materials

Article Influence of the Composition, Structure, and Physical and Chemical Properties of --Based Sorbents on Water Adsorption Ability

Ruslan Zotov 1, Eugene Meshcheryakov 2, Alesia Livanova 2, Tamara Minakova 2, Oleg Magaev 2, Lyubov Isupova 3 and Irina Kurzina 2,* 1 Salavat Catalyst Plant, Salavat 453256, Republic of Bashkortostan, Russia; [email protected] 2 Faculty of Chemistry, National Research Tomsk State University, Tomsk 634050, Russia; [email protected] (E.M.); [email protected] (A.L.); [email protected] (T.M.); [email protected] (O.M.) 3 Boreskov Institute of , Siberian Branch, Russian Academy of Sciences, Novosibirsk 630090, Russia; [email protected] * Correspondence: [email protected]; Tel.: +7-913-882-1028

Received: 29 November 2017; Accepted: 9 January 2018; Published: 14 January 2018

Abstract: Interaction between the water adsorption ability of aluminium-oxide-based sorbents and their chemical composition, - properties of the surface, and textural characteristics has been analysed. Alumina desiccants were synthesized with the centrifugal–thermal activation of followed by the hydration of the gibbsite under mild conditions. It was demonstrated that the multicyclic adsorption regeneration of samples under realistic conditions results in structural transformations and changes in the acidity of their surfaces and water adsorption ability. The modification of pseudoboehmite with alkali increases surface basicity and the dynamic capacity of adsorbents relating to water vapours. Equations have been presented that describe the adsorption and desorption processes taking place during water vapour adsorption with the materials studied.

Keywords: ; acid-base properties; kinetics; water adsorption

1. Introduction The use of active alumina as a catalyst in refining processes (e.g., reformation, hydro-treatment, and hydrocracking) and as an adsorbent, in particular, a natural gas desiccant, is widely known. The high activity of aluminium oxide in interaction with polar adsorbates (first, with water vapours) provides deep drying of associated petroleum gas to a dew point of −60 ◦C and below. Water resistance is an important positive property of aluminium oxide and often determines the selection of aluminium oxide as an adsorbent for drying and the treatment of media containing condensed moisture. The possibility of multiple temperature regeneration by means of coke burn-off ensures the long-lasting work of the adsorbent as a desiccant of olefin-containing flows [1]. Finding new, more effective aluminium-oxide-based adsorbents [2] and the use of an appropriate complex loading of desiccants into adsorbers, which allows for receiving the associated petroleum gas (APG) with the required parameters, remains relevant. It is known, for example, that loading an aluminium oxide protective layer into an adsorber extends the service life of [3,4]. The use of a more effective desiccant as a protective layer will allow for not only the protection of a bottom layer from condensed moisture but also an enhancement of the effectiveness and operational life of the adsorber. There are studies on highly effective alumina desiccant adsorbents based on low-temperature forms of aluminium oxide (η-, γ-, and χ-) received by means of the incineration of alkaline

Materials 2018, 11, 132; doi:10.3390/ma11010132 www.mdpi.com/journal/materials Materials 2018, 11, 132 2 of 10 hydration products of thermally activated alumina containing bayerite phases of 50% and more [2,5,6]. A characteristic of bayerite-containing is the formation of low-temperature phases of aluminium oxide (first, η-modification) at a temperature >300 ◦C. This provides an opportunity to obtain samples with a developed specific surface area, the required combination of surface sites, a large quantity of micropores, and therefore a larger static capacity than that of the desiccants based on γ-Al2O3 that are obtained on the basis of pseudoboehmite using the reprecipitation method [5]. The study [6] demonstrated that the dynamic capacity of pseudoboehmite-based desiccants synthesized with the centrifugal–thermal activation of gibbsite with its subsequent hydration under mild conditions is significantly lower than the dynamic capacity of desiccants based on η-Al2O3. The chemical modification of surfaces with and bases has become a more widespread method for increasing the adsorption capacity of active aluminium oxide. A good example is the adsorption capacity of acids or bases, which allows us to modify the concentration of acid–base sites and the reactivity of alumina surface hydroxyls [7]. When sulphuric acid was introduced at the preparation stage of moulding a sorbent mass, the samples received from pseudoboehmite-containing were comparable in dynamic capacity rates to bayerite-based adsorbents, and even exceeded them in static capacity rate [6]. This is related to change in composition, textural characteristics, and acid–base surface properties. Moreover, a greater modifying effect was observed in desiccants based on γ-Al2O3 that had a greater quantity of Bronsted acid sites (BAS) and potent Lewis acid sites (LAS) after the introduction of sulphate ions, and the average diameter of pores was reduced [5]. The dependence of the static capacity value at low relative humidity on the concentration of electron-withdrawing sites has been established. Impregnation with may also result in the modification and increase of the concentration of surface base sites. Interaction between structural-phase and surface characteristics of adsorbents and their sorption capacity in relation to water is relevant for study. The purpose of this work is to study the acid–base, texture, and water-absorbing properties of active adsorbents for pseudoboehmite-based samples and for samples modified with alkaline ions.

2. Materials and Methods Samples of bayerite- and pseudoboehmite-based alumina desiccants synthesized with centrifugal–thermal activation gibbsite (CTA GS) followed by its subsequent hydration under mild conditions [6], and also samples of pseudoboehmite-based aluminium oxide modified with and ions, were taken as study objects. The modification was carried out introducing and potassium hydroxide solutions at the preparation stage of moulded plastic pastes from pseudoboehmite received by means of the mild hydration of CTA GS product. Adsorbent granules had a diameter of 3.75 ± 0.15 mm at a length of 5 ± 1 mm after the completion of extrusion and the thermal treatment. A series of studies to determine adsorbent dynamic capacity values on the basis of water vapours under realistic industrial conditions has been carried out for the synthesized samples using a pilot two-reactor adsorption plant (PAP) [8]. Nine (9) adsorption–regeneration cycles were performed for each sample. At set pressure and space velocity values, gas with 100% relative humidity by water had been supplied for drying at an operating temperature of adsorption (22–27 ◦C): (1) P = 30 atm; 3 3 V(N2) = 16 m /h; V(H2O) = 11–15 mL/h; (2) P = 30 atm; V(N2) = 10 m /h; and V(H2O) = 7 mL/h. When the dew point temperature (DPT) dropped to −40 ◦C, the adsorption process was completed 3 and the plant switched automatically to the adsorbent regeneration mode (P = 25 atm, V(N2) = 2 m /h, t 360 min). The final desiccants are characterized by high stability in multiple adsorption–desorption cycles and the possibility of reaching a minimum dew point temperature of −80 ◦C in the course of drying. Samples modified with alkaline metals are characterized by a greater dynamic capacity compared to the unmodified ones [8]. A series of studies of the physical and chemical properties of the samples obtained was conducted before and after the nine cycles of vapour adsorption–desorption were conducted. Materials 2018, 11, 132 3 of 10

