This article describes an experimental protocol using electrospray-ion mobility-mass spectrometry, semi-empirical quantum calculations, and energy-resolved threshold collision-induced dissociation to measure the relative thermochemistry of the dissociation of related ternary metal complexes.
This article describes an experimental protocol using electrospray-ion mobility-mass spectrometry (ES-IM-MS) and energy-resolved threshold collision-induced dissociation (TCID) to measure the thermochemistry of the dissociation of negatively-charged [amb+M(II)+NTA]– ternary complexes into two product channels: [amb+M(II)] + NTA or [NTA+M(II)]– + amb, where M = Zn or Ni and NTA is nitrilotriacetic acid. The complexes contain one of the alternative metal binding (amb) heptapeptides with the primary structures acetyl-His1-Cys2-Gly3-Pro4-Tyr5-His6-Cys7 or acetyl-Asp1-Cys2-Gly3-Pro4-Tyr5-His6-Cys7, where the amino acids' Aa1,2,6,7 positions are the potential metal-binding sites. Geometry-optimized stationary states of the ternary complexes and their products were selected from quantum chemistry calculations (presently the PM6 semi-empirical Hamiltonian) by comparing their electronic energies and their collision cross-sections (CCS) to those measured by ES-IM-MS. From the PM6 frequency calculations, the molecular parameters of the ternary complex and its products model the energy-dependent intensities of the two product channels using a competitive TCID method to determine the threshold energies of the reactions that relate to the 0 K enthalpies of dissociation (ΔH0). Statistical mechanics thermal and entropy corrections using the PM6 rotational and vibrational frequencies provide the 298 K enthalpies of dissociation (ΔH298). These methods describe an EI-IM-MS routine that can determine thermochemistry and equilibrium constants for a range of ternary metal ion complexes.
This study describes a new technique using a commercially available ion mobility-mass spectrometer that allows the determination of the relative thermochemistry for the dissociation of an alternative metal binding (amb) ternary metal complex [amb+M(II)+NTA], where M = Zn or Ni and NTA = nitrilotriacetic acid (Figure 1). These reactions model the dissociation of the amb-tagged recombinant protein attached to the NTA-immobilized metal during immobilized metal affinity chromatography (IMAC)1,2. As an example, this method is described using the amb heptapeptide tags of amb A and H (Figure 2) (chosen from the previous studies3,4,5,6,7,8,9,10,11,12) that exhibit Zn(II) and Ni(II)-binding properties and, thus, have potential applications as purification tags. However, the described process can be used to evaluate thermochemical energies in any organometallic system. These amb peptides have metal-binding sites in the Aa1-Aa2 and Aa6-Aa7 positions that compete with the carboxylate and amine sites of the NTA. The three central amb amino acids provide a spacer (Gly3), the hinge for the two arms (Pro4), and a long-distance π-metal cation interaction (Tyr5).
The overall 1− charge state of the [amb+M(II)+NTA]– complexes is determined by the protonation state of their potential binding sites. Since there is Ni(II) or Zn(II) with the 2+ oxidation state, there must be a net of three deprotonated negatively-charged sites. The molecular modeling of the [amb+M(II)+NTA]– complexes predicts that these are two protons from the NTA and one proton from the amb (i.e., [amb-H+M(II)+NTA-2H]–). The product channels contain an ionic species and a neutral species (i.e., [NTA-3H+M(II)]– + amb or [amb-3H+M(II)]– + NTA). In the manuscript, "-3H" is excluded in the names of the complexes, but the reader should know that the -3H is implied. The instrument measures the relative intensities of the two ionic mass-to-charge (m/z) species. A major attribute of ES-IM-MS analyses is that it allows the examination of the reactivity of a specific m/z species, as utilized here and in previous amb studies3,4,5,6,7,8,9,10,11,12.
