal is proportional to the water content within thewall painting. NMR data collected with 1- and 5-mmprobeheads were processed to obtain contour plots consti-tuting a two-dimensional representation of a three-dimensional surface graph, where x and y are the

V. Di Tullio et al.

coordinates of the strip of the painted surface measured andz is the integral of the NMR echo. In this way, the contourplot was used to map the distribution of the moisture in thewall painting.

The moisture distribution maps obtained using the 1- andthe 5-mm probeheads are shown in Fig. 4a, b, respectively.In these maps, the difference in the moisture level wasshown as a gradient color, red indicates the lowest watercontent, while dark blue indicates the highest water content.

At a preliminary screening, the two maps showed a verydifferent distribution of the moisture. In the map obtainedwith the 1-mm probehead (a depth of about 0.1 cm), thedampness is more widespread both in the lower and in thehigher parts of the area investigated by NMR (see Fig. 4a).Performing the calibration procedure, it was possible to

obtain the precise amount of the moisture content. Accord-ing to this procedure, the maximum value of the integral ofthe NMR echo corresponded to 13% of the moisturecontent. Taking into account that the fully saturated waterspecimens had a moisture content of 25%, these resultsclearly showed how high the level of the dampness was inthe area of the wall painting under investigation.

In the map obtained with the 5-mm probehead (a depthof about 0.5 cm in the plaster), the distribution of themoisture gave a clear image of the front of the rising damp,from the bottom to the middle of the area investigated byNMR (see Fig. 4b). According to the calibration procedure,in this map, the maximum value of the integral of the NMRecho corresponded to about 8% of the moisture content,which was still a rather high amount.

Fig. 3 Microclimatic monitor-ing of the temperature and ofRH (relative humidity) in thearea very close to St. Clement atmass and the legend of Sisinniuswall painting during the periodJanuary 1–15, 2007

Fig. 2 Sketch of the degradedarea on the surface of the in-vestigated wall painting. LabelsA, B, and C indicate the zonesshowing a very evident damage,respectively, due to salts efflo-rescence [A1, A2], biologicalpatina [B], and encrustation [C]

Non-destructive mapping of dampness and salts

Apparently it seems that the map obtained with the 1-mmprobehead does not give significant information about thepath of the rising damp. However to understand this map,the state of conservation of the wall painting as well as theenvironmental conditions must be taken into account. In fact,the measurements at 0.1 cm of depth were performed on aslice very close to the artwork–environment interface. Forexample, in zone [A1] (see Fig. 2), the moisture contentmeasured with the 1-mm probehead was about 10–13%,

whereas the same zone measured with the 5-mm probeheadshowed moisture values of about 7–8%, which is still arather high amount. In zone [A2], the moisture content wasfound to be about 10–13% with the 1-mm probehead andabout 1–2% with the 5-mm probehead. In this case, thedifference between results obtained with the 1- and 5-mmprobehead was clearly due to the presence of superficialefflorescence of highly hygroscopic salts, which weresensitive to the high environmental relative humidity of

Fig. 4 a Map of the distribution of the moisture obtained using the 1-mm probehead. b Map of the distribution of the moisture obtainedusing the 5-mm probehead. A gradient of color is used forrepresenting the water content: the red color is associated with a very

low water content, while the blue color is associated with an highwater content. On the right side of the figures, the scale of values ofthe moisture content in percentage obtained by unilateral NMR isreported

V. Di Tullio et al.

about 90–95%. The same interpretation explains the resultsobtained with both probeheads in zone [B] affected bybiological patina, namely 10–13% with the 1-mm probeheadand 3–4% with the 5-mm probehead (see Fig. 2).

On the contrary in zone [C], the moisture contentmeasured with the 5-mm probehead was about 6–8%,whereas with the 1-mm probehead, it was found to be about2–4%. This result may be due to several factors, such as thepresence of detachment of the superficial pictorial layerfrom the mortar, or the presence of layers of hydro-repellenttreatments used in previous restoration works, or thepresence of a thick carbonate crust with low hygroscopicity.The latter explanation agrees well with previous studies[34, 35] reporting the occurrence of re-carbonation phe-nomena on the surface of the wall painting.

