rus
Issue #7-8/2023
V.V.Anashin, G.A.Gusev, A.A.Zhukov, A.A.Krasnov, V.S.Kuzminykh, P.A.Piminov, A.M.Semenov
DESCRIPTION OF THE 4+ GENERATION SYNCHROTRON RADIATION SOURCE SRF “SKIF” VACUUM SYSTEM
DESCRIPTION OF THE 4+ GENERATION SYNCHROTRON RADIATION SOURCE SRF “SKIF” VACUUM SYSTEM
DOI: 10.22184/1993-8578.2023.16.7-8.462.475
The paper considers the key decisions taken as a basis in the vacuum system design of the future synchrotron radiation source SRF "SKIF" (Koltsovo, Novosibirsk region, Russia). The focus is on the trade-off of using combined lumped pumps along with distributed pumping based on non-evaporable getters (NEG). An estimate of the time required for training the vacuum beam pipe of the SKIF storage ring is given. Designs of the main elements for the vacuum system are considered.
The paper considers the key decisions taken as a basis in the vacuum system design of the future synchrotron radiation source SRF "SKIF" (Koltsovo, Novosibirsk region, Russia). The focus is on the trade-off of using combined lumped pumps along with distributed pumping based on non-evaporable getters (NEG). An estimate of the time required for training the vacuum beam pipe of the SKIF storage ring is given. Designs of the main elements for the vacuum system are considered.
Теги: non-evaporable getter storage ring synchrotron radiation source ultra-high vacuum источник си накопитель нераспыляемый геттер сверхвысокий вакуум
INTRODUCTION
Nowadays, a synchrotron radiation source (SRS) of generation 4 + of the SRF SKIF is being developed [1]. It is based on a cyclic electron storage ring with an energy of 3 GeV with a perimeter of 476 m and an ultrasmall beam emittance of 75 pm-rad. The storage ring consists of 16 superperiods with long dispersion-free gaps (~6 m). Three of the gaps are supposed to be used for injection and RF systems, and the remaining gaps for staging specialized synchrotron radiation generators (wigglers or ondulators). In addition, 2 SR output channels from the bending magnets will be organized in each superperiod: one from a bending magnet with a small field (0.5 Tesla) for generation of UV and soft X-rays, one from a bending magnet with a large field (2 Tesla) for hard X-rays. There will be 2 RF systems in the drive: main (357 MHz) for energy loss compensation and harmonic (1071 MHz) for bunch elongation. The total current is 400 mA at 510 of the bunch (at full filling of the separatrices with a 10% gap). Injection at the experiment energy is assumed, so that a constant circulating beam current will be maintained. The injector will be a 200 MeV linear accelerator and a booster synchrotron with a perimeter of 150 m, which will perform beam acceleration from 200 MeV to 3 GeV. The injection frequency is 1 Hz. The injector is located in a separate building and is connected to an storage ring by a long transport channel. The storage ring and the large experimental hall are located in one building on a single foundation. Several specialized SR stations are moved to separate buildings.
The magneto-optical system designs of modern synchrotron radiation (SR) sources with small emittance [2] significantly limit the available aperture for vacuum beam chambers. Therefore, distributed vacuum pumps based on non-evaporable getters (NEGs), first applied in the insertion devices of the ESRF SR source [3], are gaining popularity. For example, almost all vacuum beam chambers of SR sources MAX-IV [4] and SIRIUS [5] have a thin-film NEG coating. In practice, such systems do not need any training under radiation. This means that the required dynamic (in the presence of the beam) ultrahigh vacuum can be achieved immediately after NEG activation [3, 6]. Activation itself is a heating procedure at 180÷200 °C, which requires dismantling of precision magnets (MAX IV solution) or space for installation of heaters and thermal insulation (SIRIUS solution) and a large number of mechanical compensators with RF contacts. On the other hand, compact combined vacuum pumps developed in recent years based on NEG cartridges and magnetically discharged ion-getter pumps [7], provide an opportunity to return to the classical system with concentrated pumps, but placed at a small distance; about one meter or even less.
The SKIF storage ring vacuum system will contain both distributed pumps, based on NEG coatings, and concentrated combination pumps. NEG coatings will be used in rectilinear gaps and insertion devices with low aperture chambers operating at room temperature. Concentrated pumps will be used in arches that make up 80 % of the storage perimeter and contain many precision magnetic elements. Such a solution allows to exclude the heating procedure in the arches and, consequently, to minimize the number of mechanical compensators – sources of high-frequency geometrical impedances. In addition, the use of a large number of ion-getter pumps significantly reduces dependence of the ultimate vacuum (compared to NEG) on the total level of micro-leaks, which significantly increases the reliability of the vacuum system as a whole.
