Radio Frequency Linear Accelerators for NDT Applications: Basic Overview of RF Linacs

by Harold J. Hansen*

Superman’s X-ray vision can see through anything! Most of us who have performed radiography have often wished for that kind of penetrating radiation for very thick or very dense (high atomic number) specimen material, like lead. Modern accelerators are coming closer to giving us our wish. Here are a few of the basics for this new tool in our "bag of tricks." Frank A. Iddings Tutorial Projects Editor

  

Figures 1-3
Figures 4-5

Introduction
High energy X-ray radiography can be an important part of a quality control program. In this article we will present an overview of the technology found in a typical high energy X-ray source, the radio frequency (RF) linear accelerator.

RF linear accelerators, or linacs, have many applications, including radiation therapy, medical and food product sterilization by irradiation, polymer cross linking and, of course, NDT inspection. While every application has its special design requirements, there is a core of common technology that is found in every RF linac.

In NDT, linacs are used primarily for the inspection of thick sections of materials. Linacs are also used in applications such as high energy computed tomography of specimens greater than 1 m thick and cargo container inspection. Recent developments in reliable portable linacs are opening up other applications such as field inspection of pipelines, ships, bridges, and other civil infrastructure. The replacement of isotopes (such as Co-60) by the linac is an area for growth in the future. The shorter exposure times, improved image capabilities, and greatly reduced regulatory requirements of the linac make a persuasive argument for the replacement of isotopes with a portable linac.

The linacs discussed here are those with X-ray energies from 1 to 20 MeV intended for use in NDT applications. The discussion will be in very broad terms; it will be impossible to discuss every variation in linac design. In addition, some topics have been necessarily simplified to increase the comprehensibility for a wider audience; we ask the indulgence of the few accelerator "experts" out there who may take offense at this.

---

The development of the RF linac is a classic example of basic research leading to significant, unforeseen development of commercial markets.

---

 

History
It is interesting to take a look at the history of the development of the RF linac. The particle accelerator - a category of high energy equipment which includes linear and circular machines such as betatrons, synchrotrons, cyclotrons, synchrocyclotrons, microtrons, and electron-proton RF linacs - is one of the most important tools that modern science possesses for probing the atom. The very high velocity particles produced by these accelerators can be used to break apart atoms, allowing scientists to probe the fundamental principles of matter and energy.

The development of the RF linac is a classic example of basic research leading to significant, unforeseen development of commercial markets. The driving force in the development of accelerators was the study of the atom. With the modern RF linac, the main driving force is now cancer therapy. The vast majority of commercial linac manufacturers in the world have focused their efforts on linacs for medical use. NDT RF linacs represent only a small portion of the overall linac trade.

Giving credit to the first inventor of the RF linac is difficult because it is based on the work of many scientists dating all the way back to the 19th century. One of the first linac designs was described by R. Widereo in 1929 (Livingstone, 1966). His design was limited by the lack of high power microwave sources, but it is clear from his published papers that he understood many of the principles which have led to the modern linac.

Starting in the 1950’s through the end of the 1960’s, there were quite a few companies developing commercial RF linacs. As governments started to recognize the importance of this technology, other efforts were funded at large universities and national laboratories across the US and around the world.

 

Overview
The overall block diagram of a simplified linac is shown in Figure 1. In basic terms, an RF linac uses microwave energy to create electric fields which are then used to accelerate electrons. The accelerated electrons strike a metal target to produce X-rays. Because the required RF power to set up these accelerating fields is so high (typically megawatts), the power can be produced only in very short pulses, typically 5 to 10 µs. The pulsed RF energy (typical frequencies 3 to 10 GHz), is developed in special microwave RF generator tubes. These microwave tubes require high voltage pulses and hundreds of amps of current to operate. The pulses to operate the microwave generators are created in a device called a modulator. The modulator requires high voltage DC which is produced in a high voltage supply; this is either part of the modulator or a separate system.

In order to facilitate our discussions of the RF linac, we will break it into functional subsystems and discuss each in detail. The decision as to where to divide the system is primarily one of convenience and should not be construed as the best or the only way to break up the system.

