differentially heated capillary for thermal dissociation of noncovalently bound complexes produced...

Differentially heated capillary for thermal dissociation ofnoncovalently bound complexes produced by

electrospray ionization

Fei He, Javier Ramirez, Benjamin A. Garcia, Carlito B. Lebrilla*

Department of Chemistry, University of California, Davis, CA 95616, USA

Received 6 August 1998; accepted 2 November 1998

Abstract

A heated capillary consisting of two independently heated segments is used to dissociate gas-phase complexes of peptidesandb-cyclodextrin. By fixing the temperature in the first segment and varying it in the second segment, the complex is firstdesolvated and then dissociated. The results indicate that the complex is dissociated in the gas phase rather than in the solutionphase of an intact droplet. The thermal dissociation profile, the dissociation temperature and apparent Arrhenius parametersEa

andA are obtained. These values are compared toEa obtained from blackbody infrared radiation dissociation. (Int J MassSpectrom 182/183 (1999) 261–273) © 1999 Elsevier Science B.V.

Keywords:Differentially heated capillary; Noncovalently bound complex; Cyclodextrin; Peptide

1. Introduction

The relatively small amounts of internal energydeposited in the ion during ionization makes electro-spray ionization (ESI) mass spectrometry an idealmethod for characterizing noncovalently bound com-plexes generated from solution. There are severalexamples of gas-phase complex formation by ESI MS[1–11], however characterizing the nature of theinteraction remains a difficult task. Dissociation meth-ods such as collision-induced dissociation, photodis-sociation, surface-induced dissociation, and others

that are typically used for molecular or quasimolecu-lar ions often do not work efficiently or are notsufficiently characteristic for noncovalently boundcomplexes. A method that shows considerable prom-ise employs a heated capillary found in some designsof ESI sources [12]. Smith and co-workers utilized asingle heated stage metal capillary to obtain activationenergies of multiply charged ions of melittin producedby electrospray [13,14]. These authors report thatthe apparent activation energies decrease with in-creasing charge states from1 3 to 1 6. Wysockiand co-workers also reported a thermal decompo-sition method using a heated capillary to dissociatethe protonated monomer and dimer of leucineenkephalin [15]. We extended this method to non-covalent complexes, most notably complexes ofprotonated peptides andb cyclodextrin. We re-

* Corresponding author.Dedicated to the memory of Ben Freiser to commemorate his

many seminal contributions to mass spectrometry and gas phase ionchemistry.

1387-3806/99/$20.00 © 1999 Elsevier Science B.V. All rights reservedPII S1387-3806(98)14258-6

International Journal of Mass Spectrometry 182/183 (1999) 261–273

ported the use of heated capillary dissociation(HCD) as a simple and fast method for determiningthe relative order of gas-phase stability [16]. Weshowed further that complexes can be dissociatedas the temperature of the capillary is raised until aspecific temperature is reached (the dissociationtemperature) where the complex is no longer ob-served. The dissociation temperatures of the com-plexes were compared to threshold collision-in-duced dissociation (CID) results and a directcorrelation in the results of the two dissociationmethods was found. However, in the previous reportwe could not determine whether the dissociation ofthe complex was a solution-phase or a gas-phaseevent. In other words, does the complex dissociate inthe droplets that are known to exist in the spraybefore they become gas-phase ions, or is the com-plex desolvated in the gas-phase and dissociated asa gas-phase species? To answer these questions, anew differentially heated capillary was designedand constructed. The first segment is used todesolvate the complex while the second segment isused to dissociate the complex.

Complexes of cyclodextrins in solution are be-lieved to involve noncovalent interactions. They havelong been studied in solution [17–23] but have re-cently been of interest in mass spectrometry [24–26].Cyclodextrins are cyclic oligosaccharides consistingof glucopyranose subunits. The most common cyclo-dextrins area, b, and g cyclodextrin, composed ofsix, seven, and eight glucose units, respectively. Theglucose units area–1-4 linked, forming a truncatedconical structure with a narrower and a wider rim. Thewider rim is composed of the two and three hydroxylgroups of the glucose while the narrower rim iscomposed of the six hydroxyl groups. The hydroxylpositions are sometimes alkyl derivatized to give thecompound greater solubility [27]. The rim is hydro-philic while the outer surface and the inner cavity arehydrophobic. In solution, organic compounds, partic-ularly phenyl rings, are known to include into thecyclodextrin cavity [28]. There remains a controversywhether the gas-phase complexes retain the inclusionstructure characteristic of the solution-phase species[24–26].

