implementation of a charge-sensitive ampliﬁer without a ...sep 26, 2019 · charge generally...

Implementation of a Charge-Sensitive Amplifier without a FeedbackResistor for Charge Detection Mass Spectrometry Reduces Noiseand Enables Detection of Individual Ions Carrying a Single ChargeAaron R. Todd, Andrew W. Alexander, and Martin F. Jarrold*

Chemistry Department, Indiana University, Bloomington, Indiana 47405, United States

ABSTRACT: Charge detection mass spectrometry (CDMS) depends on themeasurement of the charge induced on a cylinder by individual ions by means of acharge-sensitive amplifier. Electrical noise limits the accuracy of the chargemeasurement and the smallest charge that can be detected. Thermal noise in thefeedback resistor is a major source of electrical noise. We describe theimplementation of a charge-sensitive amplifier without a feedback resistor. Thedesign has significantly reduced 1/f noise facilitating the detection of high m/z ionsand substantially reducing the measurement time required to achieve almostperfect charge accuracy. With the new design we have been able to detect individual ions carrying a single charge. This is animportant milestone in the development of CDMS.

■ INTRODUCTIONIn conventional mass spectrometry (MS) the mass to chargeratio (m/z) is measured for gas-phase ions. For smallmolecules, where ions can be generated by electron impact,the charge state (z) is almost always one and so m/z and m aresynonymous. This changed with the development of electro-spray (ES) as an ionization method.1 ES allows much largerspecies to be placed in the gas phase and ionized. The averagecharge generally increases with the size of the ion, and largerions show a distribution of charge states which manifests as aseries of peaks in the m/z spectrum due to ions with identicalm but different z. Knowledge of the charge state is nowrequired to determine the mass. For light ions, the charge canbe deduced by resolving the isotope distribution by a high-resolution m/z measurement. When this is no longer possible,the charge state can be deduced from the spacing between thepeaks in the m/z spectrum. However, with increasing mass,heterogeneity causes the m/z peaks to broaden and eventuallycoalesce so the charge can no longer be deduced. This typicallyoccurs for ions in the 100−1000 kDa range, though there are afew examples where charge-state resolution has been obtainedfor ions with masses greater than 1 MDa.2−4

Interest in measuring accurate masses for species withmolecular weights much greater than 1 MDa has led to thedevelopment of specialized single-particle techniques wheremasses are directly determined for individual molecules.5 Onesuch approach uses nanomechanical oscillators: as singlemolecules are deposited on the oscillator, changes in theresonant frequencies are used to determine mass.6,7 However,the mass resolution is much worse than available from MS-based single-molecule techniques.5 The MS-based techniquesfall into two categories: charge shifting and direct chargemeasurement. In the charge-shifting approaches the m/z ismeasured, the charge is shifted, and then the m/z is re-measured.8−16 The charge is shifted several times, and then the

mass is determined from the m/z measurements. Thesemethods are not well-suited to determining mass distributionsbecause the measurements are relatively slow. In the directcharge measurement approaches, the m/z and z of individualions are directly determined and then multiplied to give themass. Methods in this class include quadrupole ion trapscoupled to a charge-detecting plate17,18 and TOF (time offlight) MS with kinetic energy-dependent, cryogenic detec-tors.19−22 Other methods employ a measurement of the chargeinduced on a conductor. This has been done with Fouriertransform ion cyclotron resonance (FTICR),23 where thecharge of a single trapped ion is determined from themagnitude of the charge induced on the detector plates.However, the induced charge is distributed between all theelectrodes in the cell, not just the detection ones, and so theinduced charge depends on the ion’s trajectory and can only bemeasured with an accuracy of around 10%. Single ionmeasurements on the orbitrap24,25 suffer from the sameproblem.In charge detection mass spectrometry (CDMS), the

detector is a cylinder that the ion passes through. When anion enters the cylinder, it induces a charge that dissipates whenit leaves. If the length to diameter ratio of the cylinder is largeenough, the induced charge is equal in magnitude to the chargeon the ion.26,27 The first use of CDMS was in 1960 where itwas used to determine the masses of micron-sized particles.28

In 1995, Fuerstenau and Benner coupled an electrospraysource to a CDMS detector.29 In these early experiments theCDMS detector was operated in a single pass mode where theion passes through the detection tube once. The ion’s m/z is

Received: September 26, 2019Revised: November 12, 2019Accepted: November 13, 2019Published: December 17, 2019

Research Article

pubs.acs.org/jasmsCite This: J. Am. Soc. Mass Spectrom. 2020, 31, 146−154

© 2019 American Society for MassSpectrometry. Published by the AmericanChemical Society. All rights reserved.

146 DOI: 10.1021/jasms.9b00010J. Am. Soc. Mass Spectrom. 2020, 31, 146−154

This article is made available for a limited time sponsored by ACS under the ACS Free toRead License, which permits copying and redistribution of the article for non-commercialscholarly purposes.

Dow

nloa

ded

via

IND

IAN

A U

NIV

BL

OO

MIN

GT

ON

on

July

17,

202

0 at

17:

37:2

0 (U

TC

).Se

e ht

tps:

//pub

s.ac

s.or

g/sh

arin

ggui

delin

es f

or o

ptio

ns o

n ho

w to

legi

timat

ely

shar

e pu

blis

hed

artic

les.

pubs.acs.org/jasmshttp://pubs.acs.org/action/showCitFormats?doi=10.1021/jasms.9b00010http://dx.doi.org/10.1021/jasms.9b00010http://pubs.acs.org/page/policy/freetoread/index.htmlhttp://pubs.acs.org/page/policy/freetoread/index.htmlhttp://pubs.acs.org/page/policy/freetoread/index.html

determined from the flight time through the tube and thecharge is determined from the induced charge. Theuncertainties in both the charge and m/z measurements arerelatively high for single pass CDMS, leading to a low massresolving power.To overcome the limitations of single-pass CDMS, Benner

embedded the detection cylinder in an electrostatic linear iontrap (ELIT).30 An ELIT consists of two end-caps that can beswitched between transmission and trapping mode. Intransmission mode ions pass through the trap and in trappingmode the end caps act as ion mirrors so that trapped ionsoscillate back and forth through the detection cylinder. Thelongest trapping time achieved in this early work was about 10ms (450 oscillations). The RMS uncertainty for one chargemeasurements was 50 elementary charges (e) and the chargeuncertainty improves by n−1/2, so the RMS uncertainty after450 oscillations was reduced to 2.3 e. The limit of detection(the smallest charge that could be measured) was 250 ebecause trap closure was triggered by an ion entering thedetection cylinder.In recent years there has been another resurgence of interest

in CDMS.31−36 Rodolphe and co-workers have used single-pass CDMS to measure the masses of several heterogeneoussamples including amyloid fibrils, synthetic polymers, andnanoparticles.37−39 While Williams and co-workers have takenCDMS in a different direction and obtained collisional crosssections from the ion energy decay through background gascollisions,40,41 our group has focused mainly on applications tothe analysis of viruses, gene therapy products, lipoproteins, andprotein complexes.42−56