 

High Voltage Supply
The function of the high voltage supply is to produce the high voltages required for proper modulator operation. A typical modulator requires 10 to 12 kV at power levels of 5 to 10 kW. The high voltage supply should be regulated, filtered and have some type of feedback of both the voltage and current. It must be protected against over-voltage and over-current conditions and be capable of withstanding high stresses during normal operation as well as in the event of a failure.

There are two basic approaches in the design of the high voltage supply. The traditional approach is what may be called the "brute force" method. In this approach, a large high voltage transformer is used with some type of rectification and filtering. The basic brute force high voltage supply is shown in Figure 2.

The second method for constructing the high voltage supply is by the use of high frequency switching supplies. The trend in the power supply industry has been away from linear brute force and toward switching supplies. It is now possible to get the entire high voltage supply into a 483 mm (19 in.) rack module, rather than taking up an entire rack. These high voltage supplies are available "off-the-shelf" from several manufacturers, are very reliable, and in most cases are less expensive than the brute force method. All regulation and feedback is accomplished inside the supply, greatly simplifying the rest of the design.

Regulation of the high voltage is important, as changes in high voltage result in changes in RF output and ultimately cause changes in the output of the linac. There are two basic ways of providing regulation. Direct regulation of the high voltage supply is one method. Post regulation performed on a pulse-to-pulse basis is another technique. For NDT applications this is usually sufficient to regulate at the high voltage supply.

 

Modulator
The function of the modulator is to provide high voltage pulses to the microwave generator. Almost every RF linac in production today uses some variation of the line type modulator. This design was used extensively during World War II for radar applications. They are called line modulators because the width of the output pulse is determined by an actual transmission line. Modern modulators use an artificial transmission line called a pulse-forming network. The basic circuit of a line type modulator is shown in Figure 3. The major components are the charging inductor, the pulse-forming network, the charging diode, the power switch tube, and the pulse transformer.

Operation of the line type modulator can best be understood by dividing its operation into two parts: the charging cycle and the discharging cycle. On the charging cycle, the charging inductor and the capacitance of the pulse-forming network form a resonant circuit. This resonance causes the pulse-forming network to charge up to twice the voltage supplied by the high voltage supply (20 kV for a 10 kV supply). The charging diode keeps the pulse-forming network voltage at full voltage until the discharge cycle is initiated.

The discharge cycle is initiated by conduction of the power switch. The components which operate in the discharge cycle are the power switch, pulse-forming network, and pulse transformer. The discharge cycle results in a pulse of 10 kV (with a pulse width determined by the design of the pulse-forming network) to appear across the input of the pulse transformer. With a typical ratio of 1:4 for the pulse transformer, the output pulse will be a 40 kV pulse.

In the description of the discharge cycle, we referred to the power switch; this is almost always a hydrogen thyratron. The most basic thyratron tube has three elements: the anode, cathode, and grid. (See Figure 3.) It is normally nonconducting with the grid at zero or a slightly negative voltage. When a positive pulse is applied to the grid, the tube will conduct and stay conducting until the anode current reduces to zero. Once triggered, the grid loses all control. (Early modulator designs used rotating mechanical switches similar to the distributor cap in automobile engines; it was found that these mechanical switches could not handle the repetitive switching requirement of modulator operation.) Thyratrons can handle continuous loads of hundreds of amps of current at typical modulator voltages. It is possible to substitute thyristors, which are semi-conductor equivalents of thyratrons, for some of the lower power designs.

 

Radio Frequency System
The RF system converts the high-voltage pulses from the modulator into pulsed radio-frequency energy. The RF pulses are sent to the accelerating structure to set up an electric field which is used to accelerate electrons and produce X-rays. A simplified diagram of the RF system is shown in Figure 4.

The main component of the RF system is the microwave source. There is a variety of microwave tubes for generating and amplifying microwave signals. The two most common ones used in linacs are magnetrons and klystrons.

