- Email: [email protected]
Separation of platinum and ruthenium by a sulphoxide modified polystyrene resin in laboratory column systems
Separation and Purification Technology 149 (2015) 279–287
Contents lists available at ScienceDirect
Separation and Purification Technology journal homepage: www.elsevier.com/locate/seppur
Separation of platinum and ruthenium by a sulphoxide modified polystyrene resin in laboratory column systems Michael Trautmann, Hans-Jürgen Holdt ⇑ University of Potsdam, Institute of Chemistry, Inorganic Chemistry, Karl-Liebknecht-Straße 24-25, 14476 Golm, Germany
a r t i c l e
i n f o
Article history: Received 10 March 2015 Received in revised form 13 May 2015 Accepted 21 May 2015 Available online 22 May 2015 Keywords: Solid-phase extraction Platinum group metals Sulphoxide Breakthrough curve Kinetic model
a b s t r a c t The present study deals with the adsorption performance of fixed bed columns using powdered sulphoxide modified poly(styrene-co-divinylbenzene) (d10 < 13 lm, d50 < 30 lm, d90 < 50 lm) for the separation of platinum and ruthenium from hydrochloric acidic solutions containing both metals (cPt = 20 mg/L, cRu = 10 mg/L). The influence of hydrochloric acid concentration, temperature, flow rate, flow direction, redox potential and bed height on the breakthrough characteristics was examined. Platinum was separately adsorbed mainly induced by hydrochloric acid concentration and redox potential keeping platinum as PtIV and ruthenium as RuIII. Ruthenium was separately adsorbed to 90% essentially induced by hydrochloric acid concentration, temperature and redox potential keeping platinum as PtIV and ruthenium predominantly as RuIV. Experimental data at optimised separation conditions were fitted to different kinetic models (Thomas, Yoon–Nelson, Bohart–Adams, Wolborska) to characterise the fixed bed column behaviour. Adsorption of both metals was well described by Thomas and Yoon–Nelson model with correlation coefficients R2 P 0:95 whereas Bohart–Adams and Wolborska model were less suitable. Ó 2015 Elsevier B.V. All rights reserved.
1. Introduction Handling the world energy supply is always an actual topic and becomes more and more key issue for the future of humankind. Long-term scarcity of crude oil, gas and coal associated with global environment pollution requires exploitation of renewable energy sources. The further demand of mobile devices with battery longevity leads to the development of new portable low power systems because current rechargeable batteries have reached their theoretical energy density limits. For that purpose fuel cells exhibit high efficiency and high potential for environment-friendly energy conversion leading to a significant reduction of fossil fuel use. Proton exchange membrane fuel cells (PEMFC) as well as direct liquid fuel cells (e.g. direct methanol fuel cell (DMFC), direct ethanol fuel cell (DEFC)) are prime candidates for mobile application due to their compactness and portability. According to the theoretical energy density of pressurized hydrogen (3000 Wh/L) and methanol–water mixtures (3400 Wh/L), fuel cell systems can make more energy available than Li-ion batteries (1300 Wh/L) achieving longer operating periods at the same volume [1]. A big drawback is the elemental necessity of significant amounts of costly platinum group metals (PGMs) as catalysts at which carbon supported
⇑ Corresponding author. E-mail address: [email protected] (H.-J. Holdt). http://dx.doi.org/10.1016/j.seppur.2015.05.013 1383-5866/Ó 2015 Elsevier B.V. All rights reserved.
platinum and platinum–ruthenium mixtures (1:1 atomic ratio) have emerged as best among all other catalysts [2]. Both metals are rare and mining is expensive hampering the broad commercialisation of fuel cell technology. The growing demand makes development of efficient and environmentally friendly recycling processes necessary. Catalyst recycling of exhausted fuel cells lowers manufacturing costs, reduces mining of primary sources and accordingly protects the environment. The classical route is a pyrometallurgical process. After stack dismantling, the membrane electrode assembly (MEA) is separated, shredded and incinerated followed by manifold extraction, solution and precipitation steps. Unfortunately incineration holds the adverse effect of releasing aggressive and toxic hydrofluoric acid what necessitates special linings as well as laborious gas scrubber. The hydrometallurgical route offers an alternative at which both metals are leached using strong oxidising acidic solutions [3]. Ruthenium is separated via distillation as its volatile tetroxide and platinum via precipitation or solvent extraction [4]. These processes are well established but offer disadvantages. Special design and construction materials are needed for distillation because volatile ruthenium compounds are highly explosive and extremely poisonous. Precipitation of platinum may be cheap but demands large quantities of precipitating agent and generates a lot of waste. Solvent extraction requires costly amounts of toxic and/or flammable organic solvents and consumes a lot of time due to multiple extraction steps. One useful
280
M. Trautmann, H.-J. Holdt / Separation and Purification Technology 149 (2015) 279–287
2.3. Preparation of stock solution
Fig. 1. Structure of sulphoxide modified poly(styrene-co-divinylbenzene) resin (1).
solution for the separation of platinum and ruthenium from hydrochloric acidic solutions is adsorption. Fixed bed adsorption is an efficient and effective method conducting sorption processes for industrial applications when a suitable low cost adsorbent is available. In any case, design of adsorption column needs numerous parameters to be adapted. This study deals with the solid-phase extraction (SPE) of platinum and ruthenium from artificial hydrochloric acidic solutions using powdered sulphoxide modified poly(styrene-codivinylbenzene) (d10 < 13 lm, d50 < 30 lm, d90 < 50 lm) as adsorbent (Fig. 1). Catalyst separation refers to the complex of problems which actually exist for PEMFC and DMFC recycling. This lab-scale research provides fundamental findings for prospective large-scale recovery of fuel cell catalyst similar to the contemporary industrial recycling of automobile catalyst converters. The replacement of oil addicted combustion engines by fuel cells will be inevitable due to the shortage of fossil resources. Hydrochloric acid concentration (cHCl), temperature (T), flow direction, flow rate (Q), redox potential (E0) and bed height (h) were sequentially varied and adapted to achieve separately adsorption. The experimental data at optimised conditions were fitted to Thomas, Yoon–Nelson, Bohart–Adams, Wolborska model to characterise the fixed bed column behaviour. 2. Experimental 2.1. Materials The synthesis of powdered sulphoxide modified poly(styreneco-divinylbenzene) resin 1 (1% cross-linked, sulphoxide loading: 2.2 mmol/g, d10 < 13 lm, d50 < 30 lm, d90 < 50 lm) as adsorbent was already described by us [5]. Hydrochloric acid (cHCl = 9.5 mol/L, suprapur) was obtained from Merck KGaA and used as received. Water was purified by a Milli-Q Reference ultrapure water purification system. This high purity water was used for preparation of all stock solutions and dilutions. The platinum starting material was yellow solid Na2PtCl66H2O (99.99%) from Alfa Aesar. The ruthenium starting material was black crystalline RuCl3xH2O (99.99%) from Alfa Aesar. Sodium hydroxide was used as 0.1 M NaOH Titrisol solution supplied by Merck. Anhydrous stannous(II) chloride was obtained as white crystalline SnCl2 (for synthesis) from Merck. 2.2. Instruments Fixed bed column studies were performed with SPE cartridges (Agilent technologies) 5.5 cm in height (H) and 0.5 cm in internal diameter (d). For higher temperatures a special glass coated column (self-construction) 5.5 cm in height and 0.5 cm in diameter was used. Fractions were collected by PrepFCTM fraction collector (Gilson). Different temperatures were established using a MLW thermostat U2 (VEB MLW). Different flow rates were adjusted using peristaltic pump Minipuls 3 (Gilson). Redox potential was determined using a Sentix ORP single-rod measuring cell (Pt–Ag– AgCl) containing 3 M KCl electrolyte (WTW). Infrared spectra were recorded using FTIR-Nexus (Thermo Nicolet) fitted with a Smart Orbit sample system (Diamond) for resins (transmission mode with 32 scans and a resolution of 8 cm1).
Stock solution of PtIV (cPt = 400 mg/L) was made by dissolving the appropriate amount of Na2PtCl66H2O in 9.5 M HCl (suprapur). The composition of commercial hydrated ruthenium trichloride (RuCl3xH2O) is diffuse and hydrochloric acidic solutions contain by the majority RuIV as miscellaneous monomeric as well as chloride- and oxygen-bridged polymeric complexes (e.g. [Ru(OH)nCl6n]2, [Ru2O(H2O)nCl10n]n4, [Ru2O2(H2O)nCl8n]n4 etc.) just as nitrosyl species (e.g. [RuNOCl5]2) [6]. A solution of commercial RuCl3xH2O (cRu = 200 mg/L) dissolved in hot 9.5 M HCl became deeply brown. UV/Vis spectrum revealed characteristic absorption for dimeric complexes [Ru2O2Cl4(H2O)4] and [Ru2OCl10]4. Unified stock solution of RuIII (cRu = 200 mg/L) was made of commercial RuCl3xH2O dissolved in ethanol and refluxed under argon atmosphere until the brownish solution became green [7]. Ethanol was removed under reduced pressure and the residue dissolved in 9.5 M HCl (suprapur) becoming a green solution. Argon was passed into solution to remove oxygen and preventing oxidation. Storage overnight at room temperature turned the colour to strawberry red. UV/Vis-spectrum matched with spectrum of [RuCl6]3 [8] and showed no further RuIV absorption. Stock solution of RuIV (cRu = 200 mg/L) was made by passing chlorine through a hot RuIII stock solution which became deeply brown (E0 = 1.10 V, T = 25 °C, cHCl = 9.5 mol/L). UV/Vis-spectrum matched with spectrum of [RuCl6]2 [8]. Acidity of stock solutions was standardised by titration using 0.1 M NaOH solution as titrand and phenolphthalein as indicator (intrinsic colour of ruthenium disappeared due to dilution before end-point was reached). Solutions of lower acidity and metal concentration (cPt = 20 mg/L, cRu = 10 mg/L) were obtained by appropriate dilution with deionised water as required at which equal volumes of appropriate dilutions were united. SnII stock solutions (cSn = 65 mg/L) for stannometric titrations were made freshly every time and promptly used due to oxidation by atmospheric oxygen. The appropriate amount of anhydrous SnCl2 was dissolved in a small volume of boiling 9.5 M HCl (suprapur) until a clear solution was obtained. Acidity was adjusted by filling-up with deionised water and 9.5 M HCl (suprapur) as needed. All stock solutions were standardised by ICP–OES and stored on a dark cold place. 2.4. Methods 2.4.1. Adsorption Experimental conditions in detail are listed in Table 1. Metal content of the feed solution referred to realistic leaching results. An exhausted MEA (A = 25 cm2, anode: mPt = 2 mg/cm2, mRu = 1 mg/cm2, cathode: mPt = 1 mg/cm2, NafionÒ membrane) from a direct methanol fuel cell was treated with chlorine saturated concentrated hydrochloric acid (V = 100 mL) at reflux for four
Table 1 Experimental conditions of various adsorption experiments for adsorption of platinum and ruthenium onto resin 1. parameter
experimental conditions
Hydrochloric acid (mol/L) Temperature (°C) Flow rate (mL/min) Bed height (cm) cPt (mg/L) cRu (mg/L) Column height (cm) Internal column diameter (cm)
0.1, 1.0, 3.0, 6.0, 9.5 25, 60, 90 0.1, 0.5, 1.0 0.4 (m = 50 mg), 0.8 (m = 100 mg) 20 10 5.5 0.5
M. Trautmann, H.-J. Holdt / Separation and Purification Technology 149 (2015) 279–287
281
2.5. Theoretical background Unless otherwise specified, redox potentials (in Volt [V]) refer to Ag/AgCl as reference electrode. Conversion of E0 (vs. Ag/AgCl) to E0 (vs. NHE) accords to the following relationship [9]:
E0 ðv s:Ag=AgClÞ ¼ E0 ðv s:NHEÞ 0:197
ð1Þ
Adsorption performance of platinum and ruthenium onto 1 was analysed using Thomas, Yoon–Nelson, Bohart–Adams and Wolborska model. The quality of each model was assessed using the correlation coefficient R2.