Acquiring thermochemical data for large complexes using collision-induced dissociation is a subject of significant interest13,14. Methodologies including the kinetic method are not conducive to fitting data over a range of energies, nor do they account for multi-collision environments15,16,17,18. Here, the threshold CID (TCID) method, developed using guided ion beam tandem mass spectrometry by Armentrout, Ervin, and Rodgers is applied19 to a new ES-IM-MS instrument platform utilizing traveling-wave ion guides. The TCID method allows for relative thermochemical analysis of the dissociation of the ternary complexes into their two product channels and includes a threshold law describing the transfer of collision energy between the translational energy of the reactant (ternary complex in this research) and an inert target gas (argon in this case). The method includes integration over the reactant's internal energy distribution20, the translational energy distributions between the reactant and target gas21, and the total angular momentum distributions22,23. A dissociation probability and statistical Rice-Ramsperger-Kassel-Marcus (RRKM) correction of the kinetic shifts resulting from the limited time window for observation of the products are included24. For two independent product channels, the competitive TCID method allows for the simultaneous fitting of the two competing product channels. Dissociation of the complex is through an orbiting transition state, which has the properties of the products but is held together by a locked-dipole25. The TCID method is incorporated into the CRUNCH program26, and the operation of the user interface is described here to evaluate the thermochemistry of the two dissociation channels of the ternary [amb+M(II)+NTA]– complexes. The CRUNCH program is available upon request from the developers26.
NOTE: Figure 1 shows an overview of the protocol.
1. Preparation of reagents
2. Preparation of stock solutions
3. Electrospray-ion mobility-mass spectrometry (ES-IM-MS) collision-induced dissociation (CID) analysis
4. ES-IM-MS collision cross-section (CCS) analysis
5. Analysis of ES-IM-MS CID data
6. Analysis of the average arrival times for determining collision cross-sections (CCS)
7. Computational methods
8. CRUNCH modeling
The competitive collision-induced dissociation of the [amb+M(II)+NTA]– ternary complexes of A and H into [amb+M(II)]– + NTA or [NTA+M(II)]– + amb, are shown in Figure 3. The amb is shown as either A or H and the M = Zn or Ni. The [A+Zn(II)+NTA]– ternary complex (Figure 3A) exhibits apparent thresholds of about 0.7 eV collision energy (CE) to produce [A+Zn(II)]– and about 0.9 eV to produce [NTA+Zn(II)]–. The dissociation of the [A+Ni(II)+NTA]– complex (Figure 3B) exhibits similar thresholds (~1.1 eV) for both the [NTA+Ni(II)]– and [A+Ni(II)]– products, with [NTA+Ni(II)]– increasing to 90% relative intensity, while the intensities of [A+Ni(II)]– do not rise above 18%. For the [H+Zn(II)+NTA]– ternary complex (Figure 3C), the main product is [H+Zn(II)]– rising from a threshold of about 0.6 eV to about 85% relative intensity, and at energies above 1.0 eV, the [NTA+Zn(II)]– rises to about 30%. There is also a channel for water loss from [H-H2O+Zn(II)]–. For [H+Ni(II)+NTA]– (Figure 3D), the [H+Ni(II)]– rises from a threshold of about 0.9 eV to about 40% relative intensity, while the [NTA+Ni(II)]– rises from ~1.0 eV to about 80%. Included on the graphs is the CE where the ternary complex is 50% dissociated. The Ni(II) ternary complexes require 0.31-0.37 eV higher CE than their Zn(II) ternary complex counterparts to be 50% dissociated. This suggests the Ni(II) complexes are more stable and require higher CE to dissociate, which is further investigated using the TCID technique.
Figure 4 illustrates the competitive TCID method, which allows for the simultaneous fitting of the two competing product channels.
[amb+M(II)+NTA] → [amb+M(II)]– + NTA (1)
[amb+M(II)+NTA] → [NTA+M(II)]– + amb (2)
The potential energy surface (PES) illustrates the energized ternary complex dissociating into the competing product channels and shows the PM6 geometry-optimized species used to model the dissociation of [ambH+Zn(II)+NTA]–. Included in the PES are the density of states of the ternary complex and the sum of states of the products. The 0 K threshold energies, E1 and E2, equate to the 0 K enthalpy change for reactions 1 and 2.
Figure 5 shows the structures of the other three geometry-optimized ternary complexes used in this study. These species were chosen based on their predicted electronic and zero-point energies and their agreement with the IM-MS-measured collision cross-sections (CCSHe). Table 1 shows there is an agreement between the ternary complexes LJ CCSHe and the experimental IM-MS CCSHe because they agree within their mutual uncertainties. The conformations of the [amb+M(II)] and amb were based on the findings of our previous DFT modeling3,4,5,6. The molecular parameters of these PM6 conformers were used in the TCID modeling of the energy-resolved dissociations of the ternary complexes, including their ro-vibrational frequencies for calculating their density and sum of states.