A magnetron is a microwave power oscillator which belongs to the family of electron tubes called crossed field devices. This is because it has an electric field (E) and a magnetic field (H) which are perpendicular to each other. The magnetron consists of a circular cathode inside a circular anode block. There are resonant cavities machined into the anode block. These cavities will resonant at microwave frequencies when excited by electrons interacting with the E and H fields. The operating frequency of a magnetron must be adjusted to match the frequency of the system. This is accomplished with a tuning stub which protrudes into one of the anode cavities. The frequency of the magnetron is controlled via this tuning stub by a system called the automatic frequency control.

Klystrons are also useful as microwave sources. They belong to the class of tubes called linear beam tubes. In most linac applications, the klystron is used as an amplifier, so an input signal is required. This is provided by a low power oscillator typically called an RF driver.

It is possible to use either a magnetron or a klystron in a NDT linac. The choice of which type RF generator tube is used is based partially on design requirements and partially on historical preference. Manufacturers that have a historical background in klystrons tend to use them whenever possible while those manufacturers with a magnetron background prefer magnetrons. From a technical standpoint, magnetrons are easier to drive, can be operated in any physical position, are smaller, lighter, and less expensive. Klystrons are typically capable of higher power levels and have longer lives. Klystrons require some type of RF drive and supplies for the focus solenoid(s). Klystrons are normally designed to be operated in only one physical orientation, which prevents their use in most portable applications. Klystrons are electrically "quieter" than magnetrons.

The output of the microwave source, whether klystron or magnetron, is fed into a device called an isolator. Isolators come in varying configurations and, like many microwave components, they are classified by the number of input/output ports (typically 2, 3, or 4 ports). The function of an isolator is to buffer the microwave source from the accelerating structure. The structure represents a highly dynamic load. At the beginning of the pulse, the structure is basically a short circuit; most of the RF power is reflected back toward the source. Magnetrons, to a large degree, and klystrons, to a much lesser degree, are adversely affected by this reflected power. With the isolator in-line, the reflected power is either absorbed by the isolator (2 port isolators) or redirected into RF loads (3 and 4 port isolators).

All the RF power in the system is transmitted through a special transmission line called a waveguide. A waveguide is hollow pipe, having either a circular or rectangular cross section, which is used to direct or guide the RF energy from one point or another. It usually requires pressurizing with an insulating gas (freon or sulfur hexafluoride) to withstand the high voltage stress at these power levels. The physical dimensions of a waveguide are based on the frequency of the RF energy. At the typical accelerator frequency, 3 GHz, the rectangular waveguide has an interior cross section of 32 ´ 70 mm (1.25 ´ 2.75 in.).

 

X-ray System
In the X-ray system, RF is applied to the accelerating structure to accelerate the electrons and produce X-rays. The X-ray system can be either a separate system or part of the RF system.

The main component in the X-ray system is the accelerating structure. There are many possible designs for accelerating structures; in general the structures can be classified into two separate categories: traveling wave and standing wave.

The traveling wave structure was the earliest type developed and its operation is well described in the scientific literature. In this type, the microwave energy propagates along the structure, carrying the electrons along like a "surfboard on a wave." Power is transferred from the RF wave to the electron beam; any power not transferred to the electrons exits at the end of the structure. This RF energy is either absorbed in a RF load or recycled back into the input, increasing overall efficiency of the system. There are many variations to the traveling wave design. Traveling wave structures are the easier to identify as they have two ports: one input and one output.

The standing wave structure is a variation of the traveling wave design, from which it was developed by shorting out both ends of the structure and allowing a standing wave pattern to develop. The RF is usually (although not always) fed into the center of the standing wave structure. The easiest way to identify a standing wave is that there is only one port, which is usually located in the physical center of the structure.

The choice (at the design stage) to employ either traveling wave or standing wave accelerator technology depends on several factors. As with microwave generator tubes, there is a certain historical preference involved. From a technical standpoint, traveling wave structures will be slightly longer than standing wave for a given energy and field strength. Traveling wave structures have a radial defocusing effect which must be countered by the use of an external focus solenoid along the structure. On the other hand, a properly designed standing wave structure is self focusing and thus requires no external solenoid. It is easier to vary the output energy of a traveling wave structure than a standing wave design, though both types can be varied. Since the majority of NDT linacs today use standing wave structures, we will restrict our discussion to standing wave designs.