Scheme 1. Flow chart of down flow packed bed column for fixed bed column studies at room temperature.
2.5.1. Thomas model [10] The Thomas model is frequently applied for modelling column performances, predicting breakthrough curves (BTCs) for the effluent and appraising adsorption capacity of sorbents. It assumes second-order reversible reaction kinetics, no axial dispersion as well as Langmuir (favourable) isotherm [11,12] and is generally expressed in its linear form follows:
lnððc0 =cV Þ 1Þ ¼ ððkTH qmÞ=Q V Þ ððkTH c0 VÞ=Q V Þ
ð2Þ
where kTH is the Thomas rate constant [L/mg min], q is the maximum metal adsorption capacity [mg/g], m is the mass of adsorbent [g], QV is the volumetric flow rate [L/min], V is the effluent volume [L], c0 and cV are the initial metal concentration [mg/L] at V = 0 mL and the residual concentration at volume V, respectively. 2.5.2. Yoon–Nelson model [13] The Yoon–Nelson model is relatively simple model assuming that the decrease of sorption probability of each sorbate is proportional to probability of the sorbate sorption and breakthrough on the sorbent. The linearised form of Yoon–Nelson equation is given by:
lnðct =ðc0 ct ÞÞ ¼ kYN t skYN
Scheme 2. Flow chart of down flow packed bed column for fixed bed column studies at elevated temperature.
hours. Chlorine was permanently inlet to keep redox potential at E0 = 1.10 V. At the end, leach liquor possessed a metal content of cPt = 21 mg/L and cRu = 9 mg/L. Hence, metal concentration of the feed solutions was adapted to cPt = 20 mg/L and cRu = 10 mg/L. Resin 1 (m = 100 mg) was suspended in water (V = 2.0 mL) for platinum extraction and rather in hydrochloric acid (V = 2.0 mL) conforming to the feed solution for ruthenium adsorption. Suspension was transferred to the SPE cartridge and fixed between two membranes. The entire experimental set-up for adsorption at room temperature and using downward flow is shown in Scheme 1. Feed solution was charged from the top of the column and sucked via peristaltic pump downward at a constant flow rate. For upward flow at elevated temperature, column was placed directly in the hot feed solution levelling the column temperature. Feed solution was sucked from the bottom of column via peristaltic pump upward at a constant flow rate. Set-up for adsorption at elevated temperatures and downward flow is shown in Scheme 2. Reservoir containing feed solution was pre-heated and then pumped into the cavity between heating mantle and column balancing the thermal loss. Eluates were collected using fraction collector dividing into fractions based on volume (3.0 mL/fraction). 2.4.2. Elution Supernatant feed solution was removed to a minimum level so that resin bed stayed humidified. Eluent was inlet at room temperature, at same flow rate and flow direction used for adsorption.
ð3Þ
where kYN is the Yoon–Nelson rate constant [1/min], t is time [min] and s is time required for 50% sorbate breakthrough [min]. The amount of metal adsorbed on the fixed bed is half of the total amount entering the adsorption bed within 2s period [14] and is expressed for a given bed as:
q ¼ ð0:5c0 ððQ V =1000Þ2sÞÞ=m ¼ ðc0 Q V sÞ=1000m
ð4Þ
2.5.3. Bohart–Adams model [15] The Bohart–Adams model is widely used modelling fixed bed BTCs for environmental sorption and biosorption processes [16]. It assumes that the adsorption rate basically depends on the reaction between the sorbent and the free sorption sites on the surface, a rectangular (irreversible) isotherm and a low concentration field (c < 0.15c0). Therefore this model is more suitable to describe the initial phase of the BTC [17]. The linear form is represented as:
lnðct =c0 Þ ¼ ðkBA c0 tÞ ðkBA N0 ðh=Q l ÞÞ
ð5Þ
where kBA is the Bohart–Adams rate constant [L/mg min], N0 is saturation concentration [mg/L] which is connected to q via q = N0BV/m where BV is the bed volume [L] and m the mass of adsorbent [g], h is the bed height of column [cm], Ql is the linear flow rate [cm/min], c0 and ct are the initial metal concentration [mg/L] at t = 0 min and the residual concentration at time t, respectively. 2.5.4. Wolborska model [18] The Wolborska model is also a simplified adsorption model describing breakthrough for low concentration ranges emphasising the general mass transfer for the diffusion mechanism [19]. The initial stage of BTC is controlled by film diffusion with a constant
282
M. Trautmann, H.-J. Holdt / Separation and Purification Technology 149 (2015) 279–287
kinetic coefficient and the concentration profile moves axially with a constant velocity. The final BTC as well as the width of concentration profile in the column is nearly constant [12]. The linearised equation for the mass transfer in fixed bed adsorption is represented as follows:
lnðct =c0 Þ ¼ ðba c0 Þ=N0 Þt ððba hÞ=Q l Þ
ð6Þ
where ba is the kinetic coefficient of the external mass transfer [1/h], 3. Results and discussion Sorption behaviour of PtIV from PtIV/RuIII solutions did not differ from PtIV/RuIV mixtures and is therefore not shown. 3.1. Influence of hydrochloric acid concentration Manipulating cHCl resulted in several effects. At first, proton concentration affected the protonation equilibrium of sulphoxide (R1R2SO + HCl ¢ R1R2SOH+ + Cl) and therefore the amount of electrostatic bound ions. At second, chloride concentration controls the speciation responsible for the metal complex charge (e.g. [RuCl5(H2O)]2 + Cl ¢ [RuCl6]3 + H2O) and consequently the quantity of competitor ions. The adsorption of PtIV and RuIII from PtIV/RuIII solutions as well as PtIV and RuIV from PtIV/RuIV solutions onto resin 1 was analysed using fixed bed column as a function of hydrochloric acid concentration. BTCs were obtained at cHCl = 0.1, 1.0, 3.0, 6.0 and 9.5 mol/L, cPt = 20 mg/L, cRu = 10 mg/L, Q = 1.0 mL/min, T = 25 °C and h = 0.8 cm (m = 100 mg) (Fig. 2). On increasing hydrochloric acid concentration to cHCl = 3.0 mol/L, the breakthrough volume increased coinciding with the rising sulphoxide protonation. Further increase led to a decrease due to rising concentration of competitor ions HCI 2 . Contact with cHCl = 9.5 mol/L progressively degraded resin due to acidic ether cleavage. FTIR spectrum showed a decrease of m(COC) 1090 cm1 and the appearance of two new bands at 671 cm1 (C–Cl) and 1265 cm1 (C–H from –CH2Cl). Methanol (10% v/v) was added to cHCl = 3.0 mol/L to enhance the wetting of the hydrophobic matrix but resulted no improvement. Nonetheless BTCs of PtIV were distinctively broadened what could be attributed to several reasons, e.g. Eddy diffusion, axial diffusion, dispersion, dilution, wall channelling, mass transfer effects.
Fig. 2. Effect of hydrochloric acid concentration on BTC of PtIV from PtIV/RuIII solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, T = 25 °C, Q = 1.0 mL/min, h = 0.8 cm, downflow). Symbols: (j) 0.1 mol/L, (4) 1.0 mol/L, (.) 3.0 mol/L, (h) 6.0 mol/L and (s) 9.5 mol/L.
Fig. 3. Effect of hydrochloric acid concentration on BTC of RuIII from PtIV/RuIII solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, T = 25 °C, Q = 1.0 mL/min, h = 0.8 cm, downflow). Symbols: (j) 0.1 mol/L, (4) 1.0 mol/L, (.) 3.0 mol/L, (h) 6.0 mol/L and (s) 9.5 mol/L.
As seen in Fig. 3, RuIII was not adsorbed at any hydrochloric acid concentration. RuIII forms octahedral aquachloro complexes of the general formula [RuCln(H2O)6n]3n (n = 0–6) depending on chloride concentration. At cHCl 6 0:1 mol=L, mainly cationic and neutral species exist. At cHCl P 1:0 mol=L, a crude mixture of nearly all species (n = 1–6) form at which [RuCl4(H2O)2] prevails. Distribution shifts with rising chloride concentration to more negatively charged ions so that mainly [RuCln(H2O)6n]3n (n = 5–6) predominate at cHCl P 6:0 mol=L. The highly charged hexachloro complex possesses a high charge density, high hydration energy and a huge steric hindrance. The difficulty of packing three sulphoxide molecules around one bulky ruthenate forms an obstacle leaving the anion in solution. The aquachloro complexes remained in solution because hydrogen bonding of the water ligands with water molecules around made them more dissolvable in the aqueous phase instead on the surface of the hydrophobic PS-DVB matrix. The neutral complex [RuCl3(H2O)3] could not form any electrostatic interactions whereas cationic complexes were repulsed. Replacement of chloride or water from the inner coordination sphere by sulphoxide was less successful at room temperature due to kinetic inertness of RuIII complexes. RuIV was expected to retain on resin like all divalent hexachloro complexes but was just particulately sorbed between 3 6 cHCl 6 6 mol=L (Fig. 4) at which the resin became deeply brown. BTCs exhibited a plateau-like progression with a maintaining sorption level at around 35%. The residual steady bleed of ruthenium from the beginning was inconsistent to [RuCl6]2 indicating other species. UV/Vis spectra of eluates exhibited [RuCl6]3 due to the specific absorbance at 229 nm and 349 nm. Stannometric titration revealed around 65% RuIII. Reduction through resin was excluded because nil evidences were found in FTIR spectrum for sulfone formation. Redox potential of stock solution (T = 25 °C, cHCl = 9.5 mol/L) had decreased from E0 = 1.10 V to E0 = 0.80 V during storage. This lowering was due to the continuously outgassing of chlorine during storage. RuIV is stable at E0 P 1:30 V (vs. NHE) [20] means E0 P 1:10 V (vs. Ag/AgCl) otherwise tends to reduction. Broszkiewicz [8] reported reduction of RuIV already at cHCl > 5.0 mol/L. The hexachloro complex of RuIV is only stable at high chloride concentrations greater than 9.0 mol/L [20] otherwise tends to hydrolysis which forms a jumble of monomer and dimeric chloro complexes including aqua, hydroxo and oxo groups. Those complex mixtures are hardly to handle for a well-defined adsorption.
M. Trautmann, H.-J. Holdt / Separation and Purification Technology 149 (2015) 279–287
Fig. 4. Effect of hydrochloric acid concentration on BTC of RuIII/IV from PtIV/RuIII/IV solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, T = 25 °C, Q = 1.0 mL/min, h = 0.8 cm, downflow). Symbols: (j) 0.1 mol/L, (4) 1.0 mol/L, (.) 3.0 mol/L, (h) 6.0 mol/L and (s) 9.5 mol/L.
283
Fig. 5. Effect of temperature on BTC of PtIV from PtIV/RuIII solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, cHCl = 3.0 mol/L, Q = 1.0 mL/min, h = 0.8 cm, E0 = 0.48 V, downflow). Symbols: (4) 25 °C, (.) 60 °C and (h) 90 °C.