In this paper we report the design and character-ization of the newly designed differentially heatedcapillary. This arrangement allows two segments tobe independently heated. Results of heated capillarydissociation (HCD) of cyclodextrin–peptide com-plexes using this new design are shown and comparedwith results from blackbody infrared radiation disso-ciation (BIRD) [29] obtained earlier in this laboratory[30].

2. Experimental

Bradykinin (BK, RPPGFSPFR), and three of itsanalogs: lysine1-bradykinin (L1-BK, KPPGFSPFR),desArg1-bradykinin (dR1-BK, PPGFSPFR), desArg9-bradykinin (dR9-BK, RPPGFSPF) and heptakis-tri-O-methyl-b-cyclodextrin (b-CD) were obtained fromSigma Chemical Co. (St. Louis, MO) in the highestpurity and used without further purification. Theelectrospray solutions were prepared by dissolving2:1 molar ratio of peptide and cyclodextrin in 50:50water:methanol solvent to yield ab-CD concentrationof 2.0 3 1025 M.

The experiments were carried out using a home-built electrospray Fourier transform mass spectrome-try (FTMS) equipped with an Oxford (Oxford Instru-ments, Witney, England) 5-tesla superconductingmagnet. Description of the instrument and its capa-bilities are provided in earlier publications [31,32].The analyte solution was transported from the syringepump to the electrospray needle via a 50 mm fusedsilica tubing. The solution flow rate was maintained at8 mL/min in all experiments. The charged dropletsproduced by the 3 kV electrospray needle, whose tipis approximately 5 mm from the counter electrode(capillary), drift down-field toward the front end ofthe capillary. Gas-phase ions were produced by com-pletely desolvating the droplets in the capillary. Ionstravel through a single stage quadrupole into a 2 inchcubic analyzer cell. The ions are translationallycooled by two pulses of N2 gas which increase thepressure to 1025 Torr momentarily.

262 F. He et al./International Journal of Mass Spectrometry 182/183 (1999) 261–273

2.1. Apparatus

The capillary is mounted inside a home-built elec-trospray source. The solvated ions produced from theelectrospray needle travel through the heated capillaryand a set of skimmers and lenses (Scheme 1). Onlythe relevant features of the newly designed capillarywill be described here. The capillary is made ofstainless steel with 1.9 mm o.d., 0.5 mm i.d., and 28cm length. The capillary is divided into two indepen-dently heated segments of equal length (12 cm), withthe one close to the electrospray needle (front end)designated Segment1, and the one close to the ioncyclotron resonance (ICR) cell (back end) designatedSegment2. The total heated length is 24 cm with about4 cm unheated and extends from Segment1 to atmo-sphere at the entrance of the vacuum chamber. Thelengths of the individual segments are smaller thanthat used by Wysocki and by Smith. The intensities ofthe complexes are not as great as that of the peptidesstudied by the previous researchers, and there wassome concern that the ion intensity of the complexeswould decrease too severely for the signal to beobserved in a longer capillary. Wysocki used both an18 cm and a 30 cm capillary. They found that

increasing the length to 30 cm does not vary theresults significantly. In that report, the authors alsoused two capillaries: an external segment locatedoutside the vacuum chamber, and an internal segmentlocated in the vacuum chamber. However, only thetemperature of the external capillary was varied.

Both capillary segments are resistively heated bytwo independent dc power supplies. Contact is madewith machined copper blocks that clamp on to thecapillary and soldered to power supply lines. Eachsegment has aJ type thermocouple spot welded to themiddle of the segment where the temperature ismonitored throughout the whole experiment. Becauseof the dynamic nature of the system, a temperaturegradient exists in the capillary. The temperature ismeasured in the middle of each segment, where it isthe highest for the segment.

