Inside the trap the potential is a saddle: uphill in one direction and downhill in the other. A static saddle cannot hold an ion, because the ion rolls off the downhill side. The RF voltage on the ring flips the saddle back and forth. If it flips fast enough, the ion is pushed back toward the center before it can roll away, and on average it feels a shallow bowl. That is dynamic stabilization.
The surface shows the ion’s potential energy in the r–z plane: r points toward the ring electrode and z toward the end caps. Change the RF amplitude, the frequency or the ion’s m/z and watch qz: the ion stays trapped while qz < 0.908 (with no DC voltage). Drag the ion in the small window to start it somewhere else.
Paul’s model replaces the flipping field with a mechanical saddle that spins, holding a steel ball. It works by the same mathematics: below a critical spin rate the ball rolls off; above it the ball stays near the center.
The trap is a ring electrode between two end caps, all with hyperbolic surfaces. The RF voltage V cos Ωt goes on the ring and the end caps are grounded. Inside, the potential is quadratic in all three directions, Φ ∝ αx² + βy² + γz². Laplace’s equation (∇²Φ = 0, no charge inside) forces α + β + γ = 0, so with α = β = 1 we get γ = −2: Φ ∝ x² + y² − 2z². The z curvature is twice as strong as the r curvature and opposite in sign, which is why the ring surface is r² − 2z² = r0² and the end caps are 2z² − r² = 2z0².
Play the RF slowly and watch the two parabolas: when the ring is positive, a positive ion is pushed toward the center along r and away from it along z; half a cycle later the two swap. The dashed curves show the average well the ion feels.
The diagram is plotted in az and qz, the dimensionless forms of U and V (slide 42). The z motion is stable inside one region and the r motion inside another (with ar = −az/2 and qr = −qz/2); an ion is trapped only where they overlap. The thin lines inside are lines of constant βz: the frequency of the ion’s slow secular motion along z as a fraction of half the RF frequency. Drag the dot to set V and U directly. With U = 0 the dot runs along the qz axis and leaves the stable region at qz = 0.908.
A trapped ion oscillates slowly along z and along r at the same time, each at its own secular frequency, ωu = βuΩ/2. Its path in the x–z plane is therefore a Lissajous figure. Near the qz axis βr is about half of βz, so the figure is close to a figure eight. On top of this the RF adds a small, fast wiggle; switch on “Show only the secular motion” to average it out. Drag the working point on the diagram and the β values, the frequencies and the shape of the figure all change.
Several kinds of ions are injected together. Every ion feels the same U and V, but its own az and qz depend on its m/z, so each sits at a different point on the same line through the origin of the stability diagram. Ions whose points lie inside the stable region stay at the center; the rest leave. Move U and V, or tap “Keep only” for an ion, and watch the mixture sort itself: lighter ions become unstable in z and leave through an end cap, heavier ones become unstable in r and hit the ring.
This is how the trap records a spectrum (the mass-selective instability scan). Ions are collected with U = 0 and a low RF amplitude, so all of them sit on the qz axis inside the stable region. V is then ramped. Each ion’s qz grows in proportion to V, and when it passes 0.908 the ion’s motion along z becomes unstable and it leaves through an end cap, lightest first. The m/z reaching 0.908 at any moment is proportional to V, so the detector signal against V is the mass spectrum. In a real trap about half of the ejected ions leave through the end cap facing the detector; here every ejected ion is counted. Ions leave a little after qz passes 0.908, while their swing grows, so the simulated peaks sit slightly (up to about 2%) above the true m/z; instruments calibrate this out with known compounds.
Tandem MS in a trap happens in time rather than in space. The steps follow slides 50–51. (1) Ions are collected and cooled by helium. (2) The precursor is isolated: a small AC voltage on the end caps contains the secular frequency of every other ion, so they are driven in resonance and ejected, while the precursor (left out of the waveform, the notch) stays. (3) V is lowered to make room for fragments: the precursor’s qz drops to 0.25, so the lighter fragments, which have higher qz, stay below 0.908. The low-mass cutoff is (m/z)precursor × qz/0.908. (4) A very weak AC voltage at the precursor’s own secular frequency makes it swing a little further each cycle; collisions with helium turn that motion into internal energy until it fragments (CID). (5) The fragments are scanned out as in a normal scan, or one of them is isolated and fragmented again for MS³ and MS⁴. The example is beauverolide, [M+H]⁺ m/z 488, from slide 52; its fragment intensities are approximate, read from the lecture spectra.
Several ions are injected together and the RF runs for 120 cycles. Ions whose working point (qz, az) lies inside the stable region stay trapped at the center; the rest leave, either through an end cap (unstable in z: too light for these settings) or into the ring (unstable in r: too heavy). Guess the survivor from the conditions, or set U and V so that only one ion stays. Selecting a single ion this way works near the tip of the stable region, at az ≈ 0.150, qz ≈ 0.781.
Separating neighbors one mass unit apart needs a working point very close to that tip, where the stable window is only a few percent wide. Near the tip even a small swing carries an ion out, so in a real trap about 1 mtorr of helium is added: collisions cool the ions toward the center, which is what makes unit resolution possible at higher m/z. The simulated ions here have no helium, so the neighbor rounds use light ions (up to m/z 45), where the bare trap can still separate them.
Trap with r0 = 1.00 cm and z0 = 0.707 cm (so r0² = 2z0²), end caps grounded, RF on the ring, singly charged ions. The motion is shown in slow motion. The average-potential bowl is drawn 5 times deeper than it really is, so it is visible next to the saddle.
Equations follow the lecture: φ = (φ0/r0²)(r² − 2z²), m d²z/dt² = 4e(U + V cos Ωt)z/(r0² + 2z0²), qz = 8eV/(m(r0² + 2z0²)Ω²), qr = −qz/2. The ion’s path is calculated from these equations. The average-potential depth uses Dehmelt’s approximation, Dz ≈ qzV/8, which holds for qz below about 0.4.