The best performance that has been achieved to date withcharge detection mass spectrometry is a charge RMSD of 0.196e.57 With a charge uncertainty less than around 0.2 e it isbeneficial to quantize the charge−assign the charge to thenearest integer charge state. If ions with intermediate charges(i.e., charges outside the range of n − 0.2 e and n + 0.2 e) arediscarded, the error rate for charge quantization (i.e., where theion is assigned to the wrong charge state) is less than 1 in15000. The uncertainty in the charge measurement scales ast−1/2, and it was necessary to trap ions for 3000 ms to achieve acharge uncertainty of 0.2 e. With 3 s trapping it takes a longtime to measure enough ions to generate a spectrum. Thecharge uncertainty is limited mainly by electrical noise, and solowering the noise will reduce the measurement time neededto achieve perfect charge accuracy. The limit of detection (thesmallest charge that can be detected) also depends on theelectrical noise and the best performance to date, i.e., thelowest charge that has been reliably measured is 7 e(elementary charges).58 Lowering the noise will also lowerthe limit of detection.In CDMS, the charge induced on the detection cylinder by

the ion is detected by a charge-sensitive amplifier. Figure 1ashows a schematic of a generic charge sensitive amplifierconsisting of an operational amplifier with a negative feedbackcapacitor, Cf. It produces an output voltage inverselyproportional to the integrated current at the input:

VC

i t t1

( ) dt

of 0

i∫= −(1)

It is evident from eq 1 that the gain is inversely proportionalto Cf (i.e., Cf is minimized to maximize the gain). For low noiseamplifiers, a JFET (junction field effect transistor) is usually

employed at the input (as shown in Figure 1b) and a feedbackresistor (Rf) is placed in parallel with the feedback capacitor.The feedback resistor discharges the feedback capacitor andestablishes a well-defined operating point for the amplifier.Charge-sensitive amplifiers are widely used in high energy

particle and radiation detectors. Since first proposed by Gattiand co-workers in 1956,59 there has been a continuing effort toreduce the noise and hence improve the precision and accuracyof particle and radiation parameter measurements.60−66 Thethermal noise associated with the feedback resistor is a majorcontributor to the overall noise. This leads to a noise currentgiven by

ikT f

R4

nf

=Δ

(2)

where k is the Boltzmann constant, T is the temperature, andΔf is the frequency bandwidth. The noise is minimized bymaximizing Rf. We have used values up to 100 GΩ. Thissource of noise can be completely removed by removing thefeedback resistor. However, if the feedback resistor is removed,another way must be provided to reset the amplifier and give ita well-defined operating point. A number of schemes havebeen described such as optical feedback,63 pulsed opticalfeedback,64 transistor reset,65 and pulsed source reset.66 Thelarge number of different approaches is a testament to the factthat they all have disadvantages and often introduce noise froma different source.

■ CHARGE-SENSITIVE AMPLIFIER DESIGNThe amplifier employed in this work is based on the approachof Bertuccio and co-workers.61 A simplified schematic is shownin Figure 2. In this circuit, removal of the feedback resistor ismade possible by allowing the JFET to be biased by its gateleakage current, forward biasing the JFET’s gate-channeljunction. The JFET gate/source voltage drifts to about +0.2V where the leakage current from the drain is balanced by theslightly forward biased gate-source junction. The drain currentis above the JFET’s IDSS (drain current for zero bias) value.The DC offset of the rest of the circuit is controlled by asecond negative feedback loop between the emitter of Q2 andthe base of the first transistor (Q1 in Figure 2) through R2, R3,and C1. A bootstrap circuit comprised of R4, R5, and C2 helps

Figure 1. (a) Circuit diagram for a generic charge sensitive amplifierconsisting of an operational amplifier and a feedback capacitor. (b)Low noise charge sensitive amplifier with a JFET at the input and afeedback resistor in parallel with the feedback capacitor.

Journal of the American Society for Mass Spectrometry Research Article

DOI: 10.1021/jasms.9b00010J. Am. Soc. Mass Spectrom. 2020, 31, 146−154

147

http://dx.doi.org/10.1021/jasms.9b00010

Q1 provide the voltage gain of the circuit. Q1 is driven as acommon base amplifier and Q2 acts as a common collectoramplifier that buffers the output of Q1.The feedback capacitance Cf is provided by a small wire that

sits around 2 mm from the JFET. Its value is too small to bemeasured by the equipment at our disposal. We estimate it tobe on the order of 100 fF. Cf defines the closed-loop gain at thefrequencies of interest (around 2−100 kHz). The otherfeedback path is active at lower frequencies. Because the gatevoltage is not well-defined by a feedback resistor, the JFET’stransconductance is dependent on fabrication variations and itis necessary to select the JFET by hand for optimumperformance.In previous work, we have employed a commercial charge-

sensitive preamplifier (Amptek, A250) with a 2SK152 JFET atthe input. We found that cooling the JFET to around 130 Ksignificantly reduced the noise.67 When we cooled the JFET inthe new design in Figure 2 we found that the gain variedinversely with temperaturea serious complication when anuncertainty of better than 0.2 e is required to achieve almostperfect charge accuracy. To overcome this problem, wetemperature regulated the JFET to a temperature a fewdegrees above the unregulated equilibrium value. Thetemperature regulation circuit employed a heater constructedwith 0.2 mm diameter Nichrome wire. The heater was coupledto the analog output of a temperature controller (Red Lion,T1651110). The analog output operates a silicon transistor(STMicroelectronics, D44H11 NPN), which in turn tunes theoutput of a power supply (HP 6216C) set at 10 V and 1.2A.This approach allowed the resistive heater to generate lessnoise than if driven directly by the temperature controller.