According to these results, information about the routeof the rising dump was obtained by performing the analysisusing a 5-mm probehead measuring at about 0.5 cm in theplaster, fully disregarding the signal of the surface of thewall painting. Whereas information on the state ofconservation of the wall painting (salts deposition, biolog-ical degradation, and encrustation) was obtained byperforming measurements with a 1-mm probehead, whichdetects the signal of the very first layers of the wall painting.The maps of moisture obtained using both probeheads gavethe detailed and complementary information to carry out afull and non-destructive diagnostic study of the moisturecontent and its effect on the wall painting necessary beforeplanning and performing any restoration work.

In Fig. 5, the values of the moisture contents measured at0.1 and 0.5 cm of depth are compared and reported ashistograms. Each plot reports the results obtained measur-ing the points of the matrix located at the same height fromthe floor. In this figure, the variation of the moisture contentvs. the wall depth is shown. It is worth to note thatincreasing the height from the floor, the values of themoisture content measured in the slice 0.45–0.55 cm aredefinitely lower than the corresponding ones measured inthe surface, slice 0–0.1 cm. These results point out that themeasurements in the slice 0.45–0.55 clearly evidence thecapillary rise of the moisture from the ground. Whereasthe results obtained at a depth between 0 and 0.1 cm are notcorrelated at all with the capillary rise of water, in fact, highmoisture content are also found at high heights from thefloor. It must be also noted that in many points, such as(3,1), (3,2) (C,c), (4,1), (4,5), and so on (see Fig. 5), themoisture content at the surface is definitely higher than that

�Fig. 5 The moisture contents at 0.1 and 0.5 cm of depth are comparedand reported as histograms. Each plot reports the results obtainedmeasuring the points of the matrix located at the same height (h) fromthe floor: a h=50 cm, b h=70 cm, c h=90 cm, d h=110 cm, e h=130 cm, and f h=150 cm

Non-destructive mapping of dampness and salts

measured inside evidencing a moisture gradient from inside tooutside. According to the sketch reported in Fig 2, the presenceof such a gradient might be due either to the presencehygroscopic salts and biological patina or to the presence ofcondensation phenomena, which might give rise to themoisture gradient. On the contrary in some points, such as(A,a), (A,b), (B,b), and (2,2), the moisture content at the surfaceis lower than that measured inside, evidencing a moisturegradient from outside to inside. According to the sketchreported in Fig. 2, the presence of such a gradient might bedue either to the presence of encrustation or to the occurrenceof detachment of the pictorial film from the mortar.

EDF investigation

The measurement was performed, keeping the probehead incontact with the surface for the time necessary for dataacquisition (less than 10 s).

EFD measurements showed that the wall exhibited agradient of sub-surface moisture content at a depth of about2 cm, which decreased from the bottom to the top of theinvestigated region (see Fig. 6a). The content of solublesalts, measured in terms of the salinity index SI, was higherin the first meter from the floor and showed an almostrandom behavior in the rest of the wall (see Fig. 6b). Themoisture distribution measured on the wall painting isreported in Fig. 6a. The moisture content ranged between3.5 (red) and 8% (blue), in a good agreement with themoisture content obtained at a depth of 0.5 cm byunilateral NMR (see Fig. 4b). It is worth to note that in thearea investigated by both NMR (Fig. 6c, left) and EFD(Fig. 6c, right), the moisture distribution maps obtainedwith both techniques are also in a good agreement. Notethat the upper part of the map obtained by EFD refers to anarea of the wall painting, which has not been investigatedby NMR (see Fig. 6c and Fig. 1a).

Fig. 6 Map of the distribution of moisture content (a) and of thesalinity index (b) obtained by the EFD instrument. A gradient of coloris used for representing the water content: The red color is associatedwith a very low water content, while the blue color is associated with

a high water content. c Comparison between the moisture distributionmap obtained by unilateral NMR at about 0.5 cm of depth (left side)and that obtained by EFD at about 2 cm of depth (right side)

V. Di Tullio et al.

Figure 6b shows the distribution of SI, ranging betweena minimum of 1.5 (negligible salts) in the upper part of theinvestigated region and a maximum of 5.5 (moderately highsalt concentration) in the lower part.

Conclusion

A non-destructive study consisting of microclimate moni-toring, unilateral NMR, and EFD was performed formonitoring non-destructively and in a quantitative way thedistribution of the moisture and salts in an ancient,deteriorated wall painting located in a hypogeous building,where IR thermography gives rather unreliable results.Besides, due to the preciousness of the investigated wallpainting, the drilling of solid cores for gravimetric tests washighly forbidden. Both unilateral NMR and EFD are quitenew portable, non-destructive technologies, and theircombination is absolutely new.