Sodium and potassium contents in samples were determined with inductively coupled plasma (ICP-MS) using the Agilent 7500cx (Agilent, Santa Clara, CA, USA). Thermal and gravitational (TG) tests of aluminium were performed in oxidizing medium using the NETZSCH STA 409 synchronous thermal analysis machine. During analysis, the heating rate and exposure time at the selected temperature were varied. Textural characteristics of the CTA GS product and adsorbents were determined by isotherms of adsorption at 77 K using the Asap 2400 sorptometer (Micromeritics, Norcross, GA, USA). Specific surface area was measured using the Brunauer-Emmett-Teller (BET) method, and the micropore volume using the t-method. Mesopore volume was determined by analyzing the integral pore volume distribution curve depending on radius (along adsorption branch); average pore diameter (in nm) was determined with the equation dave = 4000Vpore/A, where A is the granule surface area [10]. Determination of the crush strength of samples was carried out using a catalyst strength tester, Lintel PK-21. The study of acid–base properties of the surfaces of the samples was performed using the pH measurement method in conformity with the technique [11]. The measurement of the pH of the suspension from its formation to the achievement of electrochemical adsorption equilibrium was registered every 5–10 s according to the readings of the IPL-101 meter and a pH-meter (pH 673 M) using and standard chloride electrodes. The values of pH at 5, 10, and 15 s of the sample coming into contact with water and the pH of isoionic state of matter (pHiip), which characterizes the equilibrium state, were selected as the parameters characterizing the acid–base state of the surface. The dynamic method was used in this study to observe the adsorption of water vapours. The adsorption value was determined by the weight method using a McBenna-Bakr spring scale with a sensitivity of 2.9 × 10−3 g/mm. Prior to the performance of the adsorption measurements, each sample was trained (regenerated) at 220–240 ◦C in a nitrogen flow (extra pure grade nitrogen of impurity content not more than 10 ppm) supplied for 1 h at 5 L/h. Regeneration at temperatures up to 250 ◦C is considered to be acceptable [12], since it to the minimum reduction in the time of useful use of the adsorbent. Wet nitrogen (moisture content >80%) was supplied to the sample to perform adsorption of water vapours. The elongation of the spiral was fixed with a V-630 cathetometer (Instrument-making plant, Kharkiv region, Izyum, Ukraine). All experiments for the study of water vapour kinetics were performed at 26 ◦C and at atmospheric pressures on the aluminium oxide samples with a fraction of 0.5–1.0 mm. To exclude an impact of the advance speed of a substance on an outer granule surface, a series of experiments on water vapour adsorption at different gradually increasing flow rates up to 36/h had been conducted beforehand. It was found that if the carrier gas velocity was equal to or exceeded 27 L/h, the kinetics of water vapour adsorption did not depend on the nitrogen delivery rate. It was found the optimal adsorbent weighed amount (0.02–0.03 g) allowed us to place adsorbent granules with a fraction of 0.5–1.0 mm in a layer in a foil cup. After the completion of adsorption and the establishment of adsorption balance, dry nitrogen was supplied to the sample at 10 L/h, and the changes in the weight of the sample were recorded at preset time intervals after the water desorption.

3. Results

3.1. Chemical Composition of Desiccants According to the ICP-MS results, the sample A-1 contains ~1% of sodium, which is explained by the fact that NaOH was used as an electrolyte at the hydration stage, and potassium traces were also found (Table1). In the sample A-2, the content of sodium cations is much less, and in the modified samples A-3-Na and A-4-K, the content of the relevant modifying cation is ~2% by weight. The results of the samples before and after nine adsorption–regeneration cycles with the thermogravimetric analysis (TGA) and differential thermal analysis (DTA) methods are given in Table1. According to the literature [13], the process of water extraction from Al2O3 is carried out step-by-step. At the first stage Materials 2018, 11, 132 4 of 10