Initial testing indicated that the circuit held the temperature tobetter than 0.1 K of the specified set point once thermalequilibrium had been achieved.

■ CDMS INSTRUMENT AND METHODSThe charge-sensitive amplifier was installed into our newCDMS instrument, which incorporates a new electrostaticlinear ion trap (ELIT) with improved m/z resolution.68 Ionsare generated by nanoelectrospray (Advion Biosciences,Triversa NanoMate) in positive-ion mode.They enter the instrument through a heated metal capillary

(0.762 mm i.d., 10 cm long) and then pass through severalregions of differential pumping before reaching the ELIT. Thefirst differentially pumped region contains a FUNPET (an ionfunnel-ion carpet hybrid) designed to optimize transmissionover a broad mass range.69 The second differentially pumpedregion contains an RF hexapole with a DC offset that sets theion energy. The hexapole is followed by a segmented RFquadrupole. Ions that exit the quadrupole are focused into adual hemispherical deflection energy analyzer (HDA) whichselects a narrow band of ion kinetic energies. Ions that aretransmitted by the HDA enter the fifth differentially pumpedregion that houses the ELIT. The pressure in this region isaround 10−9 mbar. The acceleration region of an orthogonalreflectron time of flight mass spectrometer (TOFMS) islocated just before the HDA. Ions can be analyzed with theTOFMS or the acceleration region can be turned off so thations travel unimpeded to the HDA for analysis by CDMS.The ELIT consists of two end-caps that can be switched

between transmission and reflection mode. With both end-capsin reflection mode, ions oscillate back and forth through ametal cylinder placed between the end-caps. When ions are inthe cylinder, they induce a charge which is detected by thecharge-sensitive amplifier. The cylinder is long enough that theentire charge is measured.70,71 The voltage output from thecharge sensitive amplifier is amplified, digitized, and sent to acomputer through an FPGA (field-programmable gate array).The time domain signals are analyzed in real time by a Fortranprogram which uses a Gaussian window function and fastFourier transforms to determine the m/z and charge of thetrapped ions.72 Trapping events are discarded if the ions arenot trapped for the full trapping time or if a statistical analysisshowed that the m/z or charge values are not stable. Unusualshifts or excursions in the m/z or charge can arise from mass orcharge loss73 or from multiple ions with similar m/z valuesoscillating almost in phase.Simulated signals were used to characterize the limit of

detection. The signals were generated with a functiongenerator (Agilent Technologies, 33220A), passed into thevacuum chamber, and broadcast onto the detection cylinder byway of a small antenna placed through the grounded shield ofthe ELIT. This allowed the simulated signal to be picked-upand measured by the charge sensitive amplifier in the samefashion as the signal from an oscillating ion. The output fromthe charge sensitive amplifier is amplified, digitized, andtransferred to the computer for data analysis in the same wayas ion signals.Pyruvate Kinase (Lee Biosciences) was prepared at 2.5 mg/

mL in 100 mM ammonium acetate (Sigma-Aldrich, ≥99.99%)and purified by size-exclusion chromatography (Micro Bio-Spin P-6 Gel Columns in Tris Buffer, BIO-RAD). Bradykinin(Sigma-Aldrich) was dissolved in water to make a 1 mg/mLsolution. This solution was then diluted to 50μg/mL in a

Figure 2. Circuit diagram for the charge sensitive amplifier without afeedback resistor. The JFET at the input is biased by its gate leakagecurrent, forward biasing its gate-channel junction. By exploiting thismode of operation the feedback resistor can be removed to improvethe signal to noise ratio. The feedback capacitance Cf is provided by asmall wire that sits around 2 mm from the JFET. Its value is too smallto be measured by the equipment at our disposal. We estimate it to beon the order of 100 fF.



148


49:49:2 solution of water, methanol, and acetic acid.Angiotensin II (Sigma-Aldrich) was prepared in the sameway as bradykinin.

■ RESULTS AND DISCUSSIONEvaluation of Charge RMSD. Rudimentary information

about the uncertainty in the charge determination can beobtained from the noise. Measurements were performed onour prototype CDMS instrument (which has a commercialAmptek A250 charge-sensitive amplifier with a cryogenicallycooled 2SK152 JFET at the input) and our new CDMSinstrument with the new amplifier installed. In the case of thenew instrument, measurements were performed with the inputJFET at room temperature and cryogenically cooled. In eachcase, 1000 400 ms noise files were collected with theinstrument operating, except ions were not electrosprayed.The resulting files were analyzed using a fast Fourier transform(FFT). For each set of 1000 noise files, the noise in ADC bitsfound from the FFT magnitude at 10 Hz intervals from 3 kHzto 100 kHz. The frequencies were converted to m/z valuesusing74

mz

Cf 2

=(3)

where f is the ion oscillation frequency and C is a constant thatdepends on the ion energy and design of the ELIT. Finally, thenoise values in ADC bits were converted to the RMS noise inelectrons (eRMS) using known calibration factors. We equatethe eRMS determined in this way to the uncertainty in thecharge measurement (the charge RMSD). The results of thisanalysis are presented in Figure 3 which shows plots of the

charge RMSD against m/z. The black line shows theperformance of the conventional amplifier design (A250 withcooled 2SK512) on our prototype CDMS instrument. Thecharge RMSD increases as the m/z increases. According to eq3, the oscillation frequency, f, decreases as the m/z increases.Thus, the 1/f noise increases as m/z increases, and this is themain reason why the charge RMSDs increase with m/z inFigure 3.

The red line in Figure 3 shows the charge RMSDsdetermined for the new amplifier design with the input JFETat room temperature. Even with its input JFET at roomtemperature, the new amplifier design significantly out-performs the conventional design with its input JFETcryogenically cooled. The blue line in Figure 3 shows theresults for the new amplifier with its input JFET cryogenicallycooled. Cooling significantly lowers the charge RMSD. Thecharge RMSDs for the new amplifier also increase with m/z,but much less quickly than for the A250 with cooled 2SK152.At an m/z of 10000 Da the charge RMSD for the new amplifierwith cooled input JFET is 65% of the charge RMSD for theA250 with cooled 2SK152. At an m/z of 100000 Da the ratiodecreases to 51%. The relationship between the charge RMSDand the trapping time is

qt1

RMSDtrap

∝(4)

With the charge RMSD reduced by 65%, the trapping timerequired to achieve the same charge RMSD is reduced to 0.652

or 42% of the original trapping time. With the charge RMSDreduced to 51%, the trapping time can be reduced to 26% of itsoriginal value. Thus, substantial trapping time reductions areexpected to result from the improved performance of thecharge-sensitive amplifier.

