This work is divided into two studies. The first one is focused on the selection of the most adequate low-cost material to remove formic acid. The second study was the assessment of the protective capacity of the sorbent with an unstable soda glass.
Selection of sorbents
Four low-cost materials were selected to evaluate the reduction of the concentration of atmospheric formic acid. The materials tested were activated carbon, silica gel, copper, and steel. Two of them (carbon and silica) are general sorbents; however, metals are materials that react specifically with the formic acid.
Activated carbon
Activated carbon is the most common sorbent due to its availability, low cost, and good efficiency. Previous studies consider it the most effective material to reduce air pollution generated outdoors and organic acids in showcases [4, 10, 12, 13]. Grosjean et al. studied how in all the tests in a passive way, 20 g of activated carbon removed pollutants from the air by passive diffusion at a rate that exceeded by a factor of 10 the loss of pollutants on the walls of the showcase [24]. On the other hand, Schieweck observed that both under active and passive conditions, pure and impregnated activated carbons showed good adsorption efficiency for formaldehyde, formic acid, and acetic acid [5].
Activated carbon corresponds to porous carbon in which microporosity has been enhanced through activation, usually through steam oxidation [13]. The carbon is injected into a stream of hot air to create activated carbon [5]. The stream of hot air creates a large number of small pores that increase the surface area [12]. It is characterized by having a fine porous structure, a high specific surface area between 300 and 2000 m2/g and a density between 200 and 600 kg/m3 [4, 5]. In addition, its adsorption capacity varies depending on the diameter of its pores, distinguishing between micropores (< 1 nm), mesopores (1–25 nm) and macropores (> 25 nm) [4]. This material is capable of reducing polluting gases by physical adsorption on the inner surface. This mechanism is generally based on relatively weak intermolecular forces, such as van der Waals interactions. Organic compounds with a molecular weight greater than 45 g/mol are considered good adsorbates on activated carbon [5]. The water capacity of the coals in this region is close to 30% of their dry mass. This water can displace adsorbed organic molecules [13].
Nutshell carbons are particularly useful for this purpose, as they have a very suitable original porosity for activation [13]. However, a drawback of using plant materials is the existence of residual content of inorganic ash [13]. Activated carbon is available in powder, granule, foam, or fabric form. From a practical point of view, activated carbon fabric is better to install than foam, which is better to install than granulate [4]. Carbon foams are easy to cut to size; however, they tend to lose some of the carbon, leaving numerous black spots [4].
Silica gel
Silica gel is used primarily for two functions, to maintain a stable relative humidity or as a desiccant to dehumidify the air [12]. Since this material absorbs moisture from indoor air, the impact on relative humidity should be considered when using it as passive adsorbent for organic acids. Previous studies consider that silica gel does not eliminate some air pollutants generated outside but that it can be a good option to control humidity while it is used in combination with another specific sorbent [5, 12]. Another possibility to consider is that since silica gel strongly adsorbs water vapor, previously adsorbed organic molecules can be displaced [13].
Metals
Metal objects are also susceptible to organic acid attack just like glass. Corrosion caused by the vapors of these acids on different metals and alloys has been studied for years [25,26,27,28]. Taking this into account, metals could be used as sorbents when acting as sacrificial material. Lead is the most cited metal in terms of degradation; however, it is not chosen due to its toxicity since the sorbent materials studied are intended to have real applicability in museums. Bronze, iron, and copper are other materials that can be used as possible sorbents, due to the degradation they present when exposed to organic acids [3].