Figure 6 shows the convoluted CRUNCH TCID threshold fits to the energy-resolved product intensities. The convoluted fits include the available energy and angular momentum distributions of the [amb+M(II)+NTA]– + Ar reactants. The unconvoluted fits (not shown) predicted the 0 K change in enthalpies (ΔH0) for the dissociation of the ternary complex, and Table 2 shows the ΔH0 and ΔH298 (kJ/mol) for reactions 1 and 2. For the dissociation of the Zn(II) ternary complexes, both A and H exhibit ΔH0 for reaction 1, which are 31 kJ/mol and 15 kJ/mol lower than the ΔH0 for reaction 2, respectively, indicating both A and H have greater Zn(II) affinity than the NTA. The [A+Ni(II)+NTA]– ternary complex exhibits ΔH0 = 146 and 148 kJ/mol for reactions 1 and 2, respectively, indicating A and NTA have similar affinities for Ni(II). However, the dissociation of [H+Ni(II)+NTA]– shows the ΔH0 for reaction 1 is 36 kJ/mol lower than for reaction 2, indicating H has a greater Ni(II) affinity than the NTA. Overall, the [amb+Ni(II)+NTA]– complexes exhibit higher dissociation enthalpies than their [amb+Zn(II)+NTA]– counterparts, with the exception of A dissociating into [NTA+Ni(II)]–. Table 3 shows the Gibbs free energies (ΔG298) of association and the formation constants (K) for the reverse reactions:
[amb+M(II)]– + NTA → [amb+M(II)+NTA]– (3)
[NTA+M(II)]– + amb → [amb+M(II)+NTA]– (4)
Table 3 demonstrates that the formation of the Ni(II) ternary complexes is more exergonic and exhibits larger formation constants K than the Zn(II) complexes in all cases. Reaction 4 (i.e., the amb tag association with the NTA metal ion complex) is of particular interest as it represents the amb-tagged recombinant protein binding to the NTA-immobilized metal ion inside the IMAC column. Reaction 4 for the formation of [ambA+Ni(II)+NTA] exhibits the most spontaneous ΔG298 = 53.1 kJ/mol and the highest formation constant, K = 2.01 x 109.
Figure 1: Overview of the ES-IM-MS TCID method. Please click here to view a larger version of this figure.
Figure 2: The primary structures of amb A and H peptides. Color highlights the potential metal-binding sites. Please click here to view a larger version of this figure.
Figure 3: The center-of-mass, energy-resolved (eV) threshold collision-induced dissociation of [amb+M(II)+NTA]–. The energy-dependence of the product ions [amb+M(II)]– [NTA+M(II)]– and [amb-H2O+Zn(II)]– is shown. The center-of-mass collision energy, where there is 50% dissociation of the [amb+M(II)+NTA]– ternary complex, is included on the graphs. Please click here to view a larger version of this figure.
Figure 4: The model for the energy-resolved TCID method. The collisions between [ambH+Zn(II)+NTA]– + argon result in the dissociation to the [ambH+Zn(II)]– + NTA or [NTA+Zn(II)]– + ambH products. The threshold energies E1 and E2 equate to the 0 K enthalpies of dissociation (ΔH0) for the reactions [ambH+Zn(II)+NTA]– → [ambH+Zn(II)]– + NTA or [ambH+Zn(II)+NTA]– → [NTA+Zn(II)]– + ambH, respectively. Please click here to view a larger version of this figure.
Figure 5: The PM6 geometry-optimized ternary [amb+M(II)+NTA]– complexes of A and H. Conformers used in the TCID modeling of the experimental data. These conformers were selected from other candidate structures by comparing their PM6 electronic energies and how their LJ collision cross-sections (CCSHe) compared to the IM-MS measured CCSHe. Please click here to view a larger version of this figure.
Figure 6: The energy-resolved, collision-induced dissociation of [amb+M(II)+NTA]–. For species A and H, the product ions of [amb+M(II)]– and [NTA+M(II)]– with the convoluted CRUNCH threshold fits are shown. The energy (eV) values shown are the enthalpies of dissociation at 0 K for the reactions [amb+M(II)+NTA]– → [amb+M(II)]– + NTA or [amb+M(II)+NTA]– → [NTA+M(II)]– + amb. Please click here to view a larger version of this figure.