Charge-State Resolution for Pyruvate Kinase. Asdescribed above, with a charge RMSD of

Da). Signals with amplitudes leading to induced charges of 1 to10 e on the detection cylinder were broadcast from theantenna near the detection cylinder. The induced signals wereamplified, digitized, and transferred to a computer for analysisusing a slightly modified version of the Fortran code used toanalyze CDMS trapped ion data. Each time domain signal wasfirst analyzed by full event fast Fourier transform. In order toqualify as a detected event, the peak in the FFT due to thesimulated signal must rise above a threshold which is calculatedfor each file. The threshold is given by the baseline noise levelin the FFT plus six times the standard deviation of the noise inthe 40−100 kHz range in the FFT. We define the detectionefficiency as the fraction of data files where a signal wasdetected from the full event FFT. Figure 5a shows a plot of thedetection efficiencies against applied charge for signal lengthsof 100, 1500, and 3000 ms. For a 100 ms signal length, thedetection efficiency drops to around 50% for an applied chargeof 4 e and to around 1% for an applied charge of 2 e. If thesignal length is increased the signal to noise ratio in the FFTincreases and the detection efficiency increases. With a signallength of 1500 ms, the detection efficiency for an appliedcharge of 2 e increases to over 70%. A signal with an appliedcharge of 1 e is now detectable, albeit with relatively lowdetection efficiency of just under 35%. With a signal length of3000 ms the detection efficiency for an applied charge of 1 eincreases to just over 70%.Once the presence of at least one ion signal has been

established, the next step in the analysis of CDMS data is todetermine whether the ions are trapped for the full trappingtime.72 For the simulated data we know that the signal extendsthrough the whole time period. However, we want to ensurethat the program works for real signals, where the ions may notstay trapped. Light ions are much more difficult to trap forextended times than heavy ions. A small section at the

beginning of the time domain data is apodised by a Gaussianwindow function prior to performing an FFT. Initially a smallwindow is used and if a signal is not detected, the width isincreased up to a maximum value. To facilitate the detection oflow charge ions, we increased the maximum window width thatcan be employed in the windowed FFTs. If a signal is notdetected with the largest window width permitted, the file isnot processed further. When a signal is detected the window isstepped through the time domain signal and the charge andoscillation frequency at each step are averaged. If the signaldoes not persist to the end of the file, the data for that ion isstored but not included in the output file of ions that weretrapped for the full trapping period.Figure 5b shows a plot of the average measured charge

versus the applied charge for a signal length of 100 ms. The

Figure 4. CDMS results for pyruvate kinase (2.5 mg/mL in 100 mMammonium acetate) with 1500 ms trapping. (a) Charge spectrumwith a bin width of 0.1 e. The spectrum contains 1304 ions. (b) Massspectrum obtained by multiplying the m/z and integer chargedetermined for each ion and binning the resulting masses. The binwidth is 500 Da.

Figure 5. Investigation of the charge limit of detection with simulatedsignals. (a) Plot of the detection efficiency against the applied chargefor signal lengths of 100 ms (black points), 1500 ms (red points), and3000 ms (blue points). (b) Plot of the measured charge against theapplied charge for a 100 ms signal length. The points are themeasured values. The error bars show the charge RMSD. The solidline shows the expected one to one relationship between themeasured and applied charge. (c) Plots of the measured chargeagainst applied charge for signal lengths of 1500 ms (red points) and3000 ms (blue points). The charge RMSDs are comparable to orsmaller than the data points, and hence, they are not evident.



150


line shows the expected one to one dependence of themeasured and applied charges and the error bars show thecharge RMSD. For applied charges less than 5 e the measuredcharge diverges from the line and stays almost constant as theapplied charge decreases further. This happens because thedetection efficiency is low and the signal is only detected incases where the signal is boosted by the noise.Extending the signal length increases the signal to noise ratio

in the FFT, making the low charge signals easier to detect.Results are shown in Figure 5c for signal lengths of 1500 ms(red points) and 3000 ms (blue points). The charge RMSDsare now comparable to, or smaller than, the data points, andhence, the error bars are not evident in the plot. With thelonger signal lengths, the one to one relationship between theaverage measured charges and the applied charges extends tomuch lower applied charges than with a signal length of 100ms. For a signal length of 1500 ms, the average measuredcharges for applied charges greater than 2 e all lie on the line.Extending the signal length from 1500 to 3000 ms marginallydecreases the measured charge for an applied charge of 2 e (soit now also lies on the line in Figure 2c) and significantlydecreases the measured charge for an applied charge of 1 e.With a signal length of 1500 ms the +1 and +2 charge statescannot be definitively distinguished. However, they can bedistinguished with a signal length of 3000 ms, though themeasured charge for an applied charge of 1 e does not fallcompletely on the line, indicating that there is still somediscrimination for a charge of 1 e.Detection of Singly Charged Ions with Bradykinin.

Figure 6a shows the m/z spectrum recorded for bradykininwith the orthogonal time of flight (TOF) mass spectrometer.The large peak at around 1060 Da is assigned to M+H+ ions ofbradykinin (expected m/z 1061 Da) and the main peak ataround 530 Da is assigned to M + 2H+ ions (expected m/z 531Da). Figure 6b shows the CDMS m/z spectrum recorded forthe same bradykinin sample immediately after the TOF m/zspectrum in Figure 6a. The CDMS m/z spectrum, recordedwith a trapping time of 3000 ms, shows the same two peaks asthe TOF m/z spectrum in Figure 6a, though the peaks are inslightly different positions because the CDMS m/z scale wasnot precisely calibrated for this study. Figure 6c shows theCDMS mass distribution obtained by multiplying the m/z andinteger charge of each ion and then generating a histogram ofthe resulting masses. The mass distribution is dominated byions with masses around 1060 Da that contains contributionsfrom singly charged M + H+ and doubly charged M + 2H+

ions. The peaks in Figure 6b,c are relatively broad. Weattribute this, at least partly, to the elevated uncertainty in them/z values caused by collision induced frequency shifts duringthe relatively long trapping time. Much better m/z resolutioncan be obtained for heavier ions.Detection of Singly Charged Ions with Angiotensin II.

