Background

Nitrogen oxides penetrate from the outdoor to the indoor environment and can be transformed, through complex reaction pathways, into gaseous nitric (HNO3) and nitrous acids (HONO) [1]. HNO3 is a very aggressive acid in contrast with HONO. The latter is not particularly active by itself, even if it is an important source of OH radicals, one of the most important air oxidants [2]. This air pollution cause oxidation and hydrolytic degradation of archive materials such as colour changes and reduction of degree of polymerisation of paper [3], decomposition of leather [4], corrosion of metals [5, 6], and damage of colorants [7–10].

Several works report measurements of nitrous and nitric acid in various indoor environments [11–13] and recent works highlight the importance of the former species as an emerging indoor pollutant [14], but studies in cultural heritage buildings are scarce [15, 16]. A well established and known technique to collect these trace gases is represented by denuder sequence based systems [17–19], but this technique is time consuming and labour intensive. Diffusive sampling can overcome these problems since it represents an easy to use technique which exploits the spontaneous diffusion of species collected by specific absorbing media. Diffusive samplers have been used for air quality monitoring of single gases in cultural heritage buildings [20–23]. Nevertheless, when conditions which promote the formation of HONO are present (high air moisture content and surface to volume ratio), especially indoors, monitoring of NO2 by diffusive sampling can result in overestimation since the interference of the former pollutant is not negligible [24]. Therefore to overcome this problem and to account correctly for the different species a novel multipollutant diffusive sampler was designed for simultaneous sampling of three different pollutants (nitric acid, nitrous acid, and nitrogen dioxide) collected at separate sampling stages [25]. After the laboratory development, these devices were used indoors in different dwellings, as reported in [25]. The campaigns described in the following were the first applications to places of interest for cultural heritage conservation, as libraries and archives, where a certain kind of control on environmental parameters is commonly present. The aim of this study was, indeed, to test the newly developed multipollutant sampler for evaluating the air quality inside different libraries with or without HVAC (heating and ventilating and air conditioning systems). The exposure period, after these first sampling trials, could then be adjusted according to the typical range of values found for the different pollutants in the indoors of interest. These preliminary trials were necessary to avoid the saturation of the absorbing pads and the collection of insufficient analyte to be determined after the exposure period. Furthermore the sampler was also exposed outdoors to compare the pollutants concentration values and calculate indoor/outdoor ratios.

Methods

Sampling locations

Over the last 4 years the multipollutant sampler was used in different periods inside archives and libraries. Two seasonal campaigns were carried out at four different libraries: two libraries in Switzerland (the Swiss National Library in Bern and the Bibliotheque de Geneve in Geneva) and two libraries in the Czech Republic (the National Archives in Prague and the Regional Library in Teplice). All the indoor activities in the libraries are very limited.

The library in Bern is a modern building composed of a ground floor and four underground levels (1UG-4UG), the building is equipped with a heating, ventilation, and air conditioning (HVAC) system providing air filtered through a particulate filter and active carbon. An additional Purafil purification system is installed at the floor 3UG where newspaper collection is stored. Five indoor sites were selected at the different levels and an outdoor site (OUT) was chosen as well. One site was placed at the ground floor (A26) and the other four monitoring sites were placed both at the entrance, just outside of the double door insulation (WA2UG and WA3UG), and inside the second and third underground level (2UG and 3UG) in the area insulated by the double door system. The overall volume of the building in Bern is of approximately 4500 m3.

The library in Geneva is a historical building with a HVAC system equipped with particulate filters. Three indoor sampling sites were selected: the first in a compactor (number 51 in the following), the second at “La Reserve” and the third in a corridor. An outdoor site was chosen also in Geneva where the diffusive samplers were placed in a balcony (OUT). The library in Geneva has a volume of approximately 200 m3.

The archive in Prague is located in a modern building. The indoor measurements were carried out in a depository of “Bohemian tables (BT)” and in a depository of “Archives of the Czech Kingdom (ACK)”. The BT and ACK depositories have volumes of approximately 550 and 220 m3, respectively. Each depository is equipped with its own HVAC system with air recirculation. The HVAC systems in both depositories contain HEPA filters for removing airborne particles. In addition the AKC depository is equipped with filters for removing NO2 and SO2. The outdoor measurements were performed on a balcony of the building orientated toward a busy street (OUT).

The library in Teplice is equipped with double glassed windows and an electrical heating system, i.e., the only ventilation is through cracks and small openings in the building, windows and doors. The measurements were performed inside the library (IN) and just outside of the window oriented to the chateau park (OUT). The depository has a volume of approximately 450 m3.

One monitoring location in each indoor and outdoor site was selected, because other measurements confirmed that the indoor air is well mixed [26].