Subjected to previous mentioned facts, RuIV stock solution was freshly prepared adjusting E0 = 1.10 V (T = 25 °C, cHCl = 9.5 mol/L) with chlorine to completely oxidise RuIII and stabilise RuIV as [RuCl6]2. Unfortunately the high redox potential of freshly made feed solution resulted resin degradation. Sulphoxide was oxidised to sulfone and already sorbed metals completely flushed out. Further oxidants like NaClO3 and H2O2 proved inadequate for fixed bed application due to incessantly formed chlorine bubbles which ruptured the bed.
3.2. Influence of temperature Temperature is also an important parameter influencing sorption processes. The poly(styrene-co-divinylbenzene) matrix with its cross-linked polymeric chains appears as a relatively rigid construct. Diffusion into a branched and less flexible structure is difficult but improved at higher temperature because polymer density and chain rigidity decrease whereas ion mobility increase. Moreover inner-sphere substitution kinetic is governed by temperature and establishing a second sorption mechanism in addition to ion association. The effect of temperature on the adsorption of PtIV and RuIII from PtIV/RuIII solutions onto resin 1 was examined by varying the temperature to 25 °C, 60 °C and 90 °C at cHCl = 3.0 mol/L and 6.0 mol/L, at constant influent concentration cPt = 20 mg/L and cRu = 10 mg/L, Q = 1.0 mL/min, h = 0.8 cm and downward flow, as shown in Figs. 5–8. Handling cHCl = 6.0 mol/L takes the advantage of a constant boiling azeotrope offering an incessant acidity over a long heating period as well as the possibility of recycling by circuit distillation. As seen in Fig. 5 and Fig. 6, PtIV sorption decreased with rising temperature. Curve slope became steeper and breakthrough shifted to smaller effluent volumes lowering the operating capacity. This behaviour can be explained by exothermic sorption character. Previous batch experiments revealed physisorption of PtIV [5]. This kind of mechanism is always exothermic because the increasing arrangement of adsorbate on adsorbent surface reduces entropy. Concurrently, effluent concentration did not reach influent concentration at elevated temperatures merely a plateau like very long tail ran out at c/c0 < 1. This kind of tailing correlates with the diffusion into stagnant water zones. Solubility of gases in solution decrease with increasing temperature so that outgas produces microbubbles of water steam and HCl. Every rupture diversify bed package and generates zones of immobile water. Diffusion into stagnant zones is
Fig. 6. Effect of temperature on BTC of PtIV from PtIV/RuIII solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, cHCl = 6.0 mol/L, Q = 1.0 mL/min, h = 0.8 cm, E0 = 0.48 V, downflow). Symbols: (4) 25 °C, (.) 60 °C and (h) 90 °C.
Fig. 7. Effect of temperature on BTC of RuIII from PtIV/RuIII solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, cHCl = 3.0 mol/L, Q = 1.0 mL/min, h = 0.8 cm, E0 = 0.48 V, downflow). Symbols: (s) 25 °C, (.) 60 °C and (h) 90 °C.
284
M. Trautmann, H.-J. Holdt / Separation and Purification Technology 149 (2015) 279–287
Fig. 8. Effect of temperature on BTC of RuIII from PtIV/RuIII solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, cHCl = 6.0 mol/L, Q = 1.0 mL/min, h = 0.8 cm, E0 = 0.48 V, downflow). Symbols: (s) 25 °C, (.) 60 °C and (h) 90 °C.
increased due to higher ion mobility at higher temperatures. Ions which enter a stagnant zone are trapped and the only mechanism to escape is molecular diffusion. The capacity of stagnant regions is high relative to the capacity of mobile regions. Sorption at T = 90 °C was hardly to handle because operating near boiling point revealed a vigorous bubble formation. More precisely, microbubbles coalesced to macroscopic size, held by downward stream until the buoyant force exceeded. Every uplifting bubble generated voids which promoted channelling and therewith an early breakthrough because ions could move more rapidly than the average flow velocity. As indicated in Figs. 7 and 8, sorption of RuIII improved with rising temperature due to the fact that the inner-sphere substitution is strongly temperature dependent. This essentially comprises aquation at which chloride is substituted by water forming lower charged aquachloro complexes which are more prone to inner-sphere substitution by sulphoxide. The incorporation of sulphoxide into the inner coordination sphere occurred via oxygen atom due to the appearance of a new FTIR band at 874 cm1 [21]. Breakthrough behaviour at T = 90 °C and cHCl = 6.0 mol/L differed from cHCl = 3.0 mol/L due to shifted speciation distribution. Hence, T = 90 °C proved inadvisable so that T = 60 °C was favoured at which the problem of stagnant water zones remained similar to PtIV sorption.
3.3. Influence of flow direction The most common direction of reactant flow is downward through the bed especially recommended in cases of a wide variation of feed flow rate. It stabilises the bed and prevents lifting particles out of the column. Attrition and entrainment of the finest particles are minimised. Unfortunately bed compression as well as gravitation of fine particles can cause pressure drops and maldistribution of the flow. Concerning this matter, upward flow is often preferred for industrial applications especially when gas is produced (e.g. packed bed bioreactors, enzyme columns). It does not compress the bed as downward flow does whereas microbubbles escape with flow fluidizing the bed. This strictly dependents on flow velocity but minimizes phenomena like blockage or stagnant water zones. The effect of flow direction for the adsorption of PtIV and RuIII from PtIV/RuIII mixtures onto resin 1 was investigated by reversing flow (upflow) at influent concentration cPt = 20 mg/L and
Fig. 9. Effect of upflow on BTC of PtIV and RuIII from PtIV/RuIII solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, cHCl = 6.0 mol/L, T = 60 °C, Q = 1.0 mL/min, h = 0.8 cm, E0 = 0.48 V). Symbols: (s) PtIV, (.) RuIII.
cRu = 10 mg/L, cHCl = 3.0 mol/L, T = 25 °C, Q = 1.0 mL/min, E0 = 0.48 V and h = 0.8 cm. Adsorption behaviour of PtIV and RuIII was negligible affected at room temperature and is therefore not displayed. Adsorption behaviour of RuIII at cHCl = 6.0 mol/L and T = 60 °C was apparently influenced. As seen in Fig. 9, plateau-like tailing did not appear means upward flow avoided successfully formation of stagnant water zones. BTC of RuIII assumed slightly S-shape but remained non-sigmoid (with respect to c/c0 = 0.5) by ordinary tailing what can be attributed to slow sorption kinetic and excessive flow rate. Hence, upward flow was favoured over downward flow at elevated temperatures. 3.4. Influence of redox potential The redox potential has a wide influence over the oxidation state of a metal. [PtCl6]2 is relatively redox stable opposite to [RuCl6]2 which easily reduces to [RuCl6]3. Due to the easier arrangement two bulky counterions instead of three, the divalent [RuCl6]2 has a higher affinity to the resin and binds electrostatically more strongly than trivalent [RuCl6]3. To that effect the goodness of ruthenium adsorption is consequentially adjustable by manipulating the redox potential of the solution. As seen in Fig. 9, the obvious c/c0 > 0 for RuIII sorption at the beginning indicated a steady bleed of ruthenium which was caused by [RuCl6]3 and less astonishing because of the predominating species [RuCl5(H2O)]2 and [RuCl6]3 of RuIII at cHCl = 6.0 mol/L. Redox potential of the feed solution was raised from E0 = 0.48 V to E0 = 0.98 V (T = 25 °C, cHCl = 6.0 mol/L) by chlorine to oxidise both species into [RuCl6]2. This was definitely not high enough to convert RuIII completely to RuIV rather to a RuIII/IV mixture. Keeping E0 < 1.00 V was necessary to avoid resin degradation. As illustrated in Fig. 10, increasing redox potential revealed improvement of ruthenium sorption due to speciation shift whereas PtIV sorption was hardly influenced and therefore not displayed. Hence, ruthenium was enriched to 90% whereupon [RuCl6]3 bleed had to be accepted. Nonetheless BTC broadening and tailing still occurred. 3.5. Influence of flow rate Flow rate controls the contact time between adsorbate and adsorbent what is important if adsorption is governed by slow kinetics. So, residence time on the column should be long enough
M. Trautmann, H.-J. Holdt / Separation and Purification Technology 149 (2015) 279–287
285
Fig. 10. Effect of redox potential on BTC of RuIII/IV from PtIV/RuIII/IV solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, cHCl = 6.0 mol/L, T = 60 °C, Q = 1.0 mL/min, h = 0.8 cm, upflow). Symbols: (.) 0.48 V, (s) 0.98 V.
Fig. 12. Effect of flow rate on BTC of RuIII/IV from PtIV/RuIII/IV solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, cHCl = 6.0 mol/L, T = 60 °C, h = 0.8 cm, E0 = 0.98 V, upflow). Symbols: (h) 1.0 mL/min, (.) 0.5 mL/min and (s) 0.1 mL/min.
to establish adsorption equilibrium. Especially inner-sphere substitution of ruthenium aquachloro complexes reveals a slow sorption kinetic. A high flow rate performs a quick execution but also means shorter time for saturation resulting in early breakthrough, increased dispersion and poor separation. A low flow rate offers more residence time but should not be too low otherwise process becomes uninteresting for practical work and longitudinal diffusion will dominate broadening sorption zone. The effect of flow rate was studied for the adsorption of PtIV from PtIV/RuIII mixtures onto resin 1 at flow rates Q = 0.1, 0.5 and 1.0 mL/min at influent concentration cPt = 20 mg/L and cRu = 10 mg/L, cHCl = 3.0 mol/L, T = 25 °C, h = 0.8 cm and downward flow. As indicated in Fig. 11, PtIV sorption was influenced by flow rate whereas at lower ones the breakthrough occurred later resulting a higher effective adsorption. The BTC obtained at Q = 1.0 mL/min and Q = 0.5 mL/min were similar in shape but differed in breakthrough volume which was higher at lower velocity. Further lowering to Q = 0.1 mL/min subordinated dispersion, Eddy diffusion and mass transfer effects what reduced curve spreading. RuIII was not adsorbed from PtIV/RuIII mixtures at any flow rate due
to the facts mentioned in chapter 3.1 and is therefore not displayed. The effect of flow rate was examined for the adsorption of Ru from PtIV/RuIII/IV mixtures onto resin 1 at constant influent concentration cPt = 20 mg/L and cRu = 10 mg/L, cHCl = 6.0 mol/L, T = 60 °C, E0 = 0.98 V, h = 0.8 cm and upward flow. Ruthenium sorption was markedly influenced by flow rate as seen in Fig. 12. Irrespective of [RuCl6]3 bleed, breakthrough volume increased with decreasing flow rate. BTC asymmetry decreased with decreasing velocity means tailing was kinetically caused by slow sorption kinetic. Broadening decreased with decreasing velocity due to diminished influence of dispersion, diffusion and mass transfer effects. BTC gained more S-shape and steeper slope. High flow rates forced ruthenium complexes through column so that there was no adequate residence time to establish adsorption equilibrium. The effect on PtIV sorption was negligible at these conditions due to insignificant adsorption and is therefore not displayed.
3.6. Influence of bed height
Fig. 11. Effect of flow rate on BTC of PtIV from PtIV/RuIII solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, cHCl = 3.0 mol/L, T = 25 °C, h = 0.8 cm, E0 = 0.48 V, downflow). Symbols: (4) 1.0 mL/min, (s) 0.5 mL/min and (j) 0.1 mL/min.