There are several variations in the design of a standing wave structure. One way these various designs are classified is by the location of the coupler cells. An standing wave acceleration structure consists of series of resonant cavities (called accelerating cavities or cells) with a common axis: energy is fed from one cavity to another through special cavities called couplers. It is the location of these cavities in relation to the accelerating cavities that we are referring to. There are two main types of coupler designs: off-axis and on-axis (also known as inline). In general, the off-axis coupler design is slightly more efficient than an on-axis design. This higher efficiency is paid for by a higher manufacturing cost due to its complex design and more stringent tuning requirements. The on-axis design is easier to manufacture and is usually easier to tune.

While the details of the structure design are complex and beyond the scope of this paper, we can take a look at the major components of the structure. The main components (shown in Figure 5) of a structure are the RF window, the gun, the buncher, the relativistic section, and the target.

The RF window separates the high vacuum of the structure from the insulating gas in the RF wave guides. The window can be constructed of ceramic; another popular window material is glass. In some designs the window must be cooled, usually by water along the outside edge. In some designs the window is removable; this can be a desirable feature.

The gun is the source of electrons to be accelerated. There are various design options for the gun. The gun can be a simple diode gun (only two elements) or a more complicated triode gun (three elements). Some type of regulation of the gun emission current is recommended. For a diode gun, the gun is usually designed to be space charge limited, which can provide the required regulation. The gun is electrically isolated from the structure so that high voltage pulses can be applied. This high voltage accelerates or injects the electrons into the first portion of the structure.

The first section of the structure is called the buncher. Electrons produced in the gun enter the structure through the buncher. Proper acceleration of the electrons requires that the electrons enter the structure at the correct time and phase of the RF field. For an standing wave design, the buncher operation is fairly simple. The design of the buncher affects the energy distribution of the accelerated electron beam. For an NDT linac, the width of the energy distribution of the electron beam is not critical since the output X-rays will have a wide spectrum of energies anyway.

The relativistic section is where the majority of the acceleration takes place. It is called the relativistic section because electrons can reach speeds close to that of light. The accelerated electrons are directed down the structure into the target.

In the target the electrons are converted into X-rays. Linacs use a type of target called a transmission target. Electrons enter one side; X-rays exit at the other. The conversion efficiency of electron to X-ray is related to the atomic number of the target; a high atomic number is usually desirable. Since the majority of the electron beam’s energy (94 percent) is converted into heat, the target must be able to tolerate the heat or to dissipate it very fast. Tungsten is the preferred material, though gold or even uranium can be used. A practical design will sometimes use a layered target comprising either tungsten or gold brazed to copper. The target is usually water cooled to remove heat.

One important linac parameter which should be mentioned here is X-ray spot size, which determines the imaging capability of the linac. Two linacs with the same energy, radiation output, collimation, and radiographic set-up will be capable of the same image quality if they have the same X-ray spot size. There is a design limitation on minimum spot size due to scattering of electrons in the target material. In the typical thick target design (in which the thickness of the target is slightly greater than the penetration depth of the electrons), the smallest X-ray spot achievable is somewhere between 0.25 and 0.5 mm (0.001 and 0.002 in.). It is possible to get smaller spot sizes using a "thin" target design; it is important, however, to ensure that any electrons which pass though the target (these will have energies in the MeV range) are trapped in some manner, either by a magnetic field or in a low density material such as carbon.

The output of the linac experiences an effect commonly called "beaming." As the energy of the electrons hitting the target increases, the transmitted output becomes narrower and peaks along the structure axis. In some cases a special filter, called a flattening filter, is used to provide a more uniform field. While typically only transmission targets are used on linacs, it is possible to get a panoramic output at low energies (1-2 MeV). Panoramic output at higher energies is limited by the forward beaming effect.