(up to 100 ◦C), the extraction of physically adsorbed water takes place. The extraction of chemically Materials 2018, 11, 132 4 of 10 bound water takes place in the second (100–350 ◦C) and third (350–550 ◦C) stages. The removal of the residualstage quantity (up to 100 of water °С), the (quantity extraction of hydroxyls of physically remaining adsorbed on water the alumina’s takes place. surface The extraction equals ~0.1% of by weight)chemically takes place bound at water the fourth takes place stage in at the >550 second◦C. In(100–350 the reference °С) and samplethird (350–550 A-1, one °С) minimumstages. The DTA curve,removal 113 ◦C, of is the observed residual quantity in the area of water of the (quantity extraction of hydroxyls of physically remaining adsorbed on the water, alumina’s and surface at a higher temperatureequals ~0.1% (T = 505by weight)◦C), there takes is place a peak at thatthe fourth is associated stage at with>550 the°С. In degradation the reference of sample the А‐1, one phase. Accordingminimum to the DTA thermographs curve, 113 °С of, is the observed samples in the A-2 area and of A4-K, the extraction water is of being physically extracted adsorbed already water, at 52 ◦C and at a higher temperature (Т = 505 °С), there is a peak that is associated with the degradation of the and the maximum extraction level occurs at 136–153 ◦C. There are no peaks in high-temperature area boehmite phase. According to the thermographs of the samples А‐2 and А4‐К, water is being for these samples. extracted already at 52 °С and the maximum extraction level occurs at 136–153 °С. There are no peaks in high‐temperature area for these samples. Table 1. DTA and TGA data of the alumina samples. Table 1. DTA and TGA data of the alumina samples. The Weight Loss on the TG Curve, % Tmax TPT Sample ◦ The Weight◦ Loss on the TG◦ Curve, % ◦ ◦ ◦ To 240 C From 240 C to From 350 C to From 550 C to (DTA),ТmaxC (DTA),ТPTC Sample To 240 °С From◦ 240 °С to From◦ 350 °С to From◦ 550 °С to (∆m1) 350 C(∆m2) 550 C(∆m3) 900 C(∆m4) (DTA), °С (DTA), °С (Δm1) 350 °С (Δm2) 550 °С (Δm3) 900 °С (Δm4) A-1 4.8 1.0 2.2 2.0 113 505 А‐A-21 2.54.8 1.01.0 1.32.2 1.92.0 (52) 140113 –505 A-3-NaА‐2 2.32.5 0.91.0 1.01.3 1.61.9 (52)(52) 153 140 – – А‐A-4-K3‐Na 2.52.3 0.90.9 0.91.0 1.81.6 (52)(52) 136 153 – – A-1-9CА‐4‐K 6.92.5 0.80.9 3.00.9 1.71.8 (52) 86 136 504– A-2-9C 5.3 0.9 2.9 1.4 85 436 А‐1‐9C 6.9 0.8 3.0 1.7 86 504 A-3-Na-9C 9.1 1.0 2.5 1.3 81 445 A-4-K-9CА‐2‐9C 10.65.3 0.80.9 2.42.9 1.31.4 8185 444436 А‐3‐Na‐9C 9.1 1.0 2.5 1.3 81 445 ∆m1—weight change determined by TGA at a heating rate of 10◦/min to 240 ◦C integration; ∆m2—weight change А‐4‐K‐9C 10.6 0.8 2.4 1.3 81 444 determined by TGA at a heating rate of 2◦/min to 350 ◦C and holding 1 h; ∆m3—weight change determined by TGAΔm1—weight at a heating change rate of determined 10◦/min to by 900 TGA◦C at integration. a heating rate TG: of thermal 10°/min and to 240 gravitational. °С integration; TPT: ∆m2— The Phase Transitionweight Temperature. change determined by TGA at a heating rate of 2°/min to 350 °C and holding 1 h; Δm3—weight change determined by TGA at a heating rate of 10°/min to 900 °С integration. TG: thermal and The thermalgravitational. and ТPT: gravitational The Phase Transition (TG) and Temperature DTA curves for all four samples appeared to be similar after the multicyclicThe thermal tests.and gravitational The presence (TG) of and the DTA peaks curves at 436–444 for all four◦C samples and 504 appeared◦C is observed to be similar for these samplesafter on the the multicyclic DTA curve, tests. which The presence is associated of the peaks with the at 436–444 pseudoboehmite °С and 504 and°С is boehmite observed for degradation, these respectivelysamples on [14 the]. This DTA factcurve, conforms which is associated to the results with the of pseudoboehmite the X-ray phase and analysis. boehmite degradation, The assessment afterrespectively testing showed [14]. This that fact up toconforms 5% of theto the pseudoboehmite results of the X‐ray phase phase is formedanalysis. inThe samples assessment A-2, after A-3-Na, testing showed that up to 5% of the pseudoboehmite phase is formed in samples А‐2, А‐3‐Na, and and A-4-K. A‐4‐K. 3.2. Textural Characteristics of Adsorbents 3.2. Textural Characteristics of Adsorbents 2 All ofAll the of samples the samples studied studied had had comparable comparable specific specific surface surface areas areas before before testing testing (within (within 280–310 280–310 m /g). The averagem2/g). The adsorbent average adsorbent particle size particle is 100 size nm. is 100 After nm. testing After testing (nine adsorption(nine adsorption−desorption−desorption cycles), the specificcycles), surfacethe specific area surface reduces, area and reduces, this and becomes this becomes especially especially noticeable noticeable for the for samplethe sample A-2 А‐ (3082 to 201 m(3082/g) to (Table 201 m2/g), Figure (Table1 ).2, Figure 1).

(a) (b)

Figure 1. Cont. Materials 2018, 11, 132 5 of 10 Materials 2018, 11, 132 5 of 10

(c)

