Growth and Characterization of High-Quality YTiO₃ Single Crystals: Minimizing Ti⁴⁺ Containing Impurities and TiN Formation

Abstract

We report the growth of YTiO₃ single crystals using different starting materials with the nominal compositions, (1) stoichiometric YTiO₃; (2) oxygen deficient YTiO_2.925; (3) oxygen deficient YTiO_2.85, and different atmospheres, (1) 97%Ar/3%H₂; (2) Ar; (3) forming gas 95%N₂/5%H₂, using the laser floating zone growth technique. The oxygen-deficient starting materials were prepared by mixing Y₂O₃, Ti₂O₃, and Ti powder according to the YTiO_3-δ stoichiometry. The addition of Ti powder to the starting materials effectively reacts with the oxygen in the floating zone furnace chamber, reducing Ti⁴⁺ ion-containing impurities. High-quality YTiO₃ single crystals with (2 0 0) facet were grown from the starting materials corresponding to the nominal composition YTiO_2.925. YTiO₃ single crystals grown from different starting materials are characteristic of oxygen content of 3 in both pure crystals and crystals containing impurities, revealed by the same oxygen occupancy in single crystal X-ray diffraction measurements. When forming gas was used, a golden TiN coating formed on the surface of rod.

1. Introduction

RETiO₃ (RE = rare earth elements) perovskites had been intensively studied, in particular since the discovery of high T_c superconductivity in rare earth cuprate perovskites [1,2,3,4]. In all RETiO₃ compounds, except EuTiO₃, Ti ions are trivalent (Ti³⁺), with 3d¹ electronic configuration. It is well known that in high T_c cuprates, Cu²⁺ ions with 3d⁹ configuration have one unpaired spin, and the parent compounds exhibit antiferromagnetic interactions among Cu²⁺ ions. The two adjacent Cu²⁺ sites are bridged by the oxygen pπ hole, and the hopping of these oxygen pπ holes from site to site is responsible for the conductivity in high T_c cuprates [5]. A similar modification in electronic structure, governed by strong electron correlations in narrow d band, was anticipated in Mott-Hubbard insulators like RETiO₃. The magnetic interactions and conductivity in this system can be tuned by substituting RE cations with varying ionic radius, from the largest La³⁺ cations to the smallest Gd³⁺ and Y³⁺ cations. Variation in ionic radii lead to uneven tiltings of the TiO₆ octahedra, which alters the ratio between the electron bandwidth and Coulomb interaction [2,3,4,6]. For instance, LaTiO₃, with the average Ti–O–Ti bond angle θ~157° (along the ac plane and the b axis), is a G-type antiferromagnet with a Néel temperature T_N~140 K. In contrast, YTiO₃ with a more bent Ti–O–Ti bond angle θ~142° is a ferromagnet with a Curie temperature of about T_c~30 K [1,2]. The evolution of magnetic transition had been well studied in the La_xY_1-xTiO₃ solid solution as a function of La/Y ratio [7,8]. Neutron scattering measurements have thoroughly investigated the magnetic fluctuations near the antiferromagnetic (AFM) to ferromagnetic (FM) phase transition boundary at a critical doping of x = 0.7 in La_xY_1-xTiO₃ [9,10]. Notably, the one-electron bandwidth and band filling in RETiO₃ compounds can be tuned by both RE³⁺-cation-induced lattice distortions and the substitution of divalent Ca²⁺ or Sr²⁺ cations at RE sites [3]. By doping with Ca²⁺ or Sr²⁺ cations, a metal–insulator transition is introduced in La_1-xSr_xTiO₃ and Y_1-xCa_xTiO₃ systems [11,12,13,14,15,16,17,18,19,20]. In the end members SrTiO₃ and CaTiO₃, Ti⁴⁺ ions, instead of Ti³⁺ ions, balance the divalent Ca²⁺ or Sr²⁺ cations. Therefore, both Ti³⁺ and Ti⁴⁺ ions coexist in the doped compounds. Recently, by using femtosecond laser to excite the three phonon modes with center frequencies near 4, 9 and 17 THz in YTiO₃ single crystals (T_c = 27 K), transient ferromagnetism up to 80 K was realized for the 9 THz oxygen rotation mode through the time-resolved magneto-optic Kerr effect (MOKE) measurements [21]. This finding strongly suggests that the FM phase transition can potentially be tuned up to 80 K by exploiting the strong spin-lattice coupling in YTiO₃ [22,23].