Bed height is set by mass of adsorbent. Higher bed heights mean larger amounts of adsorbent providing more available binding sites. Additionally residence time increases due to longer pathways. BTCs for adsorption of PtIV from PtIV/RuIII mixtures onto resin 1 at various bed heights, at the influent concentrations of cPt = 20 mg/L and cRu = 10 mg/L, cHCl = 3.0 mol/L, T = 25 °C, Q = 0.5 mL/min, E0 = 0.48 V and downward flow are shown in Fig. 13. The corresponding BTCs of RuIII are not diagrammed because of non-adsorption. It is observed that breakthrough volume and equilibrium sorption capacity decreased with decrease of bed height due to the diminished availability of sorption sites. Higher bed height promoted broadening what can be explained by axial dispersion. The effect of bed height on the BTCs of Ru from PtIV/RuIII/IV solutions at various bed heights, at the influent concentrations of cPt = 20 mg/L and cRu = 10 mg/L, cHCl = 6.0 mol/L, T = 60 °C, Q = 0.5 mL/min, E0 = 0.98 V and upward flow is shown in Fig. 14. The corresponding BTCs of PtIV are not diagrammed due to negligible sorption. The results indicate similar to PtIV that the decrease of
286
M. Trautmann, H.-J. Holdt / Separation and Purification Technology 149 (2015) 279–287 Table 2 Results of Thomas, Yoon–Nelson, Bohart–Adams and Wolborska model for adsorption of platinum (cPt = 20 mg/L) and ruthenium (cRu = 10 mg/L) onto resin 1.
Fig. 13. Effect of bed height on BTC of PtIV from PtIV/RuIII solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, cHCl = 3.0 mol/L, T = 25 °C, Q = 0.5 mL/min, E0 = 0.48 V, downflow). Symbols: (.) h = 0.4 cm, (s) h = 0.8 cm.
Pt
Ru
Hydrochloric acid (mol/L) Temperature (°C) Flow rate (mL/min) Flow direction Bed height (cm) Redox potential (V) qexp (mg/g)
3.0 25 0.5 downflow 0.8 0.48 7.80
6.0 60 0.5 upflow 0.8 0.98 13.04
Thomas model kTH 103 (L/mg min) q (mg/g) R2
2.99 7.10 0.95
1.10 12.68 0.96
Yoon–Nelson kYN 102 (1/min) s (min) q (mg/g) R2
6.01 70.84 7.10 0.95
1.11 260.23 12.96 0.96
Bohart–Adams model kBA 103 (L/mg min) N0 103 (mg/L) q (mg/g) R2
1.25 4.29 6.74 0.78
0.59 8.69 13.64 0.86
Wolborska model ba (1/h) N0 103 (mg/L) R2
0.50 6.71 0.78
0.48 13.58 0.86
3.8. Elution
Fig. 14. Effect of bed height on BTC of Ru from PtIV/RuIII/IV solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, cHCl = 6.0 mol/L, T = 60 °C, Q = 0.5 mL/min, E0 = 0.98 V, upflow). Symbols: (j) h = 0.8 cm, (4) h = 0.4 cm.
bed height diminished available sorption sites and reduced breakthrough volume and equilibrium sorption capacity.
3.7. Column dynamics study For the purpose of characterising the fixed bed column behaviour and scaling up for industrial application, four models: Thomas, Yoon–Nelson, Bohart–Adams and Wolborska were fitted to the experimental data found for best separation conditions. Application to the data started at the concentration ratio c/c0 > 0.05 that is 5% breakthrough until c/c0 > 0.90 that is 90% breakthrough. All coefficients and relative constants were obtained using linear regression analysis and are summarised in Table 2. Platinum sorption data fitted well with Thomas (R2 = 0.95) and Yoon–Nelson (R2 = 0.95) model whereupon Bohart–Adams and Wolborska model showed a poor fit (R2 = 0.78). Ruthenium sorption data showed also a good fit to the Thomas (R2 = 0.91) and Yoon–Nelson (R2 = 0.91) model. Bohart-Adams (R2 = 0.73) and Wolborska (R2 = 0.73) model were unsuited to describe complete area of RuIII sorption.
The removal of adsorbed metal from the resin was examined using the eluents: water, hydrochloric acid (cHCl = 0.1, 3.0, 6.0 mol/L) and thiourea (Tu) (cTu = 0.5 mol/L) in hydrochloric acid (cHCl = 0.1 mol/L). It turned out that a complete pump out of the feed solution after adsorption decreased significantly the desorption yield. This could be explained by seclusion of already established flow channels forming nonadvective regions. Feed solution was removed to a minimum level so that resin stayed humidified. Platinum was eluted from the column at room temperature during washing step using hydrochloric acid and water. At cHCl 6 0.1 mol/L, sulphoxide is deprotonated eliminating any electrostatic retardation. At cHCl P 1:0 mol=L, PtIV ions are displaced by the surplus concentration of chloride ions at which elution rate increased with hydrochloric acid concentration. Desorption caused by deprotonation occurred much faster compared to the continuous displacement by competitor ions. Hence, it is advisable to wash firstly with hydrochloric acid to minimise the loss of platinum while removing ruthenium. Desorption of platinum occurred quantitative from the resin. Elution of ruthenium only took place with thiourea via nucleophilic substitution forming the neutral Ru(Tu)3 complex. Desorption rate accelerated with rising temperature due to substitution kinetic. Hydrochloric acid (cHCl = 0.1, 3.0, 6.0 mol/L) removed unattached ruthenium and platinum. Water was inadequate in case of ruthenium adsorption due to hydrolysis forming a black precipitate of ruthenium dioxide. 4. Conclusion Adsorption of platinum and ruthenium from hydrochloric acidic solution by powdered sulphoxide modified polystyrene–divinyl benzene (d50 < 30 lm) was investigated in a continuous flow fixed-bed column. The breakthrough curves were determined at various concentrations of hydrochloric acid, temperatures, flow rates, flow rate directions, redox potentials and bed heights. Metal sorption was selectable depending on parameter settings
M. Trautmann, H.-J. Holdt / Separation and Purification Technology 149 (2015) 279–287
enabling separation. Best PtIV sorption from a mixture of PtIV/RuIII was achieved at cHCl = 3.0 mol/L, T = 25 °C, Q = 0.5 mL/min, E0 = 0.48 V using downward flow. Best ruthenium sorption from a mixture of PtIV/RuIII/IV at which ruthenium was enriched to 90% occurred at cHCl = 6.0 mol/L, T = 60 °C, Q = 0.5 mL/min, E0 = 0.98 V using upward flow. Residual 10% remained in solution as [RuCl6]3 due to poor sorption affinity. The adsorbent is capable of holding a maximum of 7.80 mg Pt/g at 20 mg/L of influent concentration and 13.04 mg Ru/g at 10 mg/L of influent concentration. The column performance at these conditions was analysed using Thomas, Yoon–Nelson, Bohart–Adams and Wolborska model. Thomas and Yoon–Nelson model described adsorption behaviour of platinum and ruthenium onto 1 more reasonably than Bohart– Adams and Wolborska model due to high correlation coefficients. Elution of platinum occurred via deprotonation at cHCl 6 0:1 mol=L or expulsion at cHCl P 1:0 mol=L. Ruthenium was eluted by thiourea (cTu = 0.5 mol/L) in diluted hydrochloric acid (cHCl = 0.1 mol/L). Referring to the operating capacity, separation of platinum and ruthenium was achieved setting the above mentioned conditions. Platinum adsorbed at room temperature at which ruthenium passed solely through the column as RuIII. Elution took place with hydrochloric acid and water. Ruthenium was adsorbed to 90% at elevated temperature at which further repetitions were necessary for complete elimination from the feed solution. Elution took place with thiourea as Ru(Tu)3 complex. References [1] Gianfranco Pistoia, Batteries for Portable Devices, Elsevier (Science & Technology), Amsterdam, 2005, p. 194. [2] X. Cheng, C. Peng, Y. Ma, L. Chen, Y. Zhang, Lifetime studies of catalyst activity and microstructure in a PEMFC, in: J. Prakash, D. Chu, D. Scherson, M. Enayetullah, I. Tae Bae (Eds.), Fundamental Understanding of Electrode Processes, The electrochemical society Inc., Pennington, 2005, p. 177. [3] D. Stolten, Hydrogen and Fuel Cells: Fundamentals, Technologies and Applications, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, 2010. p. 54.
287
[4] N. Wiberg, E. Wiberg, A.F. Hollemann, Lehrbuch der Anorganischen Chemie, Walter de Gruyter, Berlin, 2007. p. 1667. [5] M. Trautmann, S. Lubahn, H.-J. Holdt, Preparation, characterisation and properties of sulphoxide modified resins for solid-phase extraction of PtIV, RuIII and RuIV from hydrochloric acid, React. Funct. Polym. 83 (2014) 84–87. [6] E.A. Seddon, K.R. Seddon, The Chemistry of Ruthenium, Elsevier Science Publishers B.V, Amsterdam, 1984. p. 159. [7] K.R. Hyde, E.W. Hooper, J. Waters, J.M. Fletcher, F- and b-Ruthenium trichloride, J. Less Common Met. 8 (1965) 428–434. [8] R.K. Broszkiewicz, Redox reactions of RuCl3 and RuCl2 6 6 : a pulse radiolysis study, Radiat. Phys. Chem. 15 (1980) 133–138. [9] R.A. Scott, C.M. Lukehart, Applications of Physical Methods to Inorganic and Bioinorganic Chemistry, Wiley & Sons, Hoboken, NJ, 2007. p. 23. [10] H.C. Thomas, Heterogeneous ion exchange in a flowing system, J. Am. Chem. Soc. 66 (1944) 1664–1666. [11] Z. Saadi, R. Saadi, R. Fazaeli, Fixed-bed adsorption dynamics of Pb (II) adsorption from aqueous solution using nanostructured c-alumina, J. Nanostructure Chem. 3 (2013) 48. [12] Z. Xu, J. Cai, B. Pan, Mathematically modeling fixed-bed adsorption in aqueous systems, J. Zhejiang Univ. Sci A (Appl Phys & Eng) 14 (2013) 155–176. [13] Y.H. Yoon, J.H. Nelson, Application of gas adsorption kinetics. I. A theoretical model for respirator cartridge service life, Am. Ind. Hyg. Assoc. J. 45 (1984) 509–516. [14] P. Sivakumar, P.N. Palanisamy, Packed bed column studies for the removal of Acid blue 92 and Basic red 29 using non-conventional adsorbent, Indian J. Chem. Technol. 16 (2009) 301–307. [15] G.S. Bohart, E.Q. Adams, Some aspects of the behavior of charcoal with respect to chlorine, J. Am. Chem. Soc. 42 (1920) 523–544. [16] K.H. Chu, Fixed bed sorption: setting the record straight on the Bohart-Adams and Thomas models, J. Hazard. Mater. 177 (2010) 1006–1012. [17] F. Pagnanelli, F. Ferella, I. De Michelis, F. Veglio, Adsorption onto activated carbon for molybdenum recovery from leach liquors of exhausted hydrotreating catalysts, Hydrometallurgy 110 (2011) 67–72. [18] A. Wolborska, Adsorption on activated carbon of p-nitrophenol from aqueous solution, Water Res. 23 (1989) 85–91. [19] P.-S. Keng, S.-L. Lee, S.-T. Ha, Y.-T. Hung, S.-T. Ong, Cheap materials to clean heavy metal polluted waters, in: E. Lichtfouse, D. Robert, J. Schwarzbauer (Eds.), Green materials for energy, Products and Depollution, Springer Science & Business Media, Heidelberg, 2013, p. 396. [20] Gmelins Handbuch der anorganischen Chemie, 8. Auflage, System No. 63, Ruthenium, Ergänzungsband, Verlag Chemie, Weinheim, 1970. [21] M. Calligaris, Structure and bonding in metal sulfoxide complexes: an update, Coord. Chem. Rev. 248 (2004) 351–375.
