601
Sov. Phys. Usp. 25(8) , Aug. 1982

FIG. 7. Isotherms of electrical resistance for the Nig0Fe10 alloy in a hydrogen atmosphere.48 1) Points, measured with an increase in pressure; 2) with a decrease in pressure. R0 is the resistance of a specimen at 25� C and P =1 bar.

also have the same metallic structure. The position of the critical points was studied for hydroge n solutions in alloys wit h 5, 10 and 15 at.% Fe. Figure 7 shows the isotherms of equilibrium values of the electrical resistance of the Ni-F e alloy with 10 at.% Fe in a hydro ge n atmosphere. At T = 150�C, th e transformatio n y� -- y2 has appreciable hysteresis, i.e., it is well know n to be a first-orde r phase transitio n accompanied by an abrupt change in the volume and solubility of hydrogen. Beginning at ~200�C , the transformatio n becomes hys teresis-free . At 250�C , the transition becomes appreciably diffuse , whil e at 300 and 350�C , the abrupt change in the resistance disappears and the R(PHz) curves assume a for m that is typical for resistance isotherms in the supercritical region. Thus, the behavior of the electrica l resistance permit s one to assum e that the curve of the transformatio n y x �y 2 on the T-PH diagram of the Ni 90 Fe 10 - H system terminates at the critical point and, in addition, 250 / 2~>' i intersect because the Curi e temperatur e of the solution is a continuous functio n of its concentration , whil e the concentratio n at these transitions changes in a discontinuous manner. For example, the curve of the Curi e points for nickel terminate s on the line of the transition yt -- y2 at ~270�C
Ponyatovskii eta/.
601


(see Fig. 5); therefore, in this case, r cr >270�C . The curve TC(PH ) for the alloy wit h 10 at.% Fe goes over into the y2 region, smoothly intersecting the continuation of the curve of the transformation y t -- y2 in the supercritical region with T~360�C ; therefore, this curve lies significantly higher than the region of criti cal phenomena and Tcr� 360�C. The evolution of the T-c projection of the phase diagram of the Ni-Fe- H system wit h a transition fro m Ni to the alloy Ni90Fe10 is shown schematically in Fig. 6. With an atomic fractio n of iron 0.15 <# Fe < 0.4 in the Ni-F e alloys, the critical temperature of stratification of the Ni-Fe- H solid solutions drops below 25�C. 46 - 4 8 In y alloys with a higher iron concentration, the hydrogen solubility for T>25� C must thus be a continuous function of pressure, whic h was confirme d experimen tally wit h alloys with 66.1 and 67.5 at.% Fe in Refs. 49 and 50, once again in the course of studying the dependences TC(PH2) (see subsection b, 1 in Sec. 4). The investigation of systems formin g wid e regions of continuous solid solutions is, naturally, most informative , since in this case it is possible to observed systematically the dependence of the properties of specimens on the hydrogen content. From this point of view , together with the Ni-F e alloys, the nickel alloys wit h Cu, Mn, Cr, and alloys of the pseudobinary Invar system Fe65(Ni1-xMnI)35 (see Ref . 51 for a discussion and refer ences) and a number of alloys based on palladium (wit h Cu, Ag, Au, Pt, Ir, and others; see reference s in Refs. 30 and 52) , for whic h both the concentration intervals of existence of continuous solid solutions of hydrogen and the mutual solubility of metallic components are high, are also convenient objects for studying the y solutions of hydrogen. The capability to for m wid e regions of continuous solid solutions wit h hydrogen is not, however, the exclusive privilege of alloys. Thus, at T=225�C , the concentration of e solutions Co- H increases monotonically wit h hydrogen pressure , reaching n = 0.51 at PH2 = 65 kbar.53 The isotherm T=300� C of the solubility of hydrogen in technetium (Fig. 8) is interesting. Two regions of supercritical anomalies (near PHz= 3 and 13 kbar) of two isomorphic phase transformations termi nating at critical points for T<300� C can be clearly seen on it. Since the investigation of precisely such nonstoichi-

ometric phases of Me-H at high pressur e is, as already noted, of considerable interest, whil e many of their properties can, for the tim e being, be studied only at atmospheric pressure and, as a rule, at low temperatures (in order to avoid losses of hydrogen fro m the specimens) , it is necessary to make some remarks concerning the interpretation of data obtained at atmospheric pressure and their relation to the measurements at hig h hydrogen pressure and high tempera tures. The first thing that comes to mind, considering that the configurational entropy should be a minimum , is that the nonstoichiometric Me- H solutions cannot be thermodynamically stable down to T = 0 K and as the temperature decreases, they must undergo either a decomposition into a phase with stoichiometric composition or they must become ordered. This situation is realized, for example, for hydrogen solutions in group V transitio n metals (see Refs. 5 and 54). In the case of hydrogen solutions based on transition metals in groups VI-VIII, an anomalous behavior of the physical properties with decreasing temperatur e is observed only for palladium hydrid e (n~ 0.6-0.8, TS 55K; see Ref . 30) and is probably related to its atomic ordering (see, for example, Refs. 55-57) , although, we must admi t that the data obtained by differen t researchers concerning the nature of this orderin g are poorly correlated. For other solutions, in the absence of a trivial stratification into isomorphic (and, as a rule , also nonstoichiometric ) phases, such phenomena have not yet been observed. It is possible that this is related to the fac t that the cor responding critical temperatures are ver y low and thermodynamically equilibrium states are not attained for kinetic reasons. Further, whe n studying high pressur e phases in Me-H systems, depending on the experimental conditions, it is necessary to distinguish clearly between two types of thermodynamic equilibrium: 1) stable equilibrium between a solid solutions and excess hydrogen, which is established wit h measurements in a hydrogen atmosphere, whe n the temperature is sufficiently high for exchange of hydrogen between the solid solution and the molecular phase; 2) metastable equilibriu m with fixed total hydrogen content in the specimen, which is established if the temperature is sufficiently low to retard the liberation of hydrogen fro m the specimen, but suffi ciently high to ensure diffusio n redistribution of hydro gen within the specime n (fo r example, wit h stratification of the solid solution into isomorphic phases). It turns out that the T-c section of the phase diagram of metastable equilibria can diffe r greatly fro m the T-c projectio n of the T-PH -c diagram of stable equilibria. For example, in the Ni- H system at room temperature and high hydrogen pressure (when conditions for stable thermodynamic equilibrium are realized), the boundary of the hump corresponding to stratification into y^ and y2 phases nj^s I, 26 - 3 2 whil e fo r T<-20� C andP= l bar44 (whe n liberation of hydrogen fro m specimens is kinetically prevented and metastable equilibrium conditions are realized), the minimu m hydrogen content in the y2 phase already constitutes �TM2in = 0.7 � 0.05. Another example is the lowering of the critical temperature fo r th e stratification y-y x + y 2 wit h a transition fro m
Ponyatovskii et al.

5

W

IS

/J 2 ,kbe r

FIG. 8. Solubility of hydrogen in technetium at T = 300" C. 1) Data In Ref. 36; 2) data in Ref. 33.
602 Sov. Phys. Usp. 25(8), Aug. 1982

602


stable to metastable equilibria in the Ni 60 Cu 40 - H system fro m rcr>100�C to r cr <-50� C (fo r more detai l see Ref . 51). Thus , establishing a relatio n betwee n the two sets of data usually obtained for high-pressur e phases (at high hydroge n pressur e and at atmospheri c pres sure) requires special analysis in each specific case. The last problem that should be examined in this section concerns the volume effect s accompanying dissolutio n of hydrogen in transitio n metals. The formatio n of hydride s at hig h hydroge n pressur e can be accompanied by an appreciable increase in the volume of the specimens V. For example, wit h the formatio n of � hydride s CrH, 3 9 MoH40 and FeH0-837 the volumes increase by =19, 20 and 16%, respectively. In the case of manganese58 and cobalt" whic h for m wid e regions of continuous e solutions wit h hydrogen , the volumes of these solutions likewise increase in proportio n to the increas e in the hydroge n concentratio n and , in addition, approximately linearly. An interestin g dependenc e V(n) at atmospher ic pressur e and room temperatur e is observed for e solutions Tc-H. According to x-ray analysis data,33 the hydrid e volume exceeds the volume of the startin g tech netium by =6.4 % and does not change over a wid e range of hydrogen concentration s 0.385 �� � 0.78. The dependenc e V(n) for y hydroge n solutions in palladium , nickel , and a large numbe r of alloys based on the m are very similar ; see Fig. 9, wherei n the depen dence s &.V0(n)= V(n) -V(0), wher e V(n) and 7(0), name ly, the volumes of the uni t cells of the metal wit h a hydroge n content of n and withou t hydrogen , are pre sented. It can be seen that the volumes of the solutions increase wit h increasin g n and, in addition , all depen dence s (where , of course, the effec t exceeds the limits of erro r in the measurements ) have a general property: for n^ 0.7-0.8, thei r slope j3 = (d/dn)&V0(n) is observed to decrease. For example, for the alloys Ni80Fe20 and Nij^Fe^g , th e slope changes fro m )3=9. 5 A3 fo r w<0. 8 t o /3~ 2.5 A3 fo r M >0. 8 (followin g Ref . 52 , wher e th e de pendence A7 0 (w ) was constructe d for palladium alloys, w e approximated th e analogous dependenc e fo r nickel iron alloys fo r w<0. 8 an d n>0. 8 b y straight line segments). The natur e of this effec t is not yet understood. Some possibilities for explaining it are opened up by the study of hydroge n solutions in alloys based on nickel, for whic h th e composition s n>l ar e obtained. Indeed,

wit h w~0. 6 and at room temperature , hydrogen in nick-el occupies octahedral interstices, 5 9 whose numbe r in a fe e lattice equals the numbe r of sites. When all these interstices are filled , the y hydrid e will have composition w = l. Therefore , y solutions wit h w> l must be structurally different fro m solutions with n~ 0.6. The only anomaly in the dependences &V0(n) for y solutions Me-H at hydrogen concentrations up to n= 1.23 (for the solution Ni80Fe20-H) occurs at w~0. 8 and it is reasonable to assume that it is related to the beginning of a structural rearrangement . As far as the rearrange ment is concerned , it may involve, for example, fillin g for w>0. 8 part of the tetrahedral interstices, whic h in a fe e lattice numbe r two per site (although in this case, due to the smaller sizes of the tetrapores, the partial volume of hydrogen, characterized by the value of /3, is mor e likely to increase rathe r than to decrease). It also cannot be excluded that for n > 1 hydrogen continues to occupy only octahedral interstices , whose numbe r relative to the numbe r of sites increases due to an increase in the numbe r of vacancies in a metal lattice for w>0.8. 2 )
4. MAGNETIC PROPERTIE S OF METAL-HYDROGEN SOLUTIONS

A characteristi c of group VI-VIII 3d metals and alloys based on the m is the existence of magnetic order and the behavior of the magnetic properties wit h hydro genation can serve as a convenient indicator of changes occuring in the electron subsystem. Experiments have shown that for transitio n metals the introduction of hydrogen does not lead to a unique change in their magnetic properties , for example, to a decrease in the temperatur e of transitio n into the magnetically ordered state, whic h is a dominan t tendency in the case of hydrogenation of rare earth metals.62 On the contrary, the effect s have turned out to be ver y diverse. In par ticular , the following are observed on hydrogenation of pur e metals. Chromium , whic h has a bcc crysta l structure , is an antiferromagne t wit h the Nee l point at TN = 312 K; the e hydrid e of chromiu m CrH0-97 is paramagnetic down to helium temperatures.63'64 The low-temperatur e modi ficatio n of manganes e (a-Mn ) is an antiferromagnet wit h TN = 100 K; the e solutions Mn-H in the range of compositions 0.65s n~s > 0.94 have a spontaneous magnetization, whic h increases monotonically as the hydrogen content in the solution increases, but even at w = 0.94 it remain s very small (sO.02 -0.05 /i B /M n atom at T=8 2 K; MB is tne Bohr magneton) , whil e the Curi e point reaches quite high values Tc> 280 K.58 Iron wit h a bcc
2