The high-quality RETiO₃ single crystals, as well as Ca²⁺- or Sr²⁺-doped samples, are highly desirable for studies of their physical property. Initially, RETiO₃ single crystals were grown with Bridgman and Czochralski methods under an argon atmosphere [24,25]. However, thermalgravimetric analysis (TGA) indicated that the crystals were slightly oxidized, with the composition of LaTiO_3.011 and CeTiO_3.013 [25]. With the development of floating zone growth techniques [26], RETiO₃ single crystals, including La_xY_1-xTiO₃, La_1-xSr_xTiO₃ and Y_1-xCa_xTiO₃, were grown [8,10,11,12,13,14,15,16,17,18,19,21,22,23,27,28]. It was realized that growing stoichiometric RETiO₃, La_1-xSr_xTiO₃ and Y_1-xCa_xTiO₃ single crystals without excess oxygen is challenging. It was observed that magnetic transition temperatures decrease with increasing oxygen content in the samples. For instance, as nominal hole doping δ increases, LaTiO_3+δ/2 undergoes a transition from an insulating phase to a metallic phase at δ~0.05, with the AFM transition disappearing at δ~0.08 [28]. In fact, with further oxidation, LaTiO_3+δ/2 can transform into pyrochlore LaTiO_3.5 (La₂Ti₂O₇) phase with increasing excess oxygen δ [29] because the unpaired 3d¹ electrons in Ti³⁺ ions are easily lost, forming Ti⁴⁺ with 3d⁰ configuration.

In this study, we thoroughly investigate the effect of growth conditions on the excess oxygen in YTiO₃ single crystals by using the laser floating zone technique. Our results show that adding Ti powder to the starting materials is critical for growing YTiO₃ single crystals with T_c~30 K, free from Ti⁴⁺-containing impurities. The Results and Discussion section is separated into the following three subsections: (1) growth from the starting materials with nominal composition YTiO₃; (2) growth from the oxygen-deficient starting materials with nominal compositions YTiO_2.925 and YTiO_2.85; (3) the formation of TiN coating.

2. Materials and Methods

The laser-heated floating zone furnace (LFZ–Compact200EX, Model: TFL–1–P, TECHNO SEARCH CORPORATION, Tokyo, Japan) consists of 5 laser heads positioned circumferentially in a horizontal plane. The laser operates at a wavelength of 940 nm (invisible). The pressure in the chamber can be increased up to 10 bar when a thick quartz tube (thickness = 5 mm) is used.

Y₂O₃, Ti₂O₃, and Ti powder were used as starting materials. The rods were prepared with the following three nominal compositions: (1) YTiO₃; (2) YTiO_2.925; (3) YTiO_2.85. The Y₂O₃, Ti₂O₃, and Ti powder were ground and mixed in an agate mortar. The mixture was pressed into rods with a diameter of ~6 mm and a length of ~100 mm using a pressing die, or loaded into a stripe balloon, and pressed under a pressure of 80 KN with a hydrostatic pressing system. The rods were sealed in quartz tubes, and sintered at 1100 °C for 24 h, as shown in Figure 1a,b. The floating zone growth conditions are summarized in

Table 1. Summary of the rod preparation conditions and of the parameters during the crystal growth by laser floating zone technique.

Starting materials	Stoichiometric ratio YTiO3	Y2O3 + Ti2O3 → 2YTiO3
	Oxygen deficient YTiO2.925	Y2O3 + 0.95Ti2O3 + 0.1Ti → 2YTiO2.925
	Oxygen deficient YTiO2.85	Y2O3 + 0.9Ti2O3 + 0.2Ti → 2YTiO2.85
Gas atmosphere	Argon
	97% Ar + 3% H2
	95% N2 + 5% H2
Growth rate	Slow 2 mm/h	The growth rate indicates the travelling speed of seed rod. The feed rod moves 5–10% faster than the seed rod to maintain the stable molten zone and uniform shape of as-grown rod.
	Middle 4–5 mm/h
	Fast 10 mm/h
Pressure	Atmosphere pressure	Gas flow rate 80–120 mL/min
	2 bar	97% Ar + 3% H2
Rotation speed	15 rpm	The upper and lower shafts rotate in an opposite direction but at the same speed.

.