The ultimate parameters of the SKIF storage ring will largely depend on the vacuum beam chambers design. Therefore, special attention is paid to selection of materials for fabrication of vacuum elements, designs of pumping ports and connecting elements to minimize resistive and geometrical impedances. The pumping system should provide a vacuum at the level of Pн = 3.3 · 10-7 Pa (in nitrogen equivalent) to achieve a "vacuum" beam lifetime of at least 10 hours at a nominal current of Iн = 0.4 А.
ESTIMATION OF TRAINING TIME UNDER RADIATION
The getter film transforms the vacuum chambers surface from a source of desorbing molecular fluxes into an absorber (sorption pump) of molecules. Therefore, as it was already mentioned, dynamic ultrahigh vacuum is achieved immediately after heating (NEG activation) on all areas with getter coating (approximately 20 % of the perimeter in the case of SKIF storage ring). Thus, the question of estimating the time of training of the inner surface under the action of SR necessary to achieve the required dynamic rarefaction level is relevant only for the SKIF arch sections where concentrated combined pumps will be applied.
At the initial design stage it is not possible to anticipate the coordinates of the available areas for concentrated pumps connection. Nevertheless, it is at the initial stage of the project that the number and parameters of the pumps must be determined with sufficient accuracy to formulate the design task. Obviously, an equidistant arrangement of identical concentrated pumps is the best first approximation. Besides, applying the Knudsen diffusion model and assuming averaging of the distributed flow of molecules from the walls of the vacuum chamber, it is easy to obtain an expression for the average quasi-equilibrium pressure:
, (1)
where q [Pa·l/m/s] – uniformly distributed gas load (desorption of molecules from the chamber walls), L [m] – distance between pumps, S [m3/s] – effective pumping speed of the port with pump, u [m4/s] – molecular conductivity of the chamber of unit length, – pump utilization factor (usually 0.2...0.5), Ps – pump inlet pressure.
It is interesting to note that varying distance between pumps from 0.5 L to 1.5 L, while maintaining the total number of pumps, leads to an increase in the average pressure by a factor of 1÷1.6 compared to the case of equidistant arrangement. The efficiency of the pumps arrangement can be taken into account by introducing an appropriate coefficient g:
. (2)
In SR sources, practically the entire surface of the vacuum beam chamber is exposed to an intense flux of photons with energies much higher than the energy of chemical bonds. Therefore, desorption under the action of SR is much higher than thermal desorption. The cross section of interaction of SR photons with molecules is 2÷3 orders of magnitude smaller than the cross section of inelastic electron-molecular interactions at electron energies of 20÷1000 eV. Therefore, gas desorption under effect of synchrotron radiation occurs in two stages: photons knock out photoelectrons from the irradiated surface (at some depth from the surface), which, in their turn, can lead to desorption of gas molecules from the surface of the vacuum chamber, both when flying from the surface and when hitting it. Besides, a part of photons is reflected and scattered over the total vacuum chamber surface. As a result, even in laboratory experiments to study desorption under the action of SR, one way or another, the entire surface of the vacuum chamber is involved [8]. This significantly complicates the interpretation and comparison of the obtained data. However, the accumulated experience makes it possible to identify the integral parameters of this complex phenomenon necessary for the calculation and design of acceletators.
The main characteristic of desorption under action of photons is the average number of molecules desorbed by one photon η [molecules/photon] – the photo-stimulated desorption coefficient. Then the flux of molecules from the walls can be written as:
, (3)
where is the photon flux [photon/(m∙s)], K ≈ 2.4 ∙ 1020 [1/(Pa ∙ m3)] is the number of molecules in a cubic meter at room temperature and 1 Pa pressure.
As a rule, the dominating gas desorbed under the SR action is hydrogen. The second in intensity is the flux of carbon monoxide (CO), which is 20–30 % of the hydrogen flux. However, since the scattering cross section of relativistic electrons is proportional to the square of the atomic number, scattering on CO molecules is dominant. Further, to simplify the analysis, only the partial pressure of CO is taken into account.