Figure 1. Values of SBET (а); pore size distribution of samples А‐1, А‐2, A‐3‐Na, and А‐4‐K (b); and А‐ Figure 1. Values of SBET (a); pore size distribution of samples A-1, A-2, A-3-Na, and A-4-K (b); and 1‐9C, А‐2‐9C, A‐3‐Na‐9C, and А‐4‐K‐9C (c). A-1-9C, A-2-9C, A-3-Na-9C, and A-4-K-9C (c). The pore volume for sample А‐1 had the least value (0.27 cm3/g) at the pore diameter of 3.3 nm, Tableand it 2. wasTexture reduced characteristics, even more the after crush the strength completion values, of the and sorption modifications−regeneration to the content cycles. of The additives pore involume the alumina was slightly samples higher before for and the aftersample pressure А‐2 (0.32 tests. cm3/g) at the pore diameter of 3.7 nm and it also was a bit smaller after testing. The pore volume for the samples А‐3‐Na and А‐4‐К modified with Na andCharacteristics K was >~0.40 cm3/g A-1 at the pore A-2 diameter A-3-Na of 5.0 and A-4-K 4.9 nm, A-1-9C respectively. A-2-9C After A-3-Na-9C the cycle A-4-K-9C tests of 2 waterSBET adsorption–desorption,± ∆, m /g 279 ± 27 the 308 pore± 30 volume 312 ± 31remained 288 ± 28nearly 265 the± 26 same 201 in± these20 samples, 230 ± 23 while 233 ± the23 The crush strength, MPa 25 25 13 17 23 24 14 13 diametersContent of modifyingincreased slightly0.98 Na; to 6.30.12 and Na; 6.2 2.10nm, Na; respectively.0.11 Na; 0.97 Na; 0.11 Na; 2.00 Na; 0.13 Na; additives, % mass. 0.04 K 0.02 K 0.03 K 2.20 K 0.04 K 0.01 K 0.02 K 2.10 K Table 2. Texture characteristics, the crush strength values, and modifications to the content of additives in the alumina samples before and after pressure tests. The pore volume for sample A-1 had the least value (0.27 cm3/g) at the pore diameter of 3.3 nm, and it wasСharacteristics reduced even А‐ more1 А‐ after the2 А‐ completion3‐Na А‐ of4‐K the sorptionА‐1‐9C −А‐regeneration2‐9C А‐3‐Na‐9C cycles. А‐4‐K The‐9C pore SBET ± Δ, m2/g 279 ± 27 308 ± 30 312 ± 31 288 ± 28 265 ± 26 201 ± 20 230 ± 23 233 ± 23 3 volumeThe was crush slightly strength, higherMPa for25 the sample25 A-213 (0.32 cm17 /g) at23 the pore24 diameter 14 of 3.7 nm13 and it also was a bitContent smaller of modifying after testing. 0.98 Na; The pore0.12 Na; volume 2.10 Na; for the0.11 Na; samples 0.97 Na; A-3-Na 0.11 andNa; A-4-K2.00 Na; modified 0.13 Na; with Na and K wasadditives, >~0.40 % mass. cm3 /g at0.04 the К pore0.02 diameter К 0.03 К of 5.02.20 and K 4.9 nm,0.04 К respectively.0.01 К After0.02 К the cycle2.10 K tests of water adsorption–desorption, the pore volume remained nearly the same in these samples, while the The pore distribution by size for each of the desiccant samples is given in Figure 1. The original diameters increased slightly to 6.3 and 6.2 nm, respectively. samples are characterized by the presence of fine mesopores in the structure, which are in the narrow The pore distribution by size for each of the desiccant samples is given in Figure1. The original distribution interval of 3.0–4.5 nm. After the multicyclic tests, the share of fine mesopores decreased samplesand arelarger characterized pores of 4.5–7.0 by thenm presenceappeared. ofA finebimodal mesopores distribution in the of the structure, second whichin the interval are in theof 5–7 narrow distributionnm was intervalobserved of for 3.0–4.5 the samples nm. After (A‐1 and the A multicyclic‐4‐К) after nine tests, cycles the sharewere completed. of fine mesopores The monitored decreased and largerchanges pores in porous of 4.5–7.0 desiccant nm appeared. structures Acould bimodal be conditioned distribution by the of the /“healing” second in the interval processes of 5–7of nm was observedthe smallest for pores the during samples regeneration. (A-1 and A-4-K) after nine cycles were completed. The monitored changes inMechanical porous desiccant and physical structures properties could of the be samples conditioned were bystudied. the sintering/“healing” It was demonstrated processesthat the of the smallestbulk pores of during the samples regeneration. А‐1 and А‐2 was 0.87 g/cm3 and 0.83 g/cm3, respectively. The modified 3 Mechanicalaluminium oxides and physical are 0.73–0.74 properties g/cm . The of thecrush samples strength were of the studied. non‐modified It was samples demonstrated А‐1 and А‐ that2 the (Table 2) was slightly higher compared to the samples modified with Na and K. After the cycle tests, bulk density of the samples A-1 and A-2 was 0.87 g/cm3 and 0.83 g/cm3, respectively. The modified the crush strength of the samples did not change within the range of the calculated error. The crush aluminium oxides are 0.73–0.74 g/cm3. The crush strength of the non-modified samples A-1 and A-2 strength of the aluminium oxide samples studied was comparable to the strength of industrial (Tablezeolites2) was of slightly the KA highertype used compared for APG drying. to the samples modified with Na and K. After the cycle tests, the crush strength of the samples did not change within the range of the calculated error. The crush strength3.3. Acid–Base of the aluminium Properties oxideof Alumina samples Desiccants studied was comparable to the strength of industrial zeolites of the KAThe type moisture used for adsorption APG drying. and water vapour adsorption are determined by the acid–base properties of the surface. It is known that pH change in the aqueous suspension of the sample within 3.3. Acid–Base Properties of Alumina Desiccants a period of time proceeds for each individual substance with an intrinsic trend of kinetic curves and Thewith moisture their different adsorption arrangement and water relative vapour to the adsorptionacidity level are of the determined original electrolyte by the acid–base (pH0) [15]. properties The of theacid–base surface. parameter It is known (рН thatof pHisoionic change state in of the matter aqueous (рНiip)) suspension had been of determined the sample from within the curves, a period of time proceeds for each individual substance with an intrinsic trend of kinetic curves and with their different arrangement relative to the acidity level of the original electrolyte (pH0)[15]. The acid–base parameter (pH of isoionic state of matter (pHiip)) had been determined from the curves, showing the Materials 2018, 11, 132 6 of 10

showing the dependence of the рН of the desiccant samples’ aqueous suspension on the period of time after the establishment of the balance. The parameter corresponds to the рН value at which, when different ions are available in solution, the equal adsorption of acid and base groups is Materialsestablished2018, 11, 132on the body surface. This parameter characterizes the relative content of acid and6 of 10 base sites on the solid body surface. The kinetic curves of the pH change of the original samples and samples after nine water dependenceadsorption of theand pHdesorption of the desiccantcycles for each samples’ sample aqueous are given suspension in Figure 2. on It can the be period observed of time that afterthe the establishmentdifferences of in the the balance.studied samples The parameter by suspension corresponds basicity toare the manifested pH value already at which, during when the differentfirst ions areseconds available of contact. in solution, The arrangement the equal of adsorption kinetic curves of acid for andall of base the samples groups is above established the neutrality on the solid bodylevel, surface. and abrupt This parameter alkalizing at characterizes initial time, an the increase relative of рН content10″, рН of15″ acidvalues, and and base the higher sites on rate the of solid bodyрН surface.change (Figure 2) indicate the presence of potent base sites that are aprotic on their surfaces. TheThe kinetic рН value curves of the of thesuspension pH change when there of the is longer original contact samples with the and aqueous samples medium after reflects nine water adsorptionthe slowly and proceeding desorption processes cycles forof dilution, each sample hydration, are and given hydrolysis in Figure of2 aluminium. It can be cations observed with that further establishment of acid−base equilibrium in the system and is characterized by the рНiit value. the differences in the studied samples by suspension basicity are manifested already during the first The tests demonstrate that an elongation of the period of contact of the samples of Al2O3 with water seconds of contact. The arrangement of kinetic curves for all of the samples is above the neutrality to 1 h and longer does not to further change in the pH of the suspension. The wide plateau at level,50–250 and abrupt s at considerably alkalizing high at initial pH values time, proves an increase the prevalence of pH10” of ,potent pH15” Bronstedvalues, sites and of the base higher nature rate of pH changeon the aluminium (Figure2) indicate oxide’s surface. the presence of potent base sites that are aprotic on their surfaces.