Figure 7a shows the CDMS m/z spectrum measured forangiotensin II with a trapping time of 3000 ms. The main peakat an m/z of around 1000 Da is attributed to the M + H+ ion ofangiotensin II. In addition, there are four peaks at lower m/zvalues. To help assign these peaks we have overlaid a chargeversus m/z scatter plot (blue points) where each point in thescatter plot represents the charge and m/z values of a singleion. There is a fairly sharp threshold at around 1.1 e in thescatter plot. This threshold occurs because singly charged ionsare close to the limit of detection and so their detectionprobabilities are enhanced by noise. This is also responsible for

the average measured charge being slightly above the line inFigure 5c for an applied charge of 1 e. For the ions in the m/zpeak at around 1000 Da there is a cluster of points withcharges slightly larger than 1 e, but no points close to 2 e, orhigher. This indicates that this peak is due exclusively to singlycharged angiotensin II ions and that there are no doublycharged dimer ions present for this sample. The lower m/z

Figure 6. CDMS detection of singly charged Bradykinin ions. (a) m/zspectrum recorded with the orthogonal time of flight (TOF) massspectrometer situated just before the HDA in the new CDMSinstrument. The spectrum shows the singly charged monomer comingin at an m/z of around 1060 Da and the doubly charged monomer atan m/z of around 530 Da. (b) Shows the CDMS m/z distributionrecorded immediately after the TOF m/z spectrum in (a). The sametwo peaks are evident. The m/z values are slightly different becausethe CDMS m/z spectrum is not precisely calibrated. The bin size is 7Da. (c) CDMS mass distribution obtained by multiplying the m/z andinteger charge determined for each ion. There is a single peak close to1060 Da. The bin size is 7 Da.



151


peaks at around 750 and 600 Da are also predominantly singlycharged. They are probably fragment ions. The peak aroundm/z 500 Da is mainly singly charged ions, however there are afew doubly charged ions. Since the m/z of this peak is veryclose to half the m/z of the peak at around 1000 Da it seemslikely that the doubly charged ions are doubly chargedangiotensin II (M + 2H+). The ions that make up the lowestm/z peak in the spectrum (at around 330 Da) are all doublycharged.Figure 7b shows the mass distribution obtained by

multiplying the measured m/z values by the integer chargeof each ion. A charge versus mass scatter plot is overlaid (bluepoints). Note that in the mass distribution the doubly chargedions with m/z values around 330 Da have moved up to 660 Daand the few doubly charged ions at m/z values around 500 Daare now included in the mass peak at around 1000 Da.These results, and the results for bradykinin described

above, demonstrate conclusively that we are now able to detectsingly charged ions by CDMS. In previous work, the chargelimit of detection (the smallest charge that could be reliablymeasured) was 7 e.58 Not all of the improvement in the limit ofdetection realized here is due to the new charge sensitiveamplifier. With the new cylindrical trap employed in this workwe were able to extend the trapping time for light ions to 3000ms. With the conical trap used in our prototype CDMSinstrument,74 such long trapping times would have beendifficult to obtain for light ions. Improvements in the Fortran

code used to analyze the data also contributed to lowering thelimit of detection.

■ CONCLUSIONSThe charge-sensitive amplifier without a feedback resistor,implemented in this work, has substantially improved perform-ance over the conventional design built around a commercialA250. The new amplifier with the input JFET at roomtemperature significantly outperforms the A250 with acryogenically cooled 2SK152 at its input. The new designreduces the amount of time required to achieve a chargeaccuracy of 0.2 e (the threshold for near perfect chargeaccuracy) by more than a factor of two for ions with m/zvalues around 10000 Da. The improvement is expected to belarger for higher m/z ions because the charge RMSD increasesless rapidly with m/z for the new amplifier than with thecommercial A250. In addition to reducing the acquisition timefor almost perfect charge accuracy, the new charge sensitiveamplifier has allowed us to reduce the limit of detection all theway down to a single charge. We would be remiss if we did notnote that there are some disadvantages associated with the newamplifier, including that the gain is temperature sensitive sowhen cryogenically cooled the input JFET must be carefullytemperature regulated.

■ AUTHOR INFORMATIONCorresponding Author*E-mail: [email protected] F. Jarrold: 0000-0001-7084-176XNotesThe authors declare the following competing financialinterest(s): M.F.J. is associated with a company developingcharge detection mass spectrometry.

■ ACKNOWLEDGMENTSThis material is based upon work supported by the NationalScience Foundation under Grant Number CHE-1531823. Wethank IU Chemistry Department Electronic InstrumentServices and Mechanical Instrument Services for their valuablecontributions.

■ REFERENCES(1) Wong, S. F.; Meng, C. K.; Fenn, J. B. Multiple charging inelectrospray ionization of poly(ethylene glycols). J. Phys. Chem. 1988,92, 546−550.(2) Tito, M. A.; Tars, K.; Valegard, K.; Hajdu, J.; Robinson, C. V.Electrospray time-of-flight mass spectrometry of the intact MS2 viruscapsid. J. Am. Chem. Soc. 2000, 122, 3550−3551.(3) Stockley, P. G.; Rolfsson, O.; Thompson, G. S.; Basnak, G.;Francese, S.; Stonehouse, N. J.; Homans, S. W.; Ashcroft, A. E. Asimple, RNA-mediated allosteric switch controls the pathway toformation of a T=3 viral capsid. J. Mol. Biol. 2007, 369, 541−552.(4) Uetrecht, C.; Versluis, C.; Watts, N. R.; Roos, W. H.; Wuite, G.J. L.; Wingfield, P. T.; Steven, A. C.; Heck, A. J. R. High-resolutionmass spectrometry of viral assemblies: Molecular composition andstability of dimorphic hepatitis B virus capsids. Proc. Natl. Acad. Sci. U.S. A. 2008, 105, 9216−9220.(5) Keifer, D. Z.; Jarrold, M. F. Single-molecule mass spectrometry.Mass Spectrom. Rev. 2017, 36, 715−733.(6) Dohn, S.; Svendsen, W.; Boisen, A. Mass and positiondetermination of attached particles on cantilever-based mass sensors.Rev. Sci. Instrum. 2007, 78, 103303.