/

/-"&&*
,, , 0

/�

+ -'

x -2 �- J 0-4 v-5
>.0 1.2 n

0.2

FIG. 9. The dependences AV 0 (n)=V(n)-V(0 ) (see Sec. 3). 1) Ni, r=29 3 K,32 2) Ni , T = 293 K,114 3, 4) Ni-Fe alloys with 20 and 67.5 at.% Fe, respectively, T = 83 K,50 5) Fe 65 Ni 6 Mn 29 , T =83 K84; the dot-dash line indicates the analogous dependence for alloys based on palladium at room temperature.52
603 Sov. Phys. Usp. 25(8), Aug. 1982

'We note that the presence of wide regions of concentrations where the volume of the solution depends weakly on the content of the injected component (as happens, for example, in Pd-H and Ni- H solutions wit h n S 0.7 or in Tc-H solutions wit h 0.385S M S 0.78 ) is not a specific property of hydrogen solutions in particular. A small (for example, Zr-C ) and even negative (Zr-N ) change in volume with increasing concentration of the injected element were already observed previously in other solutions.61 The reasons for this behavior of the volume in these systems also has not yet been established.
Ponyatovskil et al.
603


structure is a ferromagnet with r c = 1043 K, whic h has at T= 0 K a spontaneous magnetization a 0 =2.2 2 JJ.B/ atom; the c hydride of iron FeH0<8 is also a ferromagnet and, in addition, wit h nearly the same value of the magnetization a) Magnetic properties of Ni-Me alloys

3.0

9,5 fff.ff Ne, electrons/atom

10.5

FIG. 10. Spontaneous magnetization cr0 at T=0 K as a function of the average number of electrons (3d+4s ) per atomJV e for binary fee alloys.34 1) Ni-Cu; 2) Ni-Co; 3) Ni-Fe; 4) Ni-Mn; 5) Ni-Cr; 6) Ni-V; 7) Co-Cr; 8) Co-Mn.

is customarily represented in the for m of the so-called Pauling-Slater curve (Fig. 10) as a function of the average number of 3d + 4s electrons Ne (present in the isolated atoms) per atom in the alloy. Figure 10 pre sents the experimental values of a0 (agreeing well wit h data in the literature67'68) for Ni-M e alloys, serving as a basis for obtaining the solid hydrogen solutions examined in this review. The following characteristic property of the Pauling Slater curve is interesting: the values of a0 for nickel alloys wit h copper, cobalt and iron (wit h *Fe� 0.6) fall on a single straight line wit h a slope du0/dNe~ -1.05 JJ.B/electron. Nicke l is well described by the band ferromagnetis m model and, in addition, near T~0 K, it can be viewed as a strong band ferromagnet , for whic h one half of the d band (d\ wit h spins pointin g up) is filled completely, whil e the second half (d\ wit h spins pointing down) is only partially filled.7 This is shown schematically in Fig. lla. The quantity cr0 is related to the number of electrons wit h uncompensated spin, i.e., in this case, simply to the numbe r of holes p\ in the d\ subband, by the relation a0 = (l/2)gjj.Bp^ ~ iJ-xpl, since for nickel (and alloys based on it) the spectroscopic facto r g-~ 2.2 ~ 2 (se e Ref . 68) (whic h indicates the smallness of the orbital contribution , i.e., the practically pure spin origin of magnetization in these materials). The linear dependence o0(N") with

In the band theory of magnetism, the electrons, whic h partially fill the conduction band, are the carrier s of the magnetic moment. The exchange interaction causes the energy of the electrons to depend on the orientation of their spins relative to the total magnetic momen t a, which , in the firs t approximation, can be represented as a relative shif t in the electron subbands for opposite directions of spin by the energy A = / � a, wher e I is the effective exchange interaction parameter. Usually, it is assumed that in transition metals, the Ferm i level intersects the overlapping narrow d band (wit h high density of electron states) and the wid e s band (wit h low density of states), and in this case, the magnetic properties of the material are determined primaril y by the structure of the d band and the degree to which it is filled, ft should be noted that in spite of the evident roughness of these approximations, in the case of fe e alloys of 3d metals, such an approach permits describ ing quantitatively the temperature , field , pressure and concentration dependences of their macroscopic magnetic characteristics (see, for example, Refs. 76-70). The concentratio n dependences of the spontaneous magnetization a0 at T = 0 K for fe e alloys of 3d metals
604

FIG. 11. Diagram of band structures of collectivized ferro magnets at T =0 K. a) strong ferromagnets; b) weak; c) diagram illustrating flow of electrons from virtual impurity bound state d5* into the dl subband of a strong ferromagnet. E is the energy of the states; JVt and N \ are the density of states with spin pointing up and down; �F is the Fermi energy;

Sov. Phys. Usp. 25(8), Aug. 1982

Ponyatovskii ef a/.

604


slope 9a0/dN*~ -1 p.B/electron for nickel alloys wit h Cu, Co, and Fe are well explained if it is assumed67 that alloying nickel wit h these elements 1) does not lead to deformatio n of the curves for the density of electron states jft(z) andy^(e) , only shifting them in energy , relative to one another as a single entity (this approximation is called the rigid band approximation for ferro magnets; the physica l significance of this approximation and the limits of its applicability are discussed in detail in Ref. 71); 2) lets the ferromagnets remai n strong (with completely filled d! subband). Indeed, in this case, the changes in a0 accompanying alloying wil l stem primaril y fro m the change in the number of holes in the d* subband and, therefore ,
a0(
lH

ferromagnet s (4.1). This is due to the fact that screen ing of the excess impurity charge must be responsible, primarily , for part of the d band with high density of states at the Ferm i level, i.e., the d) subband; in addition , the characteristic screening radius is less than or of the order of the interatomic distance.71 There fore , the magnetic moments of the nickel matrix and of the impurit y should not depend on the impurity concentration:
| (I matrix ^ COIlSt , If 1 im p � |l matri x - ^II B ,

whic h for the average magnetic moment of the alloy M = -- 10} (iBlMe,

,)*oN ' -A^tft-M, ,

(4.1)

wher e AZ = Z N i -Z m e is the differenc e between the nuclear charges of the nickel and impurit y atoms, whic h is observed experimentally. As can be seen fro m Fig. 10, the curve of cr 0 (JV e )fo r many alloys deviates fro m the curve predicted by the rigid band model for strong collectivized ferromagnets. For Ni-F e alloys wit h xTe^ 0.63, this deviation is apparently related to the appearance of a significant num ber of holes both in the d* and in the dt subbands.72 The correspondin g band scheme is shown in Fig. lib . The fact that for collectivized ferromagnet s wit h holes in both d subbands at T = 0 K (suc h ferromagnet s are called weak) , the spontaneous magnetization may not increase, but rathe r decrease wit h increasing numbe r of holes in the d band, can be illustrated as follows. Let, as in Fig. lib , |8^(E F )/8e| > |8j^(eF)/8e , wher e � F is the Ferm i energy. Then , whe n the degree of fillin g of the d band by electrons decreases, the numbe r of holes in the di subband wil l increase more rapidly than in the d\ subband and a 0 = (l/2)gp.B(pl -pi ) . decrease. The "anomalous" dependences a0(N") of the type observed fo r Ni-C r an d Ni- V alloys, when th e sharp de viation fro m a straight line wit h slope da0/WK -1 /IB/ electron appears already at low concentration s of the alloying element , wer e explained by Friedel71 using the concept of "virtual bound states." Accordin g to Friedel, whe n an atom of a transition impurity appears in a met al, the d level of this atom, whic h is higher than the energy of the bottom of the conduction band, becomes delocalized, but in the coordinate-energy space remains a "virtual d level," namely, the region wher e 1) the amplitude of the spherical component of the wave functio n wit h / = 2 is anomalously large and 2) this amplitude corresponds to the excess charge density , which , being summe d over the entir e region, approximately equals the charge of the starting bound d state (multiply degenerate, 2Z + 1 = 5). This virtual level can split into sublevels wit h oppositely oriented spin due to the intra-atom exchange interaction , if the metal-solvent has a sufficientl y narrow conduction band (and , especially, is a ferromagne t itself). In the case of nickel , as long as the virtual dH sublevel lies below the Ferm i energy , alloying wil l lead to changes in the rigid band model for strong collectivized
605
Sov. Phys. Usp. 25(8) , Aug. 1982

(4.2 )

whic h agrees well wit h the slopes of the dependences a0(JVe) for alloys of nickel wit h chromium , vanadium and titanium. Thus, if the impurities are listed in order of decreasing atomic numbers , the followin g pattern is observed whe n nickel is alloyed wit h other 3d metals. Copper , cobalt , iron and (partially ) manganese do not change the band structur e of nicke l ver y muc h and over a wid e range of compositions of the alloys, the dependence s a0(xMe) are determine d primaril y by changes in the electro n concentratio n JVe. Wit h alloyin g by elements wit h Z^ ZCr, both changes in N" and in the band structure play an important role.
b) Ni-Me-H solutions

As already noted, in order to interpre t the propertie s of transitio n metal hydrides , for many years , and wit h variable success, two alternative models, namely the anion and proto n models, wer e used. In the firs t model, the hydroge n in the metal was assumed to be a negativel y charged ion H", whil e in the second it was assumed to be a proto n H*. Fro m the point of vie w of the rigi d band model , this meant the following. In the ani on model , narro w bands , forme d by electron levels of hydrogen , lie below the Ferm i energy of the metal-sol Ponyatovskiie f a/.
605


vent. .Since the hydrogen atom has a single electron, whil e each energy state in the band can be occupied by two electrons wit h oppositely oriented spin, these bands (one band for each hydrogen atom in a unit cell) are half filled. The missing electrons are taken fro m the conduction band of the starting metal (one electron per H atom). In the proton model, the hydrogen bands lie above the Ferm i level of the metal-solvent and the electrons fro m these bands (one electron per H atom) flow into the conduction band of the metal. The fac t that in spite of intensive study of transition metal hydride s over many decades, it was not possible to make a fina l choice even between these two diametrically opposed models is due primarily to the hydro genated metals (Pd and elements in the subgroups Ti and B having no physical properties that are uniquely related to the degree to whic h their conduction bands .are filled (this problem is discussed in detail in Ref. 73). As can be seen fro m the preceding section, in the case of hydrogenation of fe e nickel alloys, magnetic properties, whic h permit studying experimentally the status of hydrogen in these objects, could serve as such an indicator. Indeed, for example, whe n strong collectivized ferromagnets (the alloys Ni-Cu , Ni-Co , Ni-F e wit h # Fe � 0.6) are saturated wit h hydrogen, in view of the high density of states of the d\ subband at the Ferm i level, it is its degree of fillin g that wil l primaril y change. If the anion model is valid, then the spontaneous magnetization must increase wit h slope da0/dn � 1/J.B/H atom, and if the proton model is valid, a0 wil l decrease wit h slope 8 606

wit h an increase in the numbe r of protons in its inter stices. Thus, part of the electrons (TJ ~ 0.4-0.1 electrons/proton for Pd and Ni10), contributed by hydrogen atoms to the conduction band, fills the space above the Ferm i level, whil e part goes into the additional levels below it. This effec t can be qualitatively illustrated wit h the help of the diagrams in Fig. 11, assuming that as the hydrogen concentration in the metal increases the density of s states is shifted down in energy relative to the densities of d* and d* states. Calculations of the band structures of hydride s of transition metals have explained, and in many cases quantitatively, the results of the experimental study of the electron heat capacity, magnetic permeability , photoemission spectra, and other physical properties , including superconductivity , of a numbe r of Me-H systems; see Refs. 10-12, 14, and 30. In the case of NiMe-H solutions, these calculations have served as a theoretical basis fo r justifyin g th e assumptions fo r de scribing thei r magnetic properties in the "rigid d band" model,75 assuming that the change in these properties accompanying hydrogenation stems primaril y fro m the increase in the degree of fillin g of the d band in the metal-solvent by electrons and, in addition, the hydrogen must be viewed as a donor of a fractiona l numbe r of electrons TJ S 1 electrons/atom to the d band. How ever, it should be noted that such a model is not a direct result of band calculations, since the magnetic properties are also affecte d by the change in the exchange interaction, whic h is not yet possible to include in the calculations; the large volume effect s accompanying dissolution of hydrogen in Ni-M e alloys should also be kept in mind. The diversit y of the magnetic properties of Ni-M e alloys make the m convenient model objects for clarifying differen t aspects of the influence of hydrogen. Since the properties of hydrogen solutions in alloys also diffe r considerably, it makes sense to examine several Ni-Me- H systems systematically. 1) Mi-Fe- H system. Figure 12 shows the dependence ff0(n) for y solutions of hydrogen and nickel44 and Ni-F e alloys.50 The y 2 phase of Ni- H solutions is paramagnetic for T^ 4.2 K and, in addition, in the temperatur e range T<25 0 K, the minimu m solubility of hydrogen in this phase is wj>2in = 0.7. Specimens wit h n<0.1 consist of a mixture of phases (�y1+y2), whil e thei r magnetization is proportional to the content of the yx phase wit h aoCwTM")" cr"1 and decreases linearly wit h hydrogen con centration (curve 1 in Fig. 12). Stratification into y t and y2 phases was not observed in solutions of hydrogen in Ni-F e alloys wit h x7^ 0.1. As can be seen fro m Fig. 12, for alloys wit h 10, 20, and 40 at.% Fe, the spontaneous magnetization decreases wit h increasing n and, in addition, for n^> 0.8, the dependences aa(n) are nearly linear, whil e for �>0.8 , a small but systematic deviation ofCTOtoward higher values begins. The depen dence Ponyatovskii etal.
606