There are numerous studies on the degradation of copper due to its exposure to organic acids, since they are able to cause metal corrosion even at very low concentrations of acid vapor [28, 29]. López-Delgado et al. studied the corrosion it presents when exposed to both formic and acetic acid and a relative humidity of 100% [29]. Tetreault et al. also studied the formation of copper corrosion products in the presence of formic acid, obtaining that at levels above 4 ppm copper increases in weight at both 54% and 75% RH [28]. An increase in the concentration of formic acid results in a strong effect on copper corrosion. At 8 ppm of formic acid, the copper samples are covered by a thin layer of an opaque green to gray matte film [28]. At a concentration of 10 ppm, the components of the patina are mainly cuprite (Cu2O), hydroxide of copper (Cu(OH)2), and copper formate (Cu(HCOO)2), although the latter appears as a small signal [25, 28]. Above 14 ppm, copper samples show whitish surface colors. The corrosion compounds found at 14 and 140 ppm were identified respectively as copper formate and copper formate dihydrate [28]. It is important to note that the higher the concentration of formic acid to which it is exposed, the more corrosion appears at lower relative humidity.
The other metal option that arises to study it as a sorbent material is steel wool [26, 27]. This material is composed of iron, which can react with organic acids to which it is exposed in the environment. Steel wool is made up of fibers that are often used for cleaning and polishing metal or wood surfaces. Steel wool is easily solubilized when reacting with dilute acid, forming ferrous ions when reacting with protons (Reaction 3). Furthermore, ferrous ions can be oxidized to ferric ions by dissolved oxygen (Reaction 4) [27]. Similarly, adsorption of formate ions on the iron surface may occur [26].
$${\text{Fe}}^{0} \left( {\text{s}} \right) \, + {\text{ 2 H}}^{ + } \left( {{\text{aq}}} \right) \, \to {\text{ Fe}}^{{{2} + }} \left( {{\text{aq}}} \right) \, + {\text{ H}}_{{2}} \left( {\text{g}} \right).$$
(Reaction 3)
$${\text{2 Fe}}^{{{2} + }} \left( {{\text{aq}}} \right) \, + \, \raise.5ex\hbox{$\scriptstyle 1$}\kern-.1em/ \kern-.15em\lower.25ex\hbox{$\scriptstyle 2$} {\text{ O}}_{{2}} \left( {{\text{dissolved}}} \right) \, + {\text{ 2 H}}^{ + } \left( {{\text{aq}}} \right) \, \to {\text{ 2 Fe}}^{{{3} + }} \left( {{\text{aq}}} \right) \, + {\text{ H}}_{{2}} {\text{O }}\left( {\text{g}} \right).$$
(Reaction 4)
Preparation of sorbent materials
The copper was a 2491X tempered wires from the RS brand, used as an electrical material, with a cross-sectional area of 0.75 mm2 and a filament of 0.2 mm. The 24 copper wires were separated to increase the specific surface area. The cables weighted 4.63 g. The steel wool used from the Dexter company corresponds to category 00, which has a very fine fiber size (8.89–12.7 µm) within the market range. To prepare it, the fibers were separated, as in the case of copper, and 4.64 g were weighed. Activated carbon, from the Scharlau brand (CAS: 7440-44-0), is used in powder with a heavy mass of 4.66 g and of “very pure” quality. Finally, the silica gel is from the Labkem brand (Ref: SGE0-002-1K0) and is granulated with a diameter size of 2–5 mm. To prepare it, it remained in the oven at 70 °C for three days, just until the moment of introduction into the desiccator, obtaining a weight of 4.61 g. During the test, the Petri dish with renewed or regenerated silica gel was maintained at 70 °C for 90 min in an oven before their substitution in the desiccator.
Glass preparation
The composition of the soda glass prepared is based on real nineteenth century soda glasses [14] but simplified to be more susceptible to the environment. The minor components were not considered to avoid interferences, and their percentage was added to the alkaline content. The fusion was made at 1600 °C for two hours in an alumina crucible in a Termolab electric furnace, followed by an annealing at 500 °C in a Carbolite furnace.