Figure 7. CDMS detection of singly charged angiotensin II ions. (a)CDMS m/z distribution obtained using 15 Da bins. A charge versusm/z scatter plot is overlaid (blue points). Each point represents thecharge and m/z of a single ion. The charge scale is on the right. (b)Shows the CDMS mass distribution obtained using 15 Da bins. Themasses were obtained by multiplying the m/z and integer charge foreach ion. A charge versus mass scatter plot is overlaid (blue points).Each point represents the charge and mass of a single ion.



152

mailto:[email protected]://orcid.org/0000-0001-7084-176Xhttp://dx.doi.org/10.1021/jasms.9b00010

(7) Hanay, M. S.; Kelber, S.; Naik, A. K.; Chi, D.; Hentz, S.; Bullard,E. C.; Colinet, E.; Duraffourg, L.; Roukes, M. L. Single-proteinnanomechanical mass spectrometry in real time. Nat. Nanotechnol.2012, 7, 602−608.(8) Philip, M. A.; Gelbard, F.; Arnold, S. An absolute method foraerosol-particle mass and charge measurement. J. Colloid Interface Sci.1983, 91, 507−515.(9) Hars, G.; Tass, Z. Application of quadrupole ion-trap for theaccurate mass determination of submicron size charged-particles. J.Appl. Phys. 1995, 77, 4245−4250.(10) Bruce, J. E.; Cheng, X.; Bakhtiar, R.; Wu, Q.; Hofstadler, S. A.;Anderson, G. A.; Smith, R. D. Trapping, detection, and massmeasurement of individual ions in a Fourier transform ion cyclotronresonance mass spectrometer. J. Am. Chem. Soc. 1994, 116, 7839−7847.(11) Schlemmer, S.; Illemann, J.; Wellert, S.; Gerlich, D. Non-destructive high-resolution and absolute mass determination of singlecharged particles in a three-dimensional quadrupole trap. J. Appl. Phys.2001, 90, 5410−5418.(12) Peng, W.-P.; Cai, Y.; Lee, Y. T.; Chang, H. C. Laser-inducedfluorescence/ion trap as a detector for mass spectrometric analysis ofnanoparticles. Int. J. Mass Spectrom. 2003, 229, 67−76.(13) Peng, W.-P.; Yang, Y.-C.; Kang, M.-W.; Lee, Y. T.; Chang, H.-C. Measuring masses of single bacterial whole cells with a quadrupoleion trap. J. Am. Chem. Soc. 2004, 126, 11766−11767.(14) Peng, W.-P.; Lee, Y. T.; Ting, J. W.; Chang, H.-C. Averagingpeak-to-peak voltage detector for absolute mass determination ofsingle particles with quadrupole ion traps. Rev. Sci. Instrum. 2005, 76,023108.(15) Peng, W.-P.; Yang, Y.-C.; Lin, C.-W.; Chang, H.-C. Molar massand molar mass distribution of polystyrene particle size standards.Anal. Chem. 2005, 77, 7084−7089.(16) Howder, C. R.; Long, B. A.; Bell, D. M.; Furakawa, K. H.;Johnson, R. C.; Fang, Z.; Anderson, S. L. Photoluminescence ofcharged CdSe/ZnS quantum dots in the gas phase: Effects of chargeand heating on absorption and emission probabilities. ACS Nano2014, 8, 12534−12548.(17) Peng, W.-P.; Lin, H.-C.; Lin, H.-H.; Chu, M.; Yu, A. L.; Chang,H.-C.; Chen, C.-H. Charge-monitoring laser-induced acousticdesorption mass spectrometry for cell and microparticle massdistribution measurement. Angew. Chem., Int. Ed. 2007, 46, 3865−3869.(18) Nie, Z.; Cui, F.; Chu, M.; Chen, C.-H.; Chang, H.-C.; Cai, Y.Calibration of a frequency-scan quadrupole ion trap massspectrometer for microparticle mass analysis. Int. J. Mass Spectrom.2008, 270, 8−15.(19) Frank, M.; Labov, S. E.; Westmacott, G.; Benner, W. H.Energy-sensitive cryogenic detectors for high-mass biomolecule massspectrometry. Mass Spectrom. Rev. 1999, 18, 155−186.(20) Rabin, M. W.; Hilton, G. C.; Martinis, J. M. Application ofmicrocalorimeter energy measurement to biopolymer mass spectrom-etry. IEEE Trans. Appl. Supercond. 2001, 11, 242−247.(21) Wenzel, R. J.; Matter, U.; Schultheis, L.; Zenobi, R. Analysis ofmegadalton ions using cryodetection MALDI time-of-flight massspectrometry. Anal. Chem. 2005, 77, 4329−4337.(22) Plath, L. D.; Ozdemir, A.; Aksenov, A. A.; Bier, M. E.Determination of iron content and dispersity of intact ferritin bysuperconducting tunnel junction cryodetection mass spectrometry.Anal. Chem. 2015, 87, 8985−8993.(23) Chen, R.; Wu, Q.; Mitchell, D. W.; Hofstadler, S. A.;Rockwood, A. L.; Smith, R. D. Direct charge number and molecularweight determination of large individual ions by electrosprayionization Fourier transform ion cyclotron resonance mass spectrom-etry. Anal. Chem. 1994, 66, 3964−3969.(24) Makarov, A.; Denisov, E. Dynamics of ions of intact proteins inthe Orbitrap mass analyzer. J. Am. Soc. Mass Spectrom. 2009, 20,1486−1495.(25) Kafader, J. O.; Beu, S. C.; Early, B. P.; Melani, R. D.; Durbin, K.R.; Zabrouskov, V.; Makarov, A. A.; Maze, J. T.; Shinholt, D. L.; Yip,