Figure 7: The format for the CRUNCH text input file. The file contains the mean relative intensities and their standard deviations of the product ions formed as a function of center-of-mass collision energy. Please click here to view a larger version of this figure.
amb | [amb+Zn(II)+NTA]– | [amb+Ni(II)+NTA]– | ||
PM6 | Exp.a | PM6 | Exp.a | |
A | 214±2 | 214 | 219±2 | 218 |
H | 211±5 | 216 | 212±3 | 215 |
a ES-IM-MS CCSHe measurements have uncertainties of ±4 Å2. |
Table 1: Comparison of LJ collision cross-sections of the PM6 conformers of [amb+M(II)+NTA]–. Theoretical cross-sections of the PM6 conformers are compared with the experimental CCSHe measured with ES-IM-MS.
[amb+Zn(II)+NTA]– → | [amb+Ni(II)+NTA]– → | |||||||
[amb+Zn(II)]– + NTA | [NTA+Zn(II)]– + amb | [amb+Ni(II)]– + NTA | [NTA+Ni(II)]– + amb | |||||
amb | ΔH0 | ΔH298 | ΔH0 | ΔH298 | ΔH0 | ΔH298 | ΔH0 | ΔH298 |
A | 118 | 127 | 149 | 182 | 146 | 171 | 148 | 154 |
H | 96.4 | 92.3 | 111 | 115 | 125 | 140 | 161 | 216 |
Table 2: Thermochemical results from the TCID analyses. The energy-dependent reactions [amb+M(II)+NTA]– → [amb+M(II)]– + NTA or [amb+M(II)+NTA]– → [NTA+M(II)]– + amb, showing the 0 K enthalpies of dissociation (ΔH0) derived from the unconvoluted TCID model fit, and 298 K enthalpies of dissociation (ΔH298) derived from ΔH0 and statistical mechanics thermal corrections using the PM6 rotational and vibrational frequencies. Values are given in kJ/mol.
[amb+Zn(II)]– + NTA → | [NTA+Zn(II)]– + amb → | [amb+Ni(II)]– + NTA → | [NTA+Ni(II)]– + amb → | |||||
[amb+Zn(II)+NTA]– | [amb+Zn(II)+NTA]– | [amb+Ni(II)+NTA]– | [amb+Ni(II)+NTA]– | |||||
amb | ΔG298 | K | ΔG298 | K | ΔG298 | K | ΔG298 | K |
A | -34.0 | 9.05 x 105 | -21.8 | 6.59 x 103 | -45.7 | 1.01 x 108 | -53.1 | 2.01 x 109 |
H | -29.3 | 1.36 x 105 | -30.2 | 1.95 x 105 | -47.0 | 1.71 x 108 | -31.1 | 2.81 x 105 |
Table 3: Gibbs free energies of association (ΔG298) and equilibrium formation constants (K). ΔG298 and K at 298 K for the reverse reactions [amb+M(II)]– + NTA → [amb+M(II)+NTA]– and [NTA+M(II)]– + amb → [amb+M(II)+NTA]–. Derived from ΔH298 and statistical mechanics entropy calculations using the PM6 rotational and vibrational frequencies. Values for ΔG298 are in kJ/mol.
Supplementary File. Please click here to download this File.
Critical steps
ES-IM-MS threshold collision-induced dissociation (TCID) analyses. The TCID used the transfer T-wave cell in the presence of argon as the collision cell. Prior to dissociation, the precursor ions are thermalized by low-energy collisions with nitrogen gas as they pass through the ion mobility (IM) cell. This results in a more reproducible energy-resolved TCID than is achieved by using the trap as the collision cell6,40. The thermalization of the [amb+M(II)+NTA]– prior to dissociation also allows the available internal energy of the ternary complex to be characterized using 298 K temperature. The dissociation in the transfer cell also means the ternary complex and its product ions have the same average arrival times at the detector, which was useful for identifying the dissociation of the ternary complex that only occurred in the transfer cell. Other regions where dissociation can occur are the ES source (sampling cone is kept at 25 V to avoid this) or at the entrance of the IM cell. The product ions produced by the dissociation of the ternary complex in these regions have different drift times from those produced in the transfer cell because the product ions are separated from the ternary complex in the IM cell. Those product ions were excluded from the analysis. In this protocol, only the integrated arrival time distributions for the precursor and product ions that are co-aligned are used to determine their intensities. The trap bias setting is the voltage that controls the injection voltage into the IM cell, which contributes to the CID at the entrance of the IM cell. The trap bias was set at 14 V, which kept the background dissociation to a minimum while not overly affecting the overall intensities. A previous study41 determined the effective temperature (upper limit) of the peptide dimer of leucine enkephalin to be 449 K at the entrance of the IM cell. However, the effective temperature decreased rapidly as the dimer passed down the IM cell. The arrival times of the amb complexes studied here exhibited Gaussian distributions, indicating they were thermalized as they passed down the IM cell.