3. Results

3.1. HCD of cyclodextrin bradykinin complex

The temperature characteristics of one segment asthe other is heated are shown in Fig. 1. Fig. 1(a)shows the temperature response of Segment1 when

Scheme 1.

263F. He et al./International Journal of Mass Spectrometry 182/183 (1999) 261–273

Segment2 is heated. Varying the temperature of Seg-ment2 from 120–300 °C, changed the temperature ofSegment1 from about 30–50 °C. Segment1 is openeddirectly to atmosphere, however a minor heat transferstill occurs from Segment2 to Segment1 when theformer is heated even in the presence of a largematerial flow in the opposite direction. When thetemperature of Segment1 is increased from 130–

300 °C, the temperature of Segment2 rose similarlyfrom about 70–110 °C (Fig. 1(b)). That the tempera-ture in Segment2, which is downstream of Segment1,is not even higher is because of several factorsincluding the relatively poor thermal transfer betweenthe interior and exterior of the capillary—stainlesssteel has poor thermal conducting properties—and thelarge flow of ambient air into the capillary.

Fig. 1. The temperature measured on Segment1 when Segment2 (a) is heated and vice-versa (b).


Attempts were made to monitor the temperature ofthe gas by placing a thermocouple gauge directlybehind (; 1 mm) the capillary. This produced the plotshown in Fig. 2. The expansion of the gas, thoughminimal (from 100–1 Torr), may contribute to thetemperature difference between the outflow and Seg-ment2. However, it appears that the internal temper-ature is about half of that monitored externally. Moreimportantly, the response is not linear. At highertemperatures (. 400 external temperature), the in-crease in the interior temperature becomes smaller.

Fig. 3(a) shows the intensities of the complex peak[CD:BK 1 H]12, normalized to the sum of all inten-sities in the spectrum, at different temperatures ofSegment1. No heat is applied to Segment2. At lowtemperatures, the complex peak intensity increaseswith increasing temperature. At this temperature

range, desolvation occurs. At about 180 °C, the curvelevels off until 260 °C. Unless otherwise stated, alltemperatures cited will henceforth correspond to theexterior temperature of the capillary. In this “stable”region, we suspect that desolvation is complete andthe complex is further heated without undergoingdissociation. This region may correspond to a gas-phase rearrangement where the complex convertsfrom the “solution-phase” structure to a “gas-phase”structure. In solution phase, formation of the inclusioncomplex may be the driving force for interaction,whereas in the gas phase the complex is maintainedby ion-dipole interactions involving the protonatedsites on the peptides and the hydrophilic rim of thecyclodextrin.

When the temperature of Segment1 is increasedabove 260 °C, the intensity of the complex decreases.

Fig. 2. The temperature of the outflow vs the temperature on Segment2 with the thermocouple placed 1 mm from the capillary exit.


At this point the complex dissociates and the corre-sponding intensity decreases until the signal is nolonger observed. Similar behavior is obtained in theanalogous experiment where Segment2 is heated (Fig.3(b)).

To ensure that the dissociation is occurring in the

gas-phase rather than solution phase, the HCD exper-iments were performed by setting the temperature ofSegment1 to some value in the stable region, whilethe temperature on Segment2 is increased. At leasttwo scenarios for complex dissociation are proposed.The complex may either dissociate in the droplet and

Fig. 3. (a) The intensities of the complex peak [CD: BK1 H]21 (normalized to the sum of all intensities) at various temperatures of Segment1with no heat applied to Segment2. (b) The same experiment with Segment2 heated.


desolvate, or it may desolvate first and then dissociate.The latter is the desired gas-phase process. For eachcomplex, the dissociation profile such as that shownin Fig. 3 is determined to establish the temperaturewhere Segment1 should be set. During the experi-ments, all other conditions including electrosprayneedle voltage, capillary voltage, lens and skimmervoltages are held constant unless specified. The Seg-ment2 temperature is monitored until the complexpeak intensity is below 1% of the most abundant peak.This temperature is called the dissociation tempera-ture orTd (vida supra).