The phase purity of the obtained crystals was checked by the powder X-ray diffraction (XRD) measurements using a PANalytical Empyrean X-ray diffractometer (Almelo, Overijssel, The Netherlands) equipped with a long-focused sealed X-ray tube of Cu-Kα radiation (λ = 1.5418 Å) and PIXcel 1D detector. Raw pattern reduction, profile fitting, phase identification and then the Le Bail fitting for both samples were performed with HighScore plus and ICDD PDF5+ [30].

The crystallinity of the crystals was confirmed by using a Laue X-ray diffractometer in backscattering geometry (MWL120 Real-Time Back-Reflection Laue Camera System, Multiwire Laboratories, Ithaca, NY, USA). High quality single crystals with suitable size were mounted onto the goniometer head of a Rigaku Synergy-I single crystal diffractometer (Rigaku Europe SE, Neu-Isenburg, Germany), equipped with Mo micro-focused source and Bantam HyPix X-Ray detector. Unit cell refinement, data collection & reduction, and absorption correction for all data sets were performed by CrysAli^pro [31]. Structures were solved using ShelXT [32] within Olex2 [33] first, and then refined using ShelXL [34] within both Olex2 and ShelXle [35]. The final CIFs were edited with PublCIF [36] and passed IUCr checkCIF before depositing to CSD. The oxygen content in single crystals was checked through thermogravimetric analysis (TGA), which was conducted with a Setaram Labsys TGA-DSC thermal analyzer (KEP Technologies, Plan-les-Ouates, Switzerland).

Magnetization measurements were carried out using a Physical Property Measurement System (PPMS) DynaCool (Quantum Design, Inc., San Diego, CA, USA) equipped with a Vibrating Sample Magnetometer (VSM). Data were collected in both zero-field-cooling (ZFC) and field-cooling (FC) modes. For the ZFC mode, the sample was cooled down to the lowest temperature without applying a magnetic field, after which the magnetic field was applied and data were collected while warming up the sample. In the FC mode, data were collected while the sample was cooled down under an applied magnetic field.

3. Results and Discussion

3.1. Growth from the Starting Materials with the Nominal Composition YTiO₃

We initiated the growth of single crystals using rods prepared with the stoichiometric YTiO₃ composition. Figure 1c shows the XRD pattern obtained with the rods after sintering, prior to the single crystal growth. It reveals three phases, namely YTiO₃, Y₂O₃ and Ti₂O₃. No Ti⁴⁺-containing phases are present in the rods. In the literature, a single-phase powder of YTiO₃ was obtained, for instance by arc melting under argon atmosphere [20]. Polycrystalline La_xY_1-xTiO₃ (x = 0.15 and 0.3) samples were synthesized by ball milling followed by spark plasma sintering (SPS) at 1600 °C for 10 min [9]. It is well documented that the rods are at a risk of being partially oxidized when they are sintered at 1400 °C in an airtight chamber furnace under argon atmosphere [37].

After mounting the seed and feed rods in the floating zone furnace, the air in the chamber was purged by flowing argon gas over 6 h or by pumping for one hour followed by filling the chamber with a gas of mixture of 97% Ar + 3% H₂. As the laser power increases, the rods melted and joined at 1800 °C, as indicated by the pyrometer. Three growth rates were applied to grow the crystals, as summarized in Table 1. Figure 2a shows an image of YTiO₃ single crystals grown at a rate of 10 mm/h. The rod exhibits a shiny black appearance. When lower growth rates are applied (i.e., 5 and 2 mm/h), rods with similar appearance were obtained. However, they are fragile and tend to break into small pieces easily.

The Laue X-ray diffraction measurements were conducted to assess the crystallinity of the rod grown at 10 mm/h. Figure 2b,c show reflections taken from the surface of rod, perpendicular to its longitudinal direction indicating high crystallinity. Figure 2d,e present reflections from the rod cross-section. Figure 2f shows the tilted rod, with the Laue diffraction taken along this direction. The Laue spots form the two-fold symmetric pattern, as shown in Figure 2g, in agreement with the alignment of the crystal along one of the principal axes of orthorhombic YTiO₃.