ES-IM-MS collision cross sections (CCS) analyses. CCS drift times were found experimentally as the result of collisions with nitrogen. Those values were converted to helium-derived CCS drift times using a calibration curve of known standards. This is essential as the programs used to measure the CCS of the PM6 conformers require the more commonly used helium standards.
Modifications and troubleshooting of the technique
CRUNCH input text file format. The input text file suitable for the CRUNCH program is shown in Figure 7. The headers in order from top to bottom are file location and version of CRUNCH; date; number of energies; number of data series excluding the first energies column; source file; mass of the precursor complex; mass of argon; temperature of experiment; date of creation; x-data designated as −1 (the center-of-mass collision energies); and the full width at half maximum (FWHM) of the ion beam. These values must be modified for each TCID experiment. The FWHM energy spread of the ion beam and energy zero should be determined by retarding potential analysis (RPA) by scanning the CE through low voltages and monitoring the total ion current. However, under the operating conditions of the IM in the current study, the ion current signal only decreased by about 50% when the transfer CE was set to its lowest value. The ion beam energy zero and FWHM could be measured only upon additional retardation by lowering the exit IM lens. In this latter case, the FWHM of the derivative of the RPA curve gave a typical ion energy spread of 1.5 V in the lab-frame or 0.035 eV in the center-of-mass frame13.
The pressures row relates to pressure inside the collision cell but is not used here. The pressures of argon in the collision cell can be varied and the TCID data can be measured at three pressures to extrapolate to single collision conditions. However, only one pressure was used in this study, and the pressure results in multiple collisions. Developing the new platform for a single collision is an area of ongoing research. Masses relate to the two product ions whose intensities are in the columns below. Dwells can be left as default. The five columns are the center-of-mass collision energies (designated −1); the mean of the ion intensities of the species with mass 898.30 u; the standard deviations of the ion intensities of species 898.30 u; the mean of the ion intensities of the species with mass 253.53 u; and the standard deviations of the ion intensities of species 253.53 u.
Molecular Modeling
The number of conformers were narrowed initially by using models derived from previous studies9,10,11,12,13. CRUNCH fitting requires careful screening of reactants, activated molecules, and transition states to obtain accurate threshold energies. Previous research9,10,11,12,13 has included extensive screening of [amb+M(II)] conformers to obtain the structures with the parameters used in the CRUNCH modeling here. Only complexes with trans peptide bonds were used because only they are in agreement with IM-MS measured CCSHe10. The B3LYP and PM6 molecular modeling methods both predict the lowest energy [amb+M(II)]– conformer that exhibits Aa1-Cys2-Cys7 and carboxylate terminus coordination of Zn(II) or Ni(II)10,11,12,13. Familiarity with the behaviors of the known models allowed for the new conformers of [amb+M(II)+NTA]– to be determined more efficiently. To assist in conformer determination, as lower energy conformers were located by the PM6 method, they were filtered out and reassessed systematically until the most feasible, lowest energy conformers remained.
CRUNCH modeling
Time-window for observing dissociation. In this study, the 50 μs time window from the beginning of the transfer cell to the end of the TOF analyzer, where the multichannel plate detector is positioned, was used. It may be better to use the experimental time window between activation in the transfer cell and the entrance to the TOF mass analyzer because, if the activated ion dissociates during its time in the reflectron TOF, this metastable decay will be measured at a different m/z. However, in this study, the product ions observed in the mass spectra were all identifiable as the unmodified m/z species shown in Figure 3. This indicates that metastable decay was not an issue. Further research could investigate this by examining a known reaction with a high threshold and checking that the correct threshold energy is obtained using the 50 μs time window and RRKM modelling.