Sov. Phys. Usp. 25(8), Aug. 1982


decrease in a0 wit h |9a 0 /9 w > |8a�/9w| , since the effec t of pumpin g electrons fro m the dl to the d\ subband will be added to the decrease inCTOdue to filling of the subband by hydroge n electrons. Thus , it is reasonable to ascribe the observed values of the slopes 9a 0 /9w ~ -0.6 -0. 4 /i B / H atom to the change in the band structur e of Ni-Fe- H solutions occurin g wit h the penetratio n of hydrogen and accompanied by an increase in the number of holes p in the d band. The experimenta l values of 9a0/9w can be explained withi n the scope of the rigid d band model by assuming that hydrogen is a donor of a fractiona l numbe r of electrons w~0. 5 electron/H atom in the d band of Ni-F e alloys. Let us go back to Fig. 12. It is evident fro m a com parison wit h Fig. 10 that for all Ni-F e alloys wit h xve � 0.67 5 the concentratio n dependences ofCTOwit h the introductio n of hydroge n into the alloys and wit h the substitution of iron by nickel are similar: for some concentration , cr0 reaches a maximum value o^^l.Se -1.9 /j.B/atom and the n begins to decrease. The simi larity of the dependence s a0(w) for alloys wit h 66.1 and 67.5 at.% Fe and the dependence s CTO(*NI) m the Ni-F e system is even more complete , whic h can be demon strated as follows. For alloys wit h x Fe � 0.6, the magnetization decreases linearly both wit h substitution of Fe by Ni and wit h an increase in the hydrogen concen tration . Let us introduce , following Ref. 50, for each specifi c Ni-F e alloy wit h XT,,** 0.6, the coefficien t � , relating the changes in the composition of Ni-F e binary alloys and changes in n, leading to the same changes in "JFC = ,,,,, -,-\-n = --In.
(()0 0 1 dn} (A

Of,

o.e

FIG. 12. The dependence of the spontaneous magnetization (TO at T = 0 K and atmospheric pressure on the hydrogen content n in alloys Ni--F e with different concentration of iron.50 At.%Fe : 1) 0; 2) 10; 3) 20; 4) 40; 5) 60; 6) 66.1; 7) 67.5. The dashed line shows the dependence a 0 (� ) for alloys with 66.1 and 67. 5 at.% Fe, calculated using relation (4.3). The dimensions of ffo are Me/ato TM �f tn e Ni-Fe alloy.

=1.85(Lt B /ato m at n~0.3, and then decrease s monotoni cally. The firs t thin g that draw s attentio n in looking at the dependenc eCTO(M)fo r alloys wit h * Fe �0. 6 i s th e pres ence of a linea r sectio n for w S 0.8 and deviation fro m linearit y for w>0.8 . We recall that a compositio n n ~0. 7 -0.8 is singular for all y hydroge n solutions in alloys based on nicke l (and palladium) studied up to the presen t time : for � 2 0, the nature of the dependence s &V0(n) changes (see Fig. 9). The reasons for the appearance of these anomalies are not clear , so that it makes sense for the presen t to limi t the discussio n to magneti c propertie s of the solutions Ni-Fe- H wit h n s 0.8. The values of the slopes 8cr0/9�, obtained wit h a linear approximatio n to the experimental dependence s ua(n) wit h n^> 0.8 for alloys containin g 60 at.% Fe, decrease monotonically and approximately linearly in absolute magnitude, as the iro n content in the Ni-F e alloys increases (and , correspondingly , th e electro n con centratio n Ne of these alloys decreases), fro m =0. 6 HB/ H atom at * Fe =0. 1 to = 0.4 /J.B/H atom at * Fe =0.6 . As we can see, the sign and order of magnitud e of the effect s observed agree wit h the value 9aJ/9� = -lfJ. B / H atom predicte d by the proton model (i.e. , essentially the rigid band model) , but , at the same time , the dif ferenc e betwee n the experimentally obtained values of 9ff 0 /8 � and da^/dn is still appreciable. It should be noted that as long as a ferromagne t remain s stron g (i.e. , at T = 0 K, holes occur only in its d* subband), CTO~p\=p, the total numbe r of holes in the d band, and is independen t of the change in the exchange interaction and of the deformatio n of the bands, if it does not lead to a change in p. If, on the other hand, holes appear as a result of hydrogenatio n both in rft and in rft subbands, the n in the rigi d band approximation , this wil l cause a
607

\*-3)

0^

Equation (4.3 ) reflect s the experimental fact that the dependence a0(w) for these alloys can be approximately obtained fro m o0(xfe) by an appropriate change in scale along the concentratio n axis. Extrapolation of an approximately linear dependence �U Fe ) to alloys wit h xTe = 0.661 and 0.67 5 gives 5=0.185 . The dependences a0(n) for these alloys constructe d wit h such a value of the rescaling coefficien t | are show n in Fig. 12 by the dashed lines. It is eviden t that they agree wit h the experimenta l data. A n analogous correspondenc e occurs betwee n th e Curi e points of Ni-F e alloys and hydrogen solutions based on them . In the Ni 90 Fe 1 0 alloy in an inert medium , the Curi e point increases wit h pressure 7 6 wit h slope (dTc/dP)ln ~0. 5 K-kbar" 1 . In a hydroge n atmosphere , the Curie point of this alloy decreases monotonically wit h pres sure (see Fig. 5). A completely differen t pattern is observed for Invar Ni-F e alloys.49'50 The results of the determinatio n of the dependenc e of the Curi e point of the alloy wit h 67.5 at.% Fe on pressure in an inert mediu m and in hydroge n are presente d in Fig. 13. It is evident that in an inert mediu m T0 of the alloy decreases linearly wit h slope (dT c /rfP) ln = -5.05 + 0.1 K-kbar" 1 wit h pressure s up to 20 kbar. In a numbe r of papers (fo r example , in Ref . 76) , it is shown that for Ni-F e Invars the dependenc e TC(P) in an inert mediu m
Ponyatovskii ef a/.
607

Sov. Phys. Usp. 25(8) , Aug. 1982


TC,H
600

zoa

20

P, kbar

FIG. 13. The dependences of the Curie point of the Fe67>5Ni32i5 alloy on pressure in an inert medium (1--data in
Ref . 49) and in hydrogen (2--data in Ref . 49, 3--Ref . 50,

As can be seen fro m Table II and Fig. 13, the com puted values of AT� + AT� agree well wit h AT*". We note that at high pressures the contributio n of AT* to the sum ATJ + A?t *s n�t small and , thus , the agreement betwee n the latter quantity and AT'" indicates the validity of (4.3) for describing the dependences of both Te and (dT c /dP) l n of the Ni-F e Invars on thei r hydroge n content . Therefore , the correspondence (4.3 ) permits estimating the trend in the change of (dT c (�)/dP) in , whic h would be very difficul t to do experimentally . The last line of Table II presents the results of an analagous calculation for a non-Invar alloy wit h 10 at.% Fe. It is evident that Eq. (4.3 ) satisfactorily describes the behavior of TC(PH2) for thi s alloy as well. Thus, for the Ni-F e alloys in the entir e range of com positions studies 0.1�* Fe � 0.675 both the dependences cr0(n) and a0(A*Fe) as wel l as the dependences Tc(n) and r c (A# Fe ) are similar and, in addition, they have the same similarit y coefficient s �. This can formall y be represented as a similarit y of the dependences a0 and jTc on the electron concentration wit h hydrogenation and substitution of iron by nickel , respectively , and then the coefficient s of such similarit y f = (ZKi -ZFe) � = 2| will give the effectiv e numbe r of electrons introduced by hydroge n into the conduction band of the metal. The presence of this similarit y for Ni-F e alloys wit h xfe >0.6 , whic h ar e weak collectivize d ferromagnets (fo r whic h at T= 0 K holes occur both in the d\ and rft sub band 72 ) , is strong evidence of the validit y of the rigid band model for describing the magnetic properties of all fe e Ni-F e alloys and the rigid d-band model for hydroge n solutions based on them . However , the numbe r of electrons f contribute d by hydrogen to the d band, determine d fro m the magnetic measurements , is only some effective phenomenological quantity . For this reason, the study of such properties for a number of differen t alloys is a necessary step in order to clarif y the tru e role of the increase in the degree of fillin g of the d band by electrons in the change of the magnetic propertie s of hydrogen-saturated alloys. 2) Ni-Co- H system. Hydroge n solutions in fe e alloys Ni70Co3-0 and Ni40Co60, whic h are strong collectivized ferromagnets , wer e studied in Ref . 60. Dissolution of hydrogen in these alloys, as in strong collectivized NiFe ferromagnet s wit h xfe~�- 0.6, decreases a0 and Tc. The linear approximation to the experimental dependences a0(n) fo r w<0. 8 fo r single phase solutions Ni70Co30-H and Ni40Co60-H gives the slopes 9a0/8� = -0.76 and -0.71 M B / H atom, whic h agrees wit h the prediction s of the rigid d-band model. In attempting to describe the dependences a0(n) and Tc(n] for the Ni-C o alloys wit h the same coefficient s of similarit y � , introduced by a relation similar to (4.3) , such good agreement is not obtained between the computed and experimental values of ATC as for the Ni-F e alloys, but, the sign and order of magnitude of the effec t turn out to be correct. 3) Ni-Mn- H system. The dependence a^x^ �f disordered fee Ni-M n alloys is nonmonotonic (see Fig. 10). For nickel (AT e =1 0 electrons/atom) , the spontaneous magnetization o^l~ 0.616 jj.B/atom; as nickel is rePonyatovskiiera/.

4--are the computed values; see subsection a in Sec. 2).

is nearly linear (dashed line in Fig. 13) at higher pres sures as well. An increase in the hydrogen concentration in the alloy wit h an increase in pressure leads to a deviation in the dependence TC(PH ) fro m that in an inert mediu m toward higher values of Tc. For PH >15 kbar, the Curi e points of Ni 32>5 Fe 67-5 - H solutions begin to increase, reaching ~700 K at P H2 = 50 kbar. Table II presents the values of the hydrogen concentrations n on the curve of Curi e points in the alloy wit h 67.5 at.% Fe at P^= 20.35 and 51 kbar. We note that the steepest dependence TC(PH2) of the Ni 32-5 Fe 67-5 - H solution is observed in the pressure range wher e the solubility of hydroge n in the alloy increases most rapidly (~25-3 5 kbar). Assumin g that the correspondence established by Eq. (4.3) between a0 of alloys wit h and without hydrogen is also valid for TK and (rfr o /rfP) ln , it is possible to describe the dependence AT�P, namely, the difference between the values of Tc in hydrogen and in an inert medium under the same pressure. With this approach, ATC(P) wil l consist of
AT? = Tc {z
fe

- In) - Tc (x ff ) ,

(4.4)

related to the change in Te without pressure due to the introduction of hydrogen into the alloy and
(4.5 )

arising due to the differen t pressure dependence of the Curie points of the alloys with and without hydrogen. Results of this calculation are presented in Table II and in Fig. 13. The T0 for the Ni-F e alloys are taken fro m Ref. 7 2 an d (dTc/dP)u fro m Ref . 77 .