Contents lists available at ScienceDirect
Separation and Purification Technology journal homepage: www.elsevier.com/locate/seppur
Separation of platinum and ruthenium by a sulphoxide modified polystyrene resin in laboratory column systems Michael Trautmann, Hans-Jürgen Holdt ⇑ University of Potsdam, Institute of Chemistry, Inorganic Chemistry, Karl-Liebknecht-Straße 24-25, 14476 Golm, Germany
a r t i c l e
i n f o
Article history: Received 10 March 2015 Received in revised form 13 May 2015 Accepted 21 May 2015 Available online 22 May 2015 Keywords: Solid-phase extraction Platinum group metals Sulphoxide Breakthrough curve Kinetic model
a b s t r a c t The present study deals with the adsorption performance of fixed bed columns using powdered sulphoxide modified poly(styrene-co-divinylbenzene) (d10 < 13 lm, d50 < 30 lm, d90 < 50 lm) for the separation of platinum and ruthenium from hydrochloric acidic solutions containing both metals (cPt = 20 mg/L, cRu = 10 mg/L). The influence of hydrochloric acid concentration, temperature, flow rate, flow direction, redox potential and bed height on the breakthrough characteristics was examined. Platinum was separately adsorbed mainly induced by hydrochloric acid concentration and redox potential keeping platinum as PtIV and ruthenium as RuIII. Ruthenium was separately adsorbed to 90% essentially induced by hydrochloric acid concentration, temperature and redox potential keeping platinum as PtIV and ruthenium predominantly as RuIV. Experimental data at optimised separation conditions were fitted to different kinetic models (Thomas, Yoon–Nelson, Bohart–Adams, Wolborska) to characterise the fixed bed column behaviour. Adsorption of both metals was well described by Thomas and Yoon–Nelson model with correlation coefficients R2 P 0:95 whereas Bohart–Adams and Wolborska model were less suitable. Ó 2015 Elsevier B.V. All rights reserved.
1. Introduction Handling the world energy supply is always an actual topic and becomes more and more key issue for the future of humankind. Long-term scarcity of crude oil, gas and coal associated with global environment pollution requires exploitation of renewable energy sources. The further demand of mobile devices with battery longevity leads to the development of new portable low power systems because current rechargeable batteries have reached their theoretical energy density limits. For that purpose fuel cells exhibit high efficiency and high potential for environment-friendly energy conversion leading to a significant reduction of fossil fuel use. Proton exchange membrane fuel cells (PEMFC) as well as direct liquid fuel cells (e.g. direct methanol fuel cell (DMFC), direct ethanol fuel cell (DEFC)) are prime candidates for mobile application due to their compactness and portability. According to the theoretical energy density of pressurized hydrogen (3000 Wh/L) and methanol–water mixtures (3400 Wh/L), fuel cell systems can make more energy available than Li-ion batteries (1300 Wh/L) achieving longer operating periods at the same volume [1]. A big drawback is the elemental necessity of significant amounts of costly platinum group metals (PGMs) as catalysts at which carbon supported
⇑ Corresponding author. E-mail address: [email protected] (H.-J. Holdt). http://dx.doi.org/10.1016/j.seppur.2015.05.013 1383-5866/Ó 2015 Elsevier B.V. All rights reserved.
platinum and platinum–ruthenium mixtures (1:1 atomic ratio) have emerged as best among all other catalysts [2]. Both metals are rare and mining is expensive hampering the broad commercialisation of fuel cell technology. The growing demand makes development of efficient and environmentally friendly recycling processes necessary. Catalyst recycling of exhausted fuel cells lowers manufacturing costs, reduces mining of primary sources and accordingly protects the environment. The classical route is a pyrometallurgical process. After stack dismantling, the membrane electrode assembly (MEA) is separated, shredded and incinerated followed by manifold extraction, solution and precipitation steps. Unfortunately incineration holds the adverse effect of releasing aggressive and toxic hydrofluoric acid what necessitates special linings as well as laborious gas scrubber. The hydrometallurgical route offers an alternative at which both metals are leached using strong oxidising acidic solutions [3]. Ruthenium is separated via distillation as its volatile tetroxide and platinum via precipitation or solvent extraction [4]. These processes are well established but offer disadvantages. Special design and construction materials are needed for distillation because volatile ruthenium compounds are highly explosive and extremely poisonous. Precipitation of platinum may be cheap but demands large quantities of precipitating agent and generates a lot of waste. Solvent extraction requires costly amounts of toxic and/or flammable organic solvents and consumes a lot of time due to multiple extraction steps. One useful
280
M. Trautmann, H.-J. Holdt / Separation and Purification Technology 149 (2015) 279–287
2.3. Preparation of stock solution
Fig. 1. Structure of sulphoxide modified poly(styrene-co-divinylbenzene) resin (1).
solution for the separation of platinum and ruthenium from hydrochloric acidic solutions is adsorption. Fixed bed adsorption is an efficient and effective method conducting sorption processes for industrial applications when a suitable low cost adsorbent is available. In any case, design of adsorption column needs numerous parameters to be adapted. This study deals with the solid-phase extraction (SPE) of platinum and ruthenium from artificial hydrochloric acidic solutions using powdered sulphoxide modified poly(styrene-codivinylbenzene) (d10 < 13 lm, d50 < 30 lm, d90 < 50 lm) as adsorbent (Fig. 1). Catalyst separation refers to the complex of problems which actually exist for PEMFC and DMFC recycling. This lab-scale research provides fundamental findings for prospective large-scale recovery of fuel cell catalyst similar to the contemporary industrial recycling of automobile catalyst converters. The replacement of oil addicted combustion engines by fuel cells will be inevitable due to the shortage of fossil resources. Hydrochloric acid concentration (cHCl), temperature (T), flow direction, flow rate (Q), redox potential (E0) and bed height (h) were sequentially varied and adapted to achieve separately adsorption. The experimental data at optimised conditions were fitted to Thomas, Yoon–Nelson, Bohart–Adams, Wolborska model to characterise the fixed bed column behaviour. 2. Experimental 2.1. Materials The synthesis of powdered sulphoxide modified poly(styreneco-divinylbenzene) resin 1 (1% cross-linked, sulphoxide loading: 2.2 mmol/g, d10 < 13 lm, d50 < 30 lm, d90 < 50 lm) as adsorbent was already described by us [5]. Hydrochloric acid (cHCl = 9.5 mol/L, suprapur) was obtained from Merck KGaA and used as received. Water was purified by a Milli-Q Reference ultrapure water purification system. This high purity water was used for preparation of all stock solutions and dilutions. The platinum starting material was yellow solid Na2PtCl66H2O (99.99%) from Alfa Aesar. The ruthenium starting material was black crystalline RuCl3xH2O (99.99%) from Alfa Aesar. Sodium hydroxide was used as 0.1 M NaOH Titrisol solution supplied by Merck. Anhydrous stannous(II) chloride was obtained as white crystalline SnCl2 (for synthesis) from Merck. 2.2. Instruments Fixed bed column studies were performed with SPE cartridges (Agilent technologies) 5.5 cm in height (H) and 0.5 cm in internal diameter (d). For higher temperatures a special glass coated column (self-construction) 5.5 cm in height and 0.5 cm in diameter was used. Fractions were collected by PrepFCTM fraction collector (Gilson). Different temperatures were established using a MLW thermostat U2 (VEB MLW). Different flow rates were adjusted using peristaltic pump Minipuls 3 (Gilson). Redox potential was determined using a Sentix ORP single-rod measuring cell (Pt–Ag– AgCl) containing 3 M KCl electrolyte (WTW). Infrared spectra were recorded using FTIR-Nexus (Thermo Nicolet) fitted with a Smart Orbit sample system (Diamond) for resins (transmission mode with 32 scans and a resolution of 8 cm1).
Stock solution of PtIV (cPt = 400 mg/L) was made by dissolving the appropriate amount of Na2PtCl66H2O in 9.5 M HCl (suprapur). The composition of commercial hydrated ruthenium trichloride (RuCl3xH2O) is diffuse and hydrochloric acidic solutions contain by the majority RuIV as miscellaneous monomeric as well as chloride- and oxygen-bridged polymeric complexes (e.g. [Ru(OH)nCl6n]2, [Ru2O(H2O)nCl10n]n4, [Ru2O2(H2O)nCl8n]n4 etc.) just as nitrosyl species (e.g. [RuNOCl5]2) [6]. A solution of commercial RuCl3xH2O (cRu = 200 mg/L) dissolved in hot 9.5 M HCl became deeply brown. UV/Vis spectrum revealed characteristic absorption for dimeric complexes [Ru2O2Cl4(H2O)4] and [Ru2OCl10]4. Unified stock solution of RuIII (cRu = 200 mg/L) was made of commercial RuCl3xH2O dissolved in ethanol and refluxed under argon atmosphere until the brownish solution became green [7]. Ethanol was removed under reduced pressure and the residue dissolved in 9.5 M HCl (suprapur) becoming a green solution. Argon was passed into solution to remove oxygen and preventing oxidation. Storage overnight at room temperature turned the colour to strawberry red. UV/Vis-spectrum matched with spectrum of [RuCl6]3 [8] and showed no further RuIV absorption. Stock solution of RuIV (cRu = 200 mg/L) was made by passing chlorine through a hot RuIII stock solution which became deeply brown (E0 = 1.10 V, T = 25 °C, cHCl = 9.5 mol/L). UV/Vis-spectrum matched with spectrum of [RuCl6]2 [8]. Acidity of stock solutions was standardised by titration using 0.1 M NaOH solution as titrand and phenolphthalein as indicator (intrinsic colour of ruthenium disappeared due to dilution before end-point was reached). Solutions of lower acidity and metal concentration (cPt = 20 mg/L, cRu = 10 mg/L) were obtained by appropriate dilution with deionised water as required at which equal volumes of appropriate dilutions were united. SnII stock solutions (cSn = 65 mg/L) for stannometric titrations were made freshly every time and promptly used due to oxidation by atmospheric oxygen. The appropriate amount of anhydrous SnCl2 was dissolved in a small volume of boiling 9.5 M HCl (suprapur) until a clear solution was obtained. Acidity was adjusted by filling-up with deionised water and 9.5 M HCl (suprapur) as needed. All stock solutions were standardised by ICP–OES and stored on a dark cold place. 2.4. Methods 2.4.1. Adsorption Experimental conditions in detail are listed in Table 1. Metal content of the feed solution referred to realistic leaching results. An exhausted MEA (A = 25 cm2, anode: mPt = 2 mg/cm2, mRu = 1 mg/cm2, cathode: mPt = 1 mg/cm2, NafionÒ membrane) from a direct methanol fuel cell was treated with chlorine saturated concentrated hydrochloric acid (V = 100 mL) at reflux for four
Table 1 Experimental conditions of various adsorption experiments for adsorption of platinum and ruthenium onto resin 1. parameter
experimental conditions
Hydrochloric acid (mol/L) Temperature (°C) Flow rate (mL/min) Bed height (cm) cPt (mg/L) cRu (mg/L) Column height (cm) Internal column diameter (cm)
0.1, 1.0, 3.0, 6.0, 9.5 25, 60, 90 0.1, 0.5, 1.0 0.4 (m = 50 mg), 0.8 (m = 100 mg) 20 10 5.5 0.5
M. Trautmann, H.-J. Holdt / Separation and Purification Technology 149 (2015) 279–287
281
2.5. Theoretical background Unless otherwise specified, redox potentials (in Volt [V]) refer to Ag/AgCl as reference electrode. Conversion of E0 (vs. Ag/AgCl) to E0 (vs. NHE) accords to the following relationship [9]:
E0 ðv s:Ag=AgClÞ ¼ E0 ðv s:NHEÞ 0:197
ð1Þ
Adsorption performance of platinum and ruthenium onto 1 was analysed using Thomas, Yoon–Nelson, Bohart–Adams and Wolborska model. The quality of each model was assessed using the correlation coefficient R2.