Measurement campaigns

Seasonal campaigns were conducted in Switzerland during the spring/summer period of 2011 (S1) and during winter and spring/summer 2012 (W, S2). Later other measurements were performed in the Czech Republic during three campaigns in spring/summer 2013 (S1, S2) and winter 2014 (W) (Table 1).

Table 1 Monitoring scheme
Full size table

Three multipollutant samplers were exposed indoors and three outdoors for during each campaign at every location.

In parallel the exposure of single stage Analyst diffusive samplers (Marbaglass, Italy) provided measurements of NO2 and NOx during the campaigns in Switzerland. Other species (HNO3, SO2, O3 and NH3) were also monitored by the Analyst samplers in the campaigns carried out in the Czech Republic, where in addition IVL diffusive samplers (Swedish Environmental Research Institute, Sweden) for monitoring formic and acetic acids were employed.

In both cases at least 10 % of the total number of samplers was used for each campaign as field blanks. The blanks were prepared and handled along with other samplers except for the exposition to pollutants. Additionally, basic meteorological parameters including ambient temperature and relative humidity were measured at both locations in Czech Republic, both indoors and outdoors, by Tiny Tag data loggers (Gemini, UK).

Multipollutant diffusive sampler

The body of the multipollutant diffusive sampler is similar to the Analyst [27]. The internal design of the sampler, on the other hand, was changed to collect HNO3, on the first filter and NO2 and HONO in the successive absorbing pads.

A correct speciation of these compounds, intertwined through the hydrolysis mechanism according to the known pathway [28–30]:

$$2 {\text{NO}}_{ 2} + {\text{ H}}_{ 2} {\text{O }} \to {\text{ HONO }} + {\text{ HNO}}_{ 3}$$
(1)

could, indeed, only be achieved by the subsequent collection of the two species on successive reactive substrates.

Most frequently NO2, which is the predominant species in the atmosphere also due to the photolysis of HONO in the outdoor ambient air, can affect the measurements of the less abundant HONO. On the other hand, particularly indoors, when moisture and an elevated surface/volume ratio is present, NO2 measurements can be overestimated as reported in [24]. Therefore there is a mutual interference on the measurement of these two species. In this case, since nitrite is the species collected and analytically determined to get to the atmospheric concentration of these pollutants, the selectivity in the sampling phase was achieved through the implementation of successive sampling stages. After the first filter used to collect HNO3, other two successive pads were used to collect these species, assuming that HONO and part of the NO2 would both react on the second, whereas only NO2, normally the most abundant, is collected on the successive filter. During the development of this diffusive sampler trials, aimed at quantifying the relative amounts of nitrite collected on the second and third filter at different HONO/NO2 concentration ratios, were performed. The ratio of the amounts of analyte collected on the two filters at RH = 0 was determined and used in the calibration of the device.

Hence three polyethylene discs housing three active glass microfiber filters were placed along the same diffusive path of the former sampler. The first two filters perform the sequential collection of the two acidic species and the third filter is used to estimate the interference of NO2 on the HONO measurement and vice versa.

Before impregnation with the reactive solution, a washing procedure of the microfiber filters with sodium carbonate (5 %) and successive rinse (for three times) with deionized water was implemented to improve the quality of the blanks. The filters (Whatman GF/A) inside the first polyethylene disc were then coated by using a 0.1 % (w/v) aqueous solution of sodium chloride then dried in an oven at 80 °C.

The following other two glass microfiber filters, used to collect HONO and to account for the interference of NO2, were impregnated by using a 1.8 % (w/v) aqueous solution of sodium carbonate containing 1.8 % (w/v) of glycerine, they were then dried in an oven at 80 °C. Filters are placed along the diffusive path in front of the alkaline carbon filter prepared according to the procedure already described [27] and positioned at the bottom of the device for the sole NO2 sampling. An enlarged scheme of the sampler is reported in Fig. 1.

Fig. 1
figure 1

Exploded scheme of the multipollutant sampler. 1 Plastic cap (before and after exposition) or air barrier (during exposition), 2 Ring, 3 Polyethylene holder of absorbing pad impregnated with NaCl (HNO3 sampling), 4 Polyethylene holder of absorbing pad impregnated with Na2CO3 (HONO sampling), 5 Polyethylene holder of absorbing pad impregnated with Na2CO3 (interference of HONO and NO2), 6 Carbon Filter (NO2 sampling), 7, 8, 9 Plastic spacer, 10 Body of the sampler, 11 Absorbing pad (containing Na2CO3 or NaCl). Geometric features A cross sectional area (330 mm2), R maximum length of the plastic support (10 mm), r plastic support radius (6.5 mm), a diffusive path to NaCl impregnated filter (4.5 mm), b diffusive path to first Na2CO3 impregnated filter (11.3 mm), c diffusive path to second Na2CO3 impregnated filter (17.1 mm), d diffusive path to impregnated carbon filter (24.0 mm)