P. F.; Kelleher, N. L.; Compton, P. D.; Senko, M. W. STORI PlotsEnable Accurate Tracking of Individual Ion Signals. J. Am. Soc. MassSpectrom. 2019, 30, 2200−2203.(26) Shockley, W. Currents to conductors induced by a movingpoint charge. J. Appl. Phys. 1938, 9, 635−636.(27) Weinheimer, A. J. The charge induced on a conducting cylinderby a point charge and its application to the measurement of charge onprecipitation. J. Atmos. Oceanic Technol. 1988, 5, 298−304.(28) Shelton, H.; Hendricks, C. D.; Wuerker, R. F. Electrostaticacceleration of microparticles to hypervelocities. J. Appl. Phys. 1960,31, 1243−1246.(29) Fuerstenau, S. D.; Benner, H. W. Molecular weightdetermination of megadalton DNA electrospray ions using chargedetection time-of-flight mass spectrometry. Rapid Commun. MassSpectrom. 1995, 9, 1528−1538.(30) Benner, W. H. A gated electrostatic ion trap to repetitiouslymeasure the charge and m/z of large electrospray ions. Anal. Chem.1997, 69, 4162−4168.(31) Maze, J. T.; Jones, T. C.; Jarrold, M. F. Negative droplets frompositive electrospray. J. Phys. Chem. A 2006, 110, 12607−12612.(32) Mabbett, S. R.; Zilch, L. W.; Maze, J. T.; Smith, J. W.; Jarrold,M. F. Pulsed acceleration charge detection mass spectometry:Application to weighing electrosprayed droplets. Anal. Chem. 2007,79, 8431−8439.(33) Gamero-Castaño, M. Induction charge detector with multiplesensing stages. Rev. Sci. Instrum. 2007, 78, 043301.(34) Gamero-Castaño, M. Retarding potential and induction chargedetectors in tandem for measuring the charge and mass ofnanodroplets. Rev. Sci. Instrum. 2009, 80, 053301.(35) Smith, J. W.; Siegel, E. E.; Maze, J. T.; Jarrold, M. F. Imagecharge detection mass spectrometry: Pushing the envelope withsensitivity and accuracy. Anal. Chem. 2011, 83, 950−956.(36) Doussineau, T.; Bao, C. Y.; Clavier, C.; Dagany, X.; Kerleroux,M.; Antoine, R.; Dugourd, P. Infrared multiphoton dissociationtandem charge detection-mass spectrometry of single megadaltonelectrosprayed ions. Rev. Sci. Instrum. 2011, 82, 084104.(37) Doussineau, T.; Mathevon, C.; Altamura, L.; Vendrely, C.;Dugourd, P.; Forge, V.; Antoine, R. Mass determination of entireamyloid fibrils by using mass spectrometry. Angew. Chem., Int. Ed.2016, 55, 2340−2344.(38) Doussineau, T.; Kerleroux, M.; Dagany, X.; Clavier, C.;Barbaire, M.; Maurelli, J.; Antoine, R.; Dugourd, P. Chargingmegadalton poly(ethylene oxide)s by electrospray ionization. Acharge detection mass spectrometry study. Rapid Commun. MassSpectrom. 2011, 25, 617−623.(39) Doussineau, T.; Deśert, A.; Lambert, O.; Taveau, J.-C.;Lansalot, M.; Dugourd, P.; Bourgeat-Lami, E.; Ravaine, S.; Duguet, E.;Antoine, R. Charge detection mass spectrometry for the character-ization of mass and surface area of composite nanoparticles. J. Phys.Chem. C 2015, 119, 10844−10849.(40) Elliott, A. G.; Harper, C. C.; Lin, H.-W.; Susa, A. C.; Xia, Z.;Williams, E. R. Simultaneous measurements of mass and collisionalcross section of single ions with charge detection mass spectrometry.Anal. Chem. 2017, 89, 7701−7708.(41) Elliott, A. G.; Harper, C. C.; Lin, H.-W.; Williams, E. R. Mass,Mobility and MSn Measurements of Single Ions Using ChargeDetection Mass Spectrometry. Analyst 2017, 142, 2760−2769.(42) Pierson, E. E.; Keifer, D. Z.; Selzer, L.; Lee, L. S.; Contino, N.C.; Wang, J. C.; Zlotnick, A.; Jarrold, M. F. Detection of lateintermediates in virus capsid assembly by charge detection massspectrometry. J. Am. Chem. Soc. 2014, 136, 3536−3541.(43) Keifer, D. Z.; Motwani, T.; Teschke, C. M.; Jarrold, M. F.Measurement of the accurate mass of a 50 MDa infectious virus.Rapid Commun. Mass Spectrom. 2016, 30, 1957−62.(44) Pierson, E. E.; Keifer, D. Z.; Kukreja, A. A.; Wang, J. C.-Y.;Zlotnick, A.; Jarrold, M. F. Charge detection mass spectrometryidentifies preferred non-icosahedral polymorphs in the self-assemblyof woodchuck hepatitis virus capsids. J. Mol. Biol. 2016, 428, 292−300.