Scheme 1. Flow chart of down flow packed bed column for fixed bed column studies at room temperature.
2.5.1. Thomas model [10] The Thomas model is frequently applied for modelling column performances, predicting breakthrough curves (BTCs) for the effluent and appraising adsorption capacity of sorbents. It assumes second-order reversible reaction kinetics, no axial dispersion as well as Langmuir (favourable) isotherm [11,12] and is generally expressed in its linear form follows:
lnððc0 =cV Þ 1Þ ¼ ððkTH qmÞ=Q V Þ ððkTH c0 VÞ=Q V Þ
ð2Þ
where kTH is the Thomas rate constant [L/mg min], q is the maximum metal adsorption capacity [mg/g], m is the mass of adsorbent [g], QV is the volumetric flow rate [L/min], V is the effluent volume [L], c0 and cV are the initial metal concentration [mg/L] at V = 0 mL and the residual concentration at volume V, respectively. 2.5.2. Yoon–Nelson model [13] The Yoon–Nelson model is relatively simple model assuming that the decrease of sorption probability of each sorbate is proportional to probability of the sorbate sorption and breakthrough on the sorbent. The linearised form of Yoon–Nelson equation is given by:
lnðct =ðc0 ct ÞÞ ¼ kYN t skYN
Scheme 2. Flow chart of down flow packed bed column for fixed bed column studies at elevated temperature.
hours. Chlorine was permanently inlet to keep redox potential at E0 = 1.10 V. At the end, leach liquor possessed a metal content of cPt = 21 mg/L and cRu = 9 mg/L. Hence, metal concentration of the feed solutions was adapted to cPt = 20 mg/L and cRu = 10 mg/L. Resin 1 (m = 100 mg) was suspended in water (V = 2.0 mL) for platinum extraction and rather in hydrochloric acid (V = 2.0 mL) conforming to the feed solution for ruthenium adsorption. Suspension was transferred to the SPE cartridge and fixed between two membranes. The entire experimental set-up for adsorption at room temperature and using downward flow is shown in Scheme 1. Feed solution was charged from the top of the column and sucked via peristaltic pump downward at a constant flow rate. For upward flow at elevated temperature, column was placed directly in the hot feed solution levelling the column temperature. Feed solution was sucked from the bottom of column via peristaltic pump upward at a constant flow rate. Set-up for adsorption at elevated temperatures and downward flow is shown in Scheme 2. Reservoir containing feed solution was pre-heated and then pumped into the cavity between heating mantle and column balancing the thermal loss. Eluates were collected using fraction collector dividing into fractions based on volume (3.0 mL/fraction). 2.4.2. Elution Supernatant feed solution was removed to a minimum level so that resin bed stayed humidified. Eluent was inlet at room temperature, at same flow rate and flow direction used for adsorption.
ð3Þ
where kYN is the Yoon–Nelson rate constant [1/min], t is time [min] and s is time required for 50% sorbate breakthrough [min]. The amount of metal adsorbed on the fixed bed is half of the total amount entering the adsorption bed within 2s period [14] and is expressed for a given bed as:
q ¼ ð0:5c0 ððQ V =1000Þ2sÞÞ=m ¼ ðc0 Q V sÞ=1000m
ð4Þ
2.5.3. Bohart–Adams model [15] The Bohart–Adams model is widely used modelling fixed bed BTCs for environmental sorption and biosorption processes [16]. It assumes that the adsorption rate basically depends on the reaction between the sorbent and the free sorption sites on the surface, a rectangular (irreversible) isotherm and a low concentration field (c < 0.15c0). Therefore this model is more suitable to describe the initial phase of the BTC [17]. The linear form is represented as:
lnðct =c0 Þ ¼ ðkBA c0 tÞ ðkBA N0 ðh=Q l ÞÞ
ð5Þ
where kBA is the Bohart–Adams rate constant [L/mg min], N0 is saturation concentration [mg/L] which is connected to q via q = N0BV/m where BV is the bed volume [L] and m the mass of adsorbent [g], h is the bed height of column [cm], Ql is the linear flow rate [cm/min], c0 and ct are the initial metal concentration [mg/L] at t = 0 min and the residual concentration at time t, respectively. 2.5.4. Wolborska model [18] The Wolborska model is also a simplified adsorption model describing breakthrough for low concentration ranges emphasising the general mass transfer for the diffusion mechanism [19]. The initial stage of BTC is controlled by film diffusion with a constant
282
M. Trautmann, H.-J. Holdt / Separation and Purification Technology 149 (2015) 279–287
kinetic coefficient and the concentration profile moves axially with a constant velocity. The final BTC as well as the width of concentration profile in the column is nearly constant [12]. The linearised equation for the mass transfer in fixed bed adsorption is represented as follows:
lnðct =c0 Þ ¼ ðba c0 Þ=N0 Þt ððba hÞ=Q l Þ
ð6Þ
where ba is the kinetic coefficient of the external mass transfer [1/h], 3. Results and discussion Sorption behaviour of PtIV from PtIV/RuIII solutions did not differ from PtIV/RuIV mixtures and is therefore not shown. 3.1. Influence of hydrochloric acid concentration Manipulating cHCl resulted in several effects. At first, proton concentration affected the protonation equilibrium of sulphoxide (R1R2SO + HCl ¢ R1R2SOH+ + Cl) and therefore the amount of electrostatic bound ions. At second, chloride concentration controls the speciation responsible for the metal complex charge (e.g. [RuCl5(H2O)]2 + Cl ¢ [RuCl6]3 + H2O) and consequently the quantity of competitor ions. The adsorption of PtIV and RuIII from PtIV/RuIII solutions as well as PtIV and RuIV from PtIV/RuIV solutions onto resin 1 was analysed using fixed bed column as a function of hydrochloric acid concentration. BTCs were obtained at cHCl = 0.1, 1.0, 3.0, 6.0 and 9.5 mol/L, cPt = 20 mg/L, cRu = 10 mg/L, Q = 1.0 mL/min, T = 25 °C and h = 0.8 cm (m = 100 mg) (Fig. 2). On increasing hydrochloric acid concentration to cHCl = 3.0 mol/L, the breakthrough volume increased coinciding with the rising sulphoxide protonation. Further increase led to a decrease due to rising concentration of competitor ions HCI 2 . Contact with cHCl = 9.5 mol/L progressively degraded resin due to acidic ether cleavage. FTIR spectrum showed a decrease of m(COC) 1090 cm1 and the appearance of two new bands at 671 cm1 (C–Cl) and 1265 cm1 (C–H from –CH2Cl). Methanol (10% v/v) was added to cHCl = 3.0 mol/L to enhance the wetting of the hydrophobic matrix but resulted no improvement. Nonetheless BTCs of PtIV were distinctively broadened what could be attributed to several reasons, e.g. Eddy diffusion, axial diffusion, dispersion, dilution, wall channelling, mass transfer effects.
Fig. 2. Effect of hydrochloric acid concentration on BTC of PtIV from PtIV/RuIII solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, T = 25 °C, Q = 1.0 mL/min, h = 0.8 cm, downflow). Symbols: (j) 0.1 mol/L, (4) 1.0 mol/L, (.) 3.0 mol/L, (h) 6.0 mol/L and (s) 9.5 mol/L.
Fig. 3. Effect of hydrochloric acid concentration on BTC of RuIII from PtIV/RuIII solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, T = 25 °C, Q = 1.0 mL/min, h = 0.8 cm, downflow). Symbols: (j) 0.1 mol/L, (4) 1.0 mol/L, (.) 3.0 mol/L, (h) 6.0 mol/L and (s) 9.5 mol/L.
As seen in Fig. 3, RuIII was not adsorbed at any hydrochloric acid concentration. RuIII forms octahedral aquachloro complexes of the general formula [RuCln(H2O)6n]3n (n = 0–6) depending on chloride concentration. At cHCl 6 0:1 mol=L, mainly cationic and neutral species exist. At cHCl P 1:0 mol=L, a crude mixture of nearly all species (n = 1–6) form at which [RuCl4(H2O)2] prevails. Distribution shifts with rising chloride concentration to more negatively charged ions so that mainly [RuCln(H2O)6n]3n (n = 5–6) predominate at cHCl P 6:0 mol=L. The highly charged hexachloro complex possesses a high charge density, high hydration energy and a huge steric hindrance. The difficulty of packing three sulphoxide molecules around one bulky ruthenate forms an obstacle leaving the anion in solution. The aquachloro complexes remained in solution because hydrogen bonding of the water ligands with water molecules around made them more dissolvable in the aqueous phase instead on the surface of the hydrophobic PS-DVB matrix. The neutral complex [RuCl3(H2O)3] could not form any electrostatic interactions whereas cationic complexes were repulsed. Replacement of chloride or water from the inner coordination sphere by sulphoxide was less successful at room temperature due to kinetic inertness of RuIII complexes. RuIV was expected to retain on resin like all divalent hexachloro complexes but was just particulately sorbed between 3 6 cHCl 6 6 mol=L (Fig. 4) at which the resin became deeply brown. BTCs exhibited a plateau-like progression with a maintaining sorption level at around 35%. The residual steady bleed of ruthenium from the beginning was inconsistent to [RuCl6]2 indicating other species. UV/Vis spectra of eluates exhibited [RuCl6]3 due to the specific absorbance at 229 nm and 349 nm. Stannometric titration revealed around 65% RuIII. Reduction through resin was excluded because nil evidences were found in FTIR spectrum for sulfone formation. Redox potential of stock solution (T = 25 °C, cHCl = 9.5 mol/L) had decreased from E0 = 1.10 V to E0 = 0.80 V during storage. This lowering was due to the continuously outgassing of chlorine during storage. RuIV is stable at E0 P 1:30 V (vs. NHE) [20] means E0 P 1:10 V (vs. Ag/AgCl) otherwise tends to reduction. Broszkiewicz [8] reported reduction of RuIV already at cHCl > 5.0 mol/L. The hexachloro complex of RuIV is only stable at high chloride concentrations greater than 9.0 mol/L [20] otherwise tends to hydrolysis which forms a jumble of monomer and dimeric chloro complexes including aqua, hydroxo and oxo groups. Those complex mixtures are hardly to handle for a well-defined adsorption.
M. Trautmann, H.-J. Holdt / Separation and Purification Technology 149 (2015) 279–287
Fig. 4. Effect of hydrochloric acid concentration on BTC of RuIII/IV from PtIV/RuIII/IV solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, T = 25 °C, Q = 1.0 mL/min, h = 0.8 cm, downflow). Symbols: (j) 0.1 mol/L, (4) 1.0 mol/L, (.) 3.0 mol/L, (h) 6.0 mol/L and (s) 9.5 mol/L.
283
Fig. 5. Effect of temperature on BTC of PtIV from PtIV/RuIII solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, cHCl = 3.0 mol/L, Q = 1.0 mL/min, h = 0.8 cm, E0 = 0.48 V, downflow). Symbols: (4) 25 °C, (.) 60 °C and (h) 90 °C.