Because unilateral NMR detects the proton signal, thistechnique is very sensitive to the presence of moisture inwall paintings as well as in any porous material. Usingsuitable probeheads, the distribution of the moisture in thevery first layers of the painted surface as well as at a depthof about 0.5 cm in the plaster was obtained.

EFD is a very fast technique, which exploits the highdielectric contrast between a dry wall and the bulk water todetect even small amounts of moisture in the wall. Themeasurements are highly reproducible and reliable and canbe performed on any type of porous material. However, thedielectric response is not fully independent on the densityof the investigated material so that a calibration proceduremust be performed on not previously investigated materialsto obtain the suitable instrument constants. EFD techniqueallowed the obtainment of the moisture distribution at adepth of about 2 cm in the investigated wall painting andthe salinity index as well.

Collected data were processed to obtain contour plots inwhich differences in moisture content were represented ascolor gradients, allowing an easy visualization of thedistribution of the moisture and salts in the wall painting.Performing a purposely made procedure to calibrate theNMR and the EFD signal, a quantitative analysis of themoisture content was obtained. At this aim, purposelyprepared specimens were used in which the mortar wasprepared according to the ancient recipe used in theinvestigated wall painting.

This study also allowed the detection of the severalcauses of decay due to the moisture occurring in St.Clement at mass and the legend of Sisinnius wall painting.In fact, NMR measurements performed in the first layers ofthe wall painting showed the destructive action of themoisture mostly due to the interaction between the artwork

and the environment. In fact, the surface showed microbi-ological and chemical degradation, due to the microclimaticconditions and due to pollutants such as CO2, which causedre-carbonation phenomena on the surface of the wallpainting.

NMR and EFD analysis at different depth (0.5 and 2 cm)allowed the detection of the other causes of decay that is therising damp and salts content.

Finally this fully non-destructive study allowed theobtainment of complementary information useful to diag-nose the complex state of degradation of the wall painting.

This study may be easily extended to any wallpainting, especially those in hypogeous buildings, pro-vided that suitable specimens are used to reproduce thesame type of mortar constituting the wall painting to beinvestigated.

Multi-technique studies might be tuned to compileanalytical protocols for the quantitative mapping of themoisture and salts content and for evidencing the causes ofdegradation in wall paintings. Nevertheless, the type ofanalytical protocol tightly depends on the environmentalconditions and on the preciousness of the artworks.

Acknowledgments This work was partly supported by PAPER-TECH project, contract n.509095, operating within the VI EuropeanFramework. We thank the Director of the wall paintings restorationwork dr. Giulia Tamanti and the Director of the architecturalintervention arch. Francesco Sacco and for the cooperation andinterest in new methodologies. We are grateful to the restorers AnnaMaria Marinelli and Barbara Provinciali and to dr. Paola Santopadreof the Istituto Superiore per la Conservazione ed il Restauro (ISCR) ofRome for the useful discussion and suggestions during the work in St.Clement Basilica.

References

1. Proietti N, Capitani D, Cozzolino S, Valentini M, Pedemonte E,Princi E, Vicini S, Segre AL (2006) J Phys Chem B 110:23719–23728

2. Amoroso VGG, Fassina V (1983) Stone decay and conservation.Elsevier, Losanna

3. van Hees LRPJ, Binda IL, Papayianni I, Toumbakari E (2004)Mater Struct 37:664–648

4. Matteini M (1987) An assessment of Florentine methods of wallpainting conservation based on the use of mineral treatments. In:Cather S (ed) Proceedings of a Symposium of The Conservationof Wall Paintings, London, UK, pp. 13–16

5. Capitani D, Proietti N, Gobbino M, Soroldoni L, Casellato U,Valentini M, Rosina E (2009) An integrated study for mapping themoisture distribution in an ancient damaged wall painting. AnalBioanal Chem 395:2245–2253. doi:10.1007/s00216-009-3170-5

6. Rosina E, Ludwig N (1999) Optimal thermographic proceduresfor moisture analysis in buildings materials. Diagnostic ImagingTechnologies and Industrial Applications Conference, SPIEproceedings, vol 3827. International Society for Optical Engi-neering, Bellingham, pp 22–33

7. Ludwig L, Rosina E (1999) Active and passive thermography todetect moisture in building materials. Proceedings of 5th AITA,

Non-destructive mapping of dampness and salts

International Workshop on “Advanced Infrared Technology andApplications”, pp. 166–175