The glass obtained was later adapted for the accelerated aging tests. Slices and small glass blocks were cut and dried in an oven at 70 °C for 30 min to remove the water used in the process to lubricate the cutting disc. Subsequently, the slices were polished with the Buehler brand MetaServ 3000 polisher. Polishing was carried out progressively with P320, P600 and P1200 sandpaper according to FEPA standards, which correspond to a grain size of 46.2, 25.8 and 15.3 μm respectively, using ethanol or ethanol-based lubricant to lubricate the sandpaper. Finally, it was finished with a diamond paste polishing with 3 µm and 1 µm particles. On the other hand, the small glass blocks were ground and sieved following the standard UNE 400322:1999 [30]. For this, two mortars and sieves with different sizes were used, thus obtaining fractions between 0–300 μm, 300–500 μm and greater than 500 μm, which were stored in Eppendorf tubes.
Accelerated aging tests and experimental set-up
The humid and acidic environment was recreated in different desiccators. The prepared environments had 100% RH and 10 ppm of formic acid. These values were chosen in order to increase the hygroscopicity of glass and, therefore, the alteration rate [15].
To generate the acidic atmospheres, 0.549 mL of formic acid was added to 600 mL of distilled water to obtain 10 ppm following the description of Bastidas et al. [31] The different sorbents and materials were exposed during a period of 21 days, time in which the sorbents, sacrificial materials or glasses were reacted.
For the first test, the materials were deposited in Petri dishes inside the desiccators (Additional file 1: Fig. S1). In turn, the Petri dishes were supported on lower base supports to allow the formic acid and water vapors to rise more easily from the bottom of the desiccator.
For the second test, ~ 5.00 g of silica gel, a slice of the prepared glass embedded in resin, a porcelain weight boat with approximately 0.50 g of the glass fraction smaller than 300 µm and a thermohygrometer were introduced into each desiccator (Additional file 1: Fig. S1).
Characterization of the environment
The environments were monitored using thermohygrometers, and the adsorption capacity of the materials by ion exchange chromatography (IC).
The thermohygrometers, model BL-1D from the Rotronic company, were kept in the desiccators during the entire period of preparation and exposure to the environment. The thermohygrometers, with an accuracy of ± 3.0% HR and ± 0.3 °C, were programmed to record data every five minutes, obtaining the corresponding values for relative humidity and temperature.
The adsorption capacity for formic acid of the materials was evaluated by an indirect method by ion exchange chromatography. Samples of the solution from each desiccator (< 2 mL) were taken on days 0, 3, 7, 10, 14, and 21. All chromatographic analyzes were performed at room temperature using the Metrohm Advanced Compac ion chromatographic instrument (861 IC) with conductivity detector (IC-819), liquid Handling Pump Unit (IC Pump 833), sample degasser (IC-837) and an 800 Dosino Dosing Device. Data acquisition, calibration curve construction and peak integration were carried out with a Metrohm 761 data acquisition system interconnected to a computer running MagIC Net 3.3 software. The identification and quantification of formate was carried out in a Metrosep Organic Acids column (250 × 4 mm, Ø 5 µm). The mobile phase is 0.5 mM sulfuric acid and 15% acetone, with a flow rate of 0.5 mL/min. The injection volume was 20 µL and the analysis time 30 min. The standard used was Supelco's Formate Standard for IC. The calibration curve was done between 15 and 500 µl with a correlation of 0.99.
The calculation of the decrease in formic acid was carried out using the Eq. (1).
$$\Delta Concentration\left(\%\right)=\frac{\left[Formate\,Day\,X\right]-[Formate\,Day\,0]}{\left[Formate\,Day\,0\right]}\times 100.$$
(1)
Characterization of materials
Sorbents and glass samples were characterized by X-ray fluorescence (XRF), surface area analysis, gravimetry, Fourier Transformed Infrared Spectroscopy in Attenuated Total Reflectance mode (FTIR-ATR), Scanning Electron Microscopy with X-ray Energy Dispersive Spectroscopy (SEM–EDS), µ-Raman spectroscopy, and Optical Microscopy.