Subjected to previous mentioned facts, RuIV stock solution was freshly prepared adjusting E0 = 1.10 V (T = 25 °C, cHCl = 9.5 mol/L) with chlorine to completely oxidise RuIII and stabilise RuIV as [RuCl6]2. Unfortunately the high redox potential of freshly made feed solution resulted resin degradation. Sulphoxide was oxidised to sulfone and already sorbed metals completely flushed out. Further oxidants like NaClO3 and H2O2 proved inadequate for fixed bed application due to incessantly formed chlorine bubbles which ruptured the bed.
3.2. Influence of temperature Temperature is also an important parameter influencing sorption processes. The poly(styrene-co-divinylbenzene) matrix with its cross-linked polymeric chains appears as a relatively rigid construct. Diffusion into a branched and less flexible structure is difficult but improved at higher temperature because polymer density and chain rigidity decrease whereas ion mobility increase. Moreover inner-sphere substitution kinetic is governed by temperature and establishing a second sorption mechanism in addition to ion association. The effect of temperature on the adsorption of PtIV and RuIII from PtIV/RuIII solutions onto resin 1 was examined by varying the temperature to 25 °C, 60 °C and 90 °C at cHCl = 3.0 mol/L and 6.0 mol/L, at constant influent concentration cPt = 20 mg/L and cRu = 10 mg/L, Q = 1.0 mL/min, h = 0.8 cm and downward flow, as shown in Figs. 5–8. Handling cHCl = 6.0 mol/L takes the advantage of a constant boiling azeotrope offering an incessant acidity over a long heating period as well as the possibility of recycling by circuit distillation. As seen in Fig. 5 and Fig. 6, PtIV sorption decreased with rising temperature. Curve slope became steeper and breakthrough shifted to smaller effluent volumes lowering the operating capacity. This behaviour can be explained by exothermic sorption character. Previous batch experiments revealed physisorption of PtIV [5]. This kind of mechanism is always exothermic because the increasing arrangement of adsorbate on adsorbent surface reduces entropy. Concurrently, effluent concentration did not reach influent concentration at elevated temperatures merely a plateau like very long tail ran out at c/c0 < 1. This kind of tailing correlates with the diffusion into stagnant water zones. Solubility of gases in solution decrease with increasing temperature so that outgas produces microbubbles of water steam and HCl. Every rupture diversify bed package and generates zones of immobile water. Diffusion into stagnant zones is
Fig. 6. Effect of temperature on BTC of PtIV from PtIV/RuIII solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, cHCl = 6.0 mol/L, Q = 1.0 mL/min, h = 0.8 cm, E0 = 0.48 V, downflow). Symbols: (4) 25 °C, (.) 60 °C and (h) 90 °C.
Fig. 7. Effect of temperature on BTC of RuIII from PtIV/RuIII solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, cHCl = 3.0 mol/L, Q = 1.0 mL/min, h = 0.8 cm, E0 = 0.48 V, downflow). Symbols: (s) 25 °C, (.) 60 °C and (h) 90 °C.
284
M. Trautmann, H.-J. Holdt / Separation and Purification Technology 149 (2015) 279–287
Fig. 8. Effect of temperature on BTC of RuIII from PtIV/RuIII solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, cHCl = 6.0 mol/L, Q = 1.0 mL/min, h = 0.8 cm, E0 = 0.48 V, downflow). Symbols: (s) 25 °C, (.) 60 °C and (h) 90 °C.
increased due to higher ion mobility at higher temperatures. Ions which enter a stagnant zone are trapped and the only mechanism to escape is molecular diffusion. The capacity of stagnant regions is high relative to the capacity of mobile regions. Sorption at T = 90 °C was hardly to handle because operating near boiling point revealed a vigorous bubble formation. More precisely, microbubbles coalesced to macroscopic size, held by downward stream until the buoyant force exceeded. Every uplifting bubble generated voids which promoted channelling and therewith an early breakthrough because ions could move more rapidly than the average flow velocity. As indicated in Figs. 7 and 8, sorption of RuIII improved with rising temperature due to the fact that the inner-sphere substitution is strongly temperature dependent. This essentially comprises aquation at which chloride is substituted by water forming lower charged aquachloro complexes which are more prone to inner-sphere substitution by sulphoxide. The incorporation of sulphoxide into the inner coordination sphere occurred via oxygen atom due to the appearance of a new FTIR band at 874 cm1 [21]. Breakthrough behaviour at T = 90 °C and cHCl = 6.0 mol/L differed from cHCl = 3.0 mol/L due to shifted speciation distribution. Hence, T = 90 °C proved inadvisable so that T = 60 °C was favoured at which the problem of stagnant water zones remained similar to PtIV sorption.
3.3. Influence of flow direction The most common direction of reactant flow is downward through the bed especially recommended in cases of a wide variation of feed flow rate. It stabilises the bed and prevents lifting particles out of the column. Attrition and entrainment of the finest particles are minimised. Unfortunately bed compression as well as gravitation of fine particles can cause pressure drops and maldistribution of the flow. Concerning this matter, upward flow is often preferred for industrial applications especially when gas is produced (e.g. packed bed bioreactors, enzyme columns). It does not compress the bed as downward flow does whereas microbubbles escape with flow fluidizing the bed. This strictly dependents on flow velocity but minimizes phenomena like blockage or stagnant water zones. The effect of flow direction for the adsorption of PtIV and RuIII from PtIV/RuIII mixtures onto resin 1 was investigated by reversing flow (upflow) at influent concentration cPt = 20 mg/L and
Fig. 9. Effect of upflow on BTC of PtIV and RuIII from PtIV/RuIII solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, cHCl = 6.0 mol/L, T = 60 °C, Q = 1.0 mL/min, h = 0.8 cm, E0 = 0.48 V). Symbols: (s) PtIV, (.) RuIII.
cRu = 10 mg/L, cHCl = 3.0 mol/L, T = 25 °C, Q = 1.0 mL/min, E0 = 0.48 V and h = 0.8 cm. Adsorption behaviour of PtIV and RuIII was negligible affected at room temperature and is therefore not displayed. Adsorption behaviour of RuIII at cHCl = 6.0 mol/L and T = 60 °C was apparently influenced. As seen in Fig. 9, plateau-like tailing did not appear means upward flow avoided successfully formation of stagnant water zones. BTC of RuIII assumed slightly S-shape but remained non-sigmoid (with respect to c/c0 = 0.5) by ordinary tailing what can be attributed to slow sorption kinetic and excessive flow rate. Hence, upward flow was favoured over downward flow at elevated temperatures. 3.4. Influence of redox potential The redox potential has a wide influence over the oxidation state of a metal. [PtCl6]2 is relatively redox stable opposite to [RuCl6]2 which easily reduces to [RuCl6]3. Due to the easier arrangement two bulky counterions instead of three, the divalent [RuCl6]2 has a higher affinity to the resin and binds electrostatically more strongly than trivalent [RuCl6]3. To that effect the goodness of ruthenium adsorption is consequentially adjustable by manipulating the redox potential of the solution. As seen in Fig. 9, the obvious c/c0 > 0 for RuIII sorption at the beginning indicated a steady bleed of ruthenium which was caused by [RuCl6]3 and less astonishing because of the predominating species [RuCl5(H2O)]2 and [RuCl6]3 of RuIII at cHCl = 6.0 mol/L. Redox potential of the feed solution was raised from E0 = 0.48 V to E0 = 0.98 V (T = 25 °C, cHCl = 6.0 mol/L) by chlorine to oxidise both species into [RuCl6]2. This was definitely not high enough to convert RuIII completely to RuIV rather to a RuIII/IV mixture. Keeping E0 < 1.00 V was necessary to avoid resin degradation. As illustrated in Fig. 10, increasing redox potential revealed improvement of ruthenium sorption due to speciation shift whereas PtIV sorption was hardly influenced and therefore not displayed. Hence, ruthenium was enriched to 90% whereupon [RuCl6]3 bleed had to be accepted. Nonetheless BTC broadening and tailing still occurred. 3.5. Influence of flow rate Flow rate controls the contact time between adsorbate and adsorbent what is important if adsorption is governed by slow kinetics. So, residence time on the column should be long enough
M. Trautmann, H.-J. Holdt / Separation and Purification Technology 149 (2015) 279–287
285
Fig. 10. Effect of redox potential on BTC of RuIII/IV from PtIV/RuIII/IV solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, cHCl = 6.0 mol/L, T = 60 °C, Q = 1.0 mL/min, h = 0.8 cm, upflow). Symbols: (.) 0.48 V, (s) 0.98 V.
Fig. 12. Effect of flow rate on BTC of RuIII/IV from PtIV/RuIII/IV solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, cHCl = 6.0 mol/L, T = 60 °C, h = 0.8 cm, E0 = 0.98 V, upflow). Symbols: (h) 1.0 mL/min, (.) 0.5 mL/min and (s) 0.1 mL/min.
to establish adsorption equilibrium. Especially inner-sphere substitution of ruthenium aquachloro complexes reveals a slow sorption kinetic. A high flow rate performs a quick execution but also means shorter time for saturation resulting in early breakthrough, increased dispersion and poor separation. A low flow rate offers more residence time but should not be too low otherwise process becomes uninteresting for practical work and longitudinal diffusion will dominate broadening sorption zone. The effect of flow rate was studied for the adsorption of PtIV from PtIV/RuIII mixtures onto resin 1 at flow rates Q = 0.1, 0.5 and 1.0 mL/min at influent concentration cPt = 20 mg/L and cRu = 10 mg/L, cHCl = 3.0 mol/L, T = 25 °C, h = 0.8 cm and downward flow. As indicated in Fig. 11, PtIV sorption was influenced by flow rate whereas at lower ones the breakthrough occurred later resulting a higher effective adsorption. The BTC obtained at Q = 1.0 mL/min and Q = 0.5 mL/min were similar in shape but differed in breakthrough volume which was higher at lower velocity. Further lowering to Q = 0.1 mL/min subordinated dispersion, Eddy diffusion and mass transfer effects what reduced curve spreading. RuIII was not adsorbed from PtIV/RuIII mixtures at any flow rate due
to the facts mentioned in chapter 3.1 and is therefore not displayed. The effect of flow rate was examined for the adsorption of Ru from PtIV/RuIII/IV mixtures onto resin 1 at constant influent concentration cPt = 20 mg/L and cRu = 10 mg/L, cHCl = 6.0 mol/L, T = 60 °C, E0 = 0.98 V, h = 0.8 cm and upward flow. Ruthenium sorption was markedly influenced by flow rate as seen in Fig. 12. Irrespective of [RuCl6]3 bleed, breakthrough volume increased with decreasing flow rate. BTC asymmetry decreased with decreasing velocity means tailing was kinetically caused by slow sorption kinetic. Broadening decreased with decreasing velocity due to diminished influence of dispersion, diffusion and mass transfer effects. BTC gained more S-shape and steeper slope. High flow rates forced ruthenium complexes through column so that there was no adequate residence time to establish adsorption equilibrium. The effect on PtIV sorption was negligible at these conditions due to insignificant adsorption and is therefore not displayed.
3.6. Influence of bed height
Fig. 11. Effect of flow rate on BTC of PtIV from PtIV/RuIII solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, cHCl = 3.0 mol/L, T = 25 °C, h = 0.8 cm, E0 = 0.48 V, downflow). Symbols: (4) 1.0 mL/min, (s) 0.5 mL/min and (j) 0.1 mL/min.