Full size image

Chemical analysis procedure

As for the other diffusive samplers, the multipollutant samplers at the end of collection time were sealed and successively analysed according to a procedure similar to the one reported elsewhere for NO2 [27]. In this case the filters were removed from the sampler, then they were extracted by adding a solution of sodium bicarbonate and carbonate, 0.3 mM NaHCO3 and 2.7 mM Na2CO3 in a plastic vial. The solution obtained was stirred with VIBROMIX 203 EVT (from Tehtnica, Železniki, Poland) and analysed through ion chromatography (IC) (Dionex ICS 1000 equipped with AS12A column. The concentrations of analytes such as nitrate and nitrite were determined referring to calibration curves constructed with water solutions prepared by opportune dilution of stock standards (Certipur from Merck, Milan, Italy) containing 1000 mg/L of each analyte.

Results and discussion

Multipollutant diffusive sampler calibration

Some information will be given concerning the calibration of the new device performed during the laboratory trials in the development phase. The prototypes of the new sampler were exposed to standard atmospheres containing the pollutants of interest in a smog chamber. The experimental setup was equipped with reference technique for the measurement of each species, the data were used to calibrate the new device. The calibration was achieved through comparison with denuder data: NaCl coated denuders were the reference for HNO3, whereas two Na2CO3 coated denuders [31] accounted for nitrous acid.

Successively the comparison between the two different techniques gave good results also in field trials (for HNO3 R2 = 0.97; for HONO R2 = 0.96). The chemiluminescent analyser, used as a reference for NO2, was also in agreement with the newly developed diffusive sampler (R2 = 0.97).

A paired t test was performed to compare the data obtained by denuder technique and diffusive sampling [HONO: t(10) = 0.31 p = 0.75; HNO3: t(10) = 2.18 p = 0.053], and by chemiluminescent analyzer and diffusive sampling [NO2: t(10) = 1.59 p = 0.14]. At a p value of 0.05, in all cases the results are consistent with the null hypothesis that there is no difference between techniques would be accepted, and certainly for HONO there is very good agreement; for HNO3 more measurements might in retrospect have been appropriate to form a more considered view.

A comparison with the Analyst sampler for NO2 was also performed. The Analyst differs from the multipollutant sampler for the geometry, since it is basically a cylinder open on one end, whereas the multipollutant has a rather complex structure (Fig. 1). The uptake rate (Ur), which is defined as:

$$Ur = \frac{D \cdot A}{L}$$
(2)

where D is the diffusion coefficient (cm2/s) of the gaseous species, A is the cross-sectional area (cm2) and L the diffusive path (cm), is quite different for the two samplers. The ratio between the uptake rate for NO2 for the Analyst and the multipollutant is 3, being the Ur Analyst about 12 cm3/min and Ur Multipollutant about 4 cm3/min (measured at RH = 0 %, in a dry atmosphere containing NO2 without HONO). This feature of the newly developed diffusive sampler can explain the lower sensitivity to NO2 if compared to the Analyst, since in the same lapse of time the latter can collect a higher amount of analyte (nitrite), on the other hand the selectivity of multipollutant sampling is increased as already explained.

In field use of Analyst and multipollutant sampler

As a first approach, the agreement of the measurements of NO2 obtained by the Analyst sampler and by the multipollutant sampler was checked (Fig. 2). The correlation coefficient was quite good (R2 = 0.87), taking into account the whole set of measurements (both Swiss and Czech libraries and archives).

Fig. 2
figure 2

Regression between the multipollutant and the analyst sampler results of NO2 measurements

Full size image

The new multipollutant samplers and standard Analyst samplers were also compared by estimation of sensitivity given as limits of detection (LODs) and uncertainty given as relative standard deviations. The LODs were calculated as three times the standard deviation of blank samples [32]. The sensitivity of multipollutant and Analyst samplers was comparable for HNO3, but the Analyst samplers were more sensitive to low concentrations of NO2. The lowest detectable concentrations for the multipollutant samplers after one month exposure were found to be 1.4 µg/m3 for NO2, 0.2 µg/m3 for HNO3, and 0.5 µg/m3 for HONO, and for the Analyst samplers 1.0 µg/m3 for NO2 and 1.0 µg/m3 for HNO3.

To assess the reproducibility [33] relative standard deviations were considered. Standard deviations were calculated for each set of replicates during the different campaigns. The standard deviations for multipollutant samples were found to be 4 % for NO2 and 20 % for HNO3 and for Analyst samplers 5 % for NO2 and 26 % for HNO3. These results were in agreement with the variations found by other studies [34, 35] and also fulfil the ±25 % uncertainty requirement of the European Directive for indicative monitoring with diffusion samplers [36]. Higher values of the standard deviations were found for HONO measured by the multipollutant samplers (approximately 68 %). HONO measurements reproducibility was likely affected by the really low values found indoors, which in many cases were below the detection limit, therefore the dataset was eventually composed by a narrow number of data.