TABLE II. Parameters for Ni-Fe-H solutions.
*Fe "�)

PHV fcbar 20 35 51 21.5

-i n

arj. K
58 285 365 --115

4lf , A- arj+ar*' . X <"" � * 60 375 525 --120

0.675 0.675 0.675 0.1

0.041 0.735 0.99 0.395

0.018 0.136 0.183 0.1

90 160 --5

71 365 520 --145

*) The values of n �ith />H2 tr om the thir 1 column an d T=TC(PH 2 ) ar e indicated .

608

Sov. Phys. Usp. 25(8), Aug. 1982

608


similar. Let us introduce the coefficient s � = relating the changes in composition and, correspond ingly, electron concentratio n N" in the starting alloys wit h the changes in n leading to identical changes in a0. By comparin g the values of a0(n) for the Ni 80 Mn 20 - H and Ni70Mn30-H solutions, indicated by the diamondshaped symbols in Fig. 14, wit h the values of a0UMn) presented in Ref . 68 for Ni-M n alloys, we obtain � = 0.81 and 0.48 electrons/ H atom, respectively. The curves of a0(n) calculated for these values of f for the Ni80Mn20 and NL^Mn.,,, alloys are shown in Fig. 14 by the continuous lines. It is evident that they satisfactorily agree wit h the experimenta l dependences aa(n). Thus , the effec t of dissolved hydrogen onCTOof Ni-M n alloys is completely analagous to the effec t of substituting manganese by nickel. There is no such analogy for the Curi e points for alloys wit h 10 and 20 at.% Mn: an increase in the nicke l content in Ni-M n alloys increases Tc, while injecting hydrogen decreases Tc. The dependences Tc(n) and TC(#N1) again become simila r for alloys wit h 30 at.% Mn: whe n this alloy is saturated wit h hydroge n up to n = 0.85 (the value of a0 for this composition of the solution is indicated by the diamond-shaped symbols in Fig. 13b), its Curi e point increases monotonically t o ~25 0 K , whil e fo r th e Ni 83 _ 5 Mn 16 _ 5 alloy (whic h has the same value of cr0) TC=4'00 K.80

FIG. 14. at T =0 K alloys.75 calculatin

The dependences of the spontaneous magnetization a on the hydrogen content n in disordered Ni--M n 1) 10; 2) 20; 3) 30 at.% Mn; 4) values ofCT0,used for g J.

placed by manganese , a0 increases and wit h Ar Mn = 0.1, reache s th e maximu m valu e cr^^O. S /j.B/atom an d the n begin s to decrease. 6 8 The Curi e points decrease mono tonically fro m 627 K for pur e nickel to heliu m tempera ture s fo r th e alloy wit h xm~ 0.26. Th e dependence s CT O (W) , obtaine d i n Ref . 7 5 fo r hydroge n solutions i n dis ordere d Ni-M n alloys wit h 10, 20, and 30 at.% Mn are 4) Ni-Cr- H system. Summarizin g the results pre presente d in Fig. 14. sented in Sees. 1-3, we can say that the dependences CTO(W) for y solutions based on nickel alloys wit h iron , Stratificatio n into yt and y 2 phases is observed in the cobalt, and manganese are analogous to the corre Ni90Mn10-H system jus t as in the Ni- H system at atmossponding dependence s a 0 U Ni ) fo r th e starting binary al pheri c pressur e and TS 150 K. The yz phase is para loys. A simila r analogy also exists for the Curi e magneti c fo r T 3 4. 2 K , whil e th e approximatel y linear points, breaking down only for Ni-M n alloys wit h 10 and decreas e inCTOin the rang e of hydroge n concentration s 20 at.% Mn. A completely differen t behavior of mag0 � n^ 0.7 is due to the decrease in the content of the netic propertie s was discovered in Ref . 81 in Ni-C r alferromagneti c y^ phase in the two-phase mixtures (fi loys, containing �7 at.% Cr. Replacemen t of chromiu m + y 2 ) wit h increasing n. by nickel in these alloys increases both 609
Sov. Phys. Usp. 25(8), Aug.,1982 Ponyatovskii eta/.
609


tt is this picture that is observed on saturation of Ni-C r alloys with hydrogen.81 Thus, in both limiting cases, the band theory of fer romagnetism permits describing the observed dependences a0(�) and T0(n) assuming that the increase in the degree of filling of the d band of the metal by electrons whe n the metal is hydrogenated has a dominating effec t on the magnetic properties. The situation wit h alloys for which the dependences a^N") deviate fro m a straight line with slope Sa0/&N*~ -lp.B/electron only whe n the impurity content is appreciable is more complicated. Such magnets include Ni-F e alloys wit h j�: Fe >0. 6 and Ni-M n alloys wit h # Mn S 0.1 (see Fig. 10). These collectivized ferromagnets are weaker , i.e., at T = 0 K, an appreciable number of holes exist in them both in the dt and df subbands. The question as to the role played by the deformation of the band structure in the formatio n of the dependences a0GeNl) of such alloys is controversial at the present time , especially for the Ni-F e alloys, whic h in this range of concentrations have anomalous physical properties (Invar nature). Here, it is more likely that the opposite occurs: the observed analogy to the behavior ofCTO(and in most cases Tc) wit h increasing nickel content and wit h hydrogenation indicates the fact that it is the degree of filling of the d band and not its deformation , over quite a wid e range of compositions (&Ne-& f�s. 0.5 electrons/atom) that primarily determines the magnetic properties of the starting alloys (without hydrogen). Indeed, let us assume that the deviations at values of Ne, smaller than some limiting value N'lm, fro m the straight line
Mtfe)�crJ�-(tf�-7V5i)(iB, (4.6 )

change in the structure of the d bands or in the nature of the exchange interactio n in the region of the "anomalous" behavior of the dependences cr0(-^e)> but at the very least due to changes, gradually accumulating over wide ranges of concentrations of the order of the total content of iron and manganese in the alloys. It should be noted that the concentration dependence of these changes is muc h weaker i-n the case of nickel alloyed wit h iron, whe n it was necessary to have &N' = (Z Nl -Z Fe k Fe ~ 2-0.6=1. 2 electrons/atom in order that the dependence cr0(JVe) firs t deviate fro m (4.6) , than in the case of alloying wit h manganese, whe n &N* constitutes � 3-0.1 = 0.3 electrons/atom. Probably, the stronger deformatio n of the band structure of nickel whe n it is alloyed wit h manganese rather than iron is what leads to the fact that, in contrast to all the Ni-F e alloys studied, for Ni-M n alloys wit h 10 and 20 at.% Mn, an analogy is not observed in the change of the Curie points wit h composition and wit h saturation by hydrogen.
c) Fees (Ni1.xMnx)35-H solutions

The Fe65(Ni1.]tMn]()35 (% by weight ) solutions are a classical system, in whic h over a period of many years the problems of the competiion between ferro - and antiferromagneti c ordering in fe e metals was studied. This system gave the rar e possibility of observing systematically the effec t of hydrogen on both types of magnetic ordering using the same grou p of objects. The substitution of nickel by manganese in the Fe65(Ni1..eMn]()35 system decreases the Curi e point fro m T0 = 467 K for the Fe65Ni35 alloy to helium temperatures for the alloy wit h ~10% Mn. With a higher manganese content, antiferromagnetic ordering arises in the system and the Nee l point increases monotonically to TH = 442 K for the Fe65Mn35 alloy. Both ferro - and antifer romagnetically ordered alloys exhibit anomalous physical properties, characteristic of Invars (see Refs. 82 and 83). The technology that we developed for compressing hydrogen up to 70 kbar permitted obtaining y solutions with compositions up to w~ l for alloys in the entire range of concentrations fro m Fe65Ni35 to Fe^Mii^.84 The dependences a0(n) and Tc(n) for the Fe-Ni-Mn- H solutions are presented in Fig. 15. The following is observed whe n the Fe65(Ni1.xMnI)35 alloys are saturated wit h hydrogen. The Nee l points of the antiferromagnetic alloys decrease; ferromagnetic ordering arises at a definite hydrogen concentration nf; and, then , the Curie points of the ferromagneti c solutions increase monotonically. Therefore , the change in the magnetic properties of the Fe65(Ni1.IMn%)35 alloys with hydrogenation is analogous to their change wit h substitution of manganese by nickel (whic h increases the electron concentration N'). As shown in Ref. 85, whe n the Fe66Ni31Mn3 (atomic percent) alloy is saturated with hydrogen, the shape of its Mossbauer spectrum also approaches the shape of the spectrum for the Fe65Ni35 alloy. Let us introduce the similarity coefficients � for the dependences of the magnetic properties of the alloys wit h and withou t hydrogen. If it is assumed that both
Ponyatovskii et al.
610

whic h describes well the behavior of the spontaneous magnetization of strong collectivized ferromagnet s wit h a fe e lattice, for Ni-F e and Ni-M n alloys are due to sharp (occuring in the range AN"� 0.5 electrons/atom) and qualitatively differen t changes in the band struc ture. Experiments show that for both groups of alloys the influence of hydrogen onCTOof alloys wit h N*�1.8 5 /iB/atom) and Ni80Mn20-H (Fig. 14a, a^"~O.B /iB/atom) with maximum values on the corresponding curves of a0(Ne) for the Ni-F e and Ni-M n alloys (Fig. 10). Therefore , the effect of hydrogen on the band structure of the Ni-F e alloys is opposite to the effect of the substitution of nickel by iron, or in the case of the Ni-M n alloys, of substitution of nickel by manganese, i.e., in accordance with the starting assumption, it is qualitatively different . But, this is too unlikely: all the experimental and theoretical data taken together show that hydro gen must defor m the d band, forme d by the transition metal atoms, weakly and only increases the degree of filling of the band by electrons. Thus, the analysis of the effec t of hydrogen on the magnetic properties of the Ni-F e and Ni-M n alloys leads to the conclusion that the deviation of the depen dences of these alloys fro m (4.6 ) is due not to a sharp
610
Sov. Phys. Usp. 25(8), Aug. 1982


also permi t clarifyin g new aspects of this influence. The point is that the magnetic propertie s of the Fe65-(Ni1.IMn.()35 alloys82'83 and hydroge n solutions based on them84 are describe d wel l by the equations of the theor y of ver y weak collectivized ferromagnet ism.86'87 Fro m these equations, in particular , it fol lows that for wea k collectivized ferromagnet s
(4.7)

OA

FIG. 15. The dependences or 0 (� ) and Tc(n) for hydrogen solutions in fee Fe65(Ni1..l.Mn.c)35 alloys.84 Notation as in Fig. 16.

wher e kB is Boltzmann's constant, ^(EF) is the density of states at the paramagnetic Ferm i level � F , whil e vm =jf(m\t�)/Jtr(t^. In fe e ferromagnets , the Ferm i level occurs near the maximu m of the density of states jV"(z),6a whic h leads to v2^ 0- For i/ 2 � 0, the values of the factor s in Eq. (4.7 ) enclosed by the square brackets depend weakly on the magnitudes of the parameters enterin g into it, varyin g fro m 3 at v2=Q to 1 for t/2 -- -<*>. In the case of a parabolic band, this facto r equals 2. Thus, knowing Tc and cr0, for a weak collectivized ferromagnet , it is possible to estimate as wel l the den sity of states at the paramagnetic Ferm i level
l\ri-

the concentratio n dependenc e of the magneti c propertie s of the pseudobinary Fe65(N i^Mn,),^ syste m and the dependence of thes e propertie s on the hydroge n content in the alloys is determine d primaril y by the change in the degree of fillin g of the d band , the n ther e must exist a common , for all these alloys and hydroge n solutions based on them , effectiv e electro n concentratio n N* at whic h ferromagneti c orderin g arises . Assumin g that th e quantitie s f diffe r little fo r th e alloys studied , fro m the equality using the values of Ne and nf presente d in Table III, by the method of least squares we obtain (f} = 0.7 elec trons/ H atom and N j = 8.35 electrons/atom. The quantitie s N\-Ne and � = (N%-N')/n, are also presented in Table III. It is evident that the sign and orde r of mag nitud e of the f obtained in this manner agree wit h the assumptio n that the degre e of fillin g of the d band plays a dominan t role in the formatio n of the compositio n dependence s of the magneti c properties of alloys studied wit h respect to nickel and hydrogen . The results of investigation s of hydrogen solution s in Fe65(Ni1.xMn.t)35 alloys not only confir m the previously developed view s on the nature of the influenc e of hydro gen on the magneti c propertie s of transitio n metals but TABLE HI. Parameters for hydrogen solutions in MnJj j alloys
Wl.%M n
0 4. 5 10 17 24 29 35
Tc, K
Ty. K

(4.8)

The dependenc e T0(a0) for the Fe^Mi^M^)^ system is shown in Fig. 16. It is evident that the values ot Tc of the specimen s studied as a functio n of
TC,K

400

2VC . N?-.v e , Electron/ato m Electron/ato m 8.68 8.54 8.38 8.17 7.9 7 7.8 2 7.6 2
-0.33 --0.1 9 -0.03 0.1 8 0.38 0.53 0.70

�f

t. Electron/atom
_ -- -- 1.00
2,0

467 228 . __ 160 253 431 442

0-- 0.18 0.58

~

0.81 0.94

0.6 6 0.6 5 0.7 4

i,~ (A r | -- -VVnf ; A' j = 8.35 electrons/ato m in the Fe-Ni-Mn alloy .