Figure 2. Kinetic curves of the pH of the aqueous suspensions of the initial samples’ dryers and Figure 2. Kinetic curves of the pH of the aqueous suspensions of the initial samples’ dryers and dehumidifiers after nine cycles of adsorption–regeneration, ΔрН15: change in pH of the sample for 15 dehumidifiers after nine cycles of adsorption–regeneration, ∆pH : change in pH of the sample for 15 s s of contact with water. 15 of contact with water. 3.4. Kinetics of Water Vapour Adsorption on Alumina Desiccants The pHThe valuewater adsorption of the suspension dependence when on the there structure is longer and contact strength with of surface the aqueous sites may medium correspond reflects the slowlyto different proceeding kinetic parameters. processes The of dilution, kinetic interaction hydration, parameters and hydrolysis with water of vapours aluminium are needed cations to with furthercalculate establishment the hydrodynamic of acid− baseparameters equilibrium of industrial in the adsorbents system and and is characterizedthe optimal granule by the shapes pHiit of value. The testsalumina demonstrate adsorbents. that The an adsorbate elongation saturation of the of period each adsorbent of contact particle of the in samples the adsorber of Al depends2O3 with on water to 1 hthe and diffusion longer rate does of the not absorbed lead to furthermolecules change inside of in the the granules, pH of the which suspension. finally determines The wide the plateau mass at 50–250transfer s at considerably rate under a highparticular pH values hydrodynamic proves the mode. prevalence The kinetics of potent of water Bronsted vapor were sites studied of base to nature on theestimate aluminium the rate oxide’s of change surface. in the adsorption capacity of the granules in single time. Studies of water vapour kinetics were conducted to assess the rate of change with time in the adsorption capacity of 3.4. Kineticssingle granules. of Water Vapour Adsorption on Alumina Desiccants From the kinetic curves of water vapour adsorption on samples А‐1, А‐2, A‐3‐Na, and А‐4‐K Thegiven water in Figure adsorption 3, one dependencecan see that onthe theadsorption structure capacity and strength by water of surfacevapours sites under may the correspond given to differentconditions kinetic is higher parameters. for the samples The kinetic А‐3‐Na interaction and А‐4‐K parameters modified with with sodium water and vapours potassium are needed ions to calculatecompared the hydrodynamic to the original parameterssample А‐2 and of industrial the bayerite adsorbents‐based sample and ( theА‐1). optimal Moreover, granule the kinetic shapes of aluminacurves adsorbents. of the samples The with adsorbate admixtures saturation have only of eacha slight adsorbent difference. particle Unlike inthe the unmodified adsorber samples, depends on the diffusion rate of the absorbed molecules inside of the granules, which finally determines the mass transfer rate under a particular hydrodynamic mode. The kinetics of water vapor were studied to estimate the rate of change in the adsorption capacity of the granules in single time. Studies of water vapour kinetics were conducted to assess the rate of change with time in the adsorption capacity of single granules. From the kinetic curves of water vapour adsorption on samples A-1, A-2, A-3-Na, and A-4-K given in Figure3, one can see that the adsorption capacity by water vapours under the given conditions is higher for the samples A-3-Na and A-4-K modified with sodium and potassium ions compared to the Materials 2018, 11, 132 7 of 10

originalMaterials 2018 sample, 11, 132 A-2 and the bayerite-based sample (A-1). Moreover, the kinetic curves of the samples7 of 10 with admixtures have only a slight difference. Unlike the unmodified samples, these adsorbents are Materials 2018, 11, 132 7 of 10 characterizedthese adsorbents by anare irreversibility characterized effect: by an the irreversibility curves do not effect: come the to curves the initial do not state come after to desorption the initial (thestatethese difference after adsorbents desorption is ~8% are by(thecharacterized weight). difference This by is couldan ~8 irreversibility% by be explainedweight). effect: This by thecouldthe presencecurves be explained do of not the come Lewis by theto basethe presence initial sites on of thethestate surfaceLewis after base of desorption these sites compounds, on (the the difference surface which of is are these~8% capable by compounds, weight). of strongly This which could adsorbing are be explainedcapable water. of by During strongly the presence the adsorbing repeated of tests,water.the waterLewis During interactionbase the sites repeated on with the tests, thesurface surface water of theseinteraction passes compounds, through with the thewhich surface formation are passescapable of through of strongly the bonds adsorbingformation and the of processhydrogenwater. becomes During bonds the reversible. and repeated the process It tests, was water provenbecomes interaction for reversible. the sample with Itthe A-4-K was surface proven that passes no for changes throughthe sample were the observed formation А‐4‐K that in of the no trendchangeshydrogen of kinetic were bonds observed curves and for the in a processthe training trend becomes period of kinetic elongation reversible. curves of forIt thewas a training sample proven to periodfor 2 h the and elongationsample a temperature А‐ 4of‐K the that increase sample no totochanges 2502 h ◦andC. werea temperature observed inincrease the trend to 250of kinetic °С. curves for a training period elongation of the sample to 2VariousVarious h and a kinetic temperature models models increase were were toconsidered considered250 °С. for for the themathematical mathematical description description of the of kinetic the kinetic curves curves[16,17].Various [ 16It, 17was]. Itkinetic established was establishedmodels that were thatin considered the in theregion region for of the a of mathematicalsurface a surface coverage coverage description ratio ratio of of >0.4–0.6, >0.4–0.6,the kinetic the curves kinetickinetic adsorptionadsorption[16,17]. It curves curveswas established of of every every sample sample that in are arethe rectified rectified region (Figure of (Figure a surface4) in4) thein coverage the coordinates coordinates ratio of of the of>0.4–0.6, the adsorption adsorption the kinetic value value (a) and(a)adsorption and the logarithmthe logarithmcurves of of time every of (lntime sample t). In(ln this aret). article, Inrectified this wearticle, (Figure used we the4) inused method the coordinatesthe of method least squares of of the least adsorption with squares a correlation value with a coefficientcorrelation(a) and the of coefficient 0.993–0.992.logarithm of of 0.993–0.992. Thetime experimental(ln t). TheIn this experimental dataarticle, are we rectified dataused arethe in rectifiedmethod the given ofin coordinates,leastthe given squares coordinates, whenwith a the adsorptionwhencorrelation the adsorption kinetics coefficient are kinetics expressedof 0.993–0.992. are expressed by the The Roginsky–Zeldovich experimental by the Roginsky–Zeldovich data are equation: rectified equation: in the given coordinates, when the adsorption kinetics are expressed by the Roginsky–Zeldovich equation: q = B ln t + C (1) qq = = BB ln t + C C (1) (1) where a is the amount of adsorbed gas and B and C are constants. wherewhere a isa is the the amount amount of of adsorbed adsorbed gas gas and and BB andand C are constants.