By plotting intensity of the complex (normalized toall intensities) as a function of the temperature onSegment2, HCD plots are obtained. The dissociationbehavior of the bradykinin complex [CD:BK12H]21 is shown in Fig. 4. Fixing the temperature ofSegment1 to an area in the stable region as in Fig. 3and varying the temperature on Segment2 shows thatindeed the dissociation is a gas-phase process rather

than one occurring in the droplet. The intensity of thecomplex varies strongly with the temperature ofSegment2. The temperatures on Segment1 set at threedifferent temperatures (150, 200 and 250 °C) illustratesimilar dissociative behavior. More importantly,nearly identical dissociation temperatures (Td) of[CD:BK 1 2H]21 on Segment2 were obtained forthree temperatures of Segment1 (Fig. 4).

A typical spectrum for [CD:BK1 2H]21 obtainedin the stable region (T 5 200 K for both segments) isshown in Fig. 5(a). Under these conditions, theintensity of the complex is about 50% of the basepeak. Also observed are signals corresponding toammoniated, sodiated and potasiated cyclodextrin.For comparison the mass spectrum at a point near thedissociation temperature is shown in Fig. 5(b). Thetemperature on Segment1 was set to 200 °C, whichcorresponds to the middle of the stable region. Thetemperature on Segment2 was 418 °C, which wasclose to Td and the complete dissociation of the

Fig. 4. The dissociation behavior of the complex [CD: BK1 H]21 with Segment1 fixed to 150, 200, and 250 °C and Segment2 varied.


[CD:BK 1 2H]21 complex. Ions are observed corre-sponding to the fragmentation of the peptide. Thefragments are labeled according to the types ofcleavages that produce them. No fragments of thecyclodextrin are observed. These results contrasts toBIRD experiments where some fragments of thepeptides remain associated with cyclodextrin [30].The source of the fragment, whether from the com-plex or the free peptide is not immediately evident.That peptide fragmentation is observed even in thepresence of some abundance of complex indicate thatthe interactions in the complex are strong. The sum ofthe intermolecular forces that retain the complex areevidently as strong as the covalent bonds in thepeptide. We reported similar conclusions in the BIRDexperiments [30].

TheTd of four peptide complexes are tabulated inTable 1. CID threshold energies previously obtainedin this laboratory are also listed for comparison. CIDis a more established method to study gas-phasedissociation energy. Table 1 clearly shows that therelative order of HCD and CID agree well with eachother.

3.2. Determination of apparent dissociation energyEa(HCD) and the apparent preexponential factorA(HCD)

The dissociation reaction the complex undergoesas it travels through the capillary is represented by

@CD: P 1 2H]123 Products

where CD isb cyclodextrin andP is the peptide. Thelack of mass selection before the heated capillarymakes it difficult to specifically assign the products.

For a unimolecular dissociation, the rate of dissocia-tion can be written as:

kt 5 ln ~I o/I ! (1)

wherek is the rate constant for the first order reaction,t is the time the ions traverse the capillary,I o is theinitial intensity of the complex, andI is the intensityof complex at a fixed temperature. The valueI o wasdetermined by the intensities at the stable region. BothI andI o are normalized to an internal reference, whichis chosen to be the sodiated cyclodextrin peak. Thisspecies is generally unreactive in the temperaturerange of the experiments. Inserting Eq. (1) into theArrhenius Eq. (2)

ln k 5 ln A 2 ~Ea/RT! (2)

yields

ln F ln S I o

I DG 5 ln A 2Ea

RT(3)

By plotting ln [ln (I o/I )] versus 1/T, a slope equalto Ea/R is obtained as well as an intercept equal to lnA 1 ln t. Because of uncertainties in the measure-ments ofT and the complexity of the dissociationprocesses, the values obtained are apparent and aredesignated asEa(HCD) andA(HCD), respectively.