The result of the XRD measurement performed on a powder obtained by grinding part of the rod is shown in Figure 3. YTiO₃ has a GdFeO₃-type orthorhombic structure with the space group Pnma (62). The profile fitting of the XRD pattern yields lattice parameters a = 5.66971(9), b = 7.61722(9), and c = 5.33523(7) Å, respectively. In addition to the main YTiO₃ phase, additional peaks are observed at 2θ = 15.16, 28.69, 29.61, 30.71, and 50.70°, respectively. Peaks at 2θ = 15.16, 29.61, and 30.71° are assigned to cubic Y₂Ti₂O₇ and the peaks at 2θ = 28.69 and 50.70° to hexagonal Y₂TiO₅ phases, referring to the Y₂O₃-TiO₂ phase diagram [38,39] and the structure data for hexagonal RE₂TiO₅ [40]. A comparison of orthorhombic YTiO₃, cubic Y₂Ti₂O₇ and hexagonal Y₂TiO₅ structures is shown in Figure 4. It should be pointed out that the α-Y₂TiO₅ with an orthorhombic structure is stable below 1330 °C, transforming to the hexagonal β-Y₂TiO₅ between 1330 °C and 1520 °C and to a fluorite-type phase at higher temperatures [38,39].

The presence of Y₂Ti₂O₇ and Y₂TiO₅ in the rods after the floating growth process indicate Ti³⁺ is partially oxidized to Ti⁴⁺ even if hydrogen is present in the chamber. Previous reports on the growth of RETiO₃ single crystals by the floating zone method [8,10,11,12,13,14,15,16,17,18,19,21,22,23,27,28] confirm the difficulty of avoiding Ti³⁺ oxidation. The presence of TiO₂ in the Ti₂O₃ starting materials might be the cause of the formation of Ti⁴⁺ impurities in the crystals. It is also suggested that the residual oxygen in the chamber is responsible for the excess of oxygen in the crystals and the crystallization of Ti⁴⁺-containing impurities [27]. In conclusion, purging or evacuating the furnace is not efficient, as the chamber might be leaking and oxygen is inevitably introduced in the chamber during growth, which lasts for hours. A better solution is needed to impede Ti³⁺ oxidation and reduce the effect of the inevitable presence of a minute amount of TiO₂ in Ti₂O₃.

The hydrogen gas was widely used to control the oxidation of Ti³⁺ ions during the crystal growth. However, our results clearly reveal that the hydrogen gas may not work as effectively as one assumed. In fact, TiO₂ can be only slightly reduced to Ti₂O₃ under the hydrogen atmosphere at a high temperature. An effective way to reduce TiO₂ to Ti metal was realized by using argon and hydrogen plasma [41]. Once Ti³⁺ ions are oxidized and transferred to Ti⁴⁺ ions, it is hard to reverse them to Ti³⁺ ions under the hydrogen atmosphere at 1800 °C.

3.2. Growth from the Oxygen-Deficient Starting Materials with Nominal Compositions YTiO_2.925 and YTiO_2.85

The idea behind the approach is to balance the excess oxygen by using the oxygen-deficient starting materials YTiO_3-δ. In fact, this method had been successfully used in the previous work on the growth of RETiO₃ single crystals. For example, to grow LaTiO₃ and La_0.95Sr_0.05TiO₃ single crystals, Ti, TiO₂, La₂O₃, and SrTiO₃ were weighed in prescribed ratios of LaTiO_2.94 and La_0.95Sr_0.05TiO_2.94 stoichiometry [13]. In a recent report on the growth of Y_1-xLa_xTiO₃ ($x\le 0.25$) and Y_1-yCa_yTiO₃ ($y\le 0.35$) single crystals, the rods used for the crystal growth had nominal composition with oxygen deficiency Y_1-xLa(Ca)_xTiO_3-δ [27]. In the case of YTiO₃, several different values of δ in the range $0\le \delta \le 0.08$ were tried with δ = 0.04 yielding the highest T_c~30 K [27].

In this study, rods with two nominal compositions (namely YTiO_2.925 and YTiO_2.825) were used for the crystal growth, as shown in Table 1. Crystal growth from the rod with nominal composition YTiO_2.925 was performed at the growth rate of 4 and 10 mm/h. In both cases, the growth experiments were performed under an atmosphere of argon mixed with hydrogen. Figure 5a shows the rod obtained at 4 mm/h. Notably, the surface of the as-grown rod is coated by a grey crust, which can be mechanically removed using a razor blade. As observed, the rod is not completely round and two parallel flat surfaces run along the entire 70 mm length of the rod. The rod was cut into short sections and polished along the flat surfaces. XRD measurements revealed that the flat surface corresponds to the (2 0 0) plane of the YTiO₃ crystal structure, allowing for the orientation of the a-axis in the rod, as shown in Figure 5b.