FigureFigureFigure 3.3. 3. Kinetic Kinetic curvescurves curves ofof of thethe the adsorption adsorption andand desorption desorption of ofof water waterwater vapor vaporvapor on onon aluminum aluminumaluminum oxide oxideoxide samples samplessamples forforfor aa fractionafraction fraction ofof of 0.5–1.00.5–1.0 0.5–1.0 mm mm mm (conditions: (conditions:(conditions: carriercarrier gasgas adsorptionadsorption rate rate rate at at at 30 30 30 L/h, L/h, L/h, desorption desorption at 10 at 10L/h). L/h).

(a) (b) (a) (b) Figure 4. Approximations of the kinetic curves sorption of water vapor on the samples A‐1, А‐2, A‐3‐ FigureFigureNa, and 4.4. Approximations ApproximationsA‐2‐K: (a) the value of of the theof kinetic adsorption kinetic curves curves (q): sorption logarithm sorption of ofwaterof watertime vapor (ln vapor t); on (b the on) logarithm samples the samples Aof‐ 1,value А‐ A-1,2, of A-2,A ‐3‐ A-3-Na,Na,adsorption and and A‐2 A-2-K:(ln‐K: q)(a time) (thea) the(t). value value of ofadsorption adsorption (q): (q): logarithm logarithm of of time time (ln (ln t); t); ( (bb)) logarithm logarithm of valuevalue ofof adsorptionadsorption (ln(ln q)q) timetime (t). (t). The Roginsky–Zeldovich equation is typical for those processes run on proportionally non‐ uniformThe Roginsky–Zeldovich surfaces, and also for multicenterequation is chemosorption typical for those on a processesuniform surface run on in aproportionally certain region ofnon ‐ uniformthe average surfaces, surface and coverages. also for multicenter The factor B chemosorptionwas found, confirming on a uniform the nature surface of adsorbate in a certain interaction region of thewith average the adsorbent surface coverages. surface (Table The 3). factor Increase B was in found,the factor confirming B was observed the nature for the of adsorbatemodified samples, interaction with the adsorbent surface (Table 3). Increase in the factor B was observed for the modified samples,

Materials 2018, 11, 132 8 of 10

The Roginsky–Zeldovich equation is typical for those processes run on proportionally non-uniform surfaces, and also for multicenter chemosorption on a uniform surface in a certain region of the average surface coverages. The factor B was found, confirming the nature of adsorbate interaction with the adsorbent surface (Table3). Increase in the factor B was observed for the modified samples, which is associated with the growth of interaction energy between adsorbate and adsorbent and, therefore, with the availability of potent surface base sites.

Table 3. Characteristics of the kinetic curves of the adsorption and desorption of water vapor on the alumina samples.

Sample qm, g/g ads B k A-1 0.177 0.053 0.0752 A-2 0.197 0.063 0.0603 A-3-Na 0.281 0.086 0.0596 A-4-K 0.281 0.085 0.0583

qm: the maximum amount of water vapor adsorbed on the sample; B and k: coefficients of approximating the kinetic curves of adsorption and desorption, respectively.

The theoretical fundamentals of the desorption stage are developed poorly in comparison with the kinetics theory and even adsorption dynamics, so the majority of research proposes empirical decisions for the special cases containing a number of assumptions [1]. Since the external mass transfer does not have a significant impact on the processes overall at relatively high gas flow rates, and the kinetic desorption curves received in the course of gas purging tests are close to the kinetic curves in vacuum, then it is possible to apply the equation proposed in the study [1]:

ln (q/qmax) = −k × t (2) where qmax is the maximum quantity of the adsorbed substance and k is a factor (adsorption rate). According to the theory, the factor k is proportional to exp (Ed/RT), where T is the temperature in K, R is the ideal-gas constant. Ed is the activation energy of the desorption process bound with the adsorption heat with the equation: Ed = Q + Ea (3) where Q is the adsorption heat and Ea is the adsorption activation energy. The desorption kinetics curves in the region of the average coverages are rectified in the coordinates ln q−t (Figure4). The calculated factors B and K given in Table3 show that the factors B and K as well as the maximum adsorption value for the samples A-3-Na and A-4-K are close to each other. Factor B is increased in the line of A-1, A-2, A-4-K, and A-3-Na, and the factor k is decreased, which confirms an improvement in the bonding strength of adsorbed molecules with the surface. The maximum value of water vapour adsorption increases proportionally for the samples received in the course of study of the water vapour adsorption kinetics.