Scaling factors for the vibrational frequencies. The NIST-recommended scaling factors for PM6 (1.062) vibrational frequencies were used. These were satisfactory for fitting the [A+Zn(II)+NTA]–, [A+Ni(II)+NTA]–, and [H+Zn(II)+NTA]– data. For some cases where the higher energy channel is entropically favored over the lower energy channel, it may be necessary to additionally scale the frequencies of the second channel. One approach is to scale the frequencies below 900 cm−1 (as these are the least accurate) to loosen the frequencies and make the TS more entropically favored.
Optimization of parameters. Using the Yes option to Hold any parameter at present value can be helpful to fit the data successfully. For the first fit, the E0(2) is held and the model TCID is fitted to the data by optimizing the CONST, E0(1), and N variables. Once a good fit is located, the parameters option and Hold any parameter at present value can be used to hold CONST, E0(1), and N, while allowing E0(2) to optimize to the data. Finally, once E0(2) is optimized, in the parameters option, all four parameters CONST, E0(1), E0(2), and N should be allowed to optimize to the data.
Energy range for fitting the selected TCID model to the experimental data. The energy range used to fit the experimental data should reproduce as much of the experimental intensity data as possible while maintaining a good fit in the threshold region. One can start by fitting the TCID model to a small energy range at the thresholds of the experimental data. One can choose a starting energy that exhibits the background intensity just prior to the rising intensity threshold behavior. Once the TCID fit is optimized to the experimental data range, the range should be increased by 0.1 eV and the fit should be optimized again. This procedure should be repeated to fit as much of the data range as possible while maintaining the fit of the threshold region.
Thermochemical Analyses. The thermochemical results from Delta H and S at T option should be compared with a series of different energy range fits to the data to estimate the standard deviation of the TCID model fit. Fits to compare should include smaller ranges that fit the initial rising threshold intensities well with those with greater ranges that include the higher energies as well.
The authors have nothing to disclose.
This material is based upon work supported by the National Science Foundation under 1764436, NSF REU program (CHE-1659852), NSF instrument support (MRI-0821247), Physics and Astronomy Scholarship for Success (PASS) NSF project (1643567), Welch Foundation (T-0014), and computing resources from the Department of Energy (TX-W-20090427-0004-50) and L3 Communications. The authors thank Kent M. Ervin (University of Nevada – Reno) and Peter B. Armentrout (University of Utah) for sharing the CRUNCH program and for advice on fitting from PBA. The authors thank Michael T. Bower's group at the University of California – Santa Barbara for sharing the Sigma program.
Acetonitrile HPLC-grade | Fisher Scientific (www.Fishersci.com) | A998SK-4 | |
Alternative metal binding (amb) peptides | PepmicCo (www.pepmic.com) | designed peptides were synthized by order | |
Ammonium acetate (ultrapure) | VWR | 97061-014 | |
Ammonium hydroxide (trace metal grade) | Fisher Scientific (www.Fishersci.com) | A512-P500 | |
Driftscope 2.1 software program | Waters (www.waters.com) | software analysis program | |
Gaussian 09 | Gaussian | Electronic Structure Modeling Software | |
GaussView | Gaussian | Graphical Interface to Visualize Computations | |
Glacial acetic acid (Optima grade) | Fisher Scientific (www.Fishersci.com) | A465-250 | |
Ion-scaled Lennard-Jones (LJ) method | Sigma | Michael T. Bowers’ group of University of California at Santa Barbara | |
MassLynx 4.1 | Waters (www.waters.com) | software analysis program | |
Microcentrifuge Tubes | VWR | 87003-294 | 1.7 mL, polypropylene |
Microcentrifuge Tubes | VWR | 87003-298 | 2.0 mL, polypropylene |
Ni(II) nitrate hexahydrate (99% purity) | Sigma-Aldrich (www.sigmaaldrich.com) | A15540 | |
Poly-DL-alanine | Sigma-Aldrich (www.sigmaaldrich.com) | P9003-25MG | |
Waters Synapt G1 HDMS | Waters (www.waters.com) | quadrupole – ion mobility- time-of-flight mass spectrometer | |
Zn(II) nitrate hexahydrate (99%+ purity) | Alfa Aesar (www.alfa.com) | 12313 |