FIG. 16. Curie points T c as a function of the spontaneous magnetization or0 at 7" = 0 K. 1--7) y hydrogen solutions in FesfNij.^Mn, ) alloys51 (1--0; 2--4.5; 3--10; 4--17; 5--24 ; 6--29 ; 7--35 wt.% Mn) ; 8) Ni-Fe alloys, containing 67.5 at.% Fe 45 - 46 ^ ) nickel; 10) Fe65(Ni4.., Mn,)K alloys.82
Ponyatovskii eta/.
611

611

Sov. Phys. Usp. 25(8), Aug. 1982


the Ni-F e Invars. Indeed, if the possibility of random coincidences is neglected, then in order that the weak collectivized ferromagnets with identical density of states ^(EJ.) at the paramagnetic Fermi level have equal Curie points and equal values of spontaneous magnetization over a wid e range of values of X(E F ) , several conditions must be satisfied simultaneously: 1) the shape of the d bands of these ferromagnet s must be the same; 2) the degree of fillin g of the d-bands must be the same; 3) the relative shift in the dt and d* sub-bands A = la at all temperatures must be the same, i.e., the parameters / of the effective exchange interaction must be equal. Therefore , the results presented in Fig. 16 for the Fe65(N i1.JCMnJt)35 alloys support the fac t that whe n manganese is replaced by nickel in this pseudobinary system, as well as whe n the alloys are saturated with hydrogen, the change in the magnetic properties is due primary to the increase in the degree of filling of the d band and, in addition, the phenomenological coefficients f < 1 electrons/H atom for the Fe-Ni-Mn- H solutions are close to the values 77 of the number of electrons contributed by the hydroge n atoms above the Ferm i level. These results explain to a certai n extent one of the most complex problems arising in the discussion of the influence of hydrogen on the magnetic properties of transitions metals, namely, the role of volume effects: the Fe65(Nil..tMn.t)35 alloys, just as the Ni-F e alloys wit h # Fe s 0.6, are Invars, whose characteristic fea ture is the strong dependence of the magnetic properties on volume. The continuous line in Fig. 17 illustrates the depen dence a0(Ne) for fe e Ni-F e alloys at P= l bar. The dashed lines indicate the analogous dependences for P = -50 and +50 kbar, calculated assuming that the volume dependence of 00 is linear using the experimental values of (l/CT 0 )(8a 0 /9P) ln , obtained under conditions of hydrostatic compression.47'77 For ATes 9 electrons/ atom, the slope of the curve cr0(JVe) for P= -50 kbar increases sharply. For example, whe n N" decreases fro m 8.9 to 8.7 electrons/atom, a0 increases by ~0.9 /iB/atom. If the collectivized ferromagne t is strong [and the number of holes in the d bands of fe e alloys based on nickel is such that they must become strong

ferromagnets, whe n the magnetization reaches values a0(N') ~ a?1 + (N' -10)MB], a change in the degree of fill ing of the d band by AJVe= 8.9 -8.7 = 0.2 electrons/atom increases a0 by only ~0.2 MB/atom. The increase in
75

at.% fe SO 25

6.5

3.0 3.S 10,0 N8, electrons/atom

FIG. 17. Concentration dependences ofCTOfor Ni-Fe alloys. Continuous curve is for P =1 bar67'68; 1) computed curve for P = -50 kbar; 2) for P = + 50 kbar (see Sec. 3).
612

d) Applicability of the rigid d-band model for describing the magnetic properties of other Me-H solutions

1) Hydrogen solutions in fee alloys based on 3d metals. The concept of hydroge n as a donor of a fractiona l
Ponyatovskiief a/.
612

Sov. Phys. Usp. 25(8), Aug. 1982


numbe r of electron s in a weakl y deforme d d band of the metal-solvent , whic h is one of the most significan t results of the study of magneti c propertie s of y solutions Ni-Me- H makes it possible to predic t at least qualitatively the behavio r of these propertie s for a numbe r of systems. For example, as n increases , a0 and Tc of strong collectivize d ferromagnet s [Ni-Cu , JSi-Zn,Co-Fe ; for Ni-C u alloys, this is confirme d experimentally (see Ref . 89,90,51) ] an d o f alloys fo r whic h 3a 0 /8JV e > 0 wit h a lo w impurit y conten t (Ni-V,Co-W,Co-Cr) , must de crease. As a resul t of the hig h densit y of states at the Ferm i level of nicke l and cobalt, the electrons of the dissolving hydroge n must fil l the d band in the fe e alloys o f thes e metals wit h nontransitio n elements (Ni Al,NUGe,Ni-Sb , and so on) as well , again decreasing a0 and Tc. The increas e ofCTOwit h increasin g hydroge n concentratio n must , apparently , b e observed only fo r those alloys wit h 9a0/dNe>0, fo r whic h th e dependence s ff 0 (jV e ) deviat e fro m (4.6 ) only fo r a large impurit y content , i.e., it is to a large extent due to the degre e of fillin g o f th e d band (Ni-F e alloys wit h Ar Fe >0.6 ; Ni-M n with* Mn >0.1) . 2) Hydrogen solutions in 3d metals with hep lattice. A necessary conditio n for the use of the rigi d d-ban d model to describ e magneti c propertie s of Me-H solution s is knowin g how thes e propertie s depend on the degre e of fillin g of the d-band in the starting material s (withou t hydrogen) . In the case of fe e metals , such informatio n ca n b e obtained, fo r example, fro m th e de pendence s aa(Ne) and Tc(JVe) for 3d metal alloys, closely positioned in Mendeleev's table , whos e propertie s are describe d well by the rigi d band model. The concen tratio n dependence s of magneti c propertie s wit h an he p crystal lattice wer e studie d in must less detail , pri marily because of the relativel y narrow intervals of the mutual solubilit y of 3d metals in hexagonal phases. However, fo r Nez 7.7 electrons/atom, these depen dences fo r he p an d fe e alloys, apparently , ar e similar , whic h is the key for discussin g the propertie s of e hydrogen solutions at least in iron (N"=8 electrons/atom ) an d cobalt (N e = 9 electrons/atom).65 Indeed , th e values of a0 of fe e alloys of 3d metals closely positione d in the periodic table lie along a single curve as a functio n of Ne (see Fig. 10). Whe n Are decreases to =8.8 electrons / atom , a0 increases approximatel y linearl y wit h slope da0/8N*~-l ju B /electro n Ni-Cu , Ni-Co , an d Ni-F e al loys) , reaches a maximum , and begins to decrease. Fo r Ar e =8.3 5 electrons/atom , th e alloys becom e para magneti c dow n t o heliu m temperatures . The n antifer romagneti c orderin g arises an d th e Nee l point in creases monotonicall y up to TN = 440 K at N e =7.6 5 electrons/ato m (Fe65(Ni1.xMn)t)35 alloys). Available data on the magneti c propertie s of he p metals and alloys fall into an analogous scheme . He p Fe-M n alloys (Ne = 7.71-7.82 electrons/atom ) ar e antiferromagnet s wit h Nee l points =230 K.91 Extrapolation fro m the Fe-R u and Fe-Os alloys shows that at atmospheri c pressure , the metastable he p modificatio n of iron (jV e = 8 electrons/atom ) must also be an antiferromagnet , but wit h a lower Nee l point T N ~ 10 0 K. 9 2 He p cobalt (Ar e = 9 electrons/atom ) isaferromagne t wit h 613
Sov. Phys. Usp. 25(8), Aug. 1982

Alloying cobalt wit h nicke l ( 9 s A re s 9. 2 electrons /atom) leads to an approximately linear decreas e in a0 wit h slope da0/dNe~ -1 ^/electron. 93 - 9 4 Assumin g that as in the case of fe e alloys, the depen denc e cr0(Are) for alloys wit h he p lattice is determine d over quit e a wid e rang e of concentrations primaril y by changes in Ne, we shall now discus s the properties of � hydride s of cobalt and iron , described at the beginnin g of Sec. 4. Injectio n of hydroge n decreases a0 of cobalt wit h initial slope 9a 0 /9w| n=0 = -0.6 /i B / H atom , which , as can be seen fro m a compariso n wit h the dependenc e aa(N') fo r hep Co-N i alloys, agrees wit h th e prediction s of the rigid d-band model. In the case that this model is valid for describin g the magnetic propertie s of Fe-H solutions, an increase in hydroge n concentration in he p iron must lead to (as observed , for example, wit h hydrogenatio n of fe e Fe-Ni-M n alloys), a suppressio n of antiferromagneti c orderin g the appearance of spontan eous magnetizatio n at some value n = nf and furthe r to a monotonic increase in Ponyatovskii eta/.
613


of localized magnetic moments on ruthenium atoms on hydrogenation of Pd-R u alloys (local moments for 4d metal atoms in a metallic matrix wer e not previously observed95) and its interpretation withi n the scope of the rigid d band model. The experimental results are as follows. Measurements of the magnetic permeability and low-temperature heat capacity have shown that for the fe e Pd-Ru alloys studied, containing up to 4 at.% Ru (the decom position of Pd-Ru solid solutions into y + c phases pre vents obtaining homogeneous alloys wit h a large ruthen ium content), there are no local magnetic moments. When these alloys are saturated wit h hydrogen , ther e is an isomorphic phase transition y1 -- y 2 , accompanie d by an abrupt increase in the hydrogen concentration; the quantity nTM2in, the minimu m solubility of hydrogen in the y2 phase, constitutes �0.6 at T= 100 �C for the Pd-H system and decreases to =0.4 for the Pd96Ru4-H system. In the Pd-Ru- H hydrides formed , rutheniu m atoms already have local magnetic moments. The magnitude of these moments is maximum for n = w�in; fur ther increase in the hydrogen concentration in the region of homogeneity of the y 2 phase causes the m to decrease monotonically and the n to vanish (at n~ 0.8). The ranges of applicability of the rigid band model for describing the properties of 4d metal alloys are much narrower than for 3d metals.95'96 For alloys of palladium wit h its nearest neighbors in Mendeleev's table, namely, silver and rhodium, it, apparently, works satisfactorily but already for alloys wit h the next left neighbor, ruthenium , it is qualitatively incorrect. From a comparison of the concentration dependences of the coefficient of electron heat capacity (proportional to the density of states at the Ferm i level) whe n palladiu m is alloyed wit h silver and ruthenium , it may be concluded that the addition of even small (0.6 at.%) quantities of ruthenium shifts the Ferm i level relative to the d band upward in energy wit h the magnitude of this shift corresponding to an average rate of fillin g of the d band of palladium 9Ne/SxRaa 1.3 electrons/R u atom (we recall that if the rigid band model is valid, the addition of ruthenium into palladium must , on the contrary, lead to an emptying of the d band wit h 8We/ 8 *Ru = ^Ru -�zi>d= ~ 2 electrons/Ru atom). Withi n the scope of Friedel's theory,71 this means that the virtual bound d states of ruthenium atoms are appreciably above the Ferm i level of palladium, and part of the electrons in these states flow into its conduction band. Ruthenium atoms in Pd-R u alloys do not have local magnetic moments. Therefore , the virtual d levels of these atoms are not split into sublevels wit h oppositely oriented spin (this situation is schematically illustrated in Fig. 18a), since the exchange energy of such splitting is less than the energy broadening of these levels due to hybridizatio n with conduction band states. Wit h the formatio n of yz hydrides, the virtual levels of ruth enium split (Fig. 18b) and due to the differen t degree of filling of the d\ and rff sublevels, local magnetic moments appear on the rutheniu m atoms. Accordin g to Refs. 95 and 96, the splitting could be due to a decrease in the energy broadening of the virtua l levels of ruthen 614