8. Proietti N, Capitani D, Lamanna R, Presciutti F, Rossi E, SegreAL (2005) J Magn Reson 177:111–117

9. Proietti N, Capitani D, Lamanna R, Presciutti F, Rossi E, SegreAL, Botticelli G (2005) In: Baldini U, Vigato PA (eds) Thediscovery of a forgotten pictorial cycle: Casa Vasari in Florence.Polistampa Editor, Florence

10. Proietti N, Capitani D, Rossi E, Cozzolino S, Segre AL (2007) JMagn Reson 186:311–318

11. Olmi R, Bini M, Ignesti A, Priori S, Riminesi C, Felici A (2006)Meas Sci Technol 17:2281–2288

12. Olmi R, Riminesi C (2008) Il Nuovo Cimento 31 C:389–40213. Patents in Italy (2004), European Community (2005), USA

(2009)14. Bacci M, Casini A, Olmi R, Picollo M (2008) Integrated non-

invasive techniques for the diagnosis and conservation of muralpaintings and other pictorial works. Proceedings of the 31stinternational symposium on “conservation and restoration ofcultural property —study of environmental conditions surround-ing cultural properties” Tokyo, pp. 93–102

15. Bandini F, Felici A, Lanfranchi M R, Mariotti P I, Olmi R, PrioriS, Riminesi C (2008) Non-invasive technique to detect moistureand salt content on frescoes. Proceedings of the InternationalWorkshop SMW08 “In situ monitoring of monumental surfaces”,Edifir Florence, pp. 415–420

16. Ciccone B, Boddi R, Felici A, Olmi R, Priori S, Riminesi C,Vigna A (2006) ‘Monitoraggio di affreschi mediante dielettrome-tria a microonde in due chiostri fiorentini’. Proceedings of the IVNational Conference on Archeometry, Science and CulturalHeritage (in italian), Pisa, pp. 891–902.

17. Sharma S, Casanova F, Wache W, Segre AL, Blümich B (2003)Magn Reson Imaging 21:249–255

18. Blümich B, Anferova S, Kremer K, Sharma S, Hermann V, SegreAL (2003) Spectroscopy 18:18–29

19. Blümich B, Perlo J, Casanova F (2008) Spectroscopy 52:192–269

20. Eidmann G, Savelsberg R, Blümler P, Blümich B (1996) J MagnReson A 122:104–109

21. Blümich B, Blümler P, Eidmann G, Guthausen A, Haken R,Schmitz U, Saito K, Zimmer G (1998) Magn Reson Imaging16:479–484

22. Mitchell J, Blümler P, McDonald PJ (2006) Prog Nucl MagnReson Spectrosc 48:161–181

23. Fukushima E, Jackson J (1999) News Lett 490:40–4224. Blümich B, Casanova F, Perlo J, Anferova S, Anferov V, Kremer

K, Goga N, Kupferschläger K, Adams M (2005) Magn ResonImaging 23:197–201

25. Hahn EL (1950) Phys Rev 80:58026. Farrar TC, Becker ED (1971) Pulse and fourier transform NMR.

Academic, New York27. Di Zenzo S (1986) Comput Vis Graph Image Process 33:116–12528. Courant R, Herbert R, Ian S (1996) What is mathematics?: an

elementary approach to ideas and methods. Oxford UniversityPress, New York, p 344

29. ICR (1981)Water Absorption by Total Immersion. Italian Recom-mendation NORMAL 7/81. ICR Rome.

30. Olmi R, Bini M, Ignesti A, Riminesi C (2000) Meas Sci Technol11:1623–1629

31. Olmi R, Bini M, Nesti R, Pelosi G, Riminesi C (2004) JElectromagn Waves Appl 18:217–232

32. Cleveland WS (1993) Visualizing data. Hobart, Summit33. Development Core Team (2005) A language and environment

for statistical computing Foundation for Statistical Computing.http://www.R-project.org.

34. Coladonato M, Giani E, Giovagnoli A, Rinaldi R, (1996)‘Environmental study for the safety of frescos in hypogeous site:the relationship between the water quantity and the solubile saltsinto the masonry’. Proceedings of “6th International Conferenceon Non-destructive Testing. Microanalytical Methods and Envi-ronmental Evaluation for Study and Conservation of Works ofArt”, Rome

35. Giani E, Giovagnoli A, Pietrini A (2004) Conservazione eRestauro 55:74–79 (in italian)

V. Di Tullio et al.