The exact composition of the glass was analyzed with a PANalytical MagicX (PW-2424) wavelength dispersed X-ray spectrometer equipped with a rhodium tube (SUPER SHARP) of 2.4 kW. The results were treated with the quantitative silicate analysis curve after analyzing the samples as pearl (fusion of 0.3000 g sample and 5.5 g of Li2B4O7). The specific surface area was evaluated by the Monosorb Surface Area Analyzer MS-13 equipment from the Quantachrome company. The samples are degassed for 2 h in a 70:30 He:N2 gas stream at 150 °C. The measurement takes place by nitrogen adsorption at 77 K by the one-point Brunauer–Emmett–Teller (BET) method. The BET equation considers the van der Waals forces to be solely responsible for the adsorption process. From this method, the area of a solid can be determined by knowing the amount of adsorbed gas that is necessary for a monolayer to form and the area that an adsorbed molecule occupies. The specific surface area of each material is evaluated three times, obtaining the average result.
The hygroscopic capacity of each glass was determined by gravimetry in 0.5000 g of the glass fractions with Ø < 300 μm using three porcelain weight boats for each glass. Before starting the test, the empty weight boats were stored for three days in their corresponding desiccator to ensure that they are hydrated. Every week, the weight boats were weighed. Likewise, a slice of each glass was inserted into the desiccators leaving the polished face exposed. After the period of exposure to the different environments (21 days), all the weight boats and slices were stored in a desiccator with silica gel (Ø = 2–5 mm). The percentage of the increase in weight of the glass due to hydration with respect to the initial value was calculated using Eq. (2).
$$\Delta weight (\%)= \frac{Weighing\,boat\,day\,X-Weighing\,boat\,day\,0}{Weighing\,glass\,day\,0}\times 100.$$
(2)
In order to observe the changes on silica gel and glasses, Fourier transform infrared spectroscopy (FTIR-ATR) was performed. In each spectrum, eight scans were made with a sweep from 4000 to 400 cm−1. The equipment used was the PerkinElmer brand Spectrum 100 FT-IR Spectrometer, together with a PIKE Technologies brand GladiATR Attenuated Total Reflectance accessory. This technique allows characterization through reflection or absorption spectra in the infrared range of the electromagnetic spectrum. From the frequencies of the functional groups, the compounds that compose it can be identified. In addition, it is necessary to consider the advantage that it can be used for both inorganic and organic substances in solid, liquid or gaseous state.
The surface of copper, steel, and silica gel samples was observed and analyzed by scanning electron microscopy with X-ray energy dispersive spectroscopy (SEM–EDS), using the Hitachi S-4700 equipment. This technique allows morphological and microanalytical characterization through electronic images and energy dispersion X-ray microanalysis. The interaction of the electron beam on the sample generates X-rays that reach the detector and allow obtaining microanalytical information, qualitative in this case. Prior to measurement, the samples were made conductive by sputtering with carbon. Likewise, the alteration products formed on copper and silica gel were characterized by Raman spectroscopy before and after the test. Raman spectra were recorded using a confocal Raman microscope integrated with atomic force microscopy (AFM) on an ALPHA 300AR microscope from WITec. This equipment allows combining the potential of an AFM microscope, being able to obtain images of up to 3 nm of lateral resolution, with the structural and compositional characterization of the materials at a submicron scale of confocal Raman spectroscopy. The microscope is equipped with a Nd:YAG laser. The spectra obtained were analyzed with the WITec Control Plus software. The acquisition time was 3.6 s for one single spectrum and the Raman image consists of 3000 spectra with a laser excitation of 532 nm and the incident laser power of 1 mW, using a tested area of 10 µm × 10 µm. The colors in the spectra correspond to different areas in the Raman image using a filter for 172–260 cm−1.
The surface of the glass samples was observed by optical microscopy in a Zeta Systems brand optical profilometer model.