FIG. 18. Diagram illustrating the behavior of virtual bound states of ruthenium atoms In y solutions Pd--Ru--H96 a) n =0;b ) n>�TM In ; c ) n S

ium atoms due to a decrease in the degree of their hybridization wit h the conduction band states of palladium wit h the formatio n of hydrides , since in this case the virtual levels are shifted into a region of energies wit h low density of states of palladium. This conclusion is based on the experimental fact that wit h the formatio n of the y2 phase of Pd-Ru- H alloys, as measurements of the electron heat capacity have shown , the d band of palladium is entirely filled and the Ferm i level, inter secting the d sublevels of ruthenium , falls into the range of energies wher e ther e are only states of the s type wit h low density. In view of the low density of states near the Ferm i level, the increase in the hydro gen concentration (and , correspondingly, electron con centration) in the region of homogeneity of the y2 phase of Pd-Ru- H solutions leads to a rapid increase in the Ferm i energy. The virtual d states of rutheniu m atoms are gradually filled , the magnitude of the uncompen sated local moments decreases monotonically and, whe n the d sublevels are completely filled , the local moments disappear (Fig. 18c). 4) Ni-Fe- C solutions. As is evident fro m the discus sion in Sec. 2, the rigid d-band approximation , in spite of its simplicity , is useful for explaining, and sometimes predictin g as well , the propertie s of hydroge n solutions in a numbe r of transition metals and thei r alloys. The available experimental data on the effec t of carbon on the magnetic properties of Ni-F e alloys97-98 permi t examining the problem of the applicability of this approximation and to describe the properties of substitution Me-C solutions, the nearest analog to MeH solutions. Figure 19 shows the change in
3.0

9.5 10.0 / * electrons/atom

FIG. 19. The dependence of Aa0/n c of the change in the spontaneous magnetization a0 at 7' = 0 K for carbon alloyed fee Ni-Fe alloys as a function of the electron concentration of alloy s AT e .97- nc is the atomic ratio carbon metal. The dashed line shows the computed curve for |c = 3.5 electrons/C atom (see subsection d in Sec. 4).
Ponyatovskii etal.
614

Sov. Phys. Usp. 25(8), Aug. 1982


whe n carbo n is injecte d into the m wit h wc-0.01.97 It is evident tha t for alloys containing up to -55 at.% Fe (AT" s 8.9 electrons/atom) , the dissolution of carbon decreases the spontaneous magnetization and, in addition, the value of Aa c /w c depends weakly on the concentration of the Ni-F e alloys and lies in the range -3-3.5 MB/ C atom. Wit h a furthe r increase in the iron content in the alloys, the values of &a0/nc increase rap idly and become positive xYe^ 0.63 (jV� 8.74 electrons/ atom). Fro m a compariso n of the dependence a0(N") presented in Fig. 10 for Ni-F e alloys, it is evident that the initial slope of the dependences a0(nc) must depend precisel y in this manner on the concentratio n of the NiFe alloys, if the main effec t of the injectio n of carbon atoms is related (as in the case of the injectio n of hydrogen atoms) to the increase in the degree of filling of the 3d band of the metal-solvent. The value of the slope &aa/nc~ -(3-3.5) j^.B/atom for Ni-F e alloys wit h *FeS 0.55, whic h are strong collectivized ferromagnets , is interesting. Carbon, in contrast to hydrogen , has fou r valence electrons and if the rigid band model (the cation model) wer e valid for describing solid Ni-Fe- C solutions, the n the dissolution of carbon in these alloys would decrease the spontaneous magnetization wit h slope da*/dnc~ -4 /iB/C . atom. It is ver y probable that the observed differenc e between the experimental values of ACTO/WC and d
has been and continues to be devoted to the study of superconducting properties of hydrogen solutions in palladium and its alloys that it is now simply impossible not to discuss this subject in a review concerning the properties of transition metal hydrides. However, the discussion wil l be carried out on a completely dif feren t plane than in the preceeding section examining the magnetic properties of Me- H solutions. The extensive experimental and theoretical material accumulated shows that the physical nature of the appearance of superconductivity in Pd- H and Pd-Me- H systems is very complicated and it is hardly possible to expect that in the near futur e a simple model wil l appear describing even the concentration dependences of their superconducting transition temperatures Te, since in order to estimate Tc it is necessary to have a detailed knowledge of the electron and phonon spectra of the solution, whil e hydrogen strongly changes both. A detailed review and discussion of available data on the superconducting properties of Me-H solutions can be found in Refs. 8, 9, 100, and 101. We shall consider only the problem of the differen t nature of the structural instabilities in the experimentally obtained Me-H solutions, which , as the latest studies have shown, can play an important role in the appearance of superconductivity wit h anomalously high Tc at least in the case of hydrogenatio n of palladium alloys wit h previous met als. Super-conductivity arises in Pd- H solutions for n S 0.8." Further increase in hydrogen content, as shown in a numbe r of papers, leads to a monotonic increase in the superconducting transition temperature to T c = 8.8 K at w = 1 (in Fig. 20, this dependence is shown by the dashed line).100 Even higher values of Tc of the order of 13-17 K wer e obtained by implanting hydrogen into palladium alloys wit h previous metals, copper, silver , and gold.102 Both in the case of palladium and in these alloys, at a certain irradiation dose, the thin

T,,K

16

0 -1

� -2

ft T

? ?X�

X

0.2

0.4

0.6

1.0 n

5. SUPERCONDUCTIVITY OF HYDROGEN SOLUTIONS IN PALLADIUM ALLOY S

Afte r Skoskiewicz" discovered in 1972 that superconductivity exists in palladium hydride , so muc h attention
615

FIG. 20. Dependence of the superconducting transition temperature T c on the atomic ratio hydrogen/metal � . 1) For solutions Pd80Ag20-H;103 2) Pd^Cu^-H103 (signs with , arrows indicate that these specimens are not superconducting for T& 2 K); 3) Pd65Cu45-H102 (intervals are indicated in which the resistance of the specimens changes from 90 to 10% of the value of the residual resistance of the normal phase). Dashed curve shows the dependence T c (n ) for the Pd--H solutions.11"1
Ponyatovskh eta/.
615

Sov. Phys. Usp. 25(8), Aug. 1982


(-1500 A) hydrogen containing layer of the metal formed became superconducting and with further irradiation rc increased, attained a maximum value Tf and began to decrease. In its turn , the dependence of Tf on the content of the alloying element in palladium for each of .the binary systems Pd-Cu, Pd--Ag , and Pd-Au also had a maximum and, in addition, the maximum values of Tf constituted �16.6, 15.6, and 13.6 K, respectively, for the PdjsC^s, Pd70Ag30 and PdMAul6. The typical dependence Tc(w) wit h hydrogen implantation in the Pd55Cu45 alloy is presented in Fig. 20 (the accuracy of the estimates of the hydrogen concentration 6w~15%). The dependences Tc(n) for the P^^g^-H and PdMAu16-H solutions have an analogous character, only the value of the optimum hydrogen concentration at which the maximum value of Tc is attained changes (wopt = 0.7, 0.8, and 0.9, respectively, for alloys wit h Cu, Ag, and Au) . Using the technology for compressing hydrogen to high pressures, we wer e able to obtain macroscopic speciments of hydrides of palladium alloys wit h previous metals, tt turned out that if for the Pd-H alloys, obtained by implantation, the value of Tf agrees wel l with the maximum values of T0 (wit h w -- 1) for massive homogeneous specimens (see Ref . 100), then in the case of palladium-previous metal-hydrogen solutions, no such agreement is observed. The investigation of the Pd80Ag20 and Pd60Cu40 alloys (composition close to optimum for obtaining the maximum values of Te wit h hydrogen implantation), saturated with hydrogen at PH � 70 kbar, showed103 that the dependence Te(n) for tlie solid Pd80Ag20-H solutions for w � 1 is close to that for the Pd-H solutions, whil e in the Pd60Cu40-H system for w S 0.6 and T > 2 K, there is no superconductivity (see Fig. 20). Thus our results show that the anomalously high values of Tc obtained wit h hydrogen implantation in palladium alloys with precious metals are due to the char acteristics of the metastable states arising on implantation. It should be noted that at the present time there is no unified point of view as to how the alloying of palladium wit h previous metals should affect Tc in the case of the formatio n of hydrogen solutions based on the fee lattice of the metal. For example, according to the theoretical estimates in Refs. 12 and 101, the superconducting transition temperature must in this case increase, while according to the theoretical estimates in Ref. 104 and estimates made in Ref. 105 based on an experimental study of the low-temperature heat capacity of nonsuperconducting solid Pd80Ag20-H solutions, it must decrease. The possibility that the instability of the crystal lattice (due to softening of some phonon modes and, therefore, increase in the electron-phonon interaction constant) can play an important role in obtaining high Tc wit h hydrogen implantation in palladium alloys with Cu, Ag, and Au was noted by Stritzker,102'100 who discovered the interesting property that their Tf depend on the dose of implanted hydrogen 4>. For the Pd-Cu and Pd-A g alloys wit h compositions close to optimum (Tf>l5 K), the monotonic increase in Tc wit h increasing tfi was sometimes replaced by its sudden de616

crease and vanishing of superconductivity for T>1 K, while the residual resistance Rrta decreased discontinuously by 10-20%. Further increase in # led to a monotonic decrease in RTea, but the appearance of superconductivity for T>IK was no longer observed. Stritzker explained this by the formatio n of a new nonsuperconducting phase w ith a smaller value of RTtt. Thus, the reason for the appearance of superconductivity with anomalously high values of Tc accompanying implantation of hydrogen into palladium alloys wit h precious metals is closely related precisely to the structural properties of these systems. We recall that the Me-H solutions based on transition metals in the VI-VIII groups studied had only either a fe e or a hep metal sublattice. Since the palladium-precious metal alloys themselves already have a fe e structure, the structure of the hypothetical new phase, forme d wit h implantation of hydrogen, must therefor e diffe r fro m the previously observed structures. The study of the solid Pd60Cu40-H solutions, saturated wit h hydrogen at high pressure, has indeed led to the discovery of a new phase transition in them , accompanied by a tetragonal distortion of the fe e sublattice of the metal.106 The study of the isobars of the electrical resistance of the Pd60Cu40 alloy in a hydroge n atmosphere in the pressure range of 5-10 kbar showed that wit h heating above r*~220� C an irreversibl e phase transitio n occurs in the specimens and, in addition, the resistance decreases strongly in the transition process. This transition proceeds very slowly: even at T=350�C , the drif t in the resistance continues for tens of hours. An x-ray diffractio n analysis established that the metallic sublattice of Pd60Cu40-H specimens, obtained by holding at T = 200 �C < T*, 5 � PHz� 20 kbar and having the composition n^ 0.47, retains fe e symmetr y and, in addition, the value of A70 of the abrupt change in the unit-cell volume of the Pd60Cu40 alloy accompanying hy drogenation agrees satisfactorily wit h the dependence AV 0 (n ) presented in Ref. 52 (see Fig. 9) for y hydrogen solutions based on palladium and its alloys. The dif fractio n pattern for these specimens has, however , some special features: lines of the {ill} type are sharp, whil e the remaining lines are appreciably broadened, whic h can be interpreted as the result of the appearance of packing defects in the close-packed planes107 or small tetragonal or orthorhombi c distortion of the fee sublattice. Pd60Cu40-H specimens, obtained by holding at 250 � T �350� C (i.e., wit h T>T*) and P H2 * 2 0 kbar � had a metallic sublattice, whose symmetry can be described based on a fe e subcell (i n what follows, the y1 phase) with the axial ratio 0.94s c/a-S, 0.97. The hydrogen content in these specimens fell into the range 0.3 �� n ��, 0.5; the volume per metal atom had approximately the same magnitude as for y solutions Pd60Cu40-H wit h nearly the same values of n. The study of y1 specimens durin g annealing, whic h caused separation of hydrogen, led to some interesting results. When hydrogen is partially liberated afte r annealing at room temperatur e and atmospheric pres Ponyatovskii eta/ .
616