4. Discussion A comprehensive study of desiccants based on aluminium oxide modified with alkaline metals, including their physicochemical, mechanical, and physical properties and water vapour adsorption kinetics, has been conducted. The samples were studied before and after nine water adsorption–desorption cycles under maximally realistic conditions to the industrial parameters during the drying and transportation of the associated petroleum gas. It was found that the samples of bayerite- and pseudoboehmite-based alumina desiccants have potent base sites on their surface. The dynamic capacity of adsorbents with respect to water vapours increases if modified with alkali ions against the background of an enlargement of the average diameter and pore volume. This is associated with the growth of the concentration and strength of surface base sites due to the effect of alkaline cations [18]. Materials 2018, 11, 132 9 of 10

As soon as these samples were used in the nine adsorption–regeneration cycles at the pilot adsorption plant under the pressure of 30 atm, the phase composition of samples A-2, A-3-Na, and Materials 2018, 11, 132 9 of 10 A-4-K underwent minor changes: an additional pseudoboehmite phase, a pore size enlargement, asamples decrease А‐ in1 and fine А‐ pore4‐K quantity, before and and after a reduction their use inas thedesiccants specific coincides, surface area. i.e., The the total specific surface surface basicity area ofdid the not samples change. A-2 It can and be A-3-Na noted that is reduced quite a high more stability intensively of these and samples an increase was in observed the surface during acidity the ofadsorption–regeneration the samples was observed cycles after at the nine pilot adsorption–regeneration plant. cycles. The value of pHiit for the samplesThe A-1 testing and A-4-Kof the before sample and with after the their TGA use asand desiccants DTA methods coincides, demonstrated i.e., the total surfacethat the basicity peak diddisplacements not change. of It physically can be noted adsorbed that quite water a high to a lower stability temperature of these samples region (81–86 was observed °С) were during observed the adsorption–regenerationfor the samples used in cyclesmulticyclic at the drying pilot plant. with an increase in the volume of water removed at temperaturesThe testing up to of 240 the °С sample. A greater with amount the TGA of water and DTAreleased methods at these demonstrated temperatures thatwas theobserved peak ◦ displacementsin the modified of samples physically А‐3 adsorbed‐Na‐9C and water А‐4‐ toK‐ a9C lower having temperature a larger volume region and (81–86 poreC) diameter, were observed which forconfirms the samples the larger used water in multicycliccapacity of these drying samples. with an increase in the volume of water removed at ◦ temperaturesThe adsorption up to 240 kineticsC. A of greater all the amountsamples ofin waterthe region released of average at these coverage temperatures are expressed was observed by the inRoginsky–Zeldovich the modified samples equation, A-3-Na-9C which and characterizes A-4-K-9C having adsorption a larger with volume a proportionally and pore diameter, non‐uniform which confirmssurface. The the largerdesorption water is capacity carried ofout these from samples. the uniform surface. The water vapour adsorption on activeThe aluminium adsorption oxide kinetics is ofthe all thecombined samples result in the regionof three of averageprocesses: coverage chemosorption, are expressed physical by the Roginsky–Zeldovichadsorption, and capillary equation, condensation. which characterizes As follows from adsorption the data, with the a first proportionally layer is formed non-uniform on active surface.surface sites The of desorption aluminium is oxide carried due out to from dissociation the uniform chemosorption surface. The of water molecules. vapour adsorption The second on activeand the aluminium next layers oxide are is adsorbed the combined physically result ofby three means processes: of van chemosorption,der Waals strength. physical The adsorption, capillary and capillary condensation. As follows from the data, the first layer is formed on active surface sites of condensation at a relative vapour pressure below one P/P0 and at a temperature above the dew point aluminiumof the liquid oxide is typical due to dissociationfor this adsorbent. chemosorption The factors of water of the molecules. adsorption The and second desorption and the next equations layers arecalculated adsorbed are physically connected by with means the ofbinding van der energy Waals of strength. the surface The sites’ capillary water condensation adsorption. atThere a relative were vapourhigher pressureadsorption below factor one values P/P0 and set atin a the temperature Roginsky–Zeldovich above the dew equation point of for the liquidaluminium is typical oxides for thismodified adsorbent. with alkali The factors metal ofcations. the adsorption and desorption equations calculated are connected with the binding energy of the surface sites’ water adsorption. There were higher adsorption factor values set5. Conclusions in the Roginsky–Zeldovich equation for aluminium oxides modified with alkali metal cations.

5. ConclusionsAs a result of the research, it was found that the modification of an alumina‐based desiccant by alkaline cations makes it possible to increase its efficiency during the adsorption of water vapor. As As a result of the research, it was found that the modification of an alumina-based desiccant by a result of alkaline modification, a decrease in the specific surface area and an increase in the average alkaline cations makes it possible to increase its efficiency during the adsorption of water vapor. As a pore diameter were observed. The study of the kinetics of the adsorption of water vapor showed that result of alkaline modification, a decrease in the specific surface area and an increase in the average the highest adsorption rate is characteristic of adsorbents A‐1 and A‐2. Despite this fact, a high pore diameter were observed. The study of the kinetics of the adsorption of water vapor showed adsorption capacity (~0.28 g/g ads) was characteristic of the modified samples. It was revealed that that the highest adsorption rate is characteristic of adsorbents A-1 and A-2. Despite this fact, a high the multicyclic pressure tests do not significantly affect the functional characteristics of the alumina adsorption capacity (~0.28 g/g ads) was characteristic of the modified samples. It was revealed that adsorbent samples. the multicyclic pressure tests do not significantly affect the functional characteristics of the alumina adsorbentAcknowledgments: samples. The work was financially supported by project RFMEFI57517X0139 of the Ministry of education and science of Russia. Acknowledgments: The work was financially supported by project RFMEFI57517X0139 of the Ministry of educationAuthor Contributions: and science of Ruslan Russia. Zotov and Lyubov Isupova conceived and designed the experiments; Eugene Meshcheryakov and Alesia Livanova performed the experiments and analyzed the data; Tamara Minakova, AuthorOleg Magaev Contributions: and Irina KurzinaRuslan wrote Zotov the andpaper. Lyubov Isupova conceived and designed the experiments; Eugene Meshcheryakov and Alesia Livanova performed the experiments and analyzed the data; Tamara Minakova, OlegConflicts Magaev of Interest: and Irina The Kurzina authors wrote declare the no paper. conflict of interest. Conflicts of Interest: The authors declare no conflict of interest. References