The average SR flux irradiating the walls of the vacuum chamber of the SKIF storage ring is 2 ∙ 1018 [photon/(m ∙ s)] at a nominal electron beam current of 0.4 A and energy of 3 GeV. The interpole distance in the magnetic elements is 30 mm. With a technological gap of 0.5 mm and a vacuum chamber wall thickness of 1 mm, the inner diameter will be 27 mm. Thus, the molecular conductivity on CO of the chamber of unit length will be:
. (4)
Direct calculation shows that at an average pump spacing of 1 m, factor g = 1.6, factor k = 0.25 and desorption factor 5 · 10-7, the average CO pressure will be at the required level: ≈ 3 · 10–7 Pa.
It is known that the initial values of photo-desorption can be at the level of 0.005 (for CO), i.e. much higher than the required value of 5 · 10-7. However, the intensity of stimulated desorption decreases as photon dose accumulates according to the law [9]:
, (5)
where – is the initial value of the desorption coefficient, Г is the photon dose [photon/m], Г1/2 is the photon dose [photon/m], and desorption coefficient decreases by a factor of two from the initial value. At photon doses Г > Г1/2 desorption coefficient decreases according to the step law with the exponent of degree ε. According to numerous experimental data [9–19] the exponent of degree ε varies within 2/3÷1. Moreover, large values of ε correspond to large values of . It is important to note that at Г > Г1/2 the flux of desorbed molecules is proportional to , i.e., the dependence on the photon flux intensity turns out to be rather weak even at ε = 2/3 (at different sites where photon fluxes differ by a factor of 8, the gas load will differ only by a factor of 2). In addition, in a smooth chamber, the photon flux averaging is greatly facilitated by high values of reflection coefficients of photons with energies < 1 KeV at small incident angles. These two circumstances fully justify the averaging of flux of desorbing molecules from the walls of the vacuum chamber irradiated by direct, over-reflected, diffuse-scattered SR photons, and photoelectrons adopted in this paper.
Analysis of numerous experimental data [9–19], as well as the pressure dynamics in VEPP3 (BINP SB RAS) at accumulation of photon dose up to 3 · 1025 [photon/m] [20], allows us to conclude that at practically the worst value of ε = 2/3, the required degree of training (i.e., = 5 · 10–7) will be achieved at accumulation of photon dose 1 · 1025 [photon/m]. To accumulate such a dose in SKIF requires not much less than two months of continuous operation at a nominal current of 0.4 A. Obviously, it is impossible to operate with nominal current at the beginning of training due to intensive gas load (short lifetime). To calculate the real time, let us assume that during the training the electron beam current should be such that the radiation background, proportional to the losses of relativistic particles per unit time, does not exceed the nominal value. Losses of relativistic particles due to scattering on the residual gas are proportional to their current and pressure of the residual gas. Accordingly, the product I · P should not exceed In∙ Iн∙ Pн = 1.3 · 10–7 [A∙Pa]. Then the permissible current during training is determined by the ratio: I ∙ P = const = 1.3 · 10–7 [A∙Pa]. This ratio, taking into account: = 5 · 1018 · I, and the assumption Г >> Г1/2, allows us to integrate the system of equations (1–5) and calculate the training time of the SKIF drive required to reach the nominal current of 0.4 A:
(6)
Table 1 summarizes the input data for calculating the training time.
The calculation shows that the training time of the SKIF vacuum chamber is about 2.5 months, which is quite acceptable. If we assume a more severe training mode, namely, a threefold excess of radiation background due to particle losses on residual gas, the training time can be reduced by about 20 %. In this case, the training is actually divided into two stages: reaching the rated current of 0.4 A and CO pressure of 1 · 10–6 Pa (about half a month) and reaching the rated pressure of 3.3 · 10–7 Pa at constant rated current (about a month and a half).
It should be taken into account that these estimates are obtained at the most pessimistic value of the degree exponent in expression (4). Note the strong dependence of the training time on the diameter of the vacuum chamber and the average distance between the pumps. Indeed, at ε = 2/3, and taking into account (3), we obtain:
. (7)
Requirement for effective pumping speed of pumped ports:
(8)
MAIN ELEMENTS OF THE ACCUMULATOR VACUUM SYSTEM
The SKIF storage ring is divided into 16 super-periods, each of which consists of 7 girders. The scheme of the super-period of the storage ring is shown in Fig.1. The average distance between vacuum ports with pumps is about 1 m, the total number of ports is 24 pcs.
Getter pumps in the form of cartridges with heaters and ion-getter pumps are installed in the pumping ports. In general, the pumping port forms a combined pump, which is a modification of the pumps proposed in [7]. The effective pumping rate of the CO port should be not less than 55 l/s at absorbing the amount of gas, which is up to 80 % of the sorption capacity of the getter pump. CO sorption capacity: not less than 0.1 Pa∙m3. The initial value of the pumping speed of the getter pump should be not less than 800 l/s CO. The pumping speed of the ion-getter pump should be not less than 10 l/s for argon.