Sov. Phys. Usp. 25(8), Aug. 1982


sure , the metal lattice remaine d tetragonal, whil e the quantity c/a, as a rule , decreased. The alloy retained the tetragonal structur e even afte r annealing for six hours in a vacuum at 300 �C, whic h led to complete liberatio n of hydrogen , whil e the volum e per metal atom decreased and assumed a value close to that for the starting Pd60Cu40 alloy. Wit h furthe r annealing for several hours at T & 350 �C, the structur e of the alloy returned to the starting cubi c structure . The retur n to th e fe e structur e occurre d also afte r plastic deforma tion of the specime n at room temperature . These results show that at atmospheri c pressur e and T S 20 �C, the y1 structur e of the Pd60Cu40 alloy is not in thermo dynami c equilibrium . A t th e same time , th e metastability of this structur e afte r complete liberation of hydrogen and even some increase in its degree of tetra gonality support the proposition that the observed tetra gonal distortions are mainly related not to the possible phase transition s in the hydroge n sublattice (fo r example , its ordering) , but to a restructurin g of the met al sublattice itself occurrin g in the Pd60Cu40-H solution. One of the possibilities of such restructurin g is atomic ordering . The presenc e of wid e regions wit h ordered positionin g of atoms in T-c diagram s is characteristi c both of the Cu-P d syste m and a numbe r of related systems. In particular , orderin g in the regio n of compositions close to Cu 3 P d leads precisel y to tetragonal distortio n of the fe e lattic e of Cu-P d alloys.108'110 Wit h orderin g of this type , in addition to the mai n (struc tural ) lines of the fe e lattic e superstructura l lines appear whos e positio n is close to that for reflection s fro m planes wit h mixed indices. For y1 specimens , in addition to the mai n lines , 11 ver y wea k and broad lines wer e observed and mixed indices could be assigned to all of them. Thus, the results of the x-ra y diffractio n analysis permi t assertin g wit h quite a high degre e of reliabilit y that the formatio n of the y1 phase in the Pd60Cu40-H system is accompanied by orderin g of the metallic sublattice (eve n though an incomplete one , whic h is indicated by the large widt h of the superstructura l lines com pared to the widt h of the mai n lines). The appearance of packing defect s in the close-packed planes of the metalli c sublattic e of the y phase, forme d at a lower temperatur e (whic h is manifested in the broadening of all lines except {ill}), is apparently an intermediat e stage of this process. Thus, a new phase transitio n has been discovered in Pd60Cu40-H solutions , whic h does not have any analogs to previously studied hydroge n solutions based on group VI-VIII transitio n metals and thei r alloys. At the same time , it is still difficul t to give a clear answe r to the question as to whethe r or not the y'- phase arisin g at high pressur e is the nonsuperconductin g phase whic h sometime s form s on implantation of hydroge n and whos e existence accordin g to Ref . 102 in principl e can explain the anomalously high values ot Tc. However , it is possible to presen t some argument s supportin g a positive answer. The formatio n of the y' phase , jus t as the formatio n of the nonsuperconductin g phase accom panyin g implantation, is accompanied by a decrease in
617
Sov. Phys. Usp. 25(8), Aug. 1982

the electrical resistance. Strong local overheating in the process of implanting hydroge n and the high concen tration of defect s in the hydrogen containing layer pro duced create favorable conditions for occurrence of diffusio n processes. We should also mentio n Ref. 101, wherei n the possible relation of the values of the con centrations of the Pd-Cu , Pd-Ag , and Pd-A u alloys, for whic h maximu m Tc wer e observed wit h hydrogen implantation, to the mor e favorable compositions for possible ordering of alloys withou t hydrogen is pointed out. It should be noted that at the presen t tim e most superconductin g Pd-Me- H solutions are synthesized by implantation. The study of hydroge n solutions in palladium alloys wit h preciou s metals has demonstrated the necessity of including the possibility of the existence of a structura l instability in the physical descriptio n of the results obtained wit h specimen s synthesized by this specifi c method. A vivid example, illustrating the need to take into account also the effec t of radiation defects , was the discovery of superconductivit y in pur e Pd afte r irradiation by heliu m ions.111 Experiment s wer e performe d on palladium film s wit h thicknesses up to 400 A, grow n out of the vapor phase o n substrates (SiO 2 ,Al 2 O 3 , Si) , wit h temperatur e vary ing fro m 4.2 to 300 K. The energy of the He* ions was chosen to be quite hig h (130 keV) , so that they woul d definitel y pass through the palladium fil m and not be retained by it. The temperatur e of the fil m durin g irradiation did not exceed 8 K. At a definit e irradiation dose, ^~10 1 6 He* ions/cm 2 , palladium became super conducting , and the superconductin g transitio n temper ature rapidly increased and reached a value =3. 2 K. Stritzker111 interprete d the observed effec t as the result of the suppression of spin fluctuations in palladium due to a decrease in the densit y of states at the Ferm i level caused by the diffusenes s of the Ferm i surfac e resulting fro m the appearance of a large number of radiation defects. It is interesting that a necessary condition for the appearance of superconductivit y on irradiation of palladiu m by heliu m atoms is the preliminar y presenc e of defect s in the crysta l structure , arisin g on the deposition of palladium on a cold substrate. In particular , irradiation of specimens in whic h these defect s wer e annealed at 500 K generally did not lead to the appearance of superconductivity for T<: 0.1 K and, in addition, durin g the irradiatio n process, the residual resistance of such specimens increased, whil e for the unannealed specimens , it decreased. The increase in RTm durin g annealin g is a norma l phenomenon , since in this case the numbe r of scatterin g centers increases. The decrease in the residual resistance of unannealed palladiu m film s indicated, in this manner , the annealing of previously presen t defect s forme d durin g growth of these film s on a cold substrate occurrin g wit h irradia tion. The appearance of superconductivity only in unannealed specimens was , apparently, due to the fac t that these defects , themselves annealing durin g irra diation , created favorable conditions for formatio n of specifi c radiation defects . As far as the radiation dePonyatovskiief a/.
617


fects that lead to the appearance of superconductivity are concerned, they could be palladium atoms in interstitial positions or even entir e clusters of such atoms: x-ray measurements showed that irradiation by helium ions caused a strong increase in the crystal lattice constant of palladium of up to 0.4%, whic h corresponded to the presence of ~2% interstitial atoms.111
6. CONCLUSIONS

13

Working in the area of high-pressure physics, we are glad to note that the firs t results of application of highpressure methods for obtaining hydrides of transition metals and studying thei r properties are already interesting and informative. This inspires confidence that a furthe r development of this research, an increase in the pressure range and the range of objects and properties studied will lead to the discovery of many more new and unusual phenomena and wil l provide a deepe r understanding of the physical nature of hydrides. We thank V. L. Ginzburg for his interest in our work and for the suggestion that we writ e this review, E. G. Maksimov for valuable suggestions improving the paper, as wel l as S. N. Stishov, R. Z. Levitin , V. A. Somenkov, N. I. Kulikov, and B. K. Ponomarev for reading the manuscript and for valuable remarks.

*T. Greham, Proc. Eoy. Soc. 16, 422 (1868); Phil. Mag. 36, 63 (1868); C. R. Acad. Sci. 66, 1014 (1868); Ann . Chim. Phys. (Paris) 14, 315 (1868). 2 J. R. Lacher, Proc. Roy. Soc. A 161, 525 (1937). 3 T. D. Lee and C. N. Yang, Phys. Rev. 87, 410 (1952). 4 L. D. Landau and E. M. Lifshitz, Statisticheskaya Fizika, Nauka, Moscow (1976) [Engl. Transl. Statistical Physics, Pergamon Press, New York, (1979)1. 5 V . A. Somenkov, Ber. Bunsenges, Phys. Chem. 76, 733 (1972). 6 Yu . Kaganan d M. I. Klinger, J. Phys. C 7, 2791 (1974). 7 A. M. Stoneham, Ber. Bunsenges. Phys. Chem. 76, 816 (1972). 8 S. V . Vonsovskii, Yu . A . Izyumov, an d E . Z . Kurmaev , Sverkhprovodimost' perekhodnykh metallov, ikh splavov i soedinenif (Superconductivity of Transition Metals, Their Alloys and Compounds), Nauka, Moscow (1977). 9 E. G. Maksimov and O. A. Pankratov, Usp. Fiz. Nauk 116, 385 (1975) [Sov. Phys. Usp. 18, 481 (1975)1. 10 A. C. Switendick, Solid State Commun. 8, 1463 (1970); Ber. Bunsenges. Phys. Chem. 76, 535 (1972); Hydrogen in Metals, Ed. by G. Alefeld and J. Volkl, in: Topics in Applied Physics, Springer-Verlag, New York (1978), Vol. 28, p. 101. U D . A. Papaconstantopoulos and B. M. Klein , Phys. Rev . Lett. 35, 110 (1975); D. A. Papaconstantopoulos, B. M. Klein , J. S. Faulkner, and L. L. Boyer, Phys. Rev. B 18, 2784 (1978). 12 D. A. Papaconstantopoulos, B. M. Klein, E. N. Economou, and L. L. Boyer, Phys. Rev. B 17, 141 (1978); D. A. Papaconstantopoulos, E. N. Economou, B. M. Klein, and L. L. Boyer in: Proc. Intern. Conf. on Hydrogen in Metals, M'unster (1979), p. 733.
618