153


(45) Keifer, D. Z.; Motwani, T.; Teschke, C. M.; Jarrold, M. F.Acquiring structural information on virus particles via chargedetection mass spectrometry. J. Am. Soc. Mass Spectrom. 2016, 27,1028−1036.(46) Pierson, E. E.; Keifer, D. Z.; Asokan, A.; Jarrold, M. F.Resolving adeno-associated viral particle diversity with chargedetection mass spectrometry. Anal. Chem. 2016, 88, 6718−6725.(47) Delalande, L.; Tsvetkova, I. B.; Zeng, C.; Bond, K. M.; Jarrold,M. F.; Dragnea, B. Wrapping a virus in a molecular net. Nanoscale2016, 8, 16221−16228.(48) Saxena, P.; He, L.; Malyutin, A.; Datta, S. A. K.; Rein, A.; Bond,K. M.; Jarrold, M. F.; Spilotros, A.; Svergun, D.; Douglas, T.; Dragnea,B. Virus matryoshka: Virus-like particles of immature HIV-1assembled on bacteriophage-derived templates. Small 2016, 12,5862−5872.(49) Lutomski, C. A.; Lyktey, N. A.; Zhao, Z.; Pierson, E. E.;Zlotnick, A.; Jarrold, M. F. HBV capsid completion occurs througherror correction. J. Am. Chem. Soc. 2017, 139, 16932−16938.(50) Motwani, T.; Lokareddy, R. K.; Dunbar, C. A.; Cortines, J. R.;Jarrold, M. F.; Cingolani, G.; Teschke, C. M. A viral scaffoldingprotein triggers portal ring oligomerization and incorporation duringprocapsid assembly. Science Advances 2017, 3, e1700423.(51) Lee, L. S.; Brunk, N.; Haywood, D. G.; Keifer, D.; Pierson, E.;Kondylis, P.; Wang, J. C.-Y.; Jacobson, S. C.; Jarrold, M. F.; Zlotnick,A. A molecular breadboard: Removal and replacement of subunits in ahepatitis B virus capsid. Prot. Sci. 2017, 26, 2170−2180.(52) Lutomski, C. A.; Lyktey, N. A.; Pierson, E. E.; Zhao, Z.;Zlotnick, A.; Jarrold, M. F. Multiple pathways in capsid assembly. J.Am. Chem. Soc. 2018, 140, 5784−5790.(53) Dunbar, C. A.; Callaway, H. M.; Parrish, C. R.; Jarrold, M. F.Probing antibody binding to canine parvovirus with charge detectionmass spectrometry. J. Am. Chem. Soc. 2018, 140, 15701−15711.(54) Kim, S. J.; Fernandez-Martinez, J.; Nudelman, I.; Shi, Y.;Zhang, W.; Raveh, B.; Herricks, T.; Slaughter, B. D.; Hogan, J.; Upla,P.; Chemmama, I. E.; Pellarin, R.; Echeverria, I.; Shivaraju, M.;Chaudhury, A. S.; Wang, J.; Williams, R.; Unruh, J. R.; Greenberg, C.H.; Jacobs, E. Y.; Yu, Z.; de la Cruz, M. J.; Mironska, R.; Stokes, D. L.;Aitchison, J. D.; Jarrold, M. F.; Gerton, J. L.; Ludtke, S. J.; Akey, C.W.; Chait, B. T.; Sali, A.; Rout, M. P. Structure and functionalanatomy of the nuclear pore complex. Nature 2018, 555, 475−482.(55) Lutomski, C. A.; Gordon, S. M.; Remaley, A. T.; Jarrold, M. F.Resolution of lipoprotein subclasses by charge detection massspectrometry. Anal. Chem. 2018, 90, 6353−6356.(56) Dunbar, C. A.; Rayaprolu, V.; Wang, J. C.-Y.; Brown, C. J.;Leishman, E.; Jones-Burrage, S.; Trinidad, J. C.; Bradshaw, H. B.;Clemmer, D. E.; Mukhopadhyay, S.; Jarrold, M. F. Dissecting thecomponents of sindbis virus from arthropod and vertebrate hosts:Implications for infectivity differences. ACS Infect. Dis. 2019, 5, 892−902.(57) Keifer, D. Z.; Shinholt, D. L.; Jarrold, M. F. Charge detectionmass spectrometry with almost perfect charge accuracy. Anal. Chem.2015, 87, 10330−10337.(58) Pierson, E. E.; Contino, N. C.; Keifer, D. Z.; Jarrold, M. F.Charge detection mass spectrometry for single ions with anuncertainty in the charge measurement of 0.65 e. J. Am. Soc. MassSpectrom. 2015, 26, 1213−1220.(59) Cottini, C.; Gatti, E.; Giannelli, G.; Rozzi, G. Minimum noisepre-amplifier for fast ionization chambers. Nuovo Cimento 1956, 3,473−483.(60) Rehak, P.; Walton, J.; Gatti, E.; Longoni, A.; Sanpietro, M.;Kemmer, J.; Dietl, H.; Holl, P.; Klanner, R.; Lutz, G.; Wylie, A.;Becker, H. Progress in semiconductor drift detectors. Nucl. Instrum.Methods Phys. Res., Sect. A 1986, 248, 367−378.(61) Bertuccio, G.; Rehak, P.; Xi, D. M. A novel charge sensitivepreamplifier without the feedback resistor. Nucl. Instrum. MethodsPhys. Res., Sect. A 1993, 326, 71−76.(62) Bertuccio, G.; Macera, D.; Graziani, C.; Ahangarianabhari, M. ACMOS charge sensitive amplifier with sub-electron equivalent noise

charge. 2014 IEEE Nuclear Science Symposium and Medical ImagingConference (NSS/MIC); DOI: 10.1109/NSSMIC.2014.7431123.(63) Goulding, F. S.; Walton, J. T.; Malone, D. F. An opto-electronicfeedback preamplifier for high-resolution nuclear spectroscopy. Nucl.Instrum. Methods 1969, 71, 273−279.(64) Landis, D. A.; Goulding, F. S.; Jakelvic, J. M. Performance of apulsed light feedback preamplifier for semiconductor detector X-rayspectrometers. Nucl. Instrum. Methods 1970, 87, 211−213.(65) Landis, D. A.; Cork, C. P.; Madden, N. W.; Goulding, F. S.Transistor reset preamplifier for high rate high resolution spectros-copy. IEEE Trans. Nucl. Sci. 1982, NS-29, 619−624.(66) Jordanov, V. T. Charge sensitive amplifier with pulsed sourcereset. US Patent No. US6587003, 2003.(67) Contino, N. C.; Pierson, E. E.; Keifer, D. Z.; Jarrold, M. F.Charge Detection Mass Spectrometry With Resolved Charge States. J.Am. Soc. Mass Spectrom. 2013, 24, 101−108.(68) Hogan, J. A.; Jarrold, M. F. Optimized electrostatic linear iontrap for charge detection mass spectrometry. J. Am. Soc. MassSpectrom. 2018, 29, 2086−2095.(69) Draper, B. E.; Anthony, S. N.; Jarrold, M. F. The FUNPET- anew hybrid ion funnel-ion carpet atmospheric pressure interface forthe simultaneous transmission of a broad mass range. J. Am. Soc. MassSpectrom. 2018, 29, 2160−2172.(70) Shockley, W. Currents to conductors induced by a movingpoint charge. J. Appl. Phys. 1938, 9, 635−636.(71) Weinheimer, A. J. The charge induced on a conducting cylinderby a point charge and its application to the measurement of charge onprecipitation. J. Atmos. Oceanic Technol. 1988, 5, 298−304.(72) Draper, B. E.; Jarrold, M. F. Real time analysis and signaloptimization for charge detection mass spectrometry. J. Am. Soc. MassSpectrom. 2019, 30, 898−904.(73) Keifer, D. Z.; Alexander, A. W.; Jarrold, M. F. Spontaneousmass and charge losses from single multi-megadalton ions studied bycharge detection mass spectrometry. J. Am. Soc. Mass Spectrom. 2017,28, 498−506.(74) Contino, N. C.; Jarrold, M. F. Charge detection massspectrometry for single ions with a limit of detection of 30 charges.Int. J. Mass Spectrom. 2013, 345−347, 153−159.



154
http://dx.doi.org/10.1109/NSSMIC.2014.7431123http://dx.doi.org/10.1021/jasms.9b00010

implementation of a charge-sensitive ampliﬁer without a ...sep 26, 2019 · charge generally...

Documents