Bed height is set by mass of adsorbent. Higher bed heights mean larger amounts of adsorbent providing more available binding sites. Additionally residence time increases due to longer pathways. BTCs for adsorption of PtIV from PtIV/RuIII mixtures onto resin 1 at various bed heights, at the influent concentrations of cPt = 20 mg/L and cRu = 10 mg/L, cHCl = 3.0 mol/L, T = 25 °C, Q = 0.5 mL/min, E0 = 0.48 V and downward flow are shown in Fig. 13. The corresponding BTCs of RuIII are not diagrammed because of non-adsorption. It is observed that breakthrough volume and equilibrium sorption capacity decreased with decrease of bed height due to the diminished availability of sorption sites. Higher bed height promoted broadening what can be explained by axial dispersion. The effect of bed height on the BTCs of Ru from PtIV/RuIII/IV solutions at various bed heights, at the influent concentrations of cPt = 20 mg/L and cRu = 10 mg/L, cHCl = 6.0 mol/L, T = 60 °C, Q = 0.5 mL/min, E0 = 0.98 V and upward flow is shown in Fig. 14. The corresponding BTCs of PtIV are not diagrammed due to negligible sorption. The results indicate similar to PtIV that the decrease of
286
M. Trautmann, H.-J. Holdt / Separation and Purification Technology 149 (2015) 279–287 Table 2 Results of Thomas, Yoon–Nelson, Bohart–Adams and Wolborska model for adsorption of platinum (cPt = 20 mg/L) and ruthenium (cRu = 10 mg/L) onto resin 1.
Fig. 13. Effect of bed height on BTC of PtIV from PtIV/RuIII solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, cHCl = 3.0 mol/L, T = 25 °C, Q = 0.5 mL/min, E0 = 0.48 V, downflow). Symbols: (.) h = 0.4 cm, (s) h = 0.8 cm.
Pt
Ru
Hydrochloric acid (mol/L) Temperature (°C) Flow rate (mL/min) Flow direction Bed height (cm) Redox potential (V) qexp (mg/g)
3.0 25 0.5 downflow 0.8 0.48 7.80
6.0 60 0.5 upflow 0.8 0.98 13.04
Thomas model kTH 103 (L/mg min) q (mg/g) R2
2.99 7.10 0.95
1.10 12.68 0.96
Yoon–Nelson kYN 102 (1/min) s (min) q (mg/g) R2
6.01 70.84 7.10 0.95
1.11 260.23 12.96 0.96
Bohart–Adams model kBA 103 (L/mg min) N0 103 (mg/L) q (mg/g) R2
1.25 4.29 6.74 0.78
0.59 8.69 13.64 0.86
Wolborska model ba (1/h) N0 103 (mg/L) R2
0.50 6.71 0.78
0.48 13.58 0.86
3.8. Elution
Fig. 14. Effect of bed height on BTC of Ru from PtIV/RuIII/IV solutions onto resin 1 (cPt = 20 mg/L, cRu = 10 mg/L, cHCl = 6.0 mol/L, T = 60 °C, Q = 0.5 mL/min, E0 = 0.98 V, upflow). Symbols: (j) h = 0.8 cm, (4) h = 0.4 cm.
bed height diminished available sorption sites and reduced breakthrough volume and equilibrium sorption capacity.
3.7. Column dynamics study For the purpose of characterising the fixed bed column behaviour and scaling up for industrial application, four models: Thomas, Yoon–Nelson, Bohart–Adams and Wolborska were fitted to the experimental data found for best separation conditions. Application to the data started at the concentration ratio c/c0 > 0.05 that is 5% breakthrough until c/c0 > 0.90 that is 90% breakthrough. All coefficients and relative constants were obtained using linear regression analysis and are summarised in Table 2. Platinum sorption data fitted well with Thomas (R2 = 0.95) and Yoon–Nelson (R2 = 0.95) model whereupon Bohart–Adams and Wolborska model showed a poor fit (R2 = 0.78). Ruthenium sorption data showed also a good fit to the Thomas (R2 = 0.91) and Yoon–Nelson (R2 = 0.91) model. Bohart-Adams (R2 = 0.73) and Wolborska (R2 = 0.73) model were unsuited to describe complete area of RuIII sorption.
The removal of adsorbed metal from the resin was examined using the eluents: water, hydrochloric acid (cHCl = 0.1, 3.0, 6.0 mol/L) and thiourea (Tu) (cTu = 0.5 mol/L) in hydrochloric acid (cHCl = 0.1 mol/L). It turned out that a complete pump out of the feed solution after adsorption decreased significantly the desorption yield. This could be explained by seclusion of already established flow channels forming nonadvective regions. Feed solution was removed to a minimum level so that resin stayed humidified. Platinum was eluted from the column at room temperature during washing step using hydrochloric acid and water. At cHCl 6 0.1 mol/L, sulphoxide is deprotonated eliminating any electrostatic retardation. At cHCl P 1:0 mol=L, PtIV ions are displaced by the surplus concentration of chloride ions at which elution rate increased with hydrochloric acid concentration. Desorption caused by deprotonation occurred much faster compared to the continuous displacement by competitor ions. Hence, it is advisable to wash firstly with hydrochloric acid to minimise the loss of platinum while removing ruthenium. Desorption of platinum occurred quantitative from the resin. Elution of ruthenium only took place with thiourea via nucleophilic substitution forming the neutral Ru(Tu)3 complex. Desorption rate accelerated with rising temperature due to substitution kinetic. Hydrochloric acid (cHCl = 0.1, 3.0, 6.0 mol/L) removed unattached ruthenium and platinum. Water was inadequate in case of ruthenium adsorption due to hydrolysis forming a black precipitate of ruthenium dioxide. 4. Conclusion Adsorption of platinum and ruthenium from hydrochloric acidic solution by powdered sulphoxide modified polystyrene–divinyl benzene (d50 < 30 lm) was investigated in a continuous flow fixed-bed column. The breakthrough curves were determined at various concentrations of hydrochloric acid, temperatures, flow rates, flow rate directions, redox potentials and bed heights. Metal sorption was selectable depending on parameter settings
M. Trautmann, H.-J. Holdt / Separation and Purification Technology 149 (2015) 279–287
enabling separation. Best PtIV sorption from a mixture of PtIV/RuIII was achieved at cHCl = 3.0 mol/L, T = 25 °C, Q = 0.5 mL/min, E0 = 0.48 V using downward flow. Best ruthenium sorption from a mixture of PtIV/RuIII/IV at which ruthenium was enriched to 90% occurred at cHCl = 6.0 mol/L, T = 60 °C, Q = 0.5 mL/min, E0 = 0.98 V using upward flow. Residual 10% remained in solution as [RuCl6]3 due to poor sorption affinity. The adsorbent is capable of holding a maximum of 7.80 mg Pt/g at 20 mg/L of influent concentration and 13.04 mg Ru/g at 10 mg/L of influent concentration. The column performance at these conditions was analysed using Thomas, Yoon–Nelson, Bohart–Adams and Wolborska model. Thomas and Yoon–Nelson model described adsorption behaviour of platinum and ruthenium onto 1 more reasonably than Bohart– Adams and Wolborska model due to high correlation coefficients. Elution of platinum occurred via deprotonation at cHCl 6 0:1 mol=L or expulsion at cHCl P 1:0 mol=L. Ruthenium was eluted by thiourea (cTu = 0.5 mol/L) in diluted hydrochloric acid (cHCl = 0.1 mol/L). Referring to the operating capacity, separation of platinum and ruthenium was achieved setting the above mentioned conditions. Platinum adsorbed at room temperature at which ruthenium passed solely through the column as RuIII. Elution took place with hydrochloric acid and water. Ruthenium was adsorbed to 90% at elevated temperature at which further repetitions were necessary for complete elimination from the feed solution. Elution took place with thiourea as Ru(Tu)3 complex. References [1] Gianfranco Pistoia, Batteries for Portable Devices, Elsevier (Science & Technology), Amsterdam, 2005, p. 194. [2] X. Cheng, C. Peng, Y. Ma, L. Chen, Y. Zhang, Lifetime studies of catalyst activity and microstructure in a PEMFC, in: J. Prakash, D. Chu, D. Scherson, M. Enayetullah, I. Tae Bae (Eds.), Fundamental Understanding of Electrode Processes, The electrochemical society Inc., Pennington, 2005, p. 177. [3] D. Stolten, Hydrogen and Fuel Cells: Fundamentals, Technologies and Applications, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, 2010. p. 54.
287
[4] N. Wiberg, E. Wiberg, A.F. Hollemann, Lehrbuch der Anorganischen Chemie, Walter de Gruyter, Berlin, 2007. p. 1667. [5] M. Trautmann, S. Lubahn, H.-J. Holdt, Preparation, characterisation and properties of sulphoxide modified resins for solid-phase extraction of PtIV, RuIII and RuIV from hydrochloric acid, React. Funct. Polym. 83 (2014) 84–87. [6] E.A. Seddon, K.R. Seddon, The Chemistry of Ruthenium, Elsevier Science Publishers B.V, Amsterdam, 1984. p. 159. [7] K.R. Hyde, E.W. Hooper, J. Waters, J.M. Fletcher, F- and b-Ruthenium trichloride, J. Less Common Met. 8 (1965) 428–434. [8] R.K. Broszkiewicz, Redox reactions of RuCl3 and RuCl2 6 6 : a pulse radiolysis study, Radiat. Phys. Chem. 15 (1980) 133–138. [9] R.A. Scott, C.M. Lukehart, Applications of Physical Methods to Inorganic and Bioinorganic Chemistry, Wiley & Sons, Hoboken, NJ, 2007. p. 23. [10] H.C. Thomas, Heterogeneous ion exchange in a flowing system, J. Am. Chem. Soc. 66 (1944) 1664–1666. [11] Z. Saadi, R. Saadi, R. Fazaeli, Fixed-bed adsorption dynamics of Pb (II) adsorption from aqueous solution using nanostructured c-alumina, J. Nanostructure Chem. 3 (2013) 48. [12] Z. Xu, J. Cai, B. Pan, Mathematically modeling fixed-bed adsorption in aqueous systems, J. Zhejiang Univ. Sci A (Appl Phys & Eng) 14 (2013) 155–176. [13] Y.H. Yoon, J.H. Nelson, Application of gas adsorption kinetics. I. A theoretical model for respirator cartridge service life, Am. Ind. Hyg. Assoc. J. 45 (1984) 509–516. [14] P. Sivakumar, P.N. Palanisamy, Packed bed column studies for the removal of Acid blue 92 and Basic red 29 using non-conventional adsorbent, Indian J. Chem. Technol. 16 (2009) 301–307. [15] G.S. Bohart, E.Q. Adams, Some aspects of the behavior of charcoal with respect to chlorine, J. Am. Chem. Soc. 42 (1920) 523–544. [16] K.H. Chu, Fixed bed sorption: setting the record straight on the Bohart-Adams and Thomas models, J. Hazard. Mater. 177 (2010) 1006–1012. [17] F. Pagnanelli, F. Ferella, I. De Michelis, F. Veglio, Adsorption onto activated carbon for molybdenum recovery from leach liquors of exhausted hydrotreating catalysts, Hydrometallurgy 110 (2011) 67–72. [18] A. Wolborska, Adsorption on activated carbon of p-nitrophenol from aqueous solution, Water Res. 23 (1989) 85–91. [19] P.-S. Keng, S.-L. Lee, S.-T. Ha, Y.-T. Hung, S.-T. Ong, Cheap materials to clean heavy metal polluted waters, in: E. Lichtfouse, D. Robert, J. Schwarzbauer (Eds.), Green materials for energy, Products and Depollution, Springer Science & Business Media, Heidelberg, 2013, p. 396. [20] Gmelins Handbuch der anorganischen Chemie, 8. Auflage, System No. 63, Ruthenium, Ergänzungsband, Verlag Chemie, Weinheim, 1970. [21] M. Calligaris, Structure and bonding in metal sulfoxide complexes: an update, Coord. Chem. Rev. 248 (2004) 351–375.