The residence time (t) of the ion in the capillarytube was estimated. Under experimental conditions,the Reynold’s number was determined and is found tocorrespond to laminar flow [33]. The flow rate for gasinside a tube was calculated by Poiseuille’s formula

dV

dt5

~P12 2 P2

2!pr4

161hPo(4)

Table 1

[CD:BK 1 2H]21 [CD:L1-BK 1 2H]21 [CD:dR1-BK 1 2H]21 [CD:dR9-BK 1 2H]21

Ea(HCD), eV 2.65 2.49 2.22 2.16Ea(BIRD), eV 1.07 0.98 0.72 0.64DEa 1.58 1.51 1.50 1.52log A(HCD) 25.5 24.7 21 22.4log A(BIRD) 11.4 10.0 6.2 7.8Dlog A 14.2 14.7 14.8 14.6Td, °C 426 425 457 360Ecom, eV 1.2 1.2 1.5 1.0


whereV is the flow volume,t the time,P1 andP2 thepressures at two ends of the tube,Po the referencepressure (chosen asP2), h the coefficient of viscosity,

l andr the length and radius of the tube, respectively.From Eq. 4, the estimated residence time of the ion isabout 0.01 ms.

Fig. 5. The FTMS spectra of the complex [CD: BK1 H]21 with Segment1 set to 200 °C and Segment2 set to (a) 200 °C and (b) 418 °C.


The Arrhenius plots of the data presented in Fig. 5,where three Segment1 temperatures are chosen, areshown in Fig. 6. The Arrhenius plots of the threetemperatures are linear and nearly parallel. An aver-age Ea(HCD) is obtained with a deviation corre-sponding to 2.656 0.08 eV. Similarly, logA(HCD)values are also obtained that average to 25.56 0.63.

The results are compared to those obtained usingBIRD experiments on the same systems. Details ofthe BIRD results are available on a separate publica-tion [30]. The values obtained from BIRD are tabu-lated together with HCD values in Table 1. A directcomparison of the two sets of values show that theEa

values from HCD experiments are consistently largerthan the BIRD value by 1.5 eV. LogA shows the

same consistent trend with the HCD values larger thanthe BIRD values by about 14.6 eV.

4. Discussion

The characterization of the dissociation process ina heated capillary tube is complicated by severalprocesses. For further discussions on the general useof ESI to characterize noncovalent interactions thereader is referred to a review article by Smith [9]. Weidentify at least three sources that may contributeenergy to the complex: the heated tube, the expansionfrom a relatively high pressure region of the tube to alow pressure region before the skimmer, and the

Fig. 6. Arrhenius plots of the data displayed in Fig. 4. Note that the lines are nearly parallel for the three temperatures of Segment1.


collision induced dissociation because of the voltagedifferences between the capillary and the skimmer(also known as nozzle-skimmer dissociation).

The total internal energy (Etotal) of the ion as itemerges the skimmer is given by:

Etotal 5 Ethermal1 Eexpansion1 ECID (5)

where Ethermal is the energy acquired in the heatedcapillary,Eexpansionis the energy acquired during theexpansion, andECID is the energy acquired when theion collides with neutrals between the capillary andthe skimmer. Admittedly, this approach is rathersimplistic as several other processes as may occursuch as resolvation of the ion after the expansionfollowed by collision with neutrals to desolvate theion again. BothEthermalandECID increase the energyof the ion, whileEexpansiondecreases it. The effect ofECID on typical ESI spectra is readily observed.Nozzle-skimmer CID was used by Prokai et al. todissociate complexes of amino acid and cyclodextrin[34]. On our instrument, varying this potential dropchanged the Arrhenius parameters. For this reason, it

was minimized as much as possible but still providedadequate signal for analysis. Even if the skimmerzone CID affects the outcome of the Arrhenius plot, itshould lower the apparentEa relative to BIRD.Instead, theEa(HCD) values are found to be alwayshigher than theEa obtained by BIRD. We attemptedto extrapolate the value ofECID to zero energy,however this produced incoherent results. Because ofthe dynamic nature of the ionization source, the directcontribution of bothEexpansionand ECID cannot becharacterized without considerable effort.