From the two-fold symmetric Laue pattern obtained from the flat surface of the rod, shown in Figure 5c,d, the crystallographic orientation of the rod can be determined. The b/c-axis is found tilted with an angle of approximately 20° from the longitudinal axis of the rod, as confirmed by Figure 5e,f. A similar tilt angle was obtained in the previous growth using rod with the nominal YTiO₃ composition with 10mm/h growth rate.

One small disk cut from the middle part of the rod was crushed and ground into fine powder. Figure 6 shows the result of powder XRD measurements. The profile fitting of the XRD pattern confirms the absence of impurities. Pure YTiO₃ phase is grown with the lattice parameters a = 5.69075(7), b = 7.61696(7), and c = 5.33880(8) Å, respectively. Compared to the lattice parameters obtained in Figure 3, the a-lattice parameter is substantially higher for the pure crystals.

Because the evaporation of Ti was observed when crystal was grown at a rate of 4 mm/h, an additional growth was performed with similar conditions but with a faster growth rate, i.e., 10 mm/h, and with a 2 bar pressure applied in the growth chamber. The objective was to reduce the evaporation of Ti and maintain the cationic stoichiometry of the rod in the crystal. Changing these parameters results in the absence of formation of the unwanted grey crust grey on the surface of the rod. The resulting crystal rod is almost perfectly circular, without the formation of a flat surface. The powder XRD pattern, similar to the one presented in Figure 6, confirm that a pure YTiO₃ phase has been obtained.

In addition to powder X-ray diffraction measurements, YTiO₃ single crystals grown under varying conditions were characterized using single-crystal X-ray diffraction at room temperature, with results summarized in

Table 2. Results of single crystal X-ray diffraction measurements of YTiO₃ at room temperature. YTiO₃ single crystals grown under different conditions: (A) from the nominal composition YTiO₃ at a growth rate of 10 mm/h; (B) from the nominal composition YTiO_2.925 at a growth rate of 4 mm/h; (C) from the nominal composition YTiO_2.85 at a growth rate of 4 mm/h.

Batch		A	B	C
a (Å)		5.6807(3)	5.6860(3)	5.6995(1)
b (Å)		7.6204(4)	7.6108(3)	7.6297(2)
c (Å)		5.3391(2)	5.3378(2)	5.3455(1)
V (Å3)		231.13(2)	230.99(2)	232.45(1)
Bond angles (°) Ti1—O1—Ti1		143.69(9)	144.04(9)	143.35(6)
Bond angles (°) Ti1—O2—Ti1		140.97(13)	140.73(13)	140.09(10)
Atoms
Y(1)	x	0.42721(6)	0.42719(7)	0.42690(4)
	y	0.25	0.25	0.25
	z	0.02056(6)	0.02060(6)	0.02084(5)
	Occ	1 4c	1 4c	1 4c
Ti(1)	x	0	0	0
	y	0	0	0
	z	0	0	0
	Occ	1 4a	1 4a	1 4a
O(1)	x	0.04130(4)	0.04140(4)	0.04290(3)
	y	0.25	0.25	0.25
	z	−0.11860(5)	−0.11930(5)	−0.12120(3)
	Occ	1 4c	1 4c	1 4c
O(2)	x	0.19050(3)	0.19160(3)	0.19030(2)
	y	0.05800(2)	0.05750(2)	0.05883(16)
	z	0.30900(3)	0.3090(3)	0.30970(2)
	Occ	1 8d	1 8d	1 8d

. As shown, the lattice parameters obtained from powder and single-crystal X-ray diffraction measurements are comparable. The detailed structural information from single-crystal X-ray diffraction reveals minimal differences among single crystals grown under different conditions. Our findings rule out the possibility that excess oxygen atoms occupy interstitial sites (revealed by the low residual electron densities in structure refinement; See the Supplemental Materials). Moreover, our results do not support the presence of oxygen deficiency in single crystals grown from the nominal compositions YTiO_2.925 and YTiO_2.85. The TGA of the single crystal (same sample used for Figure 6) reveals that the excess oxygen is below 0.0126 in the formula YTiO_3+0.0126; See the Supplemental Materials. Combining both powder and single crystal X-ray diffraction measurements, the slight variation among single crystals grown under different conditions is attributable to the presence of impurity phases rather than to the oxygen stoichiometry. We conclude that the oxygen content in YTiO₃ single crystals is unambiguously stoichiometric, with an O content of 3.