 

Auxiliary Systems
Automatic Frequency Control
The accelerating structure operates at a very narrow range of frequencies. It is vital that the RF generator operate at the correct frequency. As the structure heats up, the operating frequency shifts, requiring the RF generator to track the change so as to maintain maximum output. Typically there is a circuit in a linac which automatically adjusts the RF generator to the correct frequency. This portion of the linac is called the automatic frequency control (AFC). There are two basic designs for AFC systems: the double tuned discriminator AFC and the quadrature hybrid phase type.

In the double tuned discriminator AFC system, there are two resonant cavities, fed off the RF system, tuned slightly above and below the desired operating frequency of the structure. These cavities are of similar design to the structure cavities and are thermally linked to the structure. The output from the tuned cavities is rectified and used to create an error signal which can adjust the frequency of the microwave generator.

The second type of AFC, the quadrature hybrid phase, uses a system which compares the phase of the forward and reflected power from the structure. The structure is a resonant circuit; thus at resonance the phase shift between forward and reflected power will be zero. Using standard microwave components, it is possible to measure this phase shift and develop an error signal to keep the system in tune. This type of circuit is more popular than the first type of AFC. Each type has its advantages and disadvantages.

 

Energy Change
It is sometimes desirable to have several output energies from a single linac. When a standing wave structure is designed, it is normally optimized to operate at one discrete energy. It is possible, by changing some parameters, to change that energy over a limited range. There are several techniques which can be used to accomplish this.

One of the ways to change the output energy is to change the beam current. The beam current has a loading effect on the structure: if the current goes up, the energy can be made to go down within certain limits. When the beam current is increased, it has the indirect effect of reducing the structure field power. It is also possible to change the energy by directly changing the structure field power.

RF field power can changed by adjusting the operating point of the microwave generator. It also can be adjusted by putting a variable mismatch in line with the RF feed to the structure. A combination of both typically works best since there is a limit as to how much you can adjust the energy using either technique alone.

 

Dose Monitoring
Monitoring of the output dose is an important factor if inspection results are going to be consistent. In most designs there is an ion chamber placed directly in the path of the output beam. This can be used to give both dose rates and counts. For factory applications, where many of the same parts and same radiographic set-up are used, an ion chamber at the output is usually sufficient. Once the required set-up has been determined, consistency between radiographs can be attained. In some applications, such as field work, you do not have the luxury of this "trial and error" method to determine the set-up requirements. In this case, an ion chamber is normally placed at the film plane to monitor the dose; this allows users to get predictable results with just a single exposure. This type of ion chamber is called a film plane dosimeter.

For any type of radiation measurement, either exposure control or safety monitoring, it is important to use the proper measurement device. Not all radiation detectors work with a linac. For example, some Geiger tube based meters will not function at all because the tube saturates on each pulse; you end up counting pulses, not radiation. If users plan to use a particular radiation meter for personnel monitoring, you must be sure that is rated for linac use.

 

Conclusion
This is an attempt to give users a better understanding of how the basic RF linac operates. Since there are books with thousands of pages which describe this technology, this attempt obviously has only scratched the surface. Linacs will continue to be an important part of the quality control process and as new markets and applications open up, users can expect to see more in everyday use.

 

References
Livingstone, M.S., The Development of High Energy Accelerators, 1966, New York, Dover, pp 92-114.

 

* L &W Research, Inc., 121 N. Plains Ind. Rd., Wallingford, CT 06492, (203) 949 0142; fax (203) 949 0148.

 

Copyright © 1998 by the American Society for Nondestructive Testing, Inc. All rights reserved.

[ Materials Evaluation ]

IRRSP, NDT Handbook, The NDT Technician and www.asnt.org are trademarks of the American Society for Nondestructive Testing, Inc. ACCP, ASNT, Level III Study Guide, Materials Evaluation, Nondestructive Testing Handbook, Research in Nondestructive Evaluation and RNDE are registered trademarks of the American Society for Nondestructive Testing, Inc. ASNT exists to create a safer world by promoting the profession and technologies of nondestructive testing.