It is believed that the uncertainty in the tempera-ture measurement contributes the most to the discrep-ancy between HCD and BIRD values. An effectivetemperature can be calculated by assuming a constantdifference inEa and lnA between the BIRD and theHCD results. Fig. 7 shows the plot of the effectivetemperature (Teffective) versus the measured tempera-ture (Tmeasured). Over the narrow temperature range(; 30 K), the plot is linear with a slope of 1.6. Thenarrow range is limited by the BIRD results. It isexpected that at a wider range, the plot will diverge

Fig. 7. Plot ofTeffective vs Tmeasured. Teffective is obtained by assuming a constant difference in lnA andEa.


from linearity as shown by the temperature measure-ments of the outflow (Fig. 2). The average differencebetweenTmeasuredandTeffective is 49.5 K. The devia-tion of the slope away from unity prohibits thedetermination ofEa directly from HCD alone.

The determination of the temperature is a difficultproblem that may diminish the utility of this method.Another problem is in defining a distinct temperature.As the tube is heated in the presence of the constantflow, a temperature gradient exist throughout thelength of the capillary. As the ions traverse throughthe capillary it will experience a gradually increasingtemperature. Furthermore, dissociation can occur allalong the capillary at different temperatures. Length-ening the capillary to ensure equilibrium between theinterior and the exterior of the capillary does not helpdefine the temperature.

5. Conclusion

The systematic nature of the error, as shown by thenearly constant values ofDEa andDlog A, suggeststhat for similar systems the HCD values may bedirectly comparable. However, the peptides representonly a small group and further studies are required todetermine what chemical factors are important andvary DEa.

Because of the difficulty in obtaining a “true”temperature, HCD may not be a reliable method forobtainingEa. Instead, we propose that the dissocia-tion temperature (orTd) be an alternative method forobtaining relative strengths of interaction. TheTd is arelative measure ofDG. By selecting a temperaturecorresponding to a relative intensity, we indirectlydefine a specificK [Eq. (6)]. T is varied untilTd isobtained which corresponds indirectly to a value ofDG

DG 5 2RT ln K (6)

The relationship betweenDG andT is simpler andis not as affected by the nonlinearity of the valueT.The valueEa is obtained from a plot of intensityversus 1/T which is more strongly affected by thenonlinear relationship between interior and exterior

temperature. For comparing relative stabilities of thecomplexes,Td may be a more reliable index ofcomparison.

Finally, a differentially heated capillary allows usto differentiate gas-phase versus solution-phase dis-sociation process. In addition, we find that the perfor-mance of the differentially heated capillary is gener-ally better than a single heated stage. Betterreproducibility is obtained with the two stages in thedetermination ofEa(HCD), A(HCD), andTd, underidentical experimental conditions.

Acknowledgement

Funding provided by the National Science Foun-dation is gratefully acknowledged.

References

[1] K.J. Light-Wahl, D.L. Springer, B.E. Winger, C.G. Edmonds,I. Camp, D.G., B.D. Thrall, R.D. Smith, J. Am. Chem. Soc.115 (1993) 803.

[2] V. Katta, B.T. Chait, J. Am. Chem. Soc. 113 (1991) 8534.[3] B. Rosinke, K. Strupat, F. Hillenkamp, J. Rosenbusch, N.

Dencher, U. Kruger, H.J. Galla, J. Mass Spectrom. 30 (1995)1462.

[4] B.L. Schwartz, D.C. Gale, R.D. Smith, A. Chilkoti, P.S.Stayton, J. Mass Spectrom. 30 (1995) 1095.

[5] P. Juhasz, K. Biemann, Carbohydr. Res. 270 (1995) 131.[6] R. Bakhtiar, R.A. Stearns, Rapid Commun. Mass Spectrom. 9

(1995) 240.[7] Y.L. Hsieh, Y.T. Li, J.D. Henion, B. Ganem, Biol. Mass

Spectrom. 23 (1994) 272.[8] M. Jaquinod, N. Potier, K. Klarskov, J.M. Reymann, O.