Figure 7a shows the magnetic transition of YTiO₃ single crystals grown under different conditions. For the crystals grown from the stoichiometric YTiO₃ starting material, T_c varies between 25 K and 27 K. These crystals contain cubic Y₂Ti₂O₇ and hexagonal Y₂TiO₅ impurity phases. For the crystals grown from the oxygen-deficient YTiO_2.925 starting material, T_c increases to 30–31 K. The crystals grown from the starting material YTiO_2.85 show the highest T_c~32 K; however, the floating zone growth with this starting material is unstable, leading to more defective and smaller crystals. The video clip available in the Supplemental Materials shows the formation of Ti bubbles in the molten zone, the bursting of which leads to the floating-zone instability during the growth using rods with the YTiO_2.85 nominal composition. Figure 7b shows the magnetic field dependence of magnetization, M vs. H. The magnetization tends to saturate above H = 0.5 T. Crystals containing impurity phases, grown from the starting material YTiO₃, exhibit a significantly lower magnetization, while pure crystals from oxygen-deficient materials have higher magnetization. Overall, single crystals grown from the starting materials YTiO_2.925 show the best crystallinity and magnetic properties. The measurements of the heat capacity of the best crystals confirm the temperature of magnetic transition. T_c varies between 29.5 and 31 K, as shown in Figure 7c.

3.3. The Formation of TiN Coating

A TiN superconductor was accidentally formed during the growth of YTiO₃ crystals using starting material with the YTiO_2.85 nominal composition and hydrogen-containing Ar atmosphere. A leak in the reactor led to the formation of a rod with a golden coating, as can be seen in the Figure 8a. The powder XRD pattern recorded on the surface of the rod confirmed the presence of TiN, as shown in Figure 8b. The growth was repeated using the same starting material but using forming gas atmosphere. The obtained rod is homogeneously coated with TiN, as shown in Figure 8c and in the XRD pattern in Figure 8d.

The possible chemical reaction pathway for the formation of TiN from Ti₂O₃ could be [42]:

Ti₂O₃ + 1/10 N₂ = 1/5 TiN + 3/5 Ti₃O₅

(1)

or Ti₂O₃ + 1/7N₂ = 2/7 TiN + 3/7 Ti₄O₇

(2)

Based on these chemical reactions, the Ti₂O₃ compound is partially oxidized to form Ti₃O₅ or Ti₄O₇. As the valence of Ti increases, the formation of TiN also contributes to the formation of Ti⁴⁺-containing impurities.

The presence of TiN in the sample is further evidenced by magnetization measurements of the rod grown under forming gas atmosphere, as shown in Figure 9. Figure 9a shows that a superconducting transition is observed in the sample at ~4 K. Figure 9b shows the magnetic hysteresis loop measured at T = 3 K, which again confirms the superconducting state in the sample. For comparison, TiN single crystals exhibit a superconducting transition temperature of 6 K, while polycrystalline TiN shows a T_c of 1.7 K [43]. In the best TiN films, the superconducting transition temperature has been measured as high as 4.8 K [44]. Since TiN has a high melting point of 2947 °C, single crystals cannot be directly grown from polycrystalline TiN rods using the floating zone method. Our study demonstrates that TiN can be efficiently formed through the reaction of Ti₂O₃ with N₂ by the floating zone method.

4. Conclusions

We have grown high-quality YTiO₃ single crystals with visible (2 0 0) facets by using the laser floating zone technique. We observed that T_c decreases with the formation of Ti⁴⁺ ion-containing impurities, such as cubic Y₂Ti₂O₇ and hexagonal Y₂TiO₅. Adding Ti powder to the starting materials to prepare the rods with the nominal compositions YTiO_3-δ effectively minimizes the effect of residual air or the air leak on the formation of Ti⁴⁺-containing impurities. Ti addition allows a better control of the oxygen stoichiometry during the crystal growth, consequently maintaining the perovskite structure of YTiO₃. Differences among crystals grown under varying conditions are attributable to impurity phases rather than crystal stoichiometry, which is confirmed as YTiO₃. We recommend using YTiO_2.925 as the optimal starting materials. The presence of larger quantity of Ti leads to unstable floating zone during the growth due to excessive Ti evaporation. Although slow growth may promote the formation of high-quality single crystals, it also increases the risk of oxidation during lengthy process. A growth rate of 4 mm/h is critical to obtain crystals with (2 0 0) facets. Growth under pressure could also mitigate the problem of air leaks. The presence of nitrogen in the chamber leads to the formation of superconductor TiN on the surface of the rods. Crystals grown in the best conditions exhibit a ferromagnetic phase transition between 29.5 and 31 K.