References1. Keltsev, N.V. Fundamentals of Adsorption Technology; Chemistry: Moscow, Russian, 1984, 595. 1.2. Keltsev,Danilevich, N.V. V.V.;Fundamentals Isupova, of L.A.; Adsorption Kagyrmanova, Technology A.P.;; Chemistry: Kharina, Moscow, I.V.; Zyuzin, Russian, D.A.; 1984; Noskov, p. 595. A.S. Highly 2. Danilevich,Effective Water V.V.; Isupova, Adsorbents L.A.; Kagyrmanova, Based on A.P.;Aluminum Kharina, I.V.;Oxide. Zyuzin, Kinet. D.A.; Catal. Noskov, 2012 A.S., Highly53, Effective632–639, Waterdoi:10.1134/S0023158412050059. Adsorbents Based on Aluminum Oxide. Kinet. Catal. 2012, 53, 632–639. [CrossRef] 3. Dabrowski,̧ , A. Adsorption—From theorytheory toto practice.practice. Adv.Adv. ColloidColloid Interface Interface Sci. Sci.2001 2001,,93 93,,135–224. 135–224. [CrossRef] 4. Sircar, S.;S.; Rao,Rao, M.B.;M.B.; Golden,Golden, T.C.T.C. DryingDrying ofof gasesgases andand liquidsliquids byby activatedactivated alumina.alumina. Stud.Stud. Surf. Sci. Catal. 1996,, 99,,629–646. 629–646. 5. Zotov, R.A.; Glazyrin, A.A.; Danilevich, V.V.; Kharina, I.V.; Zyuzin, D.A.; Volodin, A.M.; Isupova, L.A. Influence of the Temperature of Calcination of Bayerite‐Containing Aluminum Hydroxide Pellets on the Water Vapour Adsorption Capacity and Acid‐Base Properties of Alumina. Kinet. Catal. 2012, 53, 570–576, doi:10.1134/S0023158412050187. 6. Danilevich, V.V.; Isupova, L.A.; Paukshtis, E.A.; Ushakov, V.A. Effect of modifying desiccants with sulfuric acid on their physiochemical properties. Kinet. Catal. 2014, 55, 372–379, doi:10.1134/S0023158414030021.

Materials 2018, 11, 132 10 of 10

5. Zotov, R.A.; Glazyrin, A.A.; Danilevich, V.V.; Kharina, I.V.; Zyuzin, D.A.; Volodin, A.M.; Isupova, L.A. Influence of the Temperature of Calcination of Bayerite-Containing Aluminum Hydroxide Pellets on the Water Vapour Adsorption Capacity and Acid-Base Properties of Alumina. Kinet. Catal. 2012, 53, 570–576. [CrossRef] 6. Danilevich, V.V.; Isupova, L.A.; Paukshtis, E.A.; Ushakov, V.A. Effect of modifying desiccants with sulfuric acid on their physiochemical properties. Kinet. Catal. 2014, 55, 372–379. [CrossRef] 7. Fleming, H.L. Adsorption on aluminas—Current applications. Stud. Surf. Sci. Catal. 1999, 120, 561–585. 8. Zotov, R.A.; Isupova, L.A.; Danilevich, V.V.; Babina, A.A.; Sinelnikov, A.N.; Meshcheryakov, E.P.; Kurzina, I.A. Dynamic capacity dryers based on modified alumina at high pressure. Katal. v Promyshlennosti. 2016, 16, 6–13. (In Russian) [CrossRef] 9. Shefer, K.I. Analysis of Defects in Structures Hydroxides and Oxides of Aluminum Based on X-ray Data. Ph.D. Thesis, Institute of Catalysis Russian, Novosibirsk, Russian, 2008. 10. Fenelonov, V.B. Vvedenie v Fizicheskuyu Khimiyu Formirovaniya Supramolekulyarnoinstruktury Adsorbentov I Katalizatorov (Introduction to the Physical Chemistry of the Formation of the Supramolecular Structure of Adsorbents and Catalysts); Sib. Otd. Ross. Akad.Nauk: Novosibirsk, Russian, 2004. 11. Minakova, T.S. Adsorbzionnye Prozessy na Poverhnosti Tverdykh tel (Adsorption Processes on the Surface of ); Tomsk University: Tomsk, Russian, 2007. 12. Ducreux, O.; Lavigne, C.; Nedez, C. Air and Gas Drying with Activated Aluminas. 2011. Available online: http://www.cabestisrl.com.ar/AirGas%20Drying%20Brochure.pdf (accessed on 25 May 2016). 13. Alex, T.C. An insight into the changes in the thermal analysis curves of boehmite with mechanical activation. J. Therm. Anal. Calorim. 2014. [CrossRef] 14. Volkova, G.I.; Ivanov, V.G.; Kuharenko, O.A. The effect of synthesis conditions on the structure and properties of ultrafine aluminum oxyhydroxides. Chem. Sustain. Dev. 2005, 13, 427–432. 15. Nechiporenko, A.P. Donor-Acceptor Properties of Solid Surfaces of Oxides and Chalcogenides. Ph.D. Thesis, St. Petersburg, Russian, 1995. 16. Timofeev, D.P. Kinetics of Adsorption; AN SSSR: Moscow, Russian, 1962. 17. Krizhanovskaya, O.O.; Sinyaeva, L.A.; Karpov, C.I.; Selemenev, V.F.; Borodina, E.V.; Ressner, F. Kinetic models in describing the sorption of fat-soluble physiological of active substances with highly ordered inorganic -containing materials. Sorp. Chromatogram. Processes. 2014, 14, 784–794. 18. Danilevich, V.V.; Isupova, L.A.; Danilova, I.G.; Zotov, R.A.; Ushakov, V.A. Optimization of properties alumina adsorbents prepared from a hydrated product of centrifugal thermal activation of hydrargillite. Russ. J. Appl. Chem. 2016, 89, 289–299. [CrossRef]

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).