As already mentioned, special attention is paid to selection of materials for manufacturing vacuum elements, designs of pumping ports and connecting elements to minimize resistive and geometric impedances. Steps and slots in flange connections and bellows assemblies of vacuum chambers, due to their large number, can be significant sources of impedances. For these reasons, flanges of the Matsumoto-Ohtsuka design (MO-type flanges) were chosen as a flange vacuum connection [21, 22]. This connection is successfully used at the SuperKEKB collider in Japan. The scheme of flange connections of SuperKEKB vacuum chambers is shown in Fig.2 [23]. Based on the scheme, we can see the main advantages of this type of connections in comparison with ConFlat type connections: absence of gaps and steps, high current conductivity between the flange and the seal on the inner side of the vacuum chamber.
The flange material can be aluminum alloy or stainless steel, the seal material can be copper or aluminum alloy. For beam position sensors it is possible to use titanium as flange material. If it is necessary to heat the flange connection, the material of fasteners should be matched with the material of flanges according to thermal expansion coefficient (TEС). Aluminum vacuum chambers are used in the SKIF storage ring. The chamber flanges are made of aluminum alloy. The use of stainless steel flanges for welding to the chamber would require the use of bimetal: stainless steel – aluminum, which is not economical. Aluminum alloy fasteners require more delicate handling (mandatory use of torque wrenches when tightening) than standard stainless steel fasteners. Therefore, the standard fasteners were chosen for installation of vacuum chambers. However, to compensate for the difference in TEC of the fasteners and flanges, titanium rings were introduced on the aluminum alloy flange side (see Fig.3).
In this case, flange thickness has been reduced so that the total thickness does not increase significantly when the ring is taken into account. The use of titanium rings allows to use the such flange material combinations as: aluminum-aluminum, stainless steel-aluminum, and titanium-aluminum. The optimum thickness of the titanium ring can be calculated using the formula:
, (9)
where H is thickness and α is the KTR of the corresponding material, S.S. means stainless steel.
Fig.4 shows a vacuum chamber with a standard pumping port. The DN63 flanges of the evacuation port are ConFlat standard, but are made of aluminum alloy. The flanges have standard stainless steel fasteners and a TEC compensating titanium ring. This vacuum chamber passes through quadrupole and sextupole magnets. Vacuum equipment such as pumps, vacuum gauges or residual gas analyzers are connected to the evacuation ports.
The basic profile of the vacuum chamber is shown in Fig.5. The chamber is made of extruded aluminum profile. The chamber has two channels in the horizontal plane. One channel is used for water cooling, the second channel can be used for heating cable. In magnetic elements the main profile of the vacuum chamber is mechanically finalized according to the geometry of magnetic lens poles (see Fig.6, 7).
There is insufficient space between some magnetic elements of the drive to accommodate standard pump-out ports. For these places vacuum chambers with small-sized pumping ports have been developed. An example of such a chamber is shown in Fig.8.
Particular attention is paid to junction between the vacuum chamber and the pumping port connection. The vacuum chamber together with the cooling channels pass continuously through the nozzle while maintaining its profile. Inside the spigot, oval-shaped holes for pumping are made in the walls of the bundle chamber (see Fig.9, 10). The geometry of the holes should minimize the inserted impedance.
To compensate for temperature expansions, manufacturing inaccuracies, installation and exhibition of vacuum chambers, bellows mechanical compensators with reduced wave resistance are used. Appearance and design of the compensator assembly are shown in Fig.11. The design utilizes a strip spring loaded sliding contact. To prevent the strips from falling out of the working places, longitudinal and transverse bellows stroke limiters are used. The longitudinal stroke is ±5 mm.
The number of mechanical compensators in a super-period is strictly limited to only 2 pieces due to impedance minimization. Mechanical compensators are also installed between super-periods. The average distance between the nodes is 8 meters. Within this distance, several chambers with pumping ports and several beam position sensors (BPM) are installed between the chambers. The pumping ports and BPMs are mounted on exhibition supports that allow precise adjustment of the transverse position of the elements mounted on them. After adjustment, the supports are rigidly fixed in the transverse direction and have a free movement of ±5 mm in the longitudinal direction. This design ensures longitudinal mobility of the entire assembled vacuum system (with pumping ports and BPM) as a single integral chamber within the limits between the two mechanical compensators. The beam position sensors (see Fig.12) are made of titanium. The electrodes are soldered to each other with ceramic insulators directly into the housing by active soldering with hard solder.