M. Gupta and A. J. Freeman, Phys. Rev. B 17, 3029 (1978). N. I. Kulikov, V. N. Borzunov, and A. D. Zvonkov, Phys. Status Solidi B 86, 83 (1978); N. I. Kulikov, ibid. 91, 753 (1979). 15 N. A. Galktionov, Vodorod v metallakh (Hydrogen in Metals), Metallurgiya, Moscow (1967). 16 P. V. Gel'd and R. A. Ryabov, Vodorod v metallakh i splavakh (Hydrogen in Metals and Alloys), Metallurgiya, Moscow (1974). 17 M. M. Antonov, Svofstva gidridov metallov (Properties of Metal Hydrides), Naukova Dumka, Kiev (1975). 18 S. A. Shavely and D. A. Vaugham, J. Am. Chem. Soc. 71, 313 (1949). 19 B. Baranowski and M. Smialowski, J. Phys. Chem. Sol. 12, 206 (1959). 20 V. Kh. Alimov, A. E. Gorodetskii, A. P. Zakharov , and V. M. Sharapov, Dokl. Akad . Nauk SSSR 241, 595 (1978). 21 S. A. Semiletov, R. V. Baranova, Yu. P. Khodyrev, and R. M. Imamov, Kristallografiya 25, 1162 (1980) [Sov. Phys. Crystallogr. 25, 665 (1980)1. 22 B . Baranowski and R. Wisniewski, Bull. Acad. Polon. Sci. Ser. Sci. Chim. 14, 273 (1966). 23 R. Wisniewski in: III Krajowa Narada Techniki Wysokich Cisnien, Warszawa (1969), p. 62. 24 B . Baranowski and W. Bujnowski, Roczniki Chem. 44, 2271 (1970). 25 M. Krukowski and B. Baranowski, J. Less-Common Met. , 49, 385 (1976). 26 B . Baranowski, cited in Ref . 10, Vol. 29, p. 157. 27 B . Baranowski, Zs. Phys. Chem. 114, 59 (1979). 28 I. T. Belash and E. G. Ponyatovskif, Inventor's Certificate No. 741105 (USSR); Byult. Izobret. No. 22, 223 (1980). 29 V. E. Antonov, I. T. Belash, and E. G. Ponyatovskif' in: Fizika i tekhnika vysokikh davlenii (High-Pressure Physics and Technology), Naukova Dumka, Kiev (1982), Vol. 9. 30 W. Wicke and H. Brodowsky, cited in Ref. 10, Vol. 29, p. 73. 31 V. E. Antonov, I. T. Belash, E. G. Ponyatovskii, and V. I. Rashchupkin, Fiz. Met. Metalloved. 48, 75 (1979) [Phys. Met. Metallogr. (USSR). 32 V. E. Antonov, I. T. Belash, and E. G. Ponyatovskii, Dokl. Akad . Nauk SSSR 223, 1114 (1977). 33 V. I. Spitsyn, V. E. Antonov, O. A. Balakhovskii, I. T. Belash, E. G. Ponyatovskii', V. I. Rashchupkin, and V. Sh. Shekhtman, ibid. 260, 132 (1981). 34 M. Krukowski and B. Baranowski, Roczniki Chem. 49, 1183 (1975). 35 E. G. Ponyatovskii and I. T. Belash, Dokl. Akad. Nauk SSSR 224, 607 (1975). 36 V. I. Spitsyn, E. G. Ponyatovskn, V. I. Antonov, I. T. Belash, and O. A. Balakhovskii, ibid. 247, 1420 (1979). 37 V. E. Antonov, I. T. Belash, V. F. Degtyareva, E. G. Ponyatovskii, and V. I. Sheryaev, ibid. 252, 1384 (1980) [Sov. Phys. Dokl. 25, 490 (1980)1. 38 V. E. Antonov, I. T. Belash, V. F. Degtyareva, and E. G. Ponyatovskii ibid. 239, 342 (1978). 39 B. Baranowski and K. Bojarski, Roczniki Chem. 46, 525 (1972). 40 I. T. Belash, V. E. Antonov, and E. G. Ponyatovskii, Dokl. Akad . Nauk SSSR 235, 379 (1977). 41 V . E. Antonov, I. T. Belash, and E. G. Ponyatovskii ibid. 248, 635 (1979). 42 E . G. Ponyatovskii, and I. T. Belash, ibid. 229, 1171(1976). 43 T. Greham, Phil. Trans. Roy. Soc. 156, 415 (1866). 44 V. G. Tissen, V. E. Antonov , I. T. Belash, B. K. Ponomarev, and E. G. Ponyatovskii, Dokl. Akad. Nau k SSSR 242, 1340 (1978). 45 B. Baranowski and K. Bochenska, Roczniki Chem. 38, 1419 (1964). 48 B. Baranowski and S. Filipek, ibid. 47, 2165 (1975). 47 D. Bloch, Ann . de Phys. 1, 3 (1966).
14

Sov. Phys. Usp. 25(8), Aug. 1982

Ponyatovskii etal.

618


E . G. Ponyatowskii, V. E. Antonov, and I. T. Belash, Dokl. Akad . Nauk SSSH 230, 649 (1976). 49 E . G. Ponyatowski, V. E. Antonov , and I. T. Belash, Fiz. Tver. Tela 18, 3661 (1976) [Sov. Phys. Solid State 18, 2131 (1976)]. 50 V. E. Antonov, I. T. Belash, V. F. Degtyareva, B. K. Ponomarev, E. G. Ponyatovskii, and V. G. Tissen, ibid. 20, 2680 (1978) [Sov. Phys. Solid State 20, 1548 (1978)]. 51 V . E. Antonov, I. T. Belash, E. G. Ponyatovskii, and V. G. Tissen in: Fizika i tekhnika vysokikh davlenii [High-Pressure Physics and Technology], Naukova Dumka, Kiev (1981), Vol. 9, p. 3. 52 B. Baranowski, S. Majchrzak, and T. B. Flanagan, J. Phys. F: Metal Phys. 1, 258 (1971). 53 I. T. Belash, V. E. Antonov , and E. G. Ponyatovskii, Dokl. Akad. Nau k SSSR 235, 128 (1977) . 54 T. SchoberandH . Wenzl, cited in Kef . 10, Vol. 29, p. 11. 55 I. S. Anderson , C. J. Carlile, and D. K. Ross , J. Phys. C 11, L381 (1978). 56 O. Blaschke, R. Klemencic, P. Weinzierl, and O. J. Eder , Solid State Commun. 27, 1149 (1978). 5? T. E. Ellis, C. B. Satterthwaite, M. H. Mueller, and T. O. Brun , cited in Ref . 12, p. 79. 58 I. T. Belash, B. K. Ponomarev, V. G. Tissen, N. S. Afonikova, V. Sh. Shekhtman, and E. G. Ponyatovskii, Fiz. Tver. Tela 20, 422 (1978) [Sov. Phys. Solid State 20, 244 (1978)1. 59 E. O. Wollan, J. W. Cable, and W. C. Koehler, J. Phys. Chem. Solids 24, 1141 (1963). 60 V . E. Antonov , I. T. Belash, B. K. Ponomarev, E. G. Ponyatovskii, and V. G. Thiessen, Phys. Status Solid! A 57, 75 (1980). 61 R. Andrievskiian d Ya. Umanskif , Fazy vnedreniya (Interstitial Phases), Nauka, Moscow (1977). 62 W . E. Wallace , Ber. Bunsenges, Phys. Chem. 76, 832 (1972). 63 M. Hanson, H. R. Khan, A. Knodler, Ch. J. Raub , J. Less-Commo n Met. 43, 93 (1975). 64 H . R. Khan and Ch. J. Raub, ibid. 49, 399 (1976). 65 V. E. Antonov , I. T. Belash, E. G. Ponyatovskii, V. G. Thiessen, and V. I. Shiryaev, Phys. Status Solidi A 65, K43 (1981). 66 V. G. Thiessen, V. E. Antonov, I. T. Belash, V. K. Ponamarev, and E. G. Ponyatovskii, ibid. 48, K185 (1978). 67 J . Crangle and G. C. Hallam, Proc. Roy. Soc. 272, 119 (1963). 68 S. V. Vonsovskii, Magnetizm (Magnetism), Nauka, Moscow (1971). 69 M. J. Besnus, Y. Gottehrer, and G. Munschy, Phys. Status Solidi B 49, 597 (1972). 70 B. K. Ponomarev and V. G. Tissen, Zh. Eksp . Teor. Fiz. 73, 332 (1977) [Sov. Phys. JET P 46, 173 (1977)]. 71 J. Friedel, J. Phys. et Had. 19, 573 (1958); ibid. 23, 692 (1962); Nuovo Cimento Suppl. 7, 287 (1958). 72 G . T. Dubovka, Phys. Status Solidi A 24, 375 (1974). 73 J. D. Fast, Vzaimodeistvie metallov s gazami (Interaction of Metals and Gases), Metallurgizdat, Moscow (1975), Vol. 2. 74 J. Friedel, Ber. Bunsenges. Phys. Chem. 76, 828 (1972). re V. E. Antonov, I. T. Belash, and V. G. Thiessen, Phys. Status Solidi A 59, 673 (1980). 76 J. M. Leger, C. Loriers-Susse, and V. G. Vodar, Phys. Rev. B 6, 4250 (1972). 77 G . T. Dubovka and E. G. Ponyatovskii, Dokl. Akad. Nauk SSSR 206, 83 (1972). ra H. J. Schenk and H. J. Bauer, cited in Ref . 1 12, p. 350.

48

H. J. Schenk, H, J. Bauer, and B. Baranovski, Phys. Status Solidi A 52, 195 (1979), 80 M. Khansen and K. Anderko, Struktura dvoinykh splavov (Structure of Binary Alloys), Metallurgizdat, Moscow (1962). 81 G. J. Zimmerma n and H. J. Bauer, Zs. Phys. 229, 154 (1969). 82 B. K. Ponomarev and S. V. Aleksandrovich, Zh. Eksp. Teor. Fiz. 67, 1965 (1974) [Sov. Phys. JETP 40, 976 (1974)]. 83 G. T. Dubovka, E. G. Ponyatovskii, I. Ya. Georgieva, and V. E. Antonov, Phys. Status Solidi A 32, 301 (1975). 84 V. E. Antonov, I. T. Belash, B. K. Ponomarev, E. G. Ponyatovskii, and V. G. Thiessen, ibid. 52, 703 (1979). 85 T. Sohmura and F. E. Fujita, Solid State Commun. 25, 43 (1978). 86 D. M. Edward s and E. P. Wohlfarth, Proc. Roy. Soc. A 303, 127 (1968). 87 E . P. Wohlfarth, J. Appl. Phys. 39, 1061 (1968). 88 K. H. J. Buschow, Solid State Commun . 19, 421 (1976). 89 H. J. Bauer, G. Berninger, and G. Zimmermann, Zs. Naturforsch . 23a, 2023 (1968). 90 H. J. Bauer, Zs. Angew. Phys. 26, 87 (1968). 91 H. Ohno and M. Mekata, J. Phys. Soc. Jpn. 31, 102 (1971). 92 H . Ohno, ibid. 31, 92 (1971). 93 P. Weiss and R. Forrer, Ann . de Phys. 12, 279 (1929). 94 R. M. Bozorth, Ferromagnetism, D. Van Nostrand, New York (1951), p. 279. 95 E . Wicke, J. Less-Common Met. 74, 185 (1980). 96 K . Frolich, H. G. Severin, R. Hempelmann, and E. Wicke, Zs. Phys. Chem . N. F. 119, 33 (1980). 97 M. C. Cadeville, R. Caudron, and C. Lerner, J. Phys. F 4, L87 (1974). 98 M. C. Cadeville and E. P. Wohlfarth, Phys. Status Solidi A 26, K157 (1974). 99 T. Skoskiewicz, ibid. 11, K123 (1972). 100 B. Stritzker and H. Wuhl, cited in Ref . 10, Vol. 29, 243. ""B. N. Ganguly, Zs. Phys. B 22, 127 (1975). 102 B. Stritzker, Zs. Phys. 268, 261 (1974). 103 V. E. Antonov, I. T. Belash, E. G. Ponyatovskii, and V. I. Rashchupkin, Pis'ma Zh. Eksp . Teor. Fiz. 31, 422 (1980)]. 104 K. H. Benneman and J. W. Garland, Zs. Phys. 260, 367 (1973). 105 G. Wolf, J. Yanke, and K. Bohmhammel, Phys. Status Solidi A 36, K125 (1976). 106 V. G. Degtyareva, V. E. Antonov, I. T. Belash, and E. G. Ponyatovskii, ibid. 66, 77 (1981). 10? A. Guinier, Theorie et technique de la radiocristallographie, Dunod, Paris (1956). 108 W. B. Pearson, A Handbook of Lattice Spacings and Structure of Metals and Alloys in: International Series of Monographs on Metal Physics and Physical Metallurgy, Ed. by G. V. Raynor, Pergamon Press, London (1964), Vol. 4. 109 K. Schubert, B. Kiefer, M. Wilkens, and R. Haufler, Zs. Metallkunde. 46, 692 (1955). I10 D . M. Jones and E. A. Owen, Proc. Roy. Soc. B 67, 297 (1954). U1 B . Stritzker, Phys. Rev. Lett. 42, 1769 (1979). 112 B. Baranowski and K. Bojarski, Roczniki Chem. 46, 1403 (1972) . 113 V. E. Antonov, I. T. Belash, V. M. Koltygin, and E. G. Ponyatovskii, Dokl. Akad. Nau k SSSR 248, 131 (1979). 1U S. Majchrzak . Bull. Acad. Polon. Sci. Ser. Sci. Chim. 15, 485 (1967). Translated by M. E. Alferief f

79

619

Sov. Phys. Usp. 25(8), Aug. 1982

Ponyatovskii eta/.

619