Sorokine, S. Kieffer, P. Barth, V. Andriantomanga, J.F.Biellmann, A. Vandorsselaer, Eur. J. Biochem. 218 (1993) 893.

[9] R.D. Smith, K.J. Light-Wahl, Biol. Mass Spectrom. 22 (1993)493.

[10] M. Jaquinod, E. Leize, N. Potier, A.M. Albrecht, A. Shanzer,A. Vandorsselaer, Tetrahedron Lett. 34 (1993) 2771.

[11] R.D. Smith, K.J. Lightwahl, B.E. Winger, J.A. Loo, Org.Mass Spectrom. 27 (1992) 811.

[12] S.K. Chowdhury, V. Katta, B.T. Chait, Rapid Commun. MassSpectrom. 4 (1990) 81.

[13] M. Busman, A.L. Rockwood, R.D. Smith, J. Phys. Chem. 96(1992) 2397.

[14] A.L. Rockwood, M. Busman, H.R. Udseth, R.D. Smith, RapidCommun. Mass Spectrom. 5 (1991) 582.

[15] M.M.-N. (Mautner), A.R. Dongre´, A. Somogyi, V.H.Wysocki, Rapid Commun. Mass Spectrom. 9 (1995) 829.


[16] S.G. Penn, F. He, M.K. Green, C.B. Lebrilla, J. Am. Soc.Mass Spectrom. 8 (1997) 244.

[17] A. Rontoyianni, I.M. Mavridis, J. Inclusion PhenomenaMolec. Recog. Chem. 18 (1994) 211.

[18] A. Rontoyianni, I.M. Mavridis, E. Hadjoudis, A.J.M. Duisen-ber, Carbohydr. Res. 252 (1994) 19.

[19] J. Szejtli, Cyclodextrins and Their Inclusion Complexes;Akademiai Kiado, Budapest, 1982, p. 296.

[20] J. Szejtli, Pharma. Tech. Inter. 3 (1991) 16.[21] H. Hirai, Y. Shiraishi, H. Mihori, T. Kawamura, Poly. J. 27

(1995) 1064.[22] S. Letellier, B. Maupas, J.P. Gramond, F. Guyon, P. Gareil,

Anal. Chim. Acta 315 (1995) 357.[23] P.P. Korkas, E. Weber, M. Czugler, G. Narayszabo, J. Chem.

Soc.-Chem. Com. 21 (1995) 2229.[24] J.B. Cunniff, P. Vouros, J. Am. Soc. Mass Spectrom. 6 (1995)

437.[25] P. Camilleri, N.J. Haskins, A.P. New, M.R. Saunders, Rapid

Commun. Mass Spectrom. 7 (1993) 949.

[26] A. Selva, E. Redenti, M. Pasini, P. Ventura, B. Casetta, J.Mass Spectrom. 30 (1995) 219.

[27] K. Uekama, T. Irie, Drug Investigation 2 (1990) 22 (suppl.4).

[28] H.J. Schneider, T. Blatter, S. Simova, J. Am. Chem. Soc. 113(1991) 1996.

[29] P.D. Schnier, W.D. Price, R.A. Jockusch, E.R. Williams,J. Am. Chem. Soc. 118 (1996) 7178.

[30] S.G. Penn, F. He, C.B. Lebrilla, J. Phys. Chem. (1998), inpress.

[31] J.A. Carroll, S.G. Penn, S.T. Fannin, J. Wu, M.T. Cancilla,M.K. Green, C.B. Lebrilla, Anal. Chem. 68 (1996) 1798.

[32] E.E. Gard, M.K. Green, H. Warren, E.J.O. Camara, F. He,S.G. Penn, C.B. Lebrilla, Int. J. Mass Spectrom. Ion Processes158 (1996) 115.

[33] W.L. McCabe, J.C. Smith, P. Harricott, Unit Operations ofChemical Engineering, McGraw-Hill, New York, 1985.

[34] R. Ramanathan, L. Prokai, J. Am. Soc. Mass Spectrom. 6(1995) 866.


differentially heated capillary for thermal dissociation of noncovalently bound complexes produced...

Documents