CONCLUSIONS
The concept of the vacuum system of the SKIF storage synchrotron has been developed, which makes it possible to draw up a technical specification for design. It is proved that at an average distance between the concentrated pumping ports of 1 m in the storage ring arches, vacuum chambers required training time of the beam will be no more than 2.5 months. Models of the main elements of the vacuum system including specialized flange connections and diagnostic devices have been created.
PEER REVIEW INFO
Editorial board thanks the anonymous reviewer(s) for their contribution to the peer review of this work. It is also grateful for their consent to publish papers on the journal’s website and SEL eLibrary eLIBRARY.RU.
Declaration of Competing Interest. The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Nowadays, a synchrotron radiation source (SRS) of generation 4 + of the SRF SKIF is being developed [1]. It is based on a cyclic electron storage ring with an energy of 3 GeV with a perimeter of 476 m and an ultrasmall beam emittance of 75 pm-rad. The storage ring consists of 16 superperiods with long dispersion-free gaps (~6 m). Three of the gaps are supposed to be used for injection and RF systems, and the remaining gaps for staging specialized synchrotron radiation generators (wigglers or ondulators). In addition, 2 SR output channels from the bending magnets will be organized in each superperiod: one from a bending magnet with a small field (0.5 Tesla) for generation of UV and soft X-rays, one from a bending magnet with a large field (2 Tesla) for hard X-rays. There will be 2 RF systems in the drive: main (357 MHz) for energy loss compensation and harmonic (1071 MHz) for bunch elongation. The total current is 400 mA at 510 of the bunch (at full filling of the separatrices with a 10% gap). Injection at the experiment energy is assumed, so that a constant circulating beam current will be maintained. The injector will be a 200 MeV linear accelerator and a booster synchrotron with a perimeter of 150 m, which will perform beam acceleration from 200 MeV to 3 GeV. The injection frequency is 1 Hz. The injector is located in a separate building and is connected to an storage ring by a long transport channel. The storage ring and the large experimental hall are located in one building on a single foundation. Several specialized SR stations are moved to separate buildings.
The magneto-optical system designs of modern synchrotron radiation (SR) sources with small emittance [2] significantly limit the available aperture for vacuum beam chambers. Therefore, distributed vacuum pumps based on non-evaporable getters (NEGs), first applied in the insertion devices of the ESRF SR source [3], are gaining popularity. For example, almost all vacuum beam chambers of SR sources MAX-IV [4] and SIRIUS [5] have a thin-film NEG coating. In practice, such systems do not need any training under radiation. This means that the required dynamic (in the presence of the beam) ultrahigh vacuum can be achieved immediately after NEG activation [3, 6]. Activation itself is a heating procedure at 180÷200 °C, which requires dismantling of precision magnets (MAX IV solution) or space for installation of heaters and thermal insulation (SIRIUS solution) and a large number of mechanical compensators with RF contacts. On the other hand, compact combined vacuum pumps developed in recent years based on NEG cartridges and magnetically discharged ion-getter pumps [7], provide an opportunity to return to the classical system with concentrated pumps, but placed at a small distance; about one meter or even less.
The SKIF storage ring vacuum system will contain both distributed pumps, based on NEG coatings, and concentrated combination pumps. NEG coatings will be used in rectilinear gaps and insertion devices with low aperture chambers operating at room temperature. Concentrated pumps will be used in arches that make up 80 % of the storage perimeter and contain many precision magnetic elements. Such a solution allows to exclude the heating procedure in the arches and, consequently, to minimize the number of mechanical compensators – sources of high-frequency geometrical impedances. In addition, the use of a large number of ion-getter pumps significantly reduces dependence of the ultimate vacuum (compared to NEG) on the total level of micro-leaks, which significantly increases the reliability of the vacuum system as a whole.
The ultimate parameters of the SKIF storage ring will largely depend on the vacuum beam chambers design. Therefore, special attention is paid to selection of materials for fabrication of vacuum elements, designs of pumping ports and connecting elements to minimize resistive and geometrical impedances. The pumping system should provide a vacuum at the level of Pн = 3.3 · 10-7 Pa (in nitrogen equivalent) to achieve a "vacuum" beam lifetime of at least 10 hours at a nominal current of Iн = 0.4 А.
ESTIMATION OF TRAINING TIME UNDER RADIATION
The getter film transforms the vacuum chambers surface from a source of desorbing molecular fluxes into an absorber (sorption pump) of molecules. Therefore, as it was already mentioned, dynamic ultrahigh vacuum is achieved immediately after heating (NEG activation) on all areas with getter coating (approximately 20 % of the perimeter in the case of SKIF storage ring). Thus, the question of estimating the time of training of the inner surface under the action of SR necessary to achieve the required dynamic rarefaction level is relevant only for the SKIF arch sections where concentrated combined pumps will be applied.
At the initial design stage it is not possible to anticipate the coordinates of the available areas for concentrated pumps connection. Nevertheless, it is at the initial stage of the project that the number and parameters of the pumps must be determined with sufficient accuracy to formulate the design task. Obviously, an equidistant arrangement of identical concentrated pumps is the best first approximation. Besides, applying the Knudsen diffusion model and assuming averaging of the distributed flow of molecules from the walls of the vacuum chamber, it is easy to obtain an expression for the average quasi-equilibrium pressure:
, (1)
where q [Pa·l/m/s] – uniformly distributed gas load (desorption of molecules from the chamber walls), L [m] – distance between pumps, S [m3/s] – effective pumping speed of the port with pump, u [m4/s] – molecular conductivity of the chamber of unit length, – pump utilization factor (usually 0.2...0.5), Ps – pump inlet pressure.
It is interesting to note that varying distance between pumps from 0.5 L to 1.5 L, while maintaining the total number of pumps, leads to an increase in the average pressure by a factor of 1÷1.6 compared to the case of equidistant arrangement. The efficiency of the pumps arrangement can be taken into account by introducing an appropriate coefficient g:
. (2)
In SR sources, practically the entire surface of the vacuum beam chamber is exposed to an intense flux of photons with energies much higher than the energy of chemical bonds. Therefore, desorption under the action of SR is much higher than thermal desorption. The cross section of interaction of SR photons with molecules is 2÷3 orders of magnitude smaller than the cross section of inelastic electron-molecular interactions at electron energies of 20÷1000 eV. Therefore, gas desorption under effect of synchrotron radiation occurs in two stages: photons knock out photoelectrons from the irradiated surface (at some depth from the surface), which, in their turn, can lead to desorption of gas molecules from the surface of the vacuum chamber, both when flying from the surface and when hitting it. Besides, a part of photons is reflected and scattered over the total vacuum chamber surface. As a result, even in laboratory experiments to study desorption under the action of SR, one way or another, the entire surface of the vacuum chamber is involved [8]. This significantly complicates the interpretation and comparison of the obtained data. However, the accumulated experience makes it possible to identify the integral parameters of this complex phenomenon necessary for the calculation and design of acceletators.
The main characteristic of desorption under action of photons is the average number of molecules desorbed by one photon η [molecules/photon] – the photo-stimulated desorption coefficient. Then the flux of molecules from the walls can be written as:
, (3)
where is the photon flux [photon/(m∙s)], K ≈ 2.4 ∙ 1020 [1/(Pa ∙ m3)] is the number of molecules in a cubic meter at room temperature and 1 Pa pressure.
As a rule, the dominating gas desorbed under the SR action is hydrogen. The second in intensity is the flux of carbon monoxide (CO), which is 20–30 % of the hydrogen flux. However, since the scattering cross section of relativistic electrons is proportional to the square of the atomic number, scattering on CO molecules is dominant. Further, to simplify the analysis, only the partial pressure of CO is taken into account.
The average SR flux irradiating the walls of the vacuum chamber of the SKIF storage ring is 2 ∙ 1018 [photon/(m ∙ s)] at a nominal electron beam current of 0.4 A and energy of 3 GeV. The interpole distance in the magnetic elements is 30 mm. With a technological gap of 0.5 mm and a vacuum chamber wall thickness of 1 mm, the inner diameter will be 27 mm. Thus, the molecular conductivity on CO of the chamber of unit length will be:
. (4)
Direct calculation shows that at an average pump spacing of 1 m, factor g = 1.6, factor k = 0.25 and desorption factor 5 · 10-7, the average CO pressure will be at the required level: ≈ 3 · 10–7 Pa.
It is known that the initial values of photo-desorption can be at the level of 0.005 (for CO), i.e. much higher than the required value of 5 · 10-7. However, the intensity of stimulated desorption decreases as photon dose accumulates according to the law [9]:
, (5)
where – is the initial value of the desorption coefficient, Г is the photon dose [photon/m], Г1/2 is the photon dose [photon/m], and desorption coefficient decreases by a factor of two from the initial value. At photon doses Г > Г1/2 desorption coefficient decreases according to the step law with the exponent of degree ε. According to numerous experimental data [9–19] the exponent of degree ε varies within 2/3÷1. Moreover, large values of ε correspond to large values of . It is important to note that at Г > Г1/2 the flux of desorbed molecules is proportional to , i.e., the dependence on the photon flux intensity turns out to be rather weak even at ε = 2/3 (at different sites where photon fluxes differ by a factor of 8, the gas load will differ only by a factor of 2). In addition, in a smooth chamber, the photon flux averaging is greatly facilitated by high values of reflection coefficients of photons with energies < 1 KeV at small incident angles. These two circumstances fully justify the averaging of flux of desorbing molecules from the walls of the vacuum chamber irradiated by direct, over-reflected, diffuse-scattered SR photons, and photoelectrons adopted in this paper.
Analysis of numerous experimental data [9–19], as well as the pressure dynamics in VEPP3 (BINP SB RAS) at accumulation of photon dose up to 3 · 1025 [photon/m] [20], allows us to conclude that at practically the worst value of ε = 2/3, the required degree of training (i.e., = 5 · 10–7) will be achieved at accumulation of photon dose 1 · 1025 [photon/m]. To accumulate such a dose in SKIF requires not much less than two months of continuous operation at a nominal current of 0.4 A. Obviously, it is impossible to operate with nominal current at the beginning of training due to intensive gas load (short lifetime). To calculate the real time, let us assume that during the training the electron beam current should be such that the radiation background, proportional to the losses of relativistic particles per unit time, does not exceed the nominal value. Losses of relativistic particles due to scattering on the residual gas are proportional to their current and pressure of the residual gas. Accordingly, the product I · P should not exceed In∙ Iн∙ Pн = 1.3 · 10–7 [A∙Pa]. Then the permissible current during training is determined by the ratio: I ∙ P = const = 1.3 · 10–7 [A∙Pa]. This ratio, taking into account: = 5 · 1018 · I, and the assumption Г >> Г1/2, allows us to integrate the system of equations (1–5) and calculate the training time of the SKIF drive required to reach the nominal current of 0.4 A:
(6)
Table 1 summarizes the input data for calculating the training time.
The calculation shows that the training time of the SKIF vacuum chamber is about 2.5 months, which is quite acceptable. If we assume a more severe training mode, namely, a threefold excess of radiation background due to particle losses on residual gas, the training time can be reduced by about 20 %. In this case, the training is actually divided into two stages: reaching the rated current of 0.4 A and CO pressure of 1 · 10–6 Pa (about half a month) and reaching the rated pressure of 3.3 · 10–7 Pa at constant rated current (about a month and a half).
It should be taken into account that these estimates are obtained at the most pessimistic value of the degree exponent in expression (4). Note the strong dependence of the training time on the diameter of the vacuum chamber and the average distance between the pumps. Indeed, at ε = 2/3, and taking into account (3), we obtain:
. (7)
Requirement for effective pumping speed of pumped ports:
(8)
MAIN ELEMENTS OF THE ACCUMULATOR VACUUM SYSTEM
The SKIF storage ring is divided into 16 super-periods, each of which consists of 7 girders. The scheme of the super-period of the storage ring is shown in Fig.1. The average distance between vacuum ports with pumps is about 1 m, the total number of ports is 24 pcs.
Getter pumps in the form of cartridges with heaters and ion-getter pumps are installed in the pumping ports. In general, the pumping port forms a combined pump, which is a modification of the pumps proposed in [7]. The effective pumping rate of the CO port should be not less than 55 l/s at absorbing the amount of gas, which is up to 80 % of the sorption capacity of the getter pump. CO sorption capacity: not less than 0.1 Pa∙m3. The initial value of the pumping speed of the getter pump should be not less than 800 l/s CO. The pumping speed of the ion-getter pump should be not less than 10 l/s for argon.
As already mentioned, special attention is paid to selection of materials for manufacturing vacuum elements, designs of pumping ports and connecting elements to minimize resistive and geometric impedances. Steps and slots in flange connections and bellows assemblies of vacuum chambers, due to their large number, can be significant sources of impedances. For these reasons, flanges of the Matsumoto-Ohtsuka design (MO-type flanges) were chosen as a flange vacuum connection [21, 22]. This connection is successfully used at the SuperKEKB collider in Japan. The scheme of flange connections of SuperKEKB vacuum chambers is shown in Fig.2 [23]. Based on the scheme, we can see the main advantages of this type of connections in comparison with ConFlat type connections: absence of gaps and steps, high current conductivity between the flange and the seal on the inner side of the vacuum chamber.
The flange material can be aluminum alloy or stainless steel, the seal material can be copper or aluminum alloy. For beam position sensors it is possible to use titanium as flange material. If it is necessary to heat the flange connection, the material of fasteners should be matched with the material of flanges according to thermal expansion coefficient (TEС). Aluminum vacuum chambers are used in the SKIF storage ring. The chamber flanges are made of aluminum alloy. The use of stainless steel flanges for welding to the chamber would require the use of bimetal: stainless steel – aluminum, which is not economical. Aluminum alloy fasteners require more delicate handling (mandatory use of torque wrenches when tightening) than standard stainless steel fasteners. Therefore, the standard fasteners were chosen for installation of vacuum chambers. However, to compensate for the difference in TEC of the fasteners and flanges, titanium rings were introduced on the aluminum alloy flange side (see Fig.3).
In this case, flange thickness has been reduced so that the total thickness does not increase significantly when the ring is taken into account. The use of titanium rings allows to use the such flange material combinations as: aluminum-aluminum, stainless steel-aluminum, and titanium-aluminum. The optimum thickness of the titanium ring can be calculated using the formula:
, (9)
where H is thickness and α is the KTR of the corresponding material, S.S. means stainless steel.
Fig.4 shows a vacuum chamber with a standard pumping port. The DN63 flanges of the evacuation port are ConFlat standard, but are made of aluminum alloy. The flanges have standard stainless steel fasteners and a TEC compensating titanium ring. This vacuum chamber passes through quadrupole and sextupole magnets. Vacuum equipment such as pumps, vacuum gauges or residual gas analyzers are connected to the evacuation ports.
The basic profile of the vacuum chamber is shown in Fig.5. The chamber is made of extruded aluminum profile. The chamber has two channels in the horizontal plane. One channel is used for water cooling, the second channel can be used for heating cable. In magnetic elements the main profile of the vacuum chamber is mechanically finalized according to the geometry of magnetic lens poles (see Fig.6, 7).
There is insufficient space between some magnetic elements of the drive to accommodate standard pump-out ports. For these places vacuum chambers with small-sized pumping ports have been developed. An example of such a chamber is shown in Fig.8.
Particular attention is paid to junction between the vacuum chamber and the pumping port connection. The vacuum chamber together with the cooling channels pass continuously through the nozzle while maintaining its profile. Inside the spigot, oval-shaped holes for pumping are made in the walls of the bundle chamber (see Fig.9, 10). The geometry of the holes should minimize the inserted impedance.
To compensate for temperature expansions, manufacturing inaccuracies, installation and exhibition of vacuum chambers, bellows mechanical compensators with reduced wave resistance are used. Appearance and design of the compensator assembly are shown in Fig.11. The design utilizes a strip spring loaded sliding contact. To prevent the strips from falling out of the working places, longitudinal and transverse bellows stroke limiters are used. The longitudinal stroke is ±5 mm.
The number of mechanical compensators in a super-period is strictly limited to only 2 pieces due to impedance minimization. Mechanical compensators are also installed between super-periods. The average distance between the nodes is 8 meters. Within this distance, several chambers with pumping ports and several beam position sensors (BPM) are installed between the chambers. The pumping ports and BPMs are mounted on exhibition supports that allow precise adjustment of the transverse position of the elements mounted on them. After adjustment, the supports are rigidly fixed in the transverse direction and have a free movement of ±5 mm in the longitudinal direction. This design ensures longitudinal mobility of the entire assembled vacuum system (with pumping ports and BPM) as a single integral chamber within the limits between the two mechanical compensators. The beam position sensors (see Fig.12) are made of titanium. The electrodes are soldered to each other with ceramic insulators directly into the housing by active soldering with hard solder.
CONCLUSIONS
The concept of the vacuum system of the SKIF storage synchrotron has been developed, which makes it possible to draw up a technical specification for design. It is proved that at an average distance between the concentrated pumping ports of 1 m in the storage ring arches, vacuum chambers required training time of the beam will be no more than 2.5 months. Models of the main elements of the vacuum system including specialized flange connections and diagnostic devices have been created.
PEER REVIEW INFO
Editorial board thanks the anonymous reviewer(s) for their contribution to the peer review of this work. It is also grateful for their consent to publish papers on the journal’s website and SEL eLibrary eLIBRARY.RU.
Declaration of Competing Interest. The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Readers feedback