NO2 and HNO3 concentrations values at the different locations in Swiss libraries, in particular at the SNL where abatement of the pollutants was achieved through the combination of different filters, were often lower than the detection limit.

Nitric acid concentrations were very low probably due to the very high deposition velocity on available surfaces [1, 37].

The extension of the exposure period was implemented to increase the amounts of analyte collected, optimizing the sampling length according to the sensitivity of the technique. Therefore the duration of the campaigns in Switzerland was of about 20 days, whereas the later campaigns in Czech Republic were extended to about a month.

Indoor and outdoor concentrations

Indoor pollutant concentrations are influenced by indoor sources and sinks and penetration from the outdoor environment. As it can be seen from a comparison of the different environments I/O (indoor/outdoor) ratios (Tables 2, 3, 4, 5), the efficiency of the filtration systems (turned off at Bern during the second and third campaign) in removing the pollutants does not seem to produce large differences in the environmental parameters measured. For HONO, when it was possible to perform the calculation, these ratios were always higher than 1 indicating that the formation of the pollutant happens indoors. In this case the filtration systems removing NO2 should avoid the formation of this pollutant, rather than removing it from infiltrating air.

Table 2 The Swiss National Library in Bern campaigns results (µg/m3), bdl—below detection limit
Full size table
Table 3 The Bibliotheque de Geneve campaigns results (µg/m3), bdl—below detection limit
Full size table
Table 4 The National Archives in Prague campaigns results (µg/m3), bdl—below detection limit
Full size table
Table 5 The library of Regional Museum in Teplice campaigns results (µg/m3), bdl—below detection limit
Full size table

The reaction (1) is negligible in the gas phase but occurs in the presence of surfaces, where produced HNO3 remains in the surface water film, whereas HONO forms an equilibrium shifted towards the gaseous phase [38].

According to (1) the equilibrium concentration of HONO should depend on NO2 concentration and relative humidity RH.

The surface/volume ratio is also known [30] to affect both the formation of HONO and NO at different concentrations of NO2 and for different values of relative humidity.

It is in agreement with higher NO2 and HONO outdoor concentration in Prague than in Teplice. Indoor concentrations of HONO were found in the BT depository and the Teplice library, with similar indoor conditions: indoor NO2 concentrations (Tables 4, 5), RH (Table 6), and a surface/volume ratio 2.6/m and 2.9/m in Teplice and BT, respectively.

Table 6 Temperature and relative humidity
Full size table

Moreover, HONO was not detected in the ACK depository, where NO2 concentrations were lower than the detection limit of both types of dosimeters. These results are in agreement with the production of HONO indoors from NO2 by reactions on exposed surfaces which include books and manuscripts, giving rise to acid deposition of HNO3.

The outdoor concentrations of the pollutants measured in the four cities reflect the seasonal trend, being higher everywhere during wintertime due to low atmospheric layers mixing. The cities of Geneva and Prague showed the highest values (about 40 µg/m3).

The concentration values of NO2 were always lower indoors than outdoors, which indicates that there were no indoor sources. Higher indoor values of NO2 were measured at WA3UG and WA2UG, the two rooms at the library in Bern where the insulation from double doors is not provided. In these two rooms the HONO levels were also higher than in the others. The A26 Collection, where a window is sometimes opened, the concentrations of the pollutants monitored were the highest at the library in Bern. The same can be observed for the corridor at the library in Geneva.

Relatively constant values of NO2 were observed at the first indoor site in Prague (BT) during all sampling periods. Concentrations measured inside the second indoor site (ACK) were always below detection limit. The difference in NO2 concentrations was probably caused by the efficiency of filters for NO2 in the HVAC system of the AKC depository compared to the BT depository equipped only with filters for particles. Indoor NO2 concentrations measured at Teplice were comparable to concentrations in BT though outdoor concentrations were lower. This was probably due to the higher natural ventilation in this library as compared to the isolated archives.

Conclusions

The results revealed that the reproducibility of the new multipollutant sampler measurements was reasonable (according to EU directives) for NO2 and HNO3. The measured data were comparable with the standard Analyst sampler results, but differ for lower NO2 concentrations, probably due to differences in sampling and analytical procedure.

The results showed that the newly developed multipollutant sampler can be used in archives and libraries, allowing to map the pollutants distribution indoors. Due to high efficiency of insulating systems normally employed (such as double doors etc.) and filtration systems, the values recorded for the pollutants indoors are often lower than the detection limits. Hence HONO and HNO3 measurements were carried out in very “clean” environments, and the dataset eventually available was likely not wide enough to truly assess the reproducibility of the measurements